<%BANNER%>

Population Ecology and Conservation of the Snail Kite

xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101208_AAAAHZ INGEST_TIME 2010-12-08T09:12:41Z PACKAGE UFE0019653_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 130235 DFID F20101208_AAAIZC ORIGIN DEPOSITOR PATH martin_j_Page_178.jp2 GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
e7f60247c60a66eeb06df32dff577eda
SHA-1
fc861ee05dba2996d10ae2f5c79b84fd060b4f7a
122403 F20101208_AAAJDU martin_j_Page_016.jp2
8fa694795ad462e832b422e95c7487ef
d8e283f2de7570ca123aa61b64fa90d91fbad470
54632 F20101208_AAAIYO martin_j_Page_076.pro
c508599764d35b2cf028b0f3f9d8c4ad
2c802ab47135ec8316c763af44c9fb3378b1c590
48928 F20101208_AAAJEI martin_j_Page_039.jp2
37959bbc4d4faadf8b15b74b72eacae8
921a38b5821d4c568c396bceb72d5f4ecd4c2694
1053954 F20101208_AAAIZD martin_j_Page_142.tif
518fd1f141ec71e9c564a901a3c93f79
6edb3889e1e914aeaabe1a3f57f81ea12b6d81a6
117728 F20101208_AAAJDV martin_j_Page_019.jp2
c0404769ebd42a986afa9322c59686a0
6ea203d91d5bdfa1db15e5414bda9e36159c096f
118934 F20101208_AAAIYP martin_j_Page_021.jp2
51581c94738b3e6a9efabdcf97373116
64d193b055329a958a383d5fced95009f8c1571e
550590 F20101208_AAAJEJ martin_j_Page_042.jp2
3b0079553482efd50de780c5f26aa98a
287dc587986ce59961bbbfc89b6bea1995579e72
69098 F20101208_AAAIZE martin_j_Page_137.jpg
36759bbb93eda4ed8e42afd860ca4c1a
dfa08871bc02a7d760b87d288570c689f0025262
120846 F20101208_AAAJDW martin_j_Page_020.jp2
d48c0ff018b68faa8cc3c128f1ce1b4e
143ffb17cf6f02dfe253d787cff7a2a5c6b2f842
7629 F20101208_AAAIYQ martin_j_Page_107thm.jpg
3724030a41aed19fdb98864a21295da7
565ee0b638e5b4e173e2a4fc5786741f4b8a28a6
229504 F20101208_AAAJEK martin_j_Page_043.jp2
bf437189f4910725c2fdde57638c5c67
adde0cfa887885cd981b3f7b2512bc58763c5fcf
5062 F20101208_AAAIZF martin_j_Page_073thm.jpg
99cfaa8d2107c37f3db0ad9f4d34f2dc
2e3c622f73350fdad7d970a09ac320204a62a8d3
117610 F20101208_AAAJDX martin_j_Page_022.jp2
f59aa98ad532ea149838272f67e6bd8f
4565e321e012a42dd3a5560e16824463ed352d84
466794 F20101208_AAAIYR martin_j_Page_100.jp2
c382aa6e78453ed13529d82c5117f104
731a9dfaf1e42e040cdaf35b73b880b53270d97e
565487 F20101208_AAAJFA martin_j_Page_074.jp2
c080da70b16977fee1570c87f342ee2a
16b9c6e0067ec6dde9c3d15b00ddbd4d534b319b
115524 F20101208_AAAJEL martin_j_Page_046.jp2
9507e8cf2633c5b77d49df77301891c1
2ec6ca7d3612a9118f9a0719fa862e905c08802a
208210 F20101208_AAAIZG UFE0019653_00001.mets FULL
186139fbd4979fd6473f34c2902c34cc
7c0222d1007b568f5468342da6447aa54b654ae3
107370 F20101208_AAAJDY martin_j_Page_023.jp2
97433736b05dbcbc1abd7a6a2e86d36d
953151c5f2dcb8dd918f7a40053e0544ad566397
1929 F20101208_AAAIYS martin_j_Page_134.txt
cc60ad966a79ff4e00bc846f661f72a2
872a34fb878a12db01b9511f9091c94b4d6d4750
104390 F20101208_AAAJFB martin_j_Page_077.jp2
19e90926b460ef8e477a68279584918a
c6e7a60431325e356ebdfe0733126cf771211998
1051980 F20101208_AAAJEM martin_j_Page_050.jp2
0ee4d5fb6b851ee700423e7483408283
5316dcb5a9c80ea7aae73d40b76adfa3a578261c
99182 F20101208_AAAJDZ martin_j_Page_025.jp2
d23b0ff747d7dd6a8e5a53fc7f5f7707
a91cf4455da06e1d7a2ef75a6cde3f5808d03d03
2544 F20101208_AAAIYT martin_j_Page_116.txt
926873bc5eedf6a4930861dbf6df5082
38830151c6406a3be6ebf2cd84418d801d8cd32a
115784 F20101208_AAAJFC martin_j_Page_081.jp2
2664afd82440fd1581bf6cfc0ddf3c5a
d18b646066e52ed8204fb439a46070c34c356c9a
117673 F20101208_AAAJEN martin_j_Page_052.jp2
4098dd0ba028c1fc02badbac9cbef1b8
3878e3f12da89e52636d45a6b37a2671270833ce
35778 F20101208_AAAIYU martin_j_Page_045.QC.jpg
f4379779a89b66bcc4e96b505373b810
4c5b3ee09a08fec08d320ff6a2409ab9bd9a7425
110858 F20101208_AAAJFD martin_j_Page_082.jp2
b548dbd177e7f556d79a3ab520bbb44e
9766b642f7e6af1e366e9b08aa64ec225bc8319d
114620 F20101208_AAAJEO martin_j_Page_053.jp2
645bbf45989c5454bae8124670a5fbff
e8a13a8f1d8f819bb94379ded02001c38a5e4d16
23078 F20101208_AAAIZJ martin_j_Page_001.jpg
807975ae6b3e1cc70dd495c90a3ec194
5f14ce2e61cb01f7b4365ddd56000ce6ba1d8c9d
92719 F20101208_AAAIYV martin_j_Page_028.jp2
7f2ad03addecc6969a5000a0371788f3
ffddabbf134cfe33dbe6b50102058908dcf18276
117687 F20101208_AAAJFE martin_j_Page_083.jp2
8ffbd6c61c75189cbfe4553638a0745c
e0c950cf0900810ae9cef40b59a2f4abd23efa67
114235 F20101208_AAAJEP martin_j_Page_054.jp2
5cfa87020f7fb9b3166364e8bd79bfdf
5e05710e6d188c4eb5815854f898d85b0da312ae
3697 F20101208_AAAIZK martin_j_Page_002.jpg
a7a0ca75a7250cc42516206b42663a74
609e95fa602d58b5e7cb6b0d0f65c57911433348
120495 F20101208_AAAIYW martin_j_Page_018.jp2
0c4b4f85386ef843a0091e3752a732a1
49bdc0044a12253aa6efc4d21e3a3f03d5e56aec
119206 F20101208_AAAJFF martin_j_Page_085.jp2
68555ba4e0aedf252d24e62038432db7
ca82ead4c3c595b129ec470d230617844632690b
111878 F20101208_AAAJEQ martin_j_Page_055.jp2
55a04bbc7569e4f70e4407713982a076
6bdc420737783b5016c58bc21d746397d8cbb81e
103567 F20101208_AAAIZL martin_j_Page_004.jpg
bc649b0b916ae5d1c3a766779807081a
aa45f9ab89108682894f58988f61394c6ac33783
61767 F20101208_AAAIYX martin_j_Page_167.jp2
0f7edc812052408d42b374a3d72f949f
c2737c4c5465126ab394449349389c9613e8b2f2
117572 F20101208_AAAJFG martin_j_Page_087.jp2
da164130cfd9a9595d78c5bbf84e5a7d
5bf896fd78faad3015c1c9e3ff34c2470d8699c7
96205 F20101208_AAAJER martin_j_Page_056.jp2
0e036121018f85d68fcafe378011fb8f
e15bca1f09cb3b389a1aca97e87590013019f9eb
83367 F20101208_AAAIZM martin_j_Page_005.jpg
1f416c53ec219ebc08bfbaf5b7caad72
5c522b254b5846a33a1d1d50c7bb1f9496d5857b
101166 F20101208_AAAIYY martin_j_Page_023.jpg
c7c8867f2e41aa567174e18a6fa2f386
41e819222a80eca274fa8c070b53064207b3a172
96775 F20101208_AAAJFH martin_j_Page_088.jp2
0f1647fa46d4096e6cca7f0649b562c5
15b4d66759587b06a4f00835f2e5b837007fcc5f
94210 F20101208_AAAJES martin_j_Page_058.jp2
97fe03a9c25647a5ccc5c1394e173065
94ebeffdc07d08077ed4d758605c85964491e44e
133119 F20101208_AAAIZN martin_j_Page_006.jpg
fea739251b63afe73172cf433467c8cf
71182c1bc8a29ec6c4a0ad9bbf04a82b64a130f9
3165 F20101208_AAAIYZ martin_j_Page_003.jpg
55e81adf1e84177b36587a0d16485bc3
fdb1053b37323f8a616b4622e442e600571920ea
105151 F20101208_AAAJFI martin_j_Page_089.jp2
83a37362f4022599c10e9ef02d7ba262
a5855e4a72e108165d18d0be61a54e702cef992d
100925 F20101208_AAAJET martin_j_Page_059.jp2
2cb0a2a81f18297c98f522e1a09966a9
4850d97dfad1213ad574523a8a2f4b9d2b5c26c7
150641 F20101208_AAAIZO martin_j_Page_008.jpg
3f37da53643cb9a8a99ad271d36f82ed
5f41449ea323939e340ebfe78b2ab0f3ea624833
107479 F20101208_AAAJEU martin_j_Page_060.jp2
99888123a6d4094bdbfcfb1f8a4d3290
0260e4cde96cb8a16dcf33fabdb40d52eba84eca
160589 F20101208_AAAIZP martin_j_Page_009.jpg
03e0bf9d4ab2c1ddf8acf5d6e5663689
642d44ff516955fde69b64fd4e4bc6708d924e54
107959 F20101208_AAAJFJ martin_j_Page_090.jp2
7b34298b2177c86472739a8457309c07
f7d97d6d8208f76b79a7f0402899c5364dfbb4ce
116651 F20101208_AAAJEV martin_j_Page_061.jp2
5e174361bf12c9a4c1a3b40b7675ad0a
1a15935fef913e9e4fb220b0c85e78321c64e1af
36770 F20101208_AAAIZQ martin_j_Page_010.jpg
5a0c521b366ffdc29238bb460712e950
4c60626c83109703165a055105afe46fc99662e3
96365 F20101208_AAAJFK martin_j_Page_091.jp2
dae3d391c60ad6e2957bd4327c48306f
9da203d8fa70b883f6d62db394819495723e6b84
120667 F20101208_AAAJEW martin_j_Page_063.jp2
98140e389f1747147b185522b7eda461
0c57f6718312720ddfb834611fd9dd89aedd8ba3
119347 F20101208_AAAIZR martin_j_Page_011.jpg
c4ce0d52045fb6de94216f2cdc9c33ef
0242070f1cbb82946c188ecf051d14585c08641c
116753 F20101208_AAAJGA martin_j_Page_118.jp2
74beb64eb25bc0c6d117329235f1c61a
a44c673bdb6c09c39035e41cd89be191bfbe9ff9
119658 F20101208_AAAJFL martin_j_Page_095.jp2
920ddca649456fe98c4676038a12b085
c6a6cfb1ccb4cbb5fd2984d0e21b3ad05b3d5f0a
121993 F20101208_AAAJEX martin_j_Page_066.jp2
c6b6cc1125e0081f86d8cb40428a5598
e1d9a591d6e678554020c4034e4a08a9e0c72c79
107011 F20101208_AAAIZS martin_j_Page_015.jpg
6092614b0392813b00a471c548358cf8
81bb3ce4de54388937a67e7212eb641e2b7ed79f
109528 F20101208_AAAJGB martin_j_Page_120.jp2
2e0098036378e9ffe45dbcb7cce0deb3
c2aa08872864948fadba4fd402ac80ec7cf0eaec
29846 F20101208_AAAJFM martin_j_Page_096.jp2
0d8bb2f8d34e8e1d3af61548667f0ea7
5cd2c7e06208e769ed50bc384eee3b9741b1cfb4
49918 F20101208_AAAJEY martin_j_Page_071.jp2
68caed53c25554a290fef7a88cc7ce33
520da58ea7af404dab1f87e476ace3edb8de7cb7
106714 F20101208_AAAIZT martin_j_Page_017.jpg
cf8aff99573e36747cb2223122685709
1161708f713b97e997522b8523cfe419a1d23d98
89462 F20101208_AAAJGC martin_j_Page_121.jp2
b2049dbc9ba055dcf285be2110e545ca
9cb8fe399bfa5f86a89213b6cff57c57ed689c4d
149226 F20101208_AAAJFN martin_j_Page_097.jp2
166a43e413b22da9846e5852723cfe19
562d0283a9597b6ea2df27fbd400c97b6127da75
61126 F20101208_AAAJEZ martin_j_Page_072.jp2
9d5df1f214af6169774ffc6fe0583a69
892067cfdd58d54e64a423d8dec180cc89001a62
113102 F20101208_AAAIZU martin_j_Page_018.jpg
531eb8d4ff49f0eb0f4b352033a631ad
6ad6e05dbfd1d58a12c357bc9115b49260032dff
89470 F20101208_AAAJGD martin_j_Page_122.jp2
904f32a985bed62c81c363b97c5df8fd
c9416a3a42fbdb5478a6f6d8335466258dcb87d0
571464 F20101208_AAAJFO martin_j_Page_099.jp2
f1234a4981186d970faa118ea9158b56
73692aa52bfb37c693a5bad985daa73f530b9a3b
110820 F20101208_AAAIZV martin_j_Page_019.jpg
08abd404bcb7b145f845c7964365f2ec
fa4095bd71457917f8ed8e21b7a90cf650e48526
97212 F20101208_AAAJGE martin_j_Page_123.jp2
7f7687f79003e05a3812e751bfa574a4
010ae671298f13e8e27c540bd400cc728b923cdd
112229 F20101208_AAAJFP martin_j_Page_103.jp2
1a8bc146dd0910803bbd936aa34b049a
9fc6225ed6b8da62f4a35583c4345444c513076b
109875 F20101208_AAAIZW martin_j_Page_022.jpg
c1cbc08808956487bac866024edd033c
1f907a6f9d5690da720e507e6bd19760d4dd5386
105391 F20101208_AAAJGF martin_j_Page_124.jp2
135cc61b93ad0c9d7c29945ef1d7a540
055dba6a763d3c99b7ba10a7e6dc55b117f044f3
116982 F20101208_AAAJFQ martin_j_Page_104.jp2
b59abce06ff5b4e36887472b16bec484
0980abbb4bd300307be9222d11a666e2d0127bed
86051 F20101208_AAAIZX martin_j_Page_028.jpg
bb449fa83db0c6a6fdca14f56021b556
f4320e95d5d91fed6e84f301def0f998233685b9
115936 F20101208_AAAJGG martin_j_Page_126.jp2
f3c9281b46b2dc5645851a71b692aa6f
8c432e64520af106472d64a53dacc178a0db4732
108391 F20101208_AAAJFR martin_j_Page_105.jp2
8b34f854cb93c00c4f4630034f753643
a90cb7f7828d43efd2ecdb0aeac3a256cc2f2431
102304 F20101208_AAAIZY martin_j_Page_030.jpg
d4fd265263ca515d8b2875c4f318b3d4
cb14c5186966e9a814b36dc5f117a077f1f86e5b
109466 F20101208_AAAJGH martin_j_Page_127.jp2
fb84e14da8ceeea47b9968f5108e94ac
40a8c46e3dcbffad113343b7e538b09a4fd77a75
95227 F20101208_AAAJFS martin_j_Page_106.jp2
4f7f1ac5b770ebfb2b7e57b73681238d
543ed468ef20ff7a69c04d49a9eb01512ac901bc
94590 F20101208_AAAIZZ martin_j_Page_032.jpg
f9f5e307dee48de481015832e1ba1490
ecab02f90c994b295c348b81133b42840fdbeaf7
110297 F20101208_AAAJGI martin_j_Page_131.jp2
ce505bbe4406e2ef2ac35ef62847885a
f13feff23ec8f37e8a458f8eba308c39b91fd5c3
96464 F20101208_AAAJFT martin_j_Page_107.jp2
c0dd1c34eb4214729a2deda35f638919
fbc79f75980700688d2eaed4db8164ed3694852e
115890 F20101208_AAAJGJ martin_j_Page_133.jp2
70bfdd44420781d01572db9c4f621542
e666260d71af4ab0a144798caec42f637c764bf6
105494 F20101208_AAAJFU martin_j_Page_108.jp2
567c30fdf9b955728b80ce4873f3d561
a4b56a9255d85266be17fa9f29ebe8515b0b8988
114104 F20101208_AAAJFV martin_j_Page_110.jp2
c5db64b85be1b4d4da8919608aa51920
380b3b2de6cabdf1b2dbb90c4d3e61627aad19cf
106596 F20101208_AAAJGK martin_j_Page_134.jp2
0f845ac5d7a7722898b22c729357c76f
014fb087ab8db7d87b3593ec8e5f482b112bc246
84515 F20101208_AAAJFW martin_j_Page_112.jp2
53f6768b411f5bf28412cbe9a4ae3494
df5ab7c6ce09e6bdc2f40b042d369b308cace629
20250 F20101208_AAAJGL martin_j_Page_135.jp2
311a4766e87a3f34fa5f9e780030f822
fbc27c76f6ff037a14ae563130bb71d6f90efeac
99173 F20101208_AAAJFX martin_j_Page_113.jp2
1e6a6b84481c15c043a240d2ab638d30
3d0a595d94380912d5cf57c68421f91c85f0f79e
60152 F20101208_AAAJHA martin_j_Page_158.jp2
88382b987e05daa4f7da0f633e61b6a0
4de39e66d94cb7314346f4eb6e56b151ae4eb1ea
123637 F20101208_AAAJGM martin_j_Page_136.jp2
055f748497ea106c3f83e8ff5f51f13e
69577504d2e9ea2b5a4de98a56d2eaf43a616383
133253 F20101208_AAAJFY martin_j_Page_116.jp2
fa3c6999c24e031f759c2f99e21accfc
a8ae625dcd0f1560ea27b8759fee894a8130244d
54840 F20101208_AAAJHB martin_j_Page_161.jp2
c72d52fa86db273d0fb1a52f38911c52
9a19e3e8724ac0c03b1ebf6f98c8c8025180b385
68340 F20101208_AAAJGN martin_j_Page_137.jp2
21e5282c2bbb3743d564d8f797b668ad
c9b443187b7bf124cf10918d7ca5b47ac1b97359
149784 F20101208_AAAJFZ martin_j_Page_117.jp2
30376d52551ac90e40442192e5f287f2
b950ff586c13ea4f1ad3209dd32d0aa86d225a7d
93587 F20101208_AAAJHC martin_j_Page_165.jp2
5e2c52d95ba552a783e6cc75f7b73549
e569c3b11cd406f5472c67d1e940b35d9587cd4a
437245 F20101208_AAAJGO martin_j_Page_138.jp2
950b26bacc9cb0f1921215184343b4bc
5b6dfc44580c82b3e86f09bbe430e36098baaa76
45257 F20101208_AAAJHD martin_j_Page_166.jp2
d5c529c8d2e7ed7a031398d8ea3f5673
cf290b2856d6d9f58e13f3ca48a4cfb24f74a855
373516 F20101208_AAAJGP martin_j_Page_139.jp2
2b3efb6c638f624710959b8412bfabaa
15cfb65cb21a101f204123e737e3832844c7c3c4
88113 F20101208_AAAJHE martin_j_Page_169.jp2
4ab0f968daa3ddd014fd7aba735ac38b
f775aab424f1c8d1ae56f8be3a9d29f6338dc60a
485349 F20101208_AAAJGQ martin_j_Page_141.jp2
c0a54b7a43e48e640f3a97dbb82499c5
1eb3ba744f40de6dc1a52dea20304bebecbd71d5
101969 F20101208_AAAJHF martin_j_Page_170.jp2
06170ccadf0bd05f98dace12bd4bc995
a233173438b2472c61b8dad146a3f970ad5ed9fd
449636 F20101208_AAAJGR martin_j_Page_143.jp2
d77a0d94fc2b19afe6c4962ab6364394
b4b6a7ad39f4001a94e3ba72d6e67b644c2f8668
14339 F20101208_AAAJHG martin_j_Page_171.jp2
a2db6ca33ccced5a5a72f68dda9ac37a
8564a27fa161ce55e2fa00b4e5b495619ddef0c7
110085 F20101208_AAAJGS martin_j_Page_145.jp2
be6f9e3f4c919e7cfa0d0303a94138f7
06b3aee3aae3c9b8b7a8e1f686f2799acb243ba8
127160 F20101208_AAAJHH martin_j_Page_172.jp2
ba2492f498c62f5a6cd40a51e4890324
b73fb6dffd76773f907f14e697479a4cd3090562
114064 F20101208_AAAJGT martin_j_Page_146.jp2
aa3893a707ba3a462c37919c1cd5c00d
f0a558265ea53cf6a9f1f456eddb8d30fad45a09
127523 F20101208_AAAJHI martin_j_Page_174.jp2
7af3e13f2dbd4e170b68d6b26e22715b
413a5cc8d4c2825b4c73a13b66625b74a6c7de0f
124051 F20101208_AAAJGU martin_j_Page_147.jp2
1c4e1f7fd4f299b651c056774b6f2ff1
45bf9458ba682c5cd28d7bf492b78727dc0973f7
125601 F20101208_AAAJHJ martin_j_Page_175.jp2
20d83010a4fc4a68b74eebe3c7079790
0dd2ff18d891905b8d1984a8376c16ab98d047f3
126502 F20101208_AAAJGV martin_j_Page_148.jp2
67c9199c3bdfde79286d1c638c70e541
4a83dc7fae01db1c09c797ba0e1ca5923d12d13b
129980 F20101208_AAAJHK martin_j_Page_177.jp2
aeb0e5cf532aa9b5a9bc36d1b78e91d9
dfe973c989cd3c75678b001f44a392d6c9819d3e
121343 F20101208_AAAJGW martin_j_Page_149.jp2
9157b4b74623983b07b07bb205653dd6
8563f490065dc297d7296413672a8896fba6d8c1
F20101208_AAAJIA martin_j_Page_022.tif
a62e3a5edb2c74993fe1074a942b7114
04b7605513c3a5ea87a89e4fcb30203f43b46130
100866 F20101208_AAAJHL martin_j_Page_180.jp2
b9f36fdbe109e5941dd1795b339fa4cb
d750be568036c2d828d37e0b15349a0d168aeecb
109434 F20101208_AAAJGX martin_j_Page_151.jp2
783fc39887a1ecc0c9341ae68a7fd2a5
07ccdcea2c9e2d6124c820261d04c79397f23fc5
F20101208_AAAJIB martin_j_Page_023.tif
ee8d6dd97ccab1afe3a80664948a4e54
f0c921938123dfef68438358f3a9ddb519b0f712
113427 F20101208_AAAIEK martin_j_Page_132.jp2
ac31c2cd1f080914a09af551fb5b6050
c8b0523ee3f90201c0fcdb0934e54d4ea93da6c1
62957 F20101208_AAAJHM martin_j_Page_181.jp2
75dc74ad2eb1c9c087fdb8cd1d31746c
c48b336db3e88313fa0435c5d9afef04acd13433
113688 F20101208_AAAJGY martin_j_Page_153.jp2
6f94c22edbfb1803307702d32648c853
d09932bca0ba3691995524ed6aab0d929480305d
135618 F20101208_AAAIFA martin_j_Page_179.jp2
817211f80f40fe65d992cc53ca7d08ac
fa0df6fc178c9666930e1ab1b882bc70621b0081
F20101208_AAAJIC martin_j_Page_026.tif
e85e121526e7ade3a7f2bd4266177883
c5711560d42fe6d5473b28c9e5f7ce8251091398
F20101208_AAAIEL martin_j_Page_019.tif
144536489e626c545aa8c415e15fc551
dbbac86880c891e09ee4211ee8e12793b421bf3e
F20101208_AAAJHN martin_j_Page_001.tif
6f1c323cdf80958dc20c11325e8f2f3d
11f8e0f845c2801ef8c38ca8dffdb18a2d714450
77847 F20101208_AAAJGZ martin_j_Page_155.jp2
90679e64f55c6c1f038ff6360add2e58
c0ee0c73e4bb47f8ad1c5f053e5f4ba85fb22408
116484 F20101208_AAAIFB martin_j_Page_080.jp2
68435ccf026d75fdcbdb296e1c8a4a3c
22402afb894e1a62e6fd87c5fd22d7056be0fe4b
F20101208_AAAJID martin_j_Page_027.tif
3dc66b04a3b45ded21d78ea10c442acd
17fa7555583c450f5664f33fdf606ee924cd10ad
51944 F20101208_AAAIEM martin_j_Page_161.jpg
85a657ea4dcb1286c9e8677073d18cbd
5f85faa5fbcabf7fcc0d7b5c398f8f549832d824
F20101208_AAAJHO martin_j_Page_002.tif
01dd90cd5818c9215b24a50120dbe43f
69ac61f9400a526a1de397f8c891745c7b39dc8f
107161 F20101208_AAAIFC martin_j_Page_128.jpg
e0883a86da09458bac7c09127ba509c5
17fdc5490fa889775973154558bbf2482c26599d
F20101208_AAAJIE martin_j_Page_030.tif
2653f9bc7ac0e1ee987c141a120ae95a
ec90b6a284ee638cc2ebbbeedc15e740eb4702a1
60594 F20101208_AAAIEN martin_j_Page_040.jp2
b7082b30fc3b2633d7a8b2bd23c6bf46
4957f2feaf7fa79483eb97d102d9fdbb78c87b59
F20101208_AAAJHP martin_j_Page_003.tif
9d687ccc3f0e82ea1891fb64778a9c97
33f2d24c0455509016405931fc9d9d637f883af9
2181 F20101208_AAAIFD martin_j_Page_051.txt
6939ea6bb68019730612b9c64fab8ec1
f7eed9b1c8aed0ceceff0bd6ab664a60b52c253b
F20101208_AAAJIF martin_j_Page_032.tif
19637398a3ae6af0c025f9d0977e8b4f
a1a2a0f9da1ee3e420e113441cf460e94e886c00
2096 F20101208_AAAIEO martin_j_Page_145.txt
4cc0cf29730275dcfdb87cce86cedcec
8da0f224e0073204ba0c82fe2fa1f3e9c59c91b4
F20101208_AAAJHQ martin_j_Page_004.tif
720c91b01c9318d5635dc8cd7b659f21
e0a3f0528c3f7d17eccf65118da853e96fb56b4b
54483 F20101208_AAAIFE martin_j_Page_080.pro
1782e037074ab104171522bd6c2df6c6
1186a2dee1e6f9e6479bc353623179bebe31cfd5
F20101208_AAAJIG martin_j_Page_033.tif
2f152daf6582288c87d30e3703e7afd7
a5b1fee4bbc61440fd54b10984ea1cf3617444d0
32097 F20101208_AAAIEP martin_j_Page_114.QC.jpg
f9a9aef9de51cb3d451f43095850d516
d09333adbc83fd8c15b95889ac544cbae96383df
25271604 F20101208_AAAJHR martin_j_Page_006.tif
143c6cf96fe0963ec2adbea2c7aa5776
5290d94089407db35f9a9a543948391cfa1184eb
108371 F20101208_AAAIFF martin_j_Page_130.jp2
70128a4bc06494b9605cd874b9ce8559
24efb20490bdb0aa0427175c5681af2312d667e6
F20101208_AAAJIH martin_j_Page_034.tif
870280238ae1827333a7bab25a9a1a74
56649cd11e16f8f567c358d3bd1dd49962c1dbcc
100217 F20101208_AAAIEQ martin_j_Page_134.jpg
fd5cd4b8c8c9fca94f7ed8e2dfbce763
606cc285876151287918c38bf318198a770dd2a8
F20101208_AAAJHS martin_j_Page_008.tif
6affe7477018289738e2b19c090ad7d8
38f92b9657906e059cbd3950e6d622a94ab58737
53480 F20101208_AAAIFG martin_j_Page_160.pro
9eab6eb391394722e4c5b4d4c0a5d0bb
ea6f14e31d8896905a9eddf4a9d58b402f55e28c
F20101208_AAAJII martin_j_Page_036.tif
78be929bfab34e9df9efda17dda5ba19
4f2014bff70d63f2a9300681afc4cc99df132f9a
2008 F20101208_AAAIER martin_j_Page_129.txt
abe852a8147638a5f20c647e48f0d7a1
9ba5fa8761a107a773e6c70533594450f4ebbfaa
F20101208_AAAJHT martin_j_Page_009.tif
3f830eb05e26468fb7800907ef0f2830
a4a5019afe688b325c5a483bcd4d0f9ed22765ef
61503 F20101208_AAAIFH martin_j_Page_174.pro
f6fdb2d2f62ed5f74c1841cc39947183
adf84b00646a7aed5be3a3551f4374c542c27294
F20101208_AAAJIJ martin_j_Page_040.tif
2e21119f7ab65dece165a5c10fb5de6c
3330d0d3aad333157990ca382e6d5b6505a79d96
35526 F20101208_AAAIES martin_j_Page_087.QC.jpg
ce00f989a712977929de7c39b85149cc
0305fa4d6d5930bf18f6853a0ab39efbbae4f0d0
F20101208_AAAJHU martin_j_Page_010.tif
61cffcf8365d33a14662e9256618776c
cf8cced8c53d9c6f345c10f2005d33b5644335de
60493 F20101208_AAAIFI martin_j_Page_181.jpg
5678cefc76bb1620c2261364aba9f113
32366c17055df947df200447168ba1dde2f3da7e
F20101208_AAAJIK martin_j_Page_042.tif
591c36f6828c5b0d41f702201c898975
d2cd88d087579e89ac463a3937d712e0563c81eb
2146 F20101208_AAAIET martin_j_Page_015.txt
3d3abba688ce2a1a5d4f808320f9b952
0b5e786603e61b98dc244c929f3d9b9b407a418c
F20101208_AAAJHV martin_j_Page_011.tif
83d0f8cf972aa55477c22b568a2baf66
77b8ca1eca19700815066f66d2bb811a104db6db
32640 F20101208_AAAIFJ martin_j_Page_064.QC.jpg
ec5050d2ce6f1f6b7800a198d59e7a7e
614ed70538193da905ddf353502b1c3b778cc55f
F20101208_AAAJIL martin_j_Page_043.tif
8afe88229db8e411989b68cd89d99972
5f37287ff721a0c1955f12930dbdf3a0ec72c192
F20101208_AAAIEU martin_j_Page_068.tif
abe2b1cd25f8448d8203724f0f5963a5
12744fd8419431026971b412da9b95246b80c007
F20101208_AAAJHW martin_j_Page_013.tif
8c5bc7d0e77e768d5af7484269f4d721
4a4967f02e87d2d0f69ec13e6cf1f628f331d30f
F20101208_AAAJJA martin_j_Page_066.tif
998af8a6abc304e194c145d3639d4161
95bff00b8b0e0477ba8124c0b5705279d17d55a6
20357 F20101208_AAAIEV martin_j_Page_073.QC.jpg
25d360941b4d15545daf4ecf148da7e9
2a6962e560d70e52e74ca3789eb57f029a23d7db
F20101208_AAAJHX martin_j_Page_015.tif
618535f5e98e984ad8a29d1464e04312
5db86abbd59bc6430a09046656b7b3f0e2964da0
F20101208_AAAJJB martin_j_Page_067.tif
b9ca816b6f9e6c21c19b95d180da6f42
5612b2828c2fb9643d6715e95a2b84ab0242e25f
8633 F20101208_AAAIFK martin_j_Page_153thm.jpg
26d489c7b6a432dd739b93302005ac72
8a544ea6488d0501e68cb103b81da1da97d75759
F20101208_AAAJIM martin_j_Page_047.tif
0122e9d3abf2a911d5e370331af2ba27
52b89c111cdd5cc163f2687a7ae5753bf5115506
9040 F20101208_AAAIEW martin_j_Page_051thm.jpg
b6e327302327c7c1e6ccc412513e1c45
f8a6ea50554d02c0d6381f40d1854c31dd3f8536
F20101208_AAAJHY martin_j_Page_016.tif
6649809de087cfceba8ceae27981d41a
a9e84043bae62c6af63994b2c9127ac1cc787058
F20101208_AAAJJC martin_j_Page_071.tif
d4af58fbfd92594367df0fde19e10357
99a87454a0fb44e0d9e64f5ca2f192bf9e201c4d
2012 F20101208_AAAIFL martin_j_Page_086.txt
37b36250ec354e0373ce1f78b5ea4d93
8bdf35fabadfeabb244cbdbddf218420dc764409
F20101208_AAAJIN martin_j_Page_049.tif
5c9b469a77d8fc3b8198823c8b1772d9
9ee51852af866fa7bd9a886b0d49a721b5581716
32146 F20101208_AAAIEX martin_j_Page_088.QC.jpg
b392adb1db32e1ca71e760d935f639b7
9956e799b332bd141a612b543d1202f5fd926ff2
F20101208_AAAJHZ martin_j_Page_018.tif
22b8d80a78b8f10308a39df014c61705
750f2528ad661d53c9932d0ba3e4b97aa9bbe8d3
45547 F20101208_AAAIGA martin_j_Page_057.pro
99a11c5d3e30d1dd84a389e12bf87bea
5d6c6240e129f22ee791ffec59ca234e39d3afa3
F20101208_AAAJJD martin_j_Page_072.tif
2bc41cb4fd1590ec580fb26dd28e9cae
4f9aaa8e6b67a9dee80e3bef17994bba506e6134
169694 F20101208_AAAIFM martin_j_Page_007.jpg
591c9586ffed7cf3693891bb5e8f00f0
046fecb021dba042d84117f750f542f3fc6fb510
F20101208_AAAJIO martin_j_Page_050.tif
541dfeb06cdf951233a12f8b3cc6b0c4
62b2213823f66ea8d1a1b61b6f26da8b7a07e7c7
31777 F20101208_AAAIEY martin_j_Page_077.QC.jpg
330cf2a39548d5018fea6e02bbd837c6
a83e99b6f9474c9a520981662f18057c1a2a620f
51087 F20101208_AAAIGB martin_j_Page_103.pro
c6033c8d499972426b493f5a732acc2a
d530c802391b7d7325703110897bbcf1b167fb88
F20101208_AAAJJE martin_j_Page_074.tif
d1cd85c759219b4ddd16e8efe9c2adde
1e4340703ca3815434f2865b552a5f7d27362de8
1637 F20101208_AAAIFN martin_j_Page_164.txt
11d37f521ab4e038beec856dd397288e
bcfcecaadd48d8b4d8eb852ff9de6d285865e94d
F20101208_AAAJIP martin_j_Page_053.tif
08f36e6b6fc1f714fb3bcd1cf95e4f09
2f690177c7518b9cd4947af7afe8f3a5d2193788
93422 F20101208_AAAIEZ martin_j_Page_113.jpg
f08e9818a7f6c3745542735faff0d739
a53347cd17a82577be4e5ae38e4dcb65a93e8579
F20101208_AAAIGC martin_j_Page_145.tif
3d584e5825ae51aff70657d7b2bf5845
2e5ff1aa0a3f6bc4c97755fde645d06d656b389c
F20101208_AAAJJF martin_j_Page_075.tif
f9dd36a4251530a17c28f02a96cd21d9
6f80a193c6ef5bfb88c87a5feac932f7d4b3a3fa
F20101208_AAAJIQ martin_j_Page_054.tif
92c2988144df14b4c8e795a6274636f4
3e0400fc3be30ae4c5fa56e59322dcf279a7771e
34700 F20101208_AAAIFO martin_j_Page_015.QC.jpg
3416a936028845cb7e495c82e97867a2
357cb6c41198017391cdaba9c0bf49be7d50145f
F20101208_AAAIGD martin_j_Page_052.tif
704b2cd8def1992491e99a75294a2f91
06b9c6950b48a020927eb811b095af072c1efb6c
F20101208_AAAJJG martin_j_Page_079.tif
f619ad17d4c76b4daa1e6a885a47d0aa
33ab80d11bc52d092ecacb6c16cda41dac139663
F20101208_AAAJIR martin_j_Page_055.tif
190e8ea9c36bd4dff05cfd94a492564c
fbdb0023d8c9eb7e6b60e3bbbb45c25238de7e64
8973 F20101208_AAAIFP martin_j_Page_053thm.jpg
39a8d087e00d69468dd2400e99154b5f
bc161fb35aff8ea75c9f9761ed5f7911c63d4a22
53346 F20101208_AAAIGE martin_j_Page_052.pro
7a3af2f4cb847d5f333988df58514745
b60b980555b4bb83e075a0e2c0f28e8a1c76aa68
F20101208_AAAJJH martin_j_Page_080.tif
81111bf10f1ff4bf96cfb668268dbec0
a83aba42bbdffa5b9a91d778ee5d696d1b584491
F20101208_AAAJIS martin_j_Page_056.tif
f4662032ffc216f786b90d13be621f70
9b84e48450e68fe4e9f5835875325c40be940662
119575 F20101208_AAAIFQ martin_j_Page_093.jp2
471b5fa44eceee1ba1a6ef594815e927
c00a91353261593e23ff6a303939c23c4cf299db
102779 F20101208_AAAIGF martin_j_Page_086.jpg
666e09af3f1b516d75fa1b19457bfda1
f4f19afab75e4d04708c9271774af78b8c3a3d94
F20101208_AAAJJI martin_j_Page_084.tif
490fb6fd3733bd15a48fdf3d10eb52b7
844860cf569d144e6973d078c61ad8969492c596
F20101208_AAAJIT martin_j_Page_057.tif
494c8e8ac7660827c28156dec96c8cdb
dd80f7a767a0d435aac089f65136e109b92a6919
F20101208_AAAIFR martin_j_Page_037.tif
cc6412b32802700d721f252e09d0393b
2b51ad018dce53ec8a3e8774bdf79732189e4a51
388460 F20101208_AAAIGG martin_j_Page_041.jp2
214eefd83be93249e5067c83fd456eee
6a0476aeb813c438df70f22b397513984d0b8b97
F20101208_AAAJJJ martin_j_Page_086.tif
cec56a6879301270c9209baed3b22e78
8aee68bb0c23d1f8f1e4b1aeb6ee74706c788983
F20101208_AAAJIU martin_j_Page_058.tif
11deb34b0970d915b50bbd33a3aa2202
fb20c5854b83536864934f8b1ebfd7e3ccc6f167
5138 F20101208_AAAIFS martin_j_Page_171.pro
df5a3de80da41503974c8cc2fc7c554a
8c6a056f9ce5c526a9d5dc90621e02dd0404f1be
F20101208_AAAIGH martin_j_Page_176.tif
03f97dc58836469f3dcbc8aca46161dc
b1a656e4b10982d05b451eb673066a7f6ee2e4ae
F20101208_AAAJJK martin_j_Page_087.tif
9a3161e79ea14f3dbba467db9172666b
f6979acb7a4d993a3fcdd3cb0eeb4a6d1eaa9e29
F20101208_AAAJIV martin_j_Page_059.tif
7530e550d84d3968c4761cabe7bc3688
663a25bd9d82b92ff5fcfbb6e281cf6d74ee3a63
37587 F20101208_AAAIFT martin_j_Page_009.QC.jpg
90e145cbc2df68469b63a646e593bfdb
ec65b62c58d562951785c46764cab9b0bcfe6d4d
F20101208_AAAIGI martin_j_Page_029.tif
f5448efd25ab241aad5e977c407f7758
40cf29a8147b440b1fb4ec81ad84e3c73c0b9b1c
F20101208_AAAJJL martin_j_Page_091.tif
3aae368905f4984441fa0b564352eb45
3586f972cd0fe265b6912757d188b837cad294e8
F20101208_AAAJIW martin_j_Page_060.tif
8e1c957b9af35603d628e4eb2ccd5cf6
2859a589ab382fd884030a654bc3f464fabff8f3
F20101208_AAAIFU martin_j_Page_088.tif
0ee346b0d2ece4f994e7209353719982
c5e971b9dd54672560ff88b91409f03c103f715c
F20101208_AAAIGJ martin_j_Page_152.tif
80c7b43d5f59eaad6d9a7ad45b9acf7d
5922cdfed715a327a83c22e203675ccacf42780c
F20101208_AAAJKA martin_j_Page_115.tif
1279881c709674335a4633e8d1778b85
ca9f30a9fce6598e7ef9e88c9c1457ed86f4ac7b
F20101208_AAAJJM martin_j_Page_092.tif
85e0c4aaa78adbfc0bbdc855e233a1f0
20746161dcf35ba7edc686d83b30f2d96bae944e
F20101208_AAAJIX martin_j_Page_061.tif
d2bc79358ff4a8d5351f2f1194283f59
a36d7f32477b984b589daecb81994dda8ccfe4f9
97875 F20101208_AAAIFV martin_j_Page_077.jpg
299af0a7ba675f95c7920a695f2db7d0
03ab1b75f3e90fd6d98cb9148cb9dd2befd0bbb3
2144 F20101208_AAAIGK martin_j_Page_045.txt
7ff301b67bb55049a734e3de194397e3
350fd60086c794aee5fe8041b5f7cf7b6e32aa65
F20101208_AAAJKB martin_j_Page_116.tif
25d2b3877e0b44943a4a0f4d37c3a1bd
ab19e5b0e905130831da895d9de9872dce88dd68
F20101208_AAAJIY martin_j_Page_062.tif
88f833ee6bf099672d9da79ae9cd3a4d
b0e59b7c412e0ec8850478715330888e22dd83b2
100133 F20101208_AAAIFW martin_j_Page_115.jp2
f22fc1533ed062ac2434f8908b5053e6
619105c12a6289597b902f6547c9d288553310cd
F20101208_AAAJKC martin_j_Page_117.tif
bcb3e360e3fc6e8a84b64440fcbceeee
6af30b05d4b328ff5640dc222c531194db5a509b
F20101208_AAAJJN martin_j_Page_093.tif
72f6a3b1782a68d40b280ae0fad10508
408c710b3303c9c6bdcb765a14e9842478733460
F20101208_AAAJIZ martin_j_Page_064.tif
3c2689d069d6536625755c235a03937c
4a52a2ab8f94706c22c236ce5a9e7b375148dfa1
2121 F20101208_AAAIFX martin_j_Page_087.txt
f03977674eca4e713b5c63b80ecbf2a3
06152322b65c44848fbd18683b9a8899df15868e
3760 F20101208_AAAIHA martin_j_Page_136.txt
5a180eadcfa580c6d712c41a1ba91cc3
7658789d86db4a30b77148671230e21203b41f0d
36334 F20101208_AAAIGL martin_j_Page_035.QC.jpg
3539c58f1431f2c365a0703bfdda4eca
4611bd490660ada183a3a72e89f95c9fa100a69e
F20101208_AAAJKD martin_j_Page_119.tif
3dbd768b159902f2c1080f7e58db9fd2
a972bb7b1b6280eeb753e646c85e0963ec7a21b0
F20101208_AAAJJO martin_j_Page_094.tif
ebbaa5e7afe10d4490c6b9a50f8c6611
91fecd6c55303e5f42d0626f902d7c7d9f3929a3
54813 F20101208_AAAIFY martin_j_Page_019.pro
9211c6a7e81b96962edfc8c687223851
4df5f5a7fb97a56a1f3f58b15c13bfceeb5577fc
113276 F20101208_AAAIHB martin_j_Page_051.jpg
8250fd722a752592b62d9faa79dd1400
62dfe4689e8a647b185622a29aeba84e4605a73e
936 F20101208_AAAIGM martin_j_Page_101.txt
5695d08640fd13d7ccec23a79590837b
d9e5d15f2a8af0456a1d119c63dae04e840c66c5
F20101208_AAAJKE martin_j_Page_120.tif
4dac936a8fe0e5f25780218ce43daeee
ccf8e43a8752331a76ef396d9252a44f04e4d9ea
F20101208_AAAJJP martin_j_Page_095.tif
5dc896e3c72a3797dc32624bd6b0beec
5807da5768bf8f127bcf0c77306365eec06f9cba
3730 F20101208_AAAIFZ martin_j_Page_142thm.jpg
ae59e9809463aa627321d1422432be42
1948ea42e41c9d255b284bc902e06d7c20099a54
48923 F20101208_AAAIHC martin_j_Page_089.pro
529ad09356f75d74d3a43cd028d275cb
a4353fd113f112854a65666aea585cacc76aaf9c
F20101208_AAAIGN martin_j_Page_160.tif
760ccc47f1a9f81c78791618ad1b4b84
86c6593bcb3e10ed3d83876433fdd3e4c2dde4c0
F20101208_AAAJKF martin_j_Page_121.tif
4e08f1739aa26a61389f6534a1321526
44661942ed0fe8bace7367e72012006d859d4c03
F20101208_AAAJJQ martin_j_Page_096.tif
f0ccab7172ba6852bb45024a46f57299
3d6ef69e0b493a98449c580855a8ce41a8029dbb
F20101208_AAAIHD martin_j_Page_046.tif
328950a49f7691a82b9cf9956b1c204d
182bf7595bd0422a45c51b9cf3e3941611105e12
32328 F20101208_AAAIGO martin_j_Page_108.QC.jpg
c82235b6f4101d795413409c12d49f63
31a2b00d69779a2d76d912e03af1383af4260aa3
F20101208_AAAJKG martin_j_Page_124.tif
04ba9e69ff04e0e5e1ef7a13946cead2
da7c9fa104057d776965bbf5d3f89ae0ca06a175
F20101208_AAAJJR martin_j_Page_097.tif
89daa2b9e160d8d574cfdbb7e039ce54
45960ce93098de1acfea677c7261ef7f2e220989
80512 F20101208_AAAIHE martin_j_Page_111.jpg
95114575ba7a05bd0815829b2a6ed54f
d4b7f146122c6b103f810181125551b859f4aa0b
2135 F20101208_AAAIGP martin_j_Page_034.txt
747c221f104b9c4030966348b2cf6acb
f2b936cbbdc2fc8840c04165a588b767b1e3388a
F20101208_AAAJKH martin_j_Page_127.tif
2fbfb9da068a70dc7e0243274dd8120f
42888b9c977c4c35262facfd555e1dfe497b014b
F20101208_AAAJJS martin_j_Page_098.tif
8a89cbc4d8b24819029a4bc56cd863a4
3bf00ad0078b92c57a09831940444fb9a4af040d
349390 F20101208_AAAIHF martin_j_Page_098.jp2
d483cb3e68a9a630c6b823d0df8fb48f
3ead4f5d3f8ae68ce46b0a77844e548317b42c75
1881 F20101208_AAAIGQ martin_j_Page_124.txt
03b80d2e434cf4c075d88a594cffb281
0567fb24d5d241413648c96c62720db57d1674a8
F20101208_AAAJKI martin_j_Page_130.tif
a905de388e2faaca0c723532d24ceaa8
3d2d08aca24d6d53a321d2b9a0e05d9bb5ee8765
F20101208_AAAJJT martin_j_Page_100.tif
d4a2807a2f96870be191c51e2e09d8d3
637463da2c749d5bd6fc08ba31c945311226ed29
49982 F20101208_AAAIHG martin_j_Page_062.pro
e73953176c0cf80898157299271af264
8a6e0e1fc5eb03e31a29d713f2fca70d282ccc73
2148 F20101208_AAAIGR martin_j_Page_150.txt
26fd5cdf4febd5507727c072d413fb62
0bff811f6d6e3149528d05f31adf091486645601
F20101208_AAAJKJ martin_j_Page_131.tif
a73246626c623d36123476f6b5027f8f
da4c8c7e28e15592358295ed18d881087217c31e
F20101208_AAAJJU martin_j_Page_105.tif
36afbe641272be9a987009b2964a79ad
a7c0ad93d704487a1d77f6fd5c0308b4f6b5dc93
17222 F20101208_AAAIHH martin_j_Page_159.pro
96f9422dd0f413e259459acc5d90baeb
d3f5f34f93c59e4435b069e7f07a7a92c017faf1
60537 F20101208_AAAIGS martin_j_Page_175.pro
efe14addb035736593916ba7de94c7a7
ff58be6e1d7524d307cb341e9bfc1de3bdc9ccf9
F20101208_AAAJKK martin_j_Page_137.tif
25d9531cb354dbbc68fe7bc2b3a409bc
71aaa646df570619e02a4bf07de24d425b4ae143
F20101208_AAAJJV martin_j_Page_107.tif
e3821589257386786b1c41791176aa6b
69320927c76a76e5e368c7721cb53be9d39bbd46
F20101208_AAAIHI martin_j_Page_051.tif
b603d68de24626d24993cb92ee1871d9
6aa80d8b7ec1a2baf5198b24c65c02007e11b299
33971 F20101208_AAAIGT martin_j_Page_037.QC.jpg
b4cb2781ca05a43d0ce8aa61d550cfe7
118d9717890c32a5b998ac7a417a99468a638f17
F20101208_AAAJKL martin_j_Page_138.tif
9bc93741be91411be3a2078ec3bfe1ba
f15ed8fe0451ac8ce22b6b0cbf90a5df297a8388
F20101208_AAAJJW martin_j_Page_110.tif
de0d4dabdc57eef0a21dcdf73ce49e96
e94b9c47e3041d309171d544873fb4d6f1865c5c
2923 F20101208_AAAIHJ martin_j_Page_117.txt
427cae3b341c1fcf634218e518f47439
412d5147f23fcd4a7469eee839a8017d9a3ce594
1894 F20101208_AAAIGU martin_j_Page_088.txt
e29c6b17cf528962eecff74c7e6476c2
1dffa19b79b906c97ffbff7946ea77d3e8df95e7
F20101208_AAAJLA martin_j_Page_163.tif
e8d54aab58c855526e4e12ba6cf41b30
7807dfaec9c64eb67ddd4f099033ea8e9c4b1418
F20101208_AAAJKM martin_j_Page_139.tif
c30fa9eb69cd0cfb963a99f2157c14fc
e83b8cc74695f6929f333901010da1b9e1176ac9
F20101208_AAAJJX martin_j_Page_111.tif
c18857b2200afb0ca1cca881f1fe859c
42aece719a7741b472d1e72d4baae3b4f2853927
113973 F20101208_AAAIHK martin_j_Page_048.jp2
198fd2cb49416b7ddd95612e9cbf5a8a
c1dc24d88ff600e863b7f57782f7730f023fc7af
2174 F20101208_AAAIGV martin_j_Page_118.txt
d6d10c4cad1880b6960d487335c26c6d
6c1ea0512108f6bd300ec95277d0257ab20f9165
F20101208_AAAJLB martin_j_Page_164.tif
1dfc5d036d765008bb29df7be7e30f56
fa3d64db924750631b2a2cc6a936a04dd7316dc4
F20101208_AAAJKN martin_j_Page_141.tif
8e707bbcf179b7924577587f8e751845
ffc6a6aebc2141c59b81814d5338b21003eaa1d0
F20101208_AAAJJY martin_j_Page_112.tif
a45e8ae646eac37fbe78d559d45e1f7c
e4c968d3273144327f19f3e5ba84713b442a226f
8882 F20101208_AAAIHL martin_j_Page_110thm.jpg
64e67d37164ad89f5dbd3b72080a2729
caef2aa2420f9c411b6c7c64bd7456e7db11c586
7378 F20101208_AAAIGW martin_j_Page_135.pro
0883e54d74508b77027235da89eddbed
c49b5074fd0b36730593a97aa10d9fb34ac9d6c7
F20101208_AAAJLC martin_j_Page_165.tif
208e66fd5394e307a3ec03ca5a3ca0da
3ba0b0b390f601bd88bb30fe78c920b7af77192a
F20101208_AAAJJZ martin_j_Page_113.tif
28382e5d0e7c499269e910cec087383a
a2bf02bbd60d7900d463e8d8eadf38110b6af2b9
64581 F20101208_AAAIIA martin_j_Page_070.jpg
dd678372ff88cea0fc38cf4278fc1592
91b3d034d5cddebd3c58b90d9d991d7aea7608ea
1791 F20101208_AAAIGX martin_j_Page_113.txt
8557443dd75286409ba3e4c35b95715d
1c3db79b01a2861c8529c06cab8765c8194ea8a7
F20101208_AAAJLD martin_j_Page_167.tif
6aa953dc095c20cd4f4279cf44099be5
f60a6cd3b03cb0eeb436e0a191f47547b4f76b0c
8423998 F20101208_AAAJKO martin_j_Page_143.tif
1fc6d7ecab0d2c4c7d5ff48fe9de4285
9d04b53df72317bc9d0b91be5b90852830841dc4
33325 F20101208_AAAIIB martin_j_Page_145.QC.jpg
7726bacdce4f548a0f5c0f913c3ab1a2
bc6a1a618c7d909a651b8c8dd559c4f2c742a2bd
8970 F20101208_AAAIHM martin_j_Page_063thm.jpg
b4b5c2bbc8a8a8626319ab8ecd642cab
336e9f55e28c7cbfbf80dc77e84f57f366422191
115506 F20101208_AAAIGY martin_j_Page_016.jpg
d71390e8cd2c5d375a6c1d4ded858084
cfeddd7002742dbc23f8c94c5cb3e9f6d5fb9fea
F20101208_AAAJLE martin_j_Page_169.tif
95ecaed8b2fd0dd932e0ffcd8a111da0
4b927505f2eb178e149aaaca4bd725e9e7287fe1
F20101208_AAAJKP martin_j_Page_144.tif
df625e86b105788ecd6d8b677d541e29
2735e0902e96a1438b2b60b66fe8749c4e67f612
48572 F20101208_AAAIIC martin_j_Page_014.jpg
c202992854b05a68511b6077acf9d133
16e7cf5e3fe6f70189d2d6624af99a7eb7b9929c
F20101208_AAAIHN martin_j_Page_082.tif
69e9f91e23eb5d68954aa19521ea3604
4c95aee846af2a139d06b2c9967621262eb9f985
67229 F20101208_AAAIGZ martin_j_Page_011.pro
7514e50507ce384a7971468d26eb2f03
8153cc2be7f15514ad5bd28f0db21ad70a8a08dd
F20101208_AAAJLF martin_j_Page_170.tif
78b1962c6fe008b5513d76392539d95b
6084944579830f21f856d1cbe9d49e417aeda935
F20101208_AAAJKQ martin_j_Page_147.tif
665f122ee8241d8a44ee5e231f8bd38f
9c7fbbb2443a1cea5664e8fce7d5462128bef33e
32862 F20101208_AAAIID martin_j_Page_025.QC.jpg
bddcc0845a0eacede4cde7bb7fae7fa4
b2603079493bfa078eef0cb34074d4508758008e
F20101208_AAAIHO martin_j_Page_129.tif
fe2112ab0293d772e63f0cf37be73f69
93b267a86c9595a2ff6acb0dcf162b28dd64f9eb
F20101208_AAAJLG martin_j_Page_171.tif
828e76c642eb5e15b55bfa8d7088a9e7
58299f8d1ca755ec5702613f026afd1715e79928
F20101208_AAAJKR martin_j_Page_148.tif
aa7aab126f25f3aa687d2904387ad6f0
0ce24093f33646ff55b76d408a5c176c74b2d371
106237 F20101208_AAAIIE martin_j_Page_132.jpg
c3fcd17e79a82522530304caff6ac79b
8b3218739fcb090014bc67c067516bf289c3f80d
36890 F20101208_AAAIHP martin_j_Page_063.QC.jpg
8df13ef929d6323984b78844b792a0a2
7263b5c6d5c6dca2740c03039689f9d5a5a1fe8d
F20101208_AAAJLH martin_j_Page_172.tif
8ee34207997e635e1f6b7bd39f8d3e7a
2a1e241718cd82ac12682f7c4980d013f6f4f5b6
F20101208_AAAJKS martin_j_Page_149.tif
ceef39365c8143d5470f279d824642a9
45549833983f0134860c0914ce8588c1e62f01bc
109346 F20101208_AAAIIF martin_j_Page_078.jp2
dc2158e035410422b08bae7581584168
c3ab316797cc1d1ea7b70a08f7aadbfeb1d95f30
30730 F20101208_AAAIHQ martin_j_Page_115.QC.jpg
9ea1ba3e6ada2bae64dbb5c405fc933f
3de895eca7cfc1766398ca8b0c3d31208e1785b6
F20101208_AAAJLI martin_j_Page_174.tif
8f279dbe41c07c8b388a6fb082f46edf
557992db2ed53c7be97bb47a8445fc8309e9a612
F20101208_AAAJKT martin_j_Page_150.tif
895221c69df8379a811174a0ecf06935
9f0556fa156ea2c28994bc3d0cb9fd72b23a8e1c
111913 F20101208_AAAIIG martin_j_Page_086.jp2
baa550d6be41a4401a7f73d726038ea9
5f237018af5d0a716dde992d7341b1dca8fb12e8
F20101208_AAAIHR martin_j_Page_045.tif
f16163667f45df1c9c644de4b612425e
e7ba6856d1497a90dd07a4c91bf99c001714a8de
F20101208_AAAJLJ martin_j_Page_177.tif
10b1221fc278672851f12ab2dd1b4085
ac807d0b483bde2383594d78d100a5611d3c61ab
F20101208_AAAJKU martin_j_Page_154.tif
98c68285239f3995686c1961905e3763
47ad5616e933c4f0905f6df3e34236d0a3d7a51c
33006 F20101208_AAAIIH martin_j_Page_078.QC.jpg
613b1d7cb02217a3640192b5a5cda308
f278b11a577d0980d26089546b173341c5a2fbda
F20101208_AAAIHS martin_j_Page_175.tif
a413fb83d1669425845d4fb6e42136a1
267daf5a86b5f818f6caff4ec4defdc290cf93f3
F20101208_AAAJLK martin_j_Page_179.tif
328795681528007fa5e29759b75d4c41
f0ea0e0468c9cd53d5c3e14db3f7dcd0bfe120d1
F20101208_AAAJKV martin_j_Page_155.tif
826dd871c120fb5f2fa39c18eebb3206
a1e1007bb1dd53e2bd98a93aac39da9d5d61e5c8
122062 F20101208_AAAIII martin_j_Page_008.pro
a3021e7914c1e0eec8e18943ec161d41
5e2a7042da45e41b09ef94614ff80c09bde2ffad
9110 F20101208_AAAIHT martin_j_Page_021thm.jpg
cd576968f21394f4294a51112595b797
8d4f7c866ee5d1f7b64c9791a35b9b4d9d25367b
F20101208_AAAJLL martin_j_Page_180.tif
48c9b9c7a8b96a5a3a34bb0f9979bc06
9ebce41083c6362ed836e32f1cc0e41d366082c7
F20101208_AAAJKW martin_j_Page_157.tif
98bd93c3caef0a6d86316573e3be3a3d
b7f3e0531d4903c3a56f5c27f7396e5d957c6c3d
102154 F20101208_AAAIIJ martin_j_Page_079.jp2
2b348ee3db6faecadde8fbce0f6d038a
5548cbd0b1813037538314293f7c8b0d012a38b9
33298 F20101208_AAAIHU martin_j_Page_054.QC.jpg
66c6e93edf97c2314cc1db338b37a65b
66d2bf1c551f14358343a16040b8694726e02a6d
46027 F20101208_AAAJMA martin_j_Page_025.pro
f916608bacc3b4c02b91d23d28353fd5
d1ee0fc3357ffac76d0b2289e9b0b9388d529b50
F20101208_AAAJLM martin_j_Page_181.tif
893fb81c17c7136fa0ce573651014d34
c3c6f8dda27582159144a2fa80d1d5499cc519c1
F20101208_AAAJKX martin_j_Page_159.tif
defb079f483d6e20b1ce5513edd0d829
844db7cd9f8450c9a41fb7feb802fd05997696e9
1534 F20101208_AAAIHV martin_j_Page_071.txt
ff63c6a9043d65210d85fcf2d73b832e
817c51ed62f113999adc8ff11ff0265e2b12eec9
F20101208_AAAIIK martin_j_Page_085.tif
f7bdb3d9514a961eeba26964535086dd
bf72ceec4f30b3ddddde5fdbe2b1db03619bc2dd
52656 F20101208_AAAJMB martin_j_Page_026.pro
1185c5169a05b69d6a060f0c69989895
cb76d9d35ca8e6d1c74bae5463f3977802b82b83
809 F20101208_AAAJLN martin_j_Page_002.pro
929270cfe3497d56057fe2133c4e7d79
6c8e19c2a168e540419a248c65716f32ff8ba285
F20101208_AAAJKY martin_j_Page_161.tif
d9a64da45b501d1d516377679fe304b9
b69cd2981d7e3f6c591406bc6b9c1334b7d64e92
66690 F20101208_AAAIHW martin_j_Page_176.pro
1f81fd27dd36003d959d0a90e1ad1c24
c3e05493623b44a033a2f716c5b44b819988eb1b
1912 F20101208_AAAIIL martin_j_Page_070.txt
5394c6f4801e1ff229dcaaaf56ec877e
c9766d2b3ea2ebacfc03067946c2b12623246cbf
41751 F20101208_AAAJMC martin_j_Page_028.pro
79c6d72a5e42c657075f84a4b743b1c9
2c0e64471591d1d2b95e8ed758a25921333a1b0c
49520 F20101208_AAAJLO martin_j_Page_004.pro
b317e67dd945a55263c69d69f3c019ab
6026b4e8872760833a8fcac3805659647bdd5f41
F20101208_AAAJKZ martin_j_Page_162.tif
7e91a9976d97e14af0e992b78a0bbd02
fba05c8d48bd51b1e46147f16003e1c8ce2fb5a0
8622 F20101208_AAAIHX martin_j_Page_177thm.jpg
090df776cc7a50698544ce4107689065
d9e729892fb86ea387c0e4e04887e7b26032993b
50671 F20101208_AAAIJA martin_j_Page_078.pro
ae1c5625c8e8a72b2b7725d7013cde0d
1a5687f44a430508e95a8d5154ac92e7cdd3fab0
F20101208_AAAIIM martin_j_Page_044.tif
e508d630e63869c5d7d01b234206ca3e
d165bfabb8da9aadc33ed502f9fc0a3c244e6a79
53543 F20101208_AAAJMD martin_j_Page_033.pro
a9d3723551518d9ea144afa10c4ed76b
7249cd5898bfe7fc45532f3f8a76d1132dea3f4a
8308 F20101208_AAAIHY martin_j_Page_023thm.jpg
dab07e77d5912286ad5efbaa08847d72
0dbb39014109b404dfe90c1ba6df95413d255dc9
95944 F20101208_AAAIJB martin_j_Page_057.jpg
3acd73f92cd0fb5ace12e3ba7bca96f8
f66c53d2c444dc1342e8574f1c7e7afd83ebd3e1
55519 F20101208_AAAJME martin_j_Page_035.pro
4450d01916347c5bcef18d5306154d42
4336d0c50c797f75b44451646f4832ded3b9faf5
39075 F20101208_AAAJLP martin_j_Page_005.pro
eccb2470f268cceb36811b6569b39ed4
10dc481567447aa74f9c8758be3a0ad6cee4363c
68241 F20101208_AAAIHZ martin_j_Page_012.pro
f731f7f60181ecf3b82bdc9e9332bf70
1964fc28acbb22a9f6e9297aa8149d60042d9df0
2495 F20101208_AAAIJC martin_j_Page_175.txt
baf67d2b2f7bf5980662ce3255af1b07
61129678ead9d32d3c9f5a72f8f2365a911c7cf0
109131 F20101208_AAAIIN martin_j_Page_163.jp2
c005d58156474d09060453f348f93c7e
50dacfa4e229a236e98a231055e6177bc3a97bb8
27597 F20101208_AAAJMF martin_j_Page_038.pro
6c3d11ba8c4f9026bbcaf89888c75310
be219676bd8994c163ba4d53025ad9d4b8674a25
107981 F20101208_AAAJLQ martin_j_Page_009.pro
004b2002085d98a7a9820f508d785cbe
bcb0fe53ccc85f14866471b9e0b41fd6406c6422
7352 F20101208_AAAIJD martin_j_Page_001.pro
5c196286b7c6b40e0268204889ccfa31
d56b0cd72c9c74127340c808f2d36a8ac5a4f583
105177 F20101208_AAAIIO martin_j_Page_048.jpg
56dcc9c52c41fc3bb34c33df249fe405
75ceb797d3370c8219e51b4ae8306ad42bc33172
22870 F20101208_AAAJMG martin_j_Page_039.pro
78fab2f698761fae9e6cd2b3fdea5b25
7df4f8140b467c49a7ce1528a5fc0e99e111eb68
22141 F20101208_AAAJLR martin_j_Page_010.pro
531620c8992a8f35d939952cf5ecea63
b37a51ee40d24e309919e2dfed6dfc0c770293f5
83975 F20101208_AAAIJE martin_j_Page_119.jpg
839de26004793e315941536b68ebfe70
0e9aae5f155a36fa5aa3c4bd7e8e00add4f293a5
8977 F20101208_AAAIIP martin_j_Page_019thm.jpg
52a2900f6aac318a0dacb30ee0a4d295
d5cbec3592bcbea440d725cd141660d71b8a7d4c
7051 F20101208_AAAJMH martin_j_Page_041.pro
52b1d890487cfd4e2f99ce76f36f8ae2
66d214638dc133ae0d191ccf7e203085fcf5a5a2
46811 F20101208_AAAJLS martin_j_Page_013.pro
57d8e7d10dfeb0a5b1ac60c0437b49bb
9ffe63e41cf0487e12c480b2a6c5c17369154b6a
7081 F20101208_AAAIJF martin_j_Page_169thm.jpg
ea977b90e29950bc8987757979470023
95ec572b3001c3c7fb40500e6aca652d3c285cd2
104937 F20101208_AAAIIQ martin_j_Page_082.jpg
9832e8c30c4ec5f3d81aadb3cc63d31c
ecb2c353744b21fa8bf557eb3cb24818ef7256ee
24861 F20101208_AAAJMI martin_j_Page_042.pro
7922756b6b8b60c1f7c8ecb76485b772
3695051e633487368432576791d8ff553749829b
22133 F20101208_AAAJLT martin_j_Page_014.pro
883fd2e4a542d22c0b69a3bd7144181f
fd56e1c6a17c56d58f1077b0a097b50d0d90b129
F20101208_AAAIJG martin_j_Page_099.tif
e991fc8e191dbefdd5a8671b3a68f9fc
463f270f49aec422e0b318347f88d87155836fdd
F20101208_AAAIIR martin_j_Page_014.tif
f5103b5eaf58adbb981af8db026b79d1
09b0967b82edc5208de39cf80a39e075b024ac0d
6592 F20101208_AAAJMJ martin_j_Page_043.pro
3d9b3038740bcb8c8f68bddeb76c9ba5
3108c438928ab025796e9e22ab7c3cc42f7f2eda
57067 F20101208_AAAJLU martin_j_Page_016.pro
e20c405f5f1a7077eb21d658df915562
612d4ce006a1ed0a8693d30a0874d9bdea4811c9
6449 F20101208_AAAIJH martin_j_Page_135.QC.jpg
3da265b209ec090495dacd3c1e533a2c
b642a37bfe2feb8402d0739e6e8c13060efc50db
52597 F20101208_AAAIIS martin_j_Page_132.pro
6764d5377b4f2a499f959b543f10a4ee
cc375b77bb027873f76f0a6486b440088f673b4a
50634 F20101208_AAAJMK martin_j_Page_044.pro
f76b08f16170f949a40d55488e8de94d
5d8393e35ab48457bb8cf1a778c161ff3c73e605
51881 F20101208_AAAJLV martin_j_Page_017.pro
ab0809ecc1ad89702a78e4b8ec74fe3f
df32d71d57bb6c93564fdcf8fa414b92f99c92f5
51962 F20101208_AAAIJI martin_j_Page_053.pro
64e2d5c087371cc5b5b6a92bc5bde228
f0f90366c2f768b5607f8d83b21bbdff1c07eeb1
100231 F20101208_AAAIIT martin_j_Page_057.jp2
20ef21ccfc3120ff77ecc5b4d0f56704
029366f25688fdb1460cdb0341e5b252028de957
54499 F20101208_AAAJML martin_j_Page_045.pro
ee3095a216dd013107d16e6a480118f3
55dea7ce122580163cdd798d7c893f1e1a8bae1f
58026 F20101208_AAAJLW martin_j_Page_018.pro
830d7b37e466340a431cc668780daa89
0f4e3956aa739257e47964d32e3f99f3a986860f
1725 F20101208_AAAIJJ martin_j_Page_106.txt
2404f988daa6f641571ee77fbc7db60a
6dc8519c13283cb607c97325037a55dc0a4fe6ee
31356 F20101208_AAAIIU martin_j_Page_124.QC.jpg
3c16d4655a6b47c00471a8a3f56c7c04
d3d65e7aa2174c3527e04369c806d8927975b8a5
54254 F20101208_AAAJNA martin_j_Page_075.pro
a650f03a0705e50ead3fd620f7351aa7
3040f61da3d642144f98713789d23ece3ced7e6a
48439 F20101208_AAAJMM martin_j_Page_049.pro
b9cd4aed4491d4ca29a2d187bf51270a
a126cbddd5a299160905f2721a068b51c239a668
55272 F20101208_AAAJLX martin_j_Page_021.pro
cb86ae6763afb8580583f2f75d152626
18830e691c91315515b3c57128a17d156c510c6f
3818 F20101208_AAAIJK martin_j_Page_139thm.jpg
be79669281e49f1a99fd9e11342e41dc
884c8c381b8cf6454ae2c7291ea82b67f304465b
F20101208_AAAIIV martin_j_Page_134.tif
7dbc4f186d2858684f7599c4976987dc
9de5ff38e34654e0bc198e1195878ffa3a172ad8
49369 F20101208_AAAJNB martin_j_Page_077.pro
06e68dba576f84915054f0c7be9072d6
4be4c3ee0d842f7470027e0358cadc57d63044e1
53525 F20101208_AAAJMN martin_j_Page_050.pro
5a6e7117600966b630e33afd6839610b
7b613765f9aa91d636c8175dd03a3966cc31d953
54383 F20101208_AAAJLY martin_j_Page_022.pro
73e3820e7cd51c9a4d0958dcb5d6778a
4d288ed6076e5a6e89bbde1ec559b8cc3e933c54
85085 F20101208_AAAIJL martin_j_Page_029.jpg
30abe2febfc18d7234aca29c8055e9c4
52f6c0887a9495795ba260941d404f278bdce065
15278 F20101208_AAAIIW martin_j_Page_139.pro
1eb9d15c11ed44d2e2bd4303b9cc1510
6c01d2bc7eb044d96f249a2b5a0203320c02b290
46763 F20101208_AAAJNC martin_j_Page_079.pro
a67c30b2bdf366973918b95fc6ae4aa1
2e9becb48711bf10b9c07b36be7535858a88e716
55691 F20101208_AAAJMO martin_j_Page_051.pro
cb10f6d730653b2a8a4800b94faa41a5
7cb82650c1016259769a8d08014692a9d819192a
48393 F20101208_AAAJLZ martin_j_Page_023.pro
b93fb38807ddda22b8f3f60c811c1828
b180b761b49cd3710a855b0de7d8731e458ad304
5050 F20101208_AAAIKA martin_j_Page_008.txt
8398ad3f3009788146db04d1db0aaf1d
a31b8bf34a9cd8d0fbd128c7de6bbd266c082445
2234 F20101208_AAAIJM martin_j_Page_019.txt
2d54283fb98fc63f35ad906a8ea7a4e0
4bf636980484c3680ac4f1c9a0f847b90c0b2d55
36097 F20101208_AAAIIX martin_j_Page_120.QC.jpg
0d845ca766d30f307f6b554bdf72031c
1d1f19024150a927c95a318df232062ecdc8a535
51928 F20101208_AAAJND martin_j_Page_082.pro
f04e66edd421a7134347ad7e4277a616
4ee885028452025f1239c7aab5c1af1ff376a1b0
51545 F20101208_AAAJMP martin_j_Page_054.pro
89ed8e131a006a7e03c9795b167b05e3
500b154a83e0d640bec26ccd97ac96d5073b8ebb
49581 F20101208_AAAIKB martin_j_Page_030.pro
aff53efd77a4ced738eac7fb967f81a6
de5b55030565649954e224619d12848327105591
F20101208_AAAIJN martin_j_Page_125.tif
7e9ef5febf8699b5e04bbdd0eec1d47d
aca2f8c10b7adcd0d26e6c568e40e998e92a0f1a
24494 F20101208_AAAIIY martin_j_Page_168.jpg
024858c4afa3e7918f63fcbc4b43ce4a
59d86d985812f1582c97d3022f6bf88471446937
55591 F20101208_AAAJNE martin_j_Page_083.pro
0975544a8c84ccd08bd4eabe8376171a
7125355177bcc050aa1c48ece72f2e4faaaa91de
F20101208_AAAIKC martin_j_Page_078.tif
4f0bd2c602b1672bdd0bd6046ba752d5
1c31ee10616cdde1d9e894055acef363bea8e8d7
2004 F20101208_AAAIIZ martin_j_Page_062.txt
c7b8f54bf3a8d83d969ffcdbf2772bf6
ae60914d240ac087e375254ca35fcf193e7d577b
54387 F20101208_AAAJNF martin_j_Page_085.pro
c2e5fd75d3ee1e09a01b5ffacbeb9f1b
c46cb40b6467be66e04b9a6d184973647f941bda
51372 F20101208_AAAJMQ martin_j_Page_055.pro
ba73a3746cbd9f3a53faaeb130d10a0f
9120e7ab102672aa7d705ddf8fdd160be0f1d340
49332 F20101208_AAAIKD martin_j_Page_047.pro
2273644980b199bc4bb16dd43a0020b1
9dfd190ebdd0a0c8a9186de3daef2a0d71baddd1
2025 F20101208_AAAIJO martin_j_Page_061.txt
a24d907d2d009ce1c717cd8ae170a834
4c983ceddecd63a6b7dd9d11dd777167efad8685
51109 F20101208_AAAJNG martin_j_Page_086.pro
c08a2cc0c2921b5ff9bc169896eb6edd
c8b0b612cab0551ea7d2225f1e662e79bd171dd4
43243 F20101208_AAAJMR martin_j_Page_056.pro
c56b3de4193b647c4cb649a1ed9d54e9
1c2cdcf95ae8167311d93c601cad3a5d9ffdcee2
50012 F20101208_AAAIKE martin_j_Page_064.pro
0ddb704ec364486f49bce89a99f5b60c
6943444ff6c63cdf5aa083e308c13b5d1575d067
1243 F20101208_AAAIJP martin_j_Page_040.txt
51771cdab98c62b4d09fdf1863482362
dddbf1f06d3caf222582eae157f9b9becd43097c
54173 F20101208_AAAJNH martin_j_Page_087.pro
13fb40c04aad151e266f2c662bbada1c
d9e0b2c16841d1cacad33fbff3944f0b9a5e743b
44784 F20101208_AAAJMS martin_j_Page_059.pro
434296eaae3a218e744ff7c725a3b869
bfd36d36f86acf3b46160f56f6047e2cd5101f8c
32195 F20101208_AAAIKF martin_j_Page_018.QC.jpg
93f42de6e148a9be20bb363328070ee0
fb884c8fded4400828fc57868fe17a34a4946ba5
206 F20101208_AAAIJQ martin_j_Page_171.txt
2d13b07249bec9000b901a5f462d72ed
26afcca1b1130331adaa1ca1ecb1304f1d18db54
49871 F20101208_AAAJNI martin_j_Page_090.pro
1ca45bf3b423a03c22759cfcdd15ad0d
58f896556b63c0b1ab57f7cc4325b86cf4248f0b
49152 F20101208_AAAJMT martin_j_Page_060.pro
ff50eaff28c04f239c35a2cd63798fc4
2079e60c0a3a8cb6dc3bec7e839358e9db495f0d
F20101208_AAAIKG martin_j_Page_069.tif
1c2152c590d26bb16a75d636a236ef7f
10e11dc6af9a2e738ab48d1a7a612692ce3b4bb9
2169 F20101208_AAAIJR martin_j_Page_021.txt
df25c5c30d31aaa7915928c581c4f9e6
ff55b0a66556f20d1833435959f3f849c4fd4226
56784 F20101208_AAAJNJ martin_j_Page_093.pro
11eba434f4d40e005881f1e60d6b7edc
698976422bb84b2025a5537152f09a6eb65391c1
55149 F20101208_AAAJMU martin_j_Page_063.pro
cf0c256599f949e4c041486159fc6f23
2c8c880685f22f4c0d020f9a6c607c4d975f6d67
116984 F20101208_AAAIKH martin_j_Page_075.jp2
c9abbcc37b318f5232ea484d20eff48a
c31fce629ad6195d71af08ca85e67b745298119f
2368 F20101208_AAAIJS martin_j_Page_010thm.jpg
7df999af7a517f903d6b205fd877570e
6914fe56de573067132ac1b23c143dec5ad9ac79
56331 F20101208_AAAJNK martin_j_Page_095.pro
868177866a0b21298e6975340dff12ca
57b16c43551cb46b762e52c2e51619d59d64ffac
58286 F20101208_AAAJMV martin_j_Page_065.pro
d5efef7e2decf283f3b7ed44cf182c3f
aa404dafb69f3fedef528fe238a9153ba7f8239b
119434 F20101208_AAAIKI martin_j_Page_045.jp2
eb5f053cbc0dbe44c2a68d4f30482409
bf5d568321fe1d6435dccbdcc7ebd71e8def3c62
41500 F20101208_AAAIJT martin_j_Page_029.pro
4e0bb0f6bf6b67e9149b1b31c52c6d57
ada07ac7eded46ba8fab6827d9d6389cb63358e5
12192 F20101208_AAAJNL martin_j_Page_096.pro
70b845b2311899aca1c02fdeaeaaa368
9e83ae1158ef9a94d21da4dd0be1ad23e4a2bcd2
57548 F20101208_AAAJMW martin_j_Page_068.pro
58ff15a9d571f446c0d1aacc0f1c059f
3be94a693b233272621215ae495fe8fdee0a48d9
F20101208_AAAIKJ martin_j_Page_031.tif
286b413040f3e5f0f626d3d4712adc20
9669f88857238ecdcdb68d96e400505b139c1f3d
35638 F20101208_AAAIJU martin_j_Page_102.QC.jpg
e030437102e542b7b4ef31b34af7c0b5
22d65a06f688101f3dd2dc62a7bd4bb60bf18264
43469 F20101208_AAAJOA martin_j_Page_123.pro
26c97efb610c9e437d4bb05679e68fd1
c6fcc4598a94067a5386b5384a2c38166f407806
22699 F20101208_AAAJNM martin_j_Page_099.pro
1186a8b170a424cb24cf1209a574f22b
ad8d80c34d09d3c1cae90986309b2b8f7ad1c616
28377 F20101208_AAAJMX martin_j_Page_071.pro
6c9d7de2d313ea2b2e6d52ed78131663
d16baeffcd911ccc580c2b6839db0a4f40afa341
42562 F20101208_AAAIKK martin_j_Page_156.jpg
17d37a246514b5722fb16d205d8758de
bb61cc0d3c9ce572763df1e1f35a3babd824337c
110772 F20101208_AAAIJV martin_j_Page_047.jp2
75bf8738f03867e14fcd2711e49cca34
ff5d1055a3d41da73ec2d39fa6c5f59200b6efe9
47710 F20101208_AAAJOB martin_j_Page_124.pro
95cd1f4a4d47cd77ac81d0a9b3664dbc
1e1c2305a623c35a970c0607a6bb21cce831a71b
16045 F20101208_AAAJNN martin_j_Page_101.pro
c660ee1ad8d9f82bd5c586d0675b479c
d4ff724c18f12b73fa67a35d0df09d450398eb74
30862 F20101208_AAAJMY martin_j_Page_072.pro
4e786db305880e42adec1abab2d622a5
19f64008df92b183f1f8e8b68bad7cc48a361271
113665 F20101208_AAAIKL martin_j_Page_109.jp2
5f4d7f02ea48c7ddf3c720778b2a1f43
980546d2da1c2c5de8c379423d1ccae2ac5ebd43
8400 F20101208_AAAIJW martin_j_Page_174thm.jpg
7fa2bfb4812528b12590b39e280f6ce7
7c7a91803ec71a44650bd143aa37bbbadcf2b363
49767 F20101208_AAAJOC martin_j_Page_125.pro
13f33cbedee4e294f1f9125625f239fe
ad6d20e21be5ac1215370ae558bec26626bbb730
51959 F20101208_AAAJNO martin_j_Page_102.pro
d377e679eccbd79cb556f5c00591b009
5e5bb9f4ae92d2208466b4b1500c5014fd33a169
23843 F20101208_AAAJMZ martin_j_Page_074.pro
28a46b44a26d556ec5dfa4b595908ffc
1ab641cc585e8134eae37ac1aea4f1443d7aed5d
F20101208_AAAIKM martin_j_Page_073.tif
1ab19ba91d8e0a5716fc29c3022f33fc
fa005e3929bd431421ac2a8faf321753f96f7faf
11086 F20101208_AAAIJX martin_j_Page_100.pro
65ce687cf1b44723a3153e922390a562
81a2a0258101d70428b6e207f597ec49f0be7af7
104621 F20101208_AAAILA martin_j_Page_037.jpg
7a79739da86ef8c54ee6526d96603a94
aa6a05cf2fc47cfd27472b675f505e14fd0f1947
54227 F20101208_AAAJOD martin_j_Page_126.pro
4691bbfd44fae49987d4e08cd6a67409
142d9d798c4029f206c9c0f8400a66c5aa2d9ae8
53860 F20101208_AAAJNP martin_j_Page_104.pro
9b3f2a15a1e071cf694c8deab7d021ca
9b65de651ac87fe7602a1e95556e3e917383e2b4
37477 F20101208_AAAIKN martin_j_Page_093.QC.jpg
a6a5d874c3f4e1313765dad228e6b6d6
ca8d1ae2f44b425ddccf5338ff14f83eb028d5e3
112688 F20101208_AAAIJY martin_j_Page_149.jpg
4b5bf35b13e7d9e8e739936003af505d
e06b639181ef9608dd46cf6c2fa04726e59b8751
108333 F20101208_AAAILB martin_j_Page_080.jpg
e8151feb0c43f8f8ea545e2019330bf7
a2c7121b0fb7c5b817788d26686f397918634cd2
52593 F20101208_AAAJOE martin_j_Page_128.pro
235e34fa56d4ef0be40fd30d21882b0f
ae269c278d59fc8964e488eb297e32e1b67e0099
42441 F20101208_AAAJNQ martin_j_Page_106.pro
800e9858ffedb20af9d30df0bab7de65
3d8e625b1063181914c43bcd97132c9e0fdda942
1820 F20101208_AAAIKO martin_j_Page_032.txt
c8bbe3e74580f11292800210b556a359
c947b2657bcd0101773f3917d3dcafb6d175f6a2
55646 F20101208_AAAIJZ martin_j_Page_094.pro
f50f68f682ad3796177c5072db1e5261
eb220a9ec7d11326d93e3e20b8839cd4f334a3a3
F20101208_AAAILC martin_j_Page_132.tif
0c4e4d0100845aa27d99b4e9efdcdc8e
420b50a0d16a6cf0625c2a10c116b8d546db92f4
50824 F20101208_AAAJOF martin_j_Page_129.pro
2553bd25e757490205a91f9bd94a023d
af1d9ff637d4df28a02f796164ae855a2b45d407
101911 F20101208_AAAILD martin_j_Page_047.jpg
741de2767913e86c72da8fc0393f916c
083c8abe1f75cfe2e1d477696b958be44a1656df
51484 F20101208_AAAJOG martin_j_Page_131.pro
b1a2b1ac8fda1fdd2dbc4a4a538cd648
14d48d0e85d4e1f9afbbce3c594d229262b66b05
45551 F20101208_AAAJNR martin_j_Page_107.pro
51d7f17924a7d5266aff0a2bcdfe452c
35f08b74dfe4a9b9eb00eb256a2183159a084652
F20101208_AAAIKP martin_j_Page_151.tif
ccfe7ef9f87078bf1efa55e7ea4b2ceb
d2dd99498a4f8c5c04a664272f5deac0925d2a61
8896 F20101208_AAAILE martin_j_Page_035thm.jpg
c6e28fd05ba8d3ed657197665c90f555
4161453a8dd11f9b4c74b5b583c9d50e3406f97c
48801 F20101208_AAAJOH martin_j_Page_134.pro
f9c5f393e6d066693beaa33f867e5515
d926645cda25a9abd2f03b644f96067f067ae765
47331 F20101208_AAAJNS martin_j_Page_108.pro
4876c235a6bfd4156caea5e48382ec70
126771fd7c25b6adad28535148e52d21f41bd2ed
2235 F20101208_AAAIKQ martin_j_Page_147.txt
de81e94c36358596c9d3941ab0031b91
da09465dea35500ee61e3ad5b76dcabe1acb9809
659 F20101208_AAAILF martin_j_Page_138.txt
c9fec30b46404806842c3c18fe1117a3
6801db20264cb8221d2b58ca1bf5254cd4ad1fe4
40793 F20101208_AAAJOI martin_j_Page_137.pro
5a61514e531d53da7e2ba6f22db6df23
44f12d5c05f5e1183e685b7e326cb4fd12f521e4
51684 F20101208_AAAJNT martin_j_Page_109.pro
8cb4a88f7fdcba7f5e80f9913acb9809
1ef34122d1b5595b12abc5a750673834e515f3ab
9236 F20101208_AAAIKR martin_j_Page_050thm.jpg
322529d2ce01e226c54c058c26029036
6ce247ffece5dead98c8720b95441dad6aa3facb
34402 F20101208_AAAILG martin_j_Page_092.QC.jpg
6e4992b29fb160ea747290404b9118b4
efffb5f78cff4170f31b319e3c506b92c2ad6d85
27832 F20101208_AAAJOJ martin_j_Page_141.pro
c3242a5f432ca742accf9c07f4a0a1e4
45ceece7a04ebf96e5f712e3463d7e699ddeb072
52055 F20101208_AAAJNU martin_j_Page_110.pro
1e716c496d53fe7123b40ad99a246ecd
983bd124eb0f4ed9ec18d785e1e3735e31a10770
9021 F20101208_AAAIKS martin_j_Page_080thm.jpg
c380323ea70d253cf44b3b56a2a5604c
10d49c08fce1e5fed3a8d3c35457bb745f0aa401
38444 F20101208_AAAILH martin_j_Page_147.QC.jpg
f912ae2b10031be155f06c458ca77890
5be08f251364fba2bef9c4b580b1a7a78adfb386
23183 F20101208_AAAJOK martin_j_Page_144.pro
71987680c746642f7876a09939efb6f3
d90ba3d19631ae26bb11a60e7aa33fbe73a234ce
39348 F20101208_AAAJNV martin_j_Page_112.pro
a85f2a87df627126317ea43d46aa5025
1bd1ff9c0817387a67a01de1dcb9f748d75f074c
39883 F20101208_AAAIKT martin_j_Page_138.jpg
2308b4cfde5c23398027ff6693727815
0f9e52cf23cf74212996179906aeabb926d2e4e1
117642 F20101208_AAAILI martin_j_Page_035.jp2
938a8d344a4e74ee215a3c36cb3ee974
c376e84d3f4917a292c8b8afe0799281cf31e21c
52476 F20101208_AAAJOL martin_j_Page_146.pro
971283e9b5bf87941a97e4fb8a6896ab
1d21a810ac2bbcabdae3106914f6646a9eba9d73
47876 F20101208_AAAJNW martin_j_Page_114.pro
a352b8058df10c94fa8fb4bfa647eb34
f57120fc0726bdf956eb05e628d669657539d4da
93931 F20101208_AAAIKU martin_j_Page_006.pro
a5ba175b7eb9b3f8bfce4d22c5e3a496
b8c34d082c8d62189cfdfdd39828fca949d25145
98530 F20101208_AAAILJ martin_j_Page_084.jpg
33189e34c9655df916724b3f0bf5e082
8eda31a1c211bf0410b4f4086e8a8746f03b4e86
38869 F20101208_AAAJPA martin_j_Page_169.pro
e6ee5d8e16ea889f193a18caa0255c09
36cef1cd2a21d629c7810801db080d633888c95a
57115 F20101208_AAAJOM martin_j_Page_147.pro
4e3e5842fe065dc60fe8de36f67c22e4
27491b3dd767d4db570c1bac7b9d8e09c2f5d55b
46089 F20101208_AAAJNX martin_j_Page_115.pro
8a7e278347696a82d8700d7f49cba433
402384eff4b34d45262a80bd0b31471aec87bab3
F20101208_AAAIKV martin_j_Page_106.tif
aa565ac5eb5a7688e80cd5491cbe34d2
4b8fbff5708342c59f768107b9e2e4b08c0069d8
49642 F20101208_AAAILK martin_j_Page_130.pro
adf36797985e73abcd547f0d4f5a4081
8f400ef010b5863fbeeca07bc6b1a233458100f8
46212 F20101208_AAAJPB martin_j_Page_170.pro
da81d3a8a5ddd85b4f60bdd5bfef8078
b53ba6ee6b778e3cd9256f51592284d3d222872b
59241 F20101208_AAAJON martin_j_Page_148.pro
ad3eb326e1458a646e1f55223214bbad
5ece9c99aea0375e06600d98573041535ee995c9
49592 F20101208_AAAJNY martin_j_Page_120.pro
d6c0f65623149ead42d85262e780fe17
3fd9816b17aa65ec9f6555e556501b71ad4d1823
15717 F20101208_AAAIKW martin_j_Page_072.QC.jpg
92d1becc4e8053a02aec2bf58e43b15d
02ec28f5a8621c48512eb8a10982d5fff77026e0
F20101208_AAAILL martin_j_Page_022.txt
ff5809ff8e336291af074fd19c9e17c2
deb9aa336335ed265a7ac5bb797fe3d0f1c815f7
61573 F20101208_AAAJPC martin_j_Page_173.pro
8abc0ae59f09b089193d79fbe1535eb0
b1f47c8567f703e12bff99fee20c9cf3cc3aa8fa
55747 F20101208_AAAJOO martin_j_Page_149.pro
6e8f53aae23c26508b05747b8c8de41f
dbd1c98f40486f0888e40a0711694b1d5346aa5b
40806 F20101208_AAAJNZ martin_j_Page_122.pro
86f9f27000a053738dfc884f0b0777bd
829ce760da444e1fb8b34e0aac733db30d8f5d4d
48989 F20101208_AAAIKX martin_j_Page_152.pro
dbac5738737792d6bf1d7b60810f5a96
91b206671aae4fceb31c6f1183850b0fb6d3646b
2568 F20101208_AAAIMA martin_j_Page_177.txt
02a70d5cd8aab2322801bd5c180fd13d
aa75f930faa7d91d37474cfa59585a091fdf3a06
36961 F20101208_AAAILM martin_j_Page_094.QC.jpg
ea5f38407a5164d3031e758c02d45e5c
8d5ee48dd59490126fe2b28c51d7d3e167b67855
62552 F20101208_AAAJPD martin_j_Page_177.pro
358d5a418a56cae274e7773217f3be5b
faae01953fcc0fe403a8a1d75e1f0be1b6215c17
52420 F20101208_AAAJOP martin_j_Page_153.pro
32038956cb13089eb3fd478c1f7a89c0
a5523758da16b4c008461be5c88995afb35a1856
1781 F20101208_AAAIKY martin_j_Page_059.txt
5982411a5eabf60aee35f466c76bfe39
5d5d6b8608ef1142e3c208615c03ad03a065e7ea
107291 F20101208_AAAIMB martin_j_Page_152.jp2
812d7b75c82aac82a44e56dd9ddea0bd
608ec3d4e5031c597cad9b70973345180253df2b
8752 F20101208_AAAILN martin_j_Page_176thm.jpg
7f259b059063be89fbea9afdb656c263
2ecf6927dcb1024a2501510f8a48f98d2f4766e1
62561 F20101208_AAAJPE martin_j_Page_178.pro
f2005617d3ae719fe210d9c9c893e1fb
63495f636801a776d4055418a82ad2f213cd3c64
13949 F20101208_AAAJOQ martin_j_Page_154.pro
d1701d7d798ff8eaceb981dcce63954d
69fd305008f5e75563740fc831054ec4062e7d1e
99257 F20101208_AAAIKZ martin_j_Page_097.pro
6aa903be2f9281031fcf75f586a71067
7f3258a3b3292ac13bf8820eb4be0c24250e9cea
3082 F20101208_AAAIMC martin_j_Page_168thm.jpg
95ba12fc3315383ddf1abd0442bc0010
b6f086b93676bd095b203f46ce718c0ddb528584
2024 F20101208_AAAILO martin_j_Page_054.txt
3ac6ae5cd4506b0b355ecc2bbfff4592
1b091d74bac46e84d89853d6b31c421d3c7e52dd
66632 F20101208_AAAJPF martin_j_Page_179.pro
16b0397d0393749653666a921ca4926b
fcf12c1eaa6c56589258f5efac6f196a25ab9a84
65081 F20101208_AAAJOR martin_j_Page_157.pro
b09006d7c068bff78b5bd68f975c61ea
a3da4f01c4417488122e790b62b4d88934be32e7
2038 F20101208_AAAIMD martin_j_Page_082.txt
6f4f4483c5e99e3daf6446f5bc7335c2
9766addce7b43948369378ae0db2f266defe3699
108195 F20101208_AAAILP martin_j_Page_133.jpg
f54e25d94586f9e7f9f384060699ca31
4c3f2f97d29f4bf1dfb50f608c68af043bed539b
27456 F20101208_AAAJPG martin_j_Page_181.pro
05babe20c310e17689ef5b4d1fbd1d2c
3acb3af1eb442699cdf76c9b08e5c21afdc5b674
118397 F20101208_AAAIME martin_j_Page_094.jp2
a12a101484010b9169c1b9a8c30e3897
9522cef03c753bab2fc2653cd8cc28871434c6f1
418 F20101208_AAAJPH martin_j_Page_001.txt
97bee4e4ea9e9b2fed8c45ba86981388
d51a4e68c5852ed097f46cb34228b38489e279f6
42802 F20101208_AAAJOS martin_j_Page_158.pro
84be5bc6d14bba0dc8e4c15c3824d5ac
00cac0278af3b858dd0091292ee132187dfac8f8
F20101208_AAAIMF martin_j_Page_039.tif
efd9a3a0e218cae74de78fe96d5c35d1
394eb58807c276d0c0fe57cd71f3406b21d0cd5f
4768 F20101208_AAAILQ martin_j_Page_181thm.jpg
2836939d224da6db59eedd9f7858e17c
c7d8bfde519a53d2309621df2a14927154af500c
77 F20101208_AAAJPI martin_j_Page_002.txt
0653639f99f175352af51604dcc78915
d2f9a4ffbd1fb2e8a0d502da911f8e90f8690e94
40679 F20101208_AAAJOT martin_j_Page_162.pro
248071d5c9a830e6d816908c656e7105
0648369bb8a02634e1a55411a761f3d00e20e2ae
8003 F20101208_AAAIMG martin_j_Page_059thm.jpg
22bc672e097bb87e9c4eccc667af9a93
e0f4861a82f87b23e3fd0a4fbc95a8ca7acb11f7
137311 F20101208_AAAILR martin_j_Page_176.jp2
7c2be157b339446a0835d6d8fe2d0b1a
2220ddd9ac8738bf039c79dce225c53bbadd0621
1575 F20101208_AAAJPJ martin_j_Page_005.txt
78454b3faaa98d7d7d9a8e203782e9e4
6fccc1970809b7434aa229eb571fd2e4ded894da
52610 F20101208_AAAJOU martin_j_Page_163.pro
640e45887cf86089630d86fcb2ab6df5
5eb06a1a7fdb8e03b2f67c28abfc23c46ebda22c
F20101208_AAAIMH martin_j_Page_065.tif
4704fafef40030667378f1a7c0eaa1c4
c62fd1b681e72f6981383d28bf5eeb75061c684e
2061 F20101208_AAAILS martin_j_Page_128.txt
d2a87460a05060cbfeeef8229a26d8fd
7fdc098ee05ee3a0d404cb0681e7319743692105
3963 F20101208_AAAJPK martin_j_Page_006.txt
e15006932b49b73852ac2bce1a71b41a
c9f7a469707fdb267b65f7f82cd7150c2f0584e2
37638 F20101208_AAAJOV martin_j_Page_164.pro
b5b96da7ff9377d503f24a6ff3eb82a1
bd8e8246af3876dec994f916f47400c286fe08a7
113774 F20101208_AAAIMI martin_j_Page_128.jp2
a3ad0952f69c63ce98beb52825d387eb
753956bc2914fd4655a3e0286d1fa0ce2c43f0e0
41185 F20101208_AAAILT martin_j_Page_024.pro
edd558d20f6fabc39bb1979bf7ec5c06
8137330313670c6e25945b9528c8ce725ddb673b
4665 F20101208_AAAJPL martin_j_Page_007.txt
33bc2a9941f3bfc2fb21ea78ae446189
06da47e381581b7a88a9e7313ec0484b6ec0a8b2
43342 F20101208_AAAJOW martin_j_Page_165.pro
9ee3f06c8a48e4e4741e3c4132018dde
5193d7cf3ade4d1d7cea16246446335fd00ac489
F20101208_AAAIMJ martin_j_Page_038.tif
5e4cd8238c273eb472fb0719f218eaa5
b49d5c0c8e057976fd278aef230196312db20e62
8489 F20101208_AAAILU martin_j_Page_090thm.jpg
d68c93fb02cbe1e0a415c6694a571c84
32d385fbb0d2fdbd7589c7f1b46fb4944a90a477
2101 F20101208_AAAJQA martin_j_Page_033.txt
87844d7e11d6bc665d71f83a1daed88f
1db434f3c69af5b2bce8b242e912eb0f949732df
4498 F20101208_AAAJPM martin_j_Page_009.txt
4e413ace904f0235345ee45fc99159d8
68b98e015cdf66e170f3602a0a44477b13546644
19180 F20101208_AAAJOX martin_j_Page_166.pro
ebab2acbd69b90b6f87ee1d4388003d3
48c3322d8869f01f8f2c585bb56e17a49c5766ea
107873 F20101208_AAAIMK martin_j_Page_064.jp2
903e7b03e689cb8582c1ee531d385bef
45f0c4d0c4a777f03aab93cd5a494a06230d4975
28356 F20101208_AAAILV martin_j_Page_040.pro
7bac7257e46c5b48b9fdfc0fc2cc041b
ae866408e2653648d479c7a5ee351859c158baad
2064 F20101208_AAAJQB martin_j_Page_036.txt
1c33c179b8eaa4738f9099cbeff011b5
d4c799a564efcdfb97b052db6cc24e1e19fcc175
2758 F20101208_AAAJPN martin_j_Page_011.txt
59552841245a0c68d233b28178a38485
595fa548c94a8d7c4bb392010e1b63f482393ee1
31323 F20101208_AAAJOY martin_j_Page_167.pro
e1cb2af331c42c53bd87ff5a6ed64373
c252fd9077ba09b08272e47ade4e4938ccd4f056
F20101208_AAAIML martin_j_Page_108.tif
8b258ddbbc51c24f003ed0ee90e25a74
58b2e233dd6959d44edb51cfadb7e583fd945e7b
371 F20101208_AAAILW martin_j_Page_043.txt
47b524b3b8cad1077755420762bdbaff
0ade1686d0f9d27c4f617d81adf6b003c7979f93
1140 F20101208_AAAJQC martin_j_Page_038.txt
f378a9dc2c39893ba355b1e9edd0b0c8
b7b683437ce67b6cf1c0bd3b82986db2a6d9b7a8
885 F20101208_AAAJPO martin_j_Page_014.txt
fb9e15d90d5d2a6489dd3828291af2ff
3cf3eed7d96b0bf83db1d9909f07059a2d9772fe
6550 F20101208_AAAJOZ martin_j_Page_168.pro
441a9b870ca056a2d1a56c0b8c5293d4
1e662511e7098e7ffa4df7c0b3e48e720a0b8de4
9026 F20101208_AAAINA martin_j_Page_149thm.jpg
e381a5ae0afcb39a9dc5dfd7a0c0986f
5fe37e3d3f38f43bade800efb6d31cf5676c0c55
7463 F20101208_AAAIMM martin_j_Page_058thm.jpg
faf6f2168fb4a848299c33a8047be6aa
dfc5d4666c63dc9ac7d401e6636a038c91f8ebbc
1376 F20101208_AAAILX martin_j_Page_072.txt
20137e65372350e0f4778e34ef90d6b7
0af9a57789da454855e80b18288cbbdce8d8c453
1039 F20101208_AAAJQD martin_j_Page_039.txt
c043533b1d8c23fc9702976a4a503b3a
f95ef23719e391ce12303b5c0712fc20710afdd2
2267 F20101208_AAAJPP martin_j_Page_016.txt
803e4124d501a4fb8ee4e09419dd12fe
384a33d1623d16187105e56a5c574a1bcf705e6a
109468 F20101208_AAAINB martin_j_Page_104.jpg
9d5fa00e07fbe4f11402c6e05ebc3983
49ef263eca5a97674321f8fe108170001c5d521c
8846 F20101208_AAAIMN martin_j_Page_133thm.jpg
b22759a15e3e4e7b57428c757b91cfa4
962ddf46c16dbce44bdf314b89366ec05c34b239
F20101208_AAAILY martin_j_Page_024.tif
bfb7db79394c25d9871e4a7ecfc45247
ae74bb8c47a4f6f5f97b8e24c628cb3a3ee629b2
366 F20101208_AAAJQE martin_j_Page_041.txt
66b576f9496ffe223433b3b85b738b45
03913ca93b1519f0eb04093f98acb09b9da021cc
2047 F20101208_AAAJPQ martin_j_Page_017.txt
1e111ded6b9a8c3d8f62f5aab1057bdf
bf86ff5c74a6803af47c64889079d2d37cf60158
1938 F20101208_AAAINC martin_j_Page_152.txt
7182fa08dd5aac86ad5b9baa904d4d12
cc49e0b18110abc5ddafc9525c1801e8863ea92e
45232 F20101208_AAAIMO martin_j_Page_088.pro
7efc805ebd7a1285170294b69689497b
10b6ad82f11f13efde0d4b08117fc863c62c2f15
888 F20101208_AAAILZ martin_j_Page_159.txt
e8efa00e0b93bea47623d63b2b9d503f
595f1eaf0e9a67e19f3db5dbce31171547f53b35
1310 F20101208_AAAJQF martin_j_Page_042.txt
7abe9c0e2a11ad7c7a19954f8e4aae0d
9713508b023862d0f176192b318218fe04e9945f
2398 F20101208_AAAJPR martin_j_Page_018.txt
907f12607924c36e30118166374ed599
0ed90c986bd9930eff9f2858ce6914bf2fc544c7
87173 F20101208_AAAIND martin_j_Page_157.jp2
282485a7ea97366ee4252f1d57a51c63
7b7f4fc9ed0bd6ef3423a548a5a9f7eb1ef134cf
8830 F20101208_AAAIMP martin_j_Page_044thm.jpg
d758856838266c547da008e031acb368
4ec7a93f696d314a839759bfb1d9e46943d34c05
2080 F20101208_AAAJQG martin_j_Page_044.txt
9947fcb960168dc8c8c7d0fc38741abe
f38d250486710d23511c5b09851877bdae2fe405
2150 F20101208_AAAJPS martin_j_Page_020.txt
b1d0b5a175623606c8acc98fc78d8bee
31d3f3b4b27c0b40f085763ddda76b7cbe1c2e06
53566 F20101208_AAAINE martin_j_Page_133.pro
2e1bbdd900aeae4a1332d00f1a3ebf91
35b1e863e77e2ea64ddff6ac3facfe434e165d85
39989 F20101208_AAAIMQ martin_j_Page_027.pro
249a5616e7d91b6890d4d99833077120
ee85da916f491f6b7302d04549768f1a9750c482
1945 F20101208_AAAJQH martin_j_Page_047.txt
54db139290f5774b1ab653518812d093
8dd8322ac3eff57841a5fff65a423f0d545f12c8
F20101208_AAAINF martin_j_Page_166.tif
413f8324536f7d6ddd338faab95973f9
88a9e255a4afe77680dca1ecc6d1982382ddf366
2035 F20101208_AAAJQI martin_j_Page_048.txt
bbc0cf599bf92ebab00bda39bcb5dd96
4df0c4cda13e1f9022ba4c285f1a2e9c948e74a3
1909 F20101208_AAAJPT martin_j_Page_023.txt
bc6e86555f348166e478bee8c36aa3b9
c592e8b9b91028109d2247449ca63d2d60bc9959
29921 F20101208_AAAING martin_j_Page_107.QC.jpg
e79dbcce9fdde73e564eed4735d4833c
d4d7dd3d1c16fbe34b1fd94cc947d35b8b27f9b7
F20101208_AAAIMR martin_j_Page_025.tif
ab6b291c1aca676186bfd911da8be956
cdb3a9e44ff0492ae5d8b2949738b16ce00e8def
2100 F20101208_AAAJQJ martin_j_Page_050.txt
fa8095058ede42aba3644bc66ece71df
329179e61a1657ed13cda9e5c84e774b8fa6433e
1728 F20101208_AAAJPU martin_j_Page_024.txt
0a83e2c5d0ab2b56ae2805cc841b5618
85073870f4bb19e6bef6e503aea7bd395df6f4c4
F20101208_AAAIMS martin_j_Page_146.tif
6a4f4f4e989699ad53e9013016767024
859a5799f044159f9d70f120d8f7864ba68d76b3
51905 F20101208_AAAINH martin_j_Page_037.pro
2a441674da10565f5f859382740a53f3
1f1d07a17f308c24ce6848603891f7ecd8bf5a53
1731 F20101208_AAAJQK martin_j_Page_056.txt
512c46d23f5b33d3bc249257384ff4ec
1d195b64c3c18c34c99e2650017dc788d58ee17a
1848 F20101208_AAAJPV martin_j_Page_025.txt
2a124bafe4fbf6e0146dee8dde02dd1f
6fb02ae1e7e791e0c25cb5f7d5fd3b65ccc6283f
17910 F20101208_AAAIMT martin_j_Page_135.jpg
6762a844a925c45175922ed366d0bd93
9bd582c6d41db627ed79e49fb23bb2a2dc057020
47082 F20101208_AAAINI martin_j_Page_084.pro
20d6ab7b4c7fdf6b47dadf573b7cc4c3
1b40f545ce8aa0bfd48e785446a8db72acba5d64
1845 F20101208_AAAJQL martin_j_Page_057.txt
fb2377fe4030d9be241156f6a67ac894
469ba0538a02358b8ff407365d60ba77898f49e2
2087 F20101208_AAAJPW martin_j_Page_026.txt
de3bb661e7ecdab22b37962a13bca27a
ab559fcaff72d64d3e8b1b38344c7f571bd32b2e
34867 F20101208_AAAIMU martin_j_Page_030.QC.jpg
0f3702adc1fba83bb376a51a2d669272
cb7369a9bf07ee7f5c612fb17736276c21ba4b9c
34420 F20101208_AAAINJ martin_j_Page_110.QC.jpg
a857acb38a243b8d24ca0b06de1a906d
501fa765a80408d083f1f9e5cd69a727ea78f2d7
2133 F20101208_AAAJRA martin_j_Page_080.txt
dd4fef2fdf724c3ea186b0209a70db48
48dc6a9ab8c1f155ed82b2b735e9e165e135481f
1680 F20101208_AAAJQM martin_j_Page_058.txt
c81828d00b02c094fa7a82cda5e9fd4f
88027df6fcceacae1de5b320dcac8bc863a0ce55
1684 F20101208_AAAJPX martin_j_Page_027.txt
dfa59e7bafc12b5797f8b3f81bf387c5
5da1e687924328502515047baa75c90ce0e221de
35351 F20101208_AAAIMV martin_j_Page_022.QC.jpg
359bc5eba771a872889f129c2207d738
46082d44df87a6dd5612f75cdfad03e6ce88daf9
54955 F20101208_AAAINK martin_j_Page_162.jp2
87ac06674a63f6ce8ca0e19e3619d005
02dd89bba90d22054a0ce89cbfad2a65103a30b8
2134 F20101208_AAAJRB martin_j_Page_081.txt
2b8b1ce9dffb8f4cd59612765fcea7ab
76056f02a0536b370b190f798d8d61f7a7b6b6c6
2131 F20101208_AAAJQN martin_j_Page_060.txt
9abe057c9310f2ddadb11a2319e91185
993eaf9fcd0cfb5dc1023d9cf8509a3216e2f78d
1993 F20101208_AAAJPY martin_j_Page_030.txt
aad5edd624717b90cddbae5f8ce7f3e0
76494b4e4c774a2bbfae3826deb6127e48cd7a19
9139 F20101208_AAAIMW martin_j_Page_068thm.jpg
9c11340a046edf05d60c34e26747463f
3f0d8037bd8fcc19c0b0e972612764a574e139ce
42348 F20101208_AAAINL martin_j_Page_058.pro
472f1fc9cb6af938cfb787f6c600d60c
4709525463f9dff363986e967b86c570ac6ad0d0
2173 F20101208_AAAJRC martin_j_Page_083.txt
b8518c0118b62fbf7ed1d51056c05f01
fc226d7c00dc7851b1483724b23502a50ac57d41
2159 F20101208_AAAJQO martin_j_Page_063.txt
4b074d9d537bad0795f2b2bb97672b69
b475c0f246a4e7d7ac743d4af8c7a579b1b88443
2010 F20101208_AAAJPZ martin_j_Page_031.txt
5a2ac4026b7fe002f13d4ccee62f1410
2d9244b8c661d5792869f65b3a64180ce384f244
8164 F20101208_AAAIMX martin_j_Page_105thm.jpg
07d3eb9697254d501e746c6808a492e2
e61dcd055a33cfff675333b4bd7121201e32a35b
4357 F20101208_AAAIOA martin_j_Page_167thm.jpg
1e9259e993cd2b20ec081d6b8ade515e
47c01c61f26d25f1c6d1a663f25128cb548a135c
2028 F20101208_AAAINM martin_j_Page_013.txt
376bc821c718004387768a6febdffb2e
abfbf13024520d3289793dbc4cb448ab6810735a
1911 F20101208_AAAJRD martin_j_Page_084.txt
a98b9615b78c6f1368ff76fba5394882
42eef2f74defbabddac07f754f76826d6af4fe8a
1975 F20101208_AAAJQP martin_j_Page_064.txt
948e53006a49d7ef3be2869c0fb2a8ec
a1b44f4e4962feaa47b84f12afd413dc3461fa99
10959 F20101208_AAAIMY martin_j_Page_138.pro
bb291d449030d26120201fcc3b5ef10e
314607944426ac1e8aca448c733b9d15ea440f29
2057 F20101208_AAAIOB martin_j_Page_046.txt
1293f2bc42cf1ad83fdb1f38633417e9
1005351fc49e1d21ca6c0f69dad49599901fc03d
127265 F20101208_AAAINN martin_j_Page_173.jp2
6e19bcb6693ef4ad696b4a842d5cef68
f9a337fd003f7b35c56ac9c9cc4026a531aa49ee
2127 F20101208_AAAJRE martin_j_Page_085.txt
c73a0c7e1ae5cd30a897ef35293458c8
74307a2bc3de0eaa80452ffd8cd00289a2d2f101
2288 F20101208_AAAJQQ martin_j_Page_065.txt
03526f911ed8645177683b9cf4be468e
4716cdca71fdd90c0008fc701478e1909e926405
35346 F20101208_AAAIMZ martin_j_Page_044.QC.jpg
4c7b9282f79a360a570dea8aa64dd213
2102cafa9664aad5c713d2f7d724a170abc6b87c
F20101208_AAAIOC martin_j_Page_090.tif
507c4bf97904553bdb0af30682d45a60
4257acf7129abae58dd63f15e0d479d5ab819ac0
108261 F20101208_AAAINO martin_j_Page_087.jpg
bbeaf9ac5e450ac70643cfb9ff5616a5
f4aa3f63ae882c44bfb508488dba81fd500bbe06
1925 F20101208_AAAJRF martin_j_Page_089.txt
dc69c99f6109973205c22592afa11839
da3f714110b2c89578eae6759c511bdedce36074
2239 F20101208_AAAJQR martin_j_Page_066.txt
447e7fbb39f55f43e7538e8f5d9cd9da
da7bd1cf528f854a1abef5a1dd549bf8d11e4a36
8248 F20101208_AAAIOD martin_j_Page_134thm.jpg
ab2d1c4a87ee08ba4ba52dbe0ba9a59a
717f3ce54f4e1c289e521a783a318fded0dacb6d
8795 F20101208_AAAINP martin_j_Page_009thm.jpg
14f4e381d10076155af93206cf37be37
081db4a2faf41d7f476f9c809f9d15929170b684
1818 F20101208_AAAJRG martin_j_Page_091.txt
bc4f4b29bea85ef39afa12b866cc26f0
a2881f8f8eb9b0eff38d9bdb4132f30d1e6ccc8e
2126 F20101208_AAAJQS martin_j_Page_067.txt
fcd02ebb0ed15e80ccfe4b5c2120daab
b59942447ec6d1774ee167d61d4399f183ee1967
16666 F20101208_AAAIOE martin_j_Page_140.pro
ca00cd04ea0432e88a8a6da304f00ac5
a4f204e945b5217b86ece2d5ec0787bb373bc53c
112872 F20101208_AAAINQ martin_j_Page_092.jp2
894c9c84a7ac9d034325bff8151d0858
4201bef522cb7460e8b78d2f728aed69ee3c4a53
2076 F20101208_AAAJRH martin_j_Page_092.txt
0047e3a9701531fbccd59ba6be75445b
7baf14bba33f30c6fe5913021b020738cec7c3a5
2271 F20101208_AAAJQT martin_j_Page_068.txt
bb6b7208d5ef4dff0ebe6b871eb58630
7700acb0d770f5f9fb7d7a01577a93637bd33e28
352335 F20101208_AAAIOF martin_j_Page_140.jp2
3e6a86ec128e97c2e2d4b54d79c3af69
2655ff85d5621a4a0a3c8a3cc4315d5770fd6a5a
81921 F20101208_AAAINR martin_j_Page_024.jpg
7c4f8075a76a0c6158ee8022e053a8f4
65e4243d9fe67a2eeb5cea566d8452401a777e6e
2228 F20101208_AAAJRI martin_j_Page_093.txt
a80db1e936fc901145f98b07e4d7464c
ac158ad96b5cdc5876453d5ce484a9c19d0c2e33
49070 F20101208_AAAIOG martin_j_Page_144.jp2
1e5982d6e653e25c9ff38bae9d8e34e7
bcdb8e58de84f4a45b788d6be9b9d779cfa11868
2186 F20101208_AAAJRJ martin_j_Page_094.txt
95a60298edfaa6b291c10233701f3a2b
2e8c61c15290d96318970a4a45943d74db86406d
267 F20101208_AAAJQU martin_j_Page_069.txt
2842be9e95bb06abf374bd166eb4c44a
0d40320f38bf9fe40ab1424c1f244898a75e3901
2526 F20101208_AAAIOH martin_j_Page_172.txt
a4519816354ab8fcdaddc8df6f6d735b
8d3ec787ba46b77fde9cf07e8aa1a7d2d56516e2
44759 F20101208_AAAINS martin_j_Page_156.jp2
97e3442e10ed1654122bed72b01e41c7
f5321ffba63615ee0935a6a728172ce68c43ba91
F20101208_AAAJRK martin_j_Page_095.txt
d47a0947bde2804e44cee2e9019cf92d
daf046ecc81e3193b41d1118ff39e3390c83362a
1281 F20101208_AAAJQV martin_j_Page_073.txt
b81a723621fecfbfa1d5be16ad2cb162
cc8bb1aa3e31eddade403d29216a631760c4a809
1960 F20101208_AAAIOI martin_j_Page_077.txt
343ddaca3ab9a50b76b3656f56c3602b
b2f0c914b5ffd81e59edf2e0d6dc475fcddf5699
114517 F20101208_AAAINT martin_j_Page_044.jp2
96def4c3f17610c2a2765df29c6fc8d4
c896378f12d05f6c2820b2965d2524114e849188
493 F20101208_AAAJRL martin_j_Page_096.txt
1965a94f6a5a718d9769f0e0a459fd94
97a89971d99421660a65dad533aaf6de9053236a
F20101208_AAAJQW martin_j_Page_074.txt
d01ddbc841fc9852ba03489a64eb6104
86ce101e388022cbab2854b4539f1e549db4d6bf
F20101208_AAAIOJ martin_j_Page_156.tif
2a4a9452c7c95ccb7de85aa60bbe0894
6acde46710ab5732932624e4f7389e2a5e300e07
100662 F20101208_AAAINU martin_j_Page_032.jp2
e26cb66c458e57f505564f9e2ea8243f
adfd11939be80f0ba02361c78d242994b7e76b8b
1939 F20101208_AAAJSA martin_j_Page_120.txt
b22efba36d806ab6a81a70f95c6edd44
58d2b787cca566facf58d2b459e6698f646df02d
4366 F20101208_AAAJRM martin_j_Page_097.txt
a6ab8fb40ff31fd785ca23f49e913288
811396faf7ddc0017af4417134554eba5bf5cf8a
2205 F20101208_AAAJQX martin_j_Page_075.txt
c8dab7a2b85d154b3e6fc62c4f265e89
28ac7435be54f5ad6a4eaa16d26ea63123869dd5
35848 F20101208_AAAIOK martin_j_Page_021.QC.jpg
d29a2d2c7cf4b5e6fede7f7a667393a1
ca790d13cc84c32b3887abd98bd520be3b42f955
40218 F20101208_AAAINV martin_j_Page_121.pro
4bd2bbc479e3699bc126b680c1ddd640
de0a6a0614517df0f9d5a7075d07e9425aebc738
1588 F20101208_AAAJSB martin_j_Page_121.txt
a0376acdaf62cdda4d8446cc771869e2
4263b330e718447e05474a04f87a06c8067dae2e
361 F20101208_AAAJRN martin_j_Page_098.txt
12d7191a4616fe58e217bf33aa29b475
8984d42d73859b439ac3344a9de4caef784217ce
2143 F20101208_AAAJQY martin_j_Page_076.txt
d443ab340434f17deb4db3c4a0573083
ca7d089fbb96896ff139465440634106764e9d44
9553 F20101208_AAAIOL martin_j_Page_043.QC.jpg
afa9172b27f749de0336cb2d313a8fe9
da2527523fd8f8c1f62426ef818adf544e99fffc
2050 F20101208_AAAINW martin_j_Page_053.txt
39fca5934813d0c3f28a135be01a8f45
49ba2a5978a6b319b995de4d67c7cad985e63d73
1632 F20101208_AAAJSC martin_j_Page_122.txt
dc26391e65a1ef623eccdc1ab6361cd2
e53eed3035bb3cfc1eb1c465aa1c7a898ceed3a6
630 F20101208_AAAJRO martin_j_Page_100.txt
697ed7f776c7e4f266ef06739a5256d3
3a53f54e3ab22712eb8b5536d5e2efbcc4fbbb61
F20101208_AAAJQZ martin_j_Page_078.txt
8c8f6ace29cc32ff9d20f1c78646797c
76aca5da25be786ad57c12f8cf3d14173a8254ff
44541 F20101208_AAAIPA martin_j_Page_032.pro
f8bef3a03a82b7a14d71f67c200263da
8c41918757409f52c39136ca2732bfd9f8213b4d
110794 F20101208_AAAIOM martin_j_Page_129.jp2
1c889523f58a5ec4564e8d048d2feddb
333c5176388220ca0f0d83c3788b592180b46573
6820 F20101208_AAAINX martin_j_Page_005thm.jpg
0d424331a7f9a11be23936cf96dc1f06
574986b28d48724fe5c9133532b4121d5a7a796b
F20101208_AAAJSD martin_j_Page_123.txt
dd80ffd7c2a69cd364e5554362740094
e185cb4ef6d58d9543a2daddf0cb07f60685e0e4
2005 F20101208_AAAJRP martin_j_Page_103.txt
af61f36864174e5b7ab8ea9ac70ca5cc
b716098d6aae330f61500df343f52c85b639a2b5
9203 F20101208_AAAIPB martin_j_Page_066thm.jpg
59c854c38fd24f6df5e4e0bee7c962b4
8b09fd3f1eb9d08b647050e8bd04ccf302e31b3e
52721 F20101208_AAAION martin_j_Page_092.pro
41040c5cb40048aa9c853e24c68effa4
45dd23545e3bc4b56eb9b2c7c5eda423e1e8a4b0
F20101208_AAAINY martin_j_Page_063.tif
647728bac07e5346a27d39c3f5da2511
eaa37d19558f2f80b9036d7aef88d2aff4a76883
F20101208_AAAJSE martin_j_Page_127.txt
3d1196f5ae05ff851babaf4bf7725a84
ca2131ac1b81d2811b40d8dd17f9f115056b8721
2104 F20101208_AAAJRQ martin_j_Page_104.txt
8db047de15973aa6054698285f284b0b
79891196c94abb3582848751f8db1d464e3e60ca
81330 F20101208_AAAIPC martin_j_Page_157.jpg
6adf1107bd57ea3048666a526cd808c1
bb2d5151a673a418ec58f562c5bfe98d1c930d44
8813 F20101208_AAAIOO martin_j_Page_022thm.jpg
e4dc1aaabf2b52caa3893f64fdf16859
7f54a3122a8a09d0a77be6d46579d105818cf320
1965 F20101208_AAAINZ martin_j_Page_049.txt
7daeacdb66bcd83a03f53e4afe9aca8b
bb448b195e001a8ee04aa638f710c59e004b990a
2020 F20101208_AAAJSF martin_j_Page_131.txt
0f302dde8f8621f2b3f9cc9b5457b760
aefaf1e14c4c89f4d1735027b7be76ffe1d0cb52
1947 F20101208_AAAJRR martin_j_Page_105.txt
ba254a9b1417513c571a2c4757ab48f7
0b2934bca2548255dd339d6af0cc502516f2f5c9
1577 F20101208_AAAIPD martin_j_Page_069thm.jpg
844b571a02235c31e44366253367c2e4
b02397e7290da8df71b20b9848fba3512d56f7bc
8811 F20101208_AAAIOP martin_j_Page_178thm.jpg
412d3f54c249b66100ddda3d7763f6c2
6a1fb8c9c6efcbf607ab707f1830c5c24a7e194c
2088 F20101208_AAAJSG martin_j_Page_132.txt
d4b94933e51006c712fb08228e5fa9b5
75efffc31dcb6bb4830d03327d39f4a535f2b692
1859 F20101208_AAAJRS martin_j_Page_107.txt
82406a5447c71af49c34f93a84ee89a1
a4f770888006b04fbd123509c01366df380424ae
63846 F20101208_AAAIPE martin_j_Page_038.jp2
c0675f4bb5f717c58030c1005df0f644
2242b9e71e31d5bb082b38e6829952aaa8ca4f13
35245 F20101208_AAAIOQ martin_j_Page_053.QC.jpg
47612c6a05c7318dd0ed6bea417eb8e0
dd200bccf3a1cd287b8f718068d4076bdb113cfa
2094 F20101208_AAAJSH martin_j_Page_133.txt
3c7c07db23ba9640925fbc2914ffdbf3
0b0467dd82de50f6fcf0db77a5a950474a1c398f
F20101208_AAAJRT martin_j_Page_109.txt
75066709d06f7b7d73446df5c646de98
8b5b7692ae71dad8705f76cb0034ed76e8ac5e32
F20101208_AAAIPF martin_j_Page_153.tif
bdcba957aa28a92a279bb59d0721fad8
2181838cc5d126b73b8f4e525441e014989cb02f
57245 F20101208_AAAIOR martin_j_Page_066.pro
2e96fc249f35901102bd86b0fc86ccbf
6589ba9296a1e240c73bf47821b94ff42f9598ce
297 F20101208_AAAJSI martin_j_Page_135.txt
f80e0053a82a01d2c8cda9ee025e8615
b4e79a76e109a570184ddda1008b70dee6df49b6
2040 F20101208_AAAJRU martin_j_Page_110.txt
a3541c19c366211d83145c38bfc81701
0ba8e0696d775cff8db7230089163c65f6ef0b99
28251 F20101208_AAAIPG martin_j_Page_058.QC.jpg
dd344615b2da6fd64582218e2a37e301
fbb08d1be63511f795933022890dbc90faa1529c
F20101208_AAAIOS martin_j_Page_007.tif
42b67094a9424648c7b32ba27386b6e4
420e330ce81eb8e847835a470583063a4d838993
2060 F20101208_AAAJSJ martin_j_Page_137.txt
c81cff71f632effb9fc1596bbcea6f44
047e3066a7c724185fb1f613c727e0fd5f6f7597
50570 F20101208_AAAIPH martin_j_Page_145.pro
771021be3d66ae0255fe1a9a0eaf1be1
c9a10f431f0789b3cab2edcbcfebbc390533339d
891 F20101208_AAAJSK martin_j_Page_139.txt
1a5d5fe89c0ec98f0b1a2b8c736081ff
92277abfd312088abfef1c2d5f9c6440103d65fd
1707 F20101208_AAAJRV martin_j_Page_111.txt
0943037235ffaba7df3c951f6c552b2a
6cf61e843bd811485fece9aa078949c2b338b678
F20101208_AAAIPI martin_j_Page_003.txt
fb3ac3b5e6225b29fdfc15a175cf7a36
c4d53bbd167913b257c839650964058fc9e36d97
39640 F20101208_AAAIOT martin_j_Page_111.pro
6d440692a6cc2e3aaa4f8469641576a2
0112abb474a5b649f1732dc985e06dfe0f1cbdc1
1237 F20101208_AAAJSL martin_j_Page_142.txt
276fa6e22a65af6a7b3a28d71cc51d02
812e41a5ee02a54ad05d197694d514b218f8dec9
F20101208_AAAJRW martin_j_Page_112.txt
c82f2958c814583bca61d1750c47dc39
2dcb4b2ff8f35a9071f7f8cd69812a2f263da29f
120232 F20101208_AAAIPJ martin_j_Page_051.jp2
9d6e010f1d0fc853982302bf6509530a
c9973d458497bcbde115709cca8f8830059bc195
2531 F20101208_AAAIOU martin_j_Page_173.txt
01cea14f59f301cacb21741904729d7f
49023ba6af27a56d84b9a817cc57aa54cbcbd46c
2578 F20101208_AAAJTA martin_j_Page_178.txt
497baedfa44e2913d19d2ad6bb6c7c8f
174ee406e7e583897ee34f96cca5e89518c02f5a
1265 F20101208_AAAJSM martin_j_Page_144.txt
67ddaee84ecfde1aeb960be1ad637b4b
0aaaf4c7f0eb259d6568e14c57d289cf28e05294
1895 F20101208_AAAJRX martin_j_Page_114.txt
30a0b9a04f6028106ec89ce03b7ed474
00e3723d4508bb8358fb646db101cd0d32bb99c0
35000 F20101208_AAAIPK martin_j_Page_146.QC.jpg
e3ab129cc82ba289519afc4a5a8c227f
d8d3a9fb6035595b8d4ca4015323752db1ac56e6
109308 F20101208_AAAIOV martin_j_Page_034.jpg
1481a6eff526b64122b449f294d09015
ba65d780dbde7484efc50f40b443257d612e3ea9
2736 F20101208_AAAJTB martin_j_Page_179.txt
11e52501282c46ad813eb46eaa1ab555
4b388d5664cd52d52f8b3370e7fbeb479486fb1c
2066 F20101208_AAAJSN martin_j_Page_146.txt
93f9060f93162bf54c5b37e4c8c4e152
eaf9e4c2ddf5b55cdfdb4b6c19d3fe43c2e08a62
F20101208_AAAJRY martin_j_Page_115.txt
83199fe17e0bed03273b0a6bc7813553
3211ef2a43ce8a3baf71fa743d1c4b6f3e48e939
36745 F20101208_AAAIPL martin_j_Page_081.QC.jpg
e69348a20b8558d4cf07aa1370b6a7fe
aeb2757f8d4f2d1b1c73fb50c017dbb2787a8ed9
8443 F20101208_AAAIOW martin_j_Page_048thm.jpg
a5e68c3dba6e1e9c6bd65425d287aceb
a3f07a05a3a4408ea7bf0d3aef9b7674f37a1a9f
1966 F20101208_AAAJTC martin_j_Page_180.txt
0eec358002996380cffe5222790b37d9
97c29c86875bc019a201827ce8ce6a9c01d5af83
2311 F20101208_AAAJSO martin_j_Page_148.txt
c47aa5252e89fe8705507a366a12cb57
e1d313a6a15f75a6ef89c7b2984b0540fc684311
1639 F20101208_AAAJRZ martin_j_Page_119.txt
a22fbf780452c599e3db8207fa0ce633
6ba69821441f7d7dc52c455f1712f012df5af15c
52237 F20101208_AAAIPM martin_j_Page_036.pro
0281ee9f5afd89def19d68294e63a958
08229f574ac78a6995e4ba8fa01e91d8e3c665f1
133546 F20101208_AAAIOX martin_j_Page_012.jpg
ef81afed4cbfe0be750a8f8ec643c945
50a04a4648fc3b60080224a54cbb08f134450b84
547 F20101208_AAAIQA martin_j_Page_003.pro
263f482a5fd23ca630517f66a65040af
e1b29114041a547bde8853e83ec57d74c0409903
1128 F20101208_AAAJTD martin_j_Page_181.txt
92f8f871043be86ccac97c735cae98d3
2ffc9d8a55dc9f0a9c11228d713cd613e3af74fc
2183 F20101208_AAAJSP martin_j_Page_149.txt
3dd1ac9ffc57c85042e403f6c22c5ba7
eeddf5a3c73a33fa3e40dc18740af6807d107386
8997 F20101208_AAAIPN martin_j_Page_085thm.jpg
8146dc763ccd0e3179c3715f00de31ee
013e310d62b5681014774fd8c7a260a0caa41a45
47172 F20101208_AAAIOY martin_j_Page_155.pro
35f5227fae5b2a7a428d1ecc4cd28616
69d557f30b6dcaa1d3c50d3aaef74cd58f9d5ba8
64336 F20101208_AAAIQB martin_j_Page_070.jp2
5053273ddecfb5498368c418687862b5
26ee41aabfb8e7922c6293086f16568ae2a28c02
2034 F20101208_AAAJTE martin_j_Page_001thm.jpg
4c0beeca77ab94302d50ff2def6f84d8
9dc8683de25a2969a4b84458bdc482592e058939
1985 F20101208_AAAJSQ martin_j_Page_151.txt
bb602bbc8bd895e08c98df14808f054f
176cf3465ff354bb6c6e8eb58ef78225e073275e
72486 F20101208_AAAIPO martin_j_Page_117.pro
7d719ab43d9fbbc6ed5f2e47a1891e96
af976bbcf3b8a3e9851f62ae43b9691c6f0ec929
50725 F20101208_AAAIOZ martin_j_Page_031.pro
5c72ef70e335b2fd4337dddd9572fbba
6e2234d021723ef6ed806070cef02852887906a0
F20101208_AAAIQC martin_j_Page_048.tif
a7ddee6803de0f2475dec06166d6d269
c38ee9f0731f010b1df9bcc5592430a33e25e1a1
1261633 F20101208_AAAJTF martin_j.pdf
b8133ed3ff1f57886c9912a99b498aa0
07fe83493b7366c7b2c05f93af9fdcae766f8249
563 F20101208_AAAJSR martin_j_Page_154.txt
52f2949cbd2c1f6ca12869a714203712
13a63cb05633305abe0289c532087a8f49411f36
14248 F20101208_AAAIPP martin_j_Page_166.QC.jpg
d8870617d81f956c2cd6e3e00adc1b8e
7537a7bb2e5c30942988b814eb6f61551bd9f102
110401 F20101208_AAAIQD martin_j_Page_021.jpg
76c4b8bd577e3236241e4c6b1a39d3b1
d125be41aaec2a15f000023f60d1280911fc1c0d
269726 F20101208_AAAJTG UFE0019653_00001.xml
c5949600672009b6bd0a8cc4823be552
528f71672fd04882e62753d11da19208c10d9efd
784 F20101208_AAAJSS martin_j_Page_156.txt
c82cba3405644f5c37bfa15ff4863453
c6392991a5448e5ca752893dd84ed452d16f343d
14154 F20101208_AAAIPQ martin_j_Page_144.QC.jpg
e08fb66e320eb7ab71daaa408d4ab45f
2c87feef38d3e2de913b1c7c728f3fff3100c2c9
993 F20101208_AAAIQE martin_j_Page_010.txt
83e5f0c88cc05557e4b14000b8aef17a
61d069caf3f3020806490e7a2010d4f3687afaec
7591 F20101208_AAAJTH martin_j_Page_001.QC.jpg
3df872422c99f28b00f319baf780b828
6654933344202461bbe7e4c3ba3278b12eb36707
3569 F20101208_AAAJST martin_j_Page_157.txt
8c59c04cbd9946cc92087078ecb79ec0
3257c96e4f72e9aacfe6cd3f89ef6696dc8a6f85
7481 F20101208_AAAIPR martin_j_Page_106thm.jpg
1d1b420b244dbc29b8afd13d24c69d37
2fbfe44388dc1b1a8116d81a758570a1daddf260
47114 F20101208_AAAIQF martin_j_Page_180.pro
4a01091cf42574ecbf32dbe090cac07a
6e4f054d4c0c267ea256deaf73209627c7fbdfec
486 F20101208_AAAJTI martin_j_Page_002thm.jpg
3ea551c6c277343d8a0d859a1d779d37
56e7d08107c2a1ab2d0be2513f54f50a821bb2ee
2362 F20101208_AAAJSU martin_j_Page_163.txt
c7cd39286dd40ce6fc1ac8a56187a0a9
e4ef162dbcc7355f95b9e487b19e496b9fd8f32b
F20101208_AAAIPS martin_j_Page_133.tif
e783bb1667682c32024e7ea0bdb9ca45
3dbfdb8701befe9c533955599d4c2afaca6e4a00
35941 F20101208_AAAIQG martin_j_Page_150.QC.jpg
ad884948eb38e1898de21f1e1acef704
682ce6b9d0a5daae3a688c198870586cd5927ad2
1166 F20101208_AAAJTJ martin_j_Page_002.QC.jpg
1c08b933a129150d5f68d22bfd87e219
0ee3047f6be3303158ab49dfe29d3b4fc5aa74ba
1461 F20101208_AAAJSV martin_j_Page_167.txt
77d63b62c29ba554ef68413a19ef72cc
54d7e190fcd1719bbdbb8e0375be370ee57e2de6
F20101208_AAAIPT martin_j_Page_126.tif
837e8a1964681149d2dd84453e497817
b5a5684dc531f2b5e775d4b2ab83b1624536776e
58192 F20101208_AAAIQH martin_j_Page_038.jpg
c3c8b39f5f29047fa2e42eba0e8edf71
c77b0cf5016345331dc5590a537b1e88cf7e11b7
448 F20101208_AAAJTK martin_j_Page_003thm.jpg
c4e07de5814be0b22795df7c65b09f04
8ce1b618a8688ca42eff4f016893fa0f8e6c0a52
7860 F20101208_AAAIQI martin_j_Page_013thm.jpg
ed775d08a1d7165b4873b1a725f6b70b
40ed1279d8dc6875673553a113cf0a6610617e78
8666 F20101208_AAAJTL martin_j_Page_004thm.jpg
2dd9b2c132a9f4af5620767bf72b46a4
28e510afbf310f47ac403e0904dc6d791f0293c3
F20101208_AAAJSW martin_j_Page_169.txt
1c531b73d012fc7ea22e938b6b24228c
231c2495de115f2e7654ac3eed02f6b0643ae691
52410 F20101208_AAAIPU martin_j_Page_046.pro
8122affffca819a10cddd9cf15559d9c
f45c47bd52e5aad6456d0d9c3d3afdc5957142c6
29054 F20101208_AAAIQJ martin_j_Page_056.QC.jpg
3db8a2f571730f361082302478762584
668ea00aa42ad67b53470ccb74a1e39c2355b107
8372 F20101208_AAAJUA martin_j_Page_015thm.jpg
32c307025538661270575cd76b06c22f
9c3aef568711fa5d29a3a741195f79879269badf
32892 F20101208_AAAJTM martin_j_Page_004.QC.jpg
760321aad6976e8c76f58208237a3bb6
d935255870978e6ea9b1d96cee5b3edd586f0f0e
1899 F20101208_AAAJSX martin_j_Page_170.txt
6ad63989caaa98be53054e154e05f61b
f0b9807da85ecada596cdbf0f69336b175c51b05
1888 F20101208_AAAIPV martin_j_Page_165.txt
643e1c7785b41ba1e71612d8dc19a65f
c7a3b1684d0ff6d36bbdc1bfd1b62f2170205bda
41322 F20101208_AAAIQK martin_j_Page_140.jpg
0f542c11e3228572be0d052c123578f6
c31890bb3bc3aa60c94d8bb1665a76203d5ff781
9159 F20101208_AAAJUB martin_j_Page_016thm.jpg
7161fb75999d8a60b0eabb927440d10b
8ba128a3b062e064355e58f6b66ba8c8838a9fb5
27245 F20101208_AAAJTN martin_j_Page_005.QC.jpg
d77a67fa64cf0c592089ae9a19fa7cc2
6955be5132a928f5aab428752fe381828a80f4b3
2536 F20101208_AAAJSY martin_j_Page_174.txt
cf2c5fe8223fa20d203895804cceea61
7a1bead70faed6532448b1cb15fcc73ee18e4693
F20101208_AAAIPW martin_j_Page_114.tif
85ca7983c949dc43253fd377c5fed95b
8ddbfd62a4263c5a5bb3e867a88ad29e22a79d0e
F20101208_AAAIQL martin_j_Page_168.tif
722005e0a4c05de90bce37e3e87877fb
9cd81a3cd250443b99e08cb9d1d9e6b395a7d0ff
38628 F20101208_AAAJUC martin_j_Page_016.QC.jpg
91e3e141132a45223accdb5930a95efe
f041c189c8ef8edc67e15c8e421c1a7fa6048626
7152 F20101208_AAAJTO martin_j_Page_006thm.jpg
a63ca3453b5f87e417a9ebb260ef729e
05cf7878a45a2e4d163fb2336efbbcdb9dac3ea9
2730 F20101208_AAAJSZ martin_j_Page_176.txt
c02999a9de5315c2b06d29911e2edc62
c121bd59a9f73f4a5c383683a2d310288355c324
93156 F20101208_AAAIPX martin_j_Page_115.jpg
aa55f90f50eda120ce7e2aff8476aaa6
224b5560d2af2c6ee78e9a2c254d90830343c2c4
111793 F20101208_AAAIRA martin_j_Page_062.jp2
62f65ca1177ee5b6e5ab221b3ea5350b
b8e3c5831a79a6ba3ddd367ee1e9377e17df62c0
34606 F20101208_AAAIQM martin_j_Page_173.QC.jpg
54bc5f3637f7874c025e8ed1fe346bc5
ed82890a0c151b10056debc050da13dfbde61928
8396 F20101208_AAAJUD martin_j_Page_017thm.jpg
6a5c51c6842c2164d12593e05a12a203
b2a09ec586a93d209c5eac2b562231bcc9ea7809
30594 F20101208_AAAJTP martin_j_Page_006.QC.jpg
1aad0b13227a8faeff1d31296f45becf
03169ae664179e2a4961f76e24d161c9867a1686
97864 F20101208_AAAIPY martin_j_Page_013.jpg
acce8bcdcc85472e76c65c716d268748
66e14b0b171949972afd1345eedd2086fe1b603d
105116 F20101208_AAAIRB martin_j_Page_114.jp2
8b9b625e8fe03f7ec6f3e589a4063082
565735b7e48f61e9e0df3df6be55d786305b3417
122755 F20101208_AAAIQN martin_j_Page_174.jpg
88f0bfb1dd00fe4dda86f916383c220d
c5a00bf45e95c0dd2142c72d41534f951df84f35
34848 F20101208_AAAJUE martin_j_Page_017.QC.jpg
168b03c8ec841ab255e428d7006ff1bc
66f601c3d0aaf2841602bf8081dcba1bdc3d1e24
38119 F20101208_AAAJTQ martin_j_Page_007.QC.jpg
1eb145c80ce3f5ea93f47a59aebed40d
1d2abcaeffc069b92b6fd7986bd6ceef8419ef27
116513 F20101208_AAAIPZ martin_j_Page_067.jp2
df762223f9ece772dd81d1f006ab92b4
9bdc23d844971afa8d044c70f41908dd80cb591e
9318 F20101208_AAAIRC martin_j_Page_148thm.jpg
a3e1195948dfc253bf4f6c42de0560f0
6b2bd8a4401b93261f7fc75a10121e5de2bed132
82323 F20101208_AAAIQO martin_j_Page_136.pro
26fd7ad44b5e4984342e99c1a48050e5
6e18d6f26ae48840c05527308ad683f3729858d5
5855 F20101208_AAAKAA martin_j_Page_157thm.jpg
af09d95f3eb50c3bc69ddda21280cb36
c69d6956599592928bac845fc2d938400f15f4e7
7975 F20101208_AAAJUF martin_j_Page_018thm.jpg
c5e26eb75efacf01b95b36f0f0aea587
d9eddfe28bdb163d3b9c6b98878c9e57509611ea
7478 F20101208_AAAJTR martin_j_Page_008thm.jpg
49f265faa8ce978f32bd42c070b6b120
89ef13b51b70b9fad7ae67d662de33969a095dda
33924 F20101208_AAAIRD martin_j_Page_153.QC.jpg
b92d1e19fc48d8efb8ac5b9e49125e8a
bab9809d41c00a67cee7b45ade3e4c880372383e
F20101208_AAAIQP martin_j_Page_041.tif
f3ba3436b37074761b387b1bd6dcf46f
55d147097964887519d3c5c80593393ef32def72
21912 F20101208_AAAKAB martin_j_Page_157.QC.jpg
ed3c8a287a1e02cf0f9d1c1de59cb12e
ef89b55ec8af00e7a41aecb5264e9ec457513bb5
36913 F20101208_AAAJUG martin_j_Page_019.QC.jpg
22544e4e5868045e673950a159eae34f
17d2b19446d4629edf166afb23ffe72964835227
32676 F20101208_AAAJTS martin_j_Page_008.QC.jpg
e0c5b18b8bec8dff96fb5f766f799615
b2fb631f760c96ec8dd7659194f20cd637a242cb
118807 F20101208_AAAIRE martin_j_Page_076.jp2
60359b89e6939a4099b45339cb3a4343
08c3e4c8473d692882d73602e6a0722f6085d9c9
F20101208_AAAIQQ martin_j_Page_135.tif
79456582a7a2d8e24d50c10b0d71b9d8
631d228417d4caedb14b9240b8986e7ae5e9a48f
4558 F20101208_AAAKAC martin_j_Page_158thm.jpg
3fb2c5192c00874631c4841ebd991d35
04cd5c4f1ce7121ca27e4609f1b6fb3748d443a6
8905 F20101208_AAAJUH martin_j_Page_020thm.jpg
45e5011a639e23e1bebf9bd0c10cfbcd
6de226966ba575070d5dd54db488511a039de00f
9747 F20101208_AAAJTT martin_j_Page_010.QC.jpg
32be04d7e69b91bf8e9328434e412865
e17149c16c91ef7691bbae6cd59a3990417ab126
26816 F20101208_AAAIRF martin_j_Page_161.pro
2f24f126e9440fa93ef78c7f0c694d59
e2d9f1a599bef6144f88ecd78df7cce6c6dd8bab
112389 F20101208_AAAIQR martin_j_Page_020.jpg
4aaf604990c484be4f7656238ad80618
4d2038b0efb06d607aa924ac7ed5a110ed79ab2b
8878 F20101208_AAAJUI martin_j_Page_026thm.jpg
0204c835d6d28e0d24422e02e07aa821
2b3a2188d2df83c71f30073c0f093b79b8336013
33440 F20101208_AAAJTU martin_j_Page_011.QC.jpg
b05cea2b46eadf2254771e4b8c95a38b
3e9de91edc04f02455d28915b8d90793cf86d350
102886 F20101208_AAAIRG martin_j_Page_084.jp2
5e8af163e672942c5c26e3584d9c368a
ef329a033153ae05122746556fff37b1ce92036e
F20101208_AAAIQS martin_j_Page_123.tif
c2010ee4558ae3a66f58b0ea16138fef
2693fd2b9e08b5c7f2a5980a8fac46835183d9d9
16624 F20101208_AAAKAD martin_j_Page_158.QC.jpg
bc273413cd9f2d161037e87eb9e27330
123cc663cf47c6fdfcced51714abbfcaf7a043be
34362 F20101208_AAAJUJ martin_j_Page_026.QC.jpg
da24866ec359321e1f4f37fc73ace1e4
4c9512a69d64c1d1aabef4e2d81e6bd1c270aed7
8962 F20101208_AAAJTV martin_j_Page_012thm.jpg
510ae0898bd18acf8f66585d4e4ee2d6
884e6847d00454f5968eafc74b41cfed921bf9e4
341 F20101208_AAAIRH martin_j_Page_168.txt
44c389411da87d006cb474cfd5adef8e
9ae92dd5afebb874eeab276863dfeab5eda99625
1409 F20101208_AAAIQT martin_j_Page_099.txt
6a9a51adfa09bf418eb81e9925d2c615
1e1c2ce9032b2a6a08d7112980490c21016f3a44
2571 F20101208_AAAKAE martin_j_Page_159thm.jpg
cf7df571228f74ebe72e2cb850ce178f
2b970b2231814007ba4181959c1d406e7aeba0f0
6967 F20101208_AAAJUK martin_j_Page_027thm.jpg
4b71ee23c0d165b8a39b02d0a6bb7347
896068897ec7fb04cda24ac57933aac62e984c20
37555 F20101208_AAAJTW martin_j_Page_012.QC.jpg
7534b76a1cdf022d8f7cbea2658e0396
fde4aba9c5a8914369f582f7cf5cfd8f7c15f29e
4236 F20101208_AAAIRI martin_j_Page_100thm.jpg
0a0e5210945c4b20086bf17c497e5911
4b530911e068ecd84b1ef1bcbfca67738f841c50
F20101208_AAAIQU martin_j_Page_017.tif
7f65f1d67b1b4ea3b51beeaa922ea5da
5cd144ace881742528686a0782084701aac8120f
8479 F20101208_AAAKAF martin_j_Page_159.QC.jpg
c3bebbefe39ab23fe7f54be18876eda8
71f4c56bdf2a0381c2f655749ce97d51ae6d5288
28375 F20101208_AAAJUL martin_j_Page_027.QC.jpg
083a4c8cf0fdd0ca7774e6a721ebcc16
9a679313ea5435af807c94cab7bdf7a78485a006
33376 F20101208_AAAIRJ martin_j_Page_062.QC.jpg
c0f6456e3e1019c4ef91deef80a3b485
fc4843f8fb6c5d7305a2b4d98a70da38b257493e
8098 F20101208_AAAKAG martin_j_Page_160thm.jpg
62260ac929c7caa70a634d759c28c6e2
400434a6298123ca511b11ccdad72fc3a80d66c5
15706 F20101208_AAAJVA martin_j_Page_039.QC.jpg
ed81b68cb2b5c16ef5d008a65cb49dfa
94bbeb7c3e1cd8ff2b275b142315087726a462fe
7166 F20101208_AAAJUM martin_j_Page_028thm.jpg
7dd373c8f2f344b49aa49cc5a3e268d3
815626f609f30c491148338fe8901157a3d65b16
30674 F20101208_AAAJTX martin_j_Page_013.QC.jpg
594039e78457899e5ec4aabd3052845c
b92c5fff00133490876440bd80be85b57f5854a4
37641 F20101208_AAAIRK martin_j_Page_070.pro
d70b990e3561671023ea03712a3ed9b3
ed5d17bfb767882274a424453fa7c4450e9d49c0
1872 F20101208_AAAIQV martin_j_Page_108.txt
b1d396dd824f0b8f8a83960b25346f9c
97da9c999bdfc9523b970ab13760a904a1a450b7
32651 F20101208_AAAKAH martin_j_Page_160.QC.jpg
8bae19a2124410c08be6e1fefede6539
bffa7c234e483faf1b7e073922125ff6aa5d0c05
4098 F20101208_AAAJVB martin_j_Page_040thm.jpg
eb2eac262964594a1f65d0b70b867176
e8cd4c9c1c51ed0a15841de987548a78814b4dbd
7492 F20101208_AAAJUN martin_j_Page_029thm.jpg
f18e561487608347a2e55d4f810db619
5ebdf6e9d50a41611d76bcd7199a3a9190a4c818
4272 F20101208_AAAJTY martin_j_Page_014thm.jpg
a106288a6945cfc5a725d3625b14e698
1385100cdeff4656e2904528957da8cf3ff49c55
8355 F20101208_AAAIRL martin_j_Page_047thm.jpg
76dd866b959f434ba7c4e92e11de95b4
45265e43da33361c3d408e094a52be914bf732b6
108877 F20101208_AAAIQW martin_j_Page_118.jpg
7a7233babed55cf133ca372c33815d8a
6f3b0b012816fe8ff8a9792c598620af3de179b3
15354 F20101208_AAAKAI martin_j_Page_161.QC.jpg
9ca291dc2d8826796e60d1b07d09e10a
5d3893fce2d87de2cfc8494160d56434a6c48001
17794 F20101208_AAAJVC martin_j_Page_040.QC.jpg
acbc1f4e2459452497946b84533db079
78a56251f9a1034ed3b38b9d99008d2be9f5e6e6
8765 F20101208_AAAJUO martin_j_Page_031thm.jpg
c2337487894f2e065437bc36b575ba73
2fe1bbaf56386636d622216fab8160bb51f9b3d7
16016 F20101208_AAAJTZ martin_j_Page_014.QC.jpg
e5f6b9cd99a2ba916b5e20428c3f8aed
3a7e3f3baf75d70d70fe672d6e2b37304630c366
8406 F20101208_AAAISA martin_j_Page_030thm.jpg
747cdbb0852897e186d13142ed61e729
1079dc392daa04113386fb9bd75ac98b8d9cf5e3
51293 F20101208_AAAIRM martin_j_Page_048.pro
a44e6e71dd293039eb0c47989539696f
c2c394f6b58fede1237a021cb08332ed72f79ceb
8740 F20101208_AAAIQX martin_j_Page_126thm.jpg
fd7d915f870de1c1309e0307b14b0150
bd1e70390af835c47462ede2644abe293fa7ad5d
5673 F20101208_AAAKAJ martin_j_Page_162thm.jpg
2b58ac91031606d1e3bc2aed821a5a1d
f3f49f4ff2c969d90069e39c37b7bc7416b3e7fa
13007 F20101208_AAAJVD martin_j_Page_041.QC.jpg
763b928c263db99c58c6432881330279
c65e7bc6405e29f50809f9c0a33d6fd99624467c
36301 F20101208_AAAJUP martin_j_Page_031.QC.jpg
4bdaa4b9c296f856e3641b84c58ba24e
3e19d8bb8afe7cc195020c7b5faf919c9d23e34f
F20101208_AAAISB martin_j_Page_158.tif
f5bd61194f32a4684882cb5f6e702557
8422357c5789fbc35b76335002b00d425c180ecc
90303 F20101208_AAAIRN martin_j_Page_119.jp2
6da727b534297d76bafb5f5a026f8e92
fc6bcef8704ef1fec7aec35fd6add10167add9f4
F20101208_AAAIQY martin_j_Page_118.tif
75277bde75448218297ceecbb872ca91
1c522972f1fd8104caaa9319525879b0fdc70706
8462 F20101208_AAAKAK martin_j_Page_163thm.jpg
5117b693d62f90f044603c6a1c1c1da9
3f5d09780fccaa5ad123f33124e666aba5583642
5380 F20101208_AAAJVE martin_j_Page_042thm.jpg
f650a67599d5ce8163b5411a69640242
211f17e1d1e8a688635a950ffa3c770c2fd4edec
8218 F20101208_AAAJUQ martin_j_Page_032thm.jpg
573d34a907d6f0b98eccb4c313e3ff11
a9af9547a7f710b0b2db7ec7c306fc0827e2abb6
45544 F20101208_AAAISC martin_j_Page_142.jpg
94cdbc3933313d3466559eea32a78372
7aaf3ec85d6178a413462b7311f10744b5e5c724
4164 F20101208_AAAIRO martin_j_Page_041thm.jpg
8f506d7b0446398481f369713762f117
8e4bf3887f29dec0cb818c0379e3e449cba30f9e
1946 F20101208_AAAIQZ martin_j_Page_130.txt
78e9940d11da6765d9bc05d4a95feae1
fe38e075e7d94718d9e2aa8cfa31f4dc5e81e5f5
33981 F20101208_AAAKBA martin_j_Page_175.QC.jpg
3f32482bf6b81943b96b7bbcf721fec4
e8047f6384949b072a485b4adf72ec8c77b12990
33684 F20101208_AAAKAL martin_j_Page_163.QC.jpg
4c0eb0cf47c0ab7c888950548a1e8c5e
4fd5849ec32c330284702a3c484fcb59b6172f24
3075 F20101208_AAAJVF martin_j_Page_043thm.jpg
d013813424d9f684d9d6a20494bee223
374efb2910783b0940b940762b0c10ac2e97e2df
30448 F20101208_AAAJUR martin_j_Page_032.QC.jpg
281294307716e2c070affe4aa81f2c12
b8853f7f931f17c9b8134eb197a7623e3eb74d76
19897 F20101208_AAAISD martin_j_Page_143.pro
c5bbcea2abbccb008ea05e2cf33d74b1
5ea37afa77aa226b0641ce9c63f5981cfb7aaa06
5891 F20101208_AAAIRP martin_j_Page_069.QC.jpg
33441541243a3d4bdec860fbc43c9cab
99d88d2a66dcea8fd6694d760fb7c38148baac03
36443 F20101208_AAAKBB martin_j_Page_176.QC.jpg
a2abf2fc33700fc0ee6633294570266c
c711479be768ae7299a393ee6734b9bd0eb7fc0f
4594 F20101208_AAAKAM martin_j_Page_164thm.jpg
602e5bff5fab6e3a0d0f6421d0bb23d6
7b1b735cbc79fab1a1b20df5e6c1b29c5a586af6
9016 F20101208_AAAJVG martin_j_Page_045thm.jpg
93a0a6e95a4ea8f6ec7852c3c4de1caa
4eb97c2d0febf872b00c215543af71cae26f1730
35303 F20101208_AAAJUS martin_j_Page_033.QC.jpg
12e95ddaeaa4e47257bc347297ad244a
36a9905c16ae5bb495f058cc5a039fb87ab48941
111175 F20101208_AAAIRQ martin_j_Page_075.jpg
947fd212bd405488ba322ecfeca37f94
aa397f0eb139b9bf06e45a307c58108a115bc321
32210 F20101208_AAAISE martin_j_Page_159.jp2
bf3fbcc334a1ea3fe12c13d590c34cb8
bfdd5cc382858fad3acc7c907934a2d0dfc0e161
34732 F20101208_AAAKBC martin_j_Page_177.QC.jpg
702b4f938fca04feeca9b3ca72828f25
40c652431f39caa19ee4b4195647ef5b8f70c48b
18456 F20101208_AAAKAN martin_j_Page_164.QC.jpg
8e78dc3b98f3d7fbf4afef46a84e94f4
4cd7687764571aeae5a8dd70937b09f3cca70b0a
35017 F20101208_AAAJVH martin_j_Page_046.QC.jpg
328623cf97fd441a8fc43c502b55a33c
c7db3365b6d73b3d854c56c14c4df16701b752a7
8730 F20101208_AAAJUT martin_j_Page_034thm.jpg
d6c266abc9a6d18f7991d9a5b35e394e
578de9999695cfd991615263ca30ee8c18f46132
100988 F20101208_AAAIRR martin_j_Page_049.jpg
976ea598bb8c4670cd4ff02fbf0a68c1
a0e2025488e9a1a9b61091c346f0606b4f8201d6
4437 F20101208_AAAISF martin_j_Page_070thm.jpg
f50f9c9711c06dd41d554b2600fa13b8
05b8123f4daa8519a44300bdcea32a5425d7c6aa
34077 F20101208_AAAKBD martin_j_Page_178.QC.jpg
7878e457177bcd4f006d9e3a08c84d16
36cefb14adcbdbbe513fa95d0634d1f01487fdb4
7257 F20101208_AAAKAO martin_j_Page_165thm.jpg
becbab5169bab9deb3990f9421fa1e72
47b6e6b117e60e0b5230fcd80f459a0e0abcf9f1
33448 F20101208_AAAJVI martin_j_Page_047.QC.jpg
aaec8738057ab9749c0d0d894c69ba80
e6254efbf02e0d5bf3a429fcff4e804b7d2add8a
35708 F20101208_AAAJUU martin_j_Page_034.QC.jpg
03eb3f2aef43a556cc9d024b0e37e60e
a5226feafed400decc56ac01bf588908883c4f19
13427 F20101208_AAAIRS martin_j_Page_142.QC.jpg
0c736837941973ece419aa57552695bd
1a4c4a5ce4f93b23247aec1300d98a079d92256d
F20101208_AAAISG martin_j_Page_089.tif
fd4ca07fc5a7be0bb8a5e265a0eb742e
e8bca5bbdff28f546f70f2e9589f9a368fd26197
27326 F20101208_AAAKAP martin_j_Page_165.QC.jpg
89c1a05a4f3ba4eaa249b185a71da2ba
7077342a2ccfafeb9ddd3c5a87bed40375655c1f
8238 F20101208_AAAJVJ martin_j_Page_049thm.jpg
aca1222172527eb929d93cab00483506
405c16f9e9adae217edeead9ef6c4fa24b3b62cc
8638 F20101208_AAAJUV martin_j_Page_036thm.jpg
c7acc1eab8899d8f04102fec76e04adf
f2cd7a27813afa52e2c159083abb7ab6f321da45
7937 F20101208_AAAIRT martin_j_Page_151thm.jpg
5061164704deb2b2fa8386d032c476a6
f21b8ada7afe61e179afa92d0a2a257eabc6879a
51750 F20101208_AAAISH martin_j_Page_015.pro
9839005e6cfca3f7b5251b98d477fee3
1622d30802813093e242f5b31c8a5d45f1c7e346
8932 F20101208_AAAKBE martin_j_Page_179thm.jpg
c48f6c67f435042789d654dcaf88e8c7
0bbd1d16c37d962b04b4557e41d710769aee8c7b
3488 F20101208_AAAKAQ martin_j_Page_166thm.jpg
0f210fb883fa7bad113d42d244d3361e
84a05b8b176b31427543f9a708c58e0bcb29dd41
33291 F20101208_AAAJVK martin_j_Page_049.QC.jpg
869a0680dbf1773b44942a3769db7bf8
e533433ea4f8024b99657852d0fb9a140ad6d8ce
34522 F20101208_AAAJUW martin_j_Page_036.QC.jpg
cf3e441de34cb4b7e85d7fac9e972476
b9d14b07d29a856ce6e74cf35e5208e3518f097a
37335 F20101208_AAAIRU martin_j_Page_117.QC.jpg
50a9421d9d8eca8ef2627fecea8a8535
fcd0a1b182ef57ad4a9433beb4de023a9aec8d99
45314 F20101208_AAAISI martin_j_Page_113.pro
3a3211e339871fb5c9964b5fe7f18031
121f236d2c74bae731079ad862ed4469b41d4b73
36160 F20101208_AAAKBF martin_j_Page_179.QC.jpg
ca669a538b32f7fbe09050add65b5370
0d889609b191c165955ed24696623c037e6c3bed
17398 F20101208_AAAKAR martin_j_Page_167.QC.jpg
be3ab1cfe3d6b1b6749b9e8723677383
e39ac5f4734c97dcb19c1b2eded5dc89011621ff
37780 F20101208_AAAJVL martin_j_Page_050.QC.jpg
d87ea91cbe5bec70e252ad8ba7f36e8a
4a0eca2b94310e3cbc373c713e09ce1e8278d732
8409 F20101208_AAAJUX martin_j_Page_037thm.jpg
34da41d1f2301a0ce644ec45cb61859c
fa65868cf1ce66302f9dbdf2073734cbbcd6d5d2
34096 F20101208_AAAIRV martin_j_Page_154.jp2
889be6c8079915ad0a76cd53f955c6c0
1c12880d086970c386f3bef5edd41b681ff77f61
101039 F20101208_AAAISJ martin_j_Page_152.jpg
5eaef881e67533f27bf51caea1b57a8b
3d5722ffd222c7d11b4bcd0acb05030dc1e9d943
6595 F20101208_AAAKBG martin_j_Page_180thm.jpg
603aaff7c63c9f8c39c057d456cb569d
ba0d3d27be24617e2fb0cede854de0d6c0094c04
24112 F20101208_AAAKAS martin_j_Page_169.QC.jpg
d11966c40284ce2afe1d6aa0b06bdd9e
76f560639348e6ed8ff0d104dd243e6912b14d80
34975 F20101208_AAAJWA martin_j_Page_067.QC.jpg
cb8e5fff62c88f66374e447a3754551a
0be9f5856546885dc52cbf388f9c26783cd112aa
38432 F20101208_AAAJVM martin_j_Page_051.QC.jpg
129ace9a070d2fcd9be86101a7eb2f04
3aeb897b137c6b5647519d41946863f2a219c9bc
87881 F20101208_AAAISK martin_j_Page_024.jp2
9c39294a07f3ec033ec8f5f520a8b49a
ce2560e209aa949e3551728faa96cf7e36d3cc5c
19431 F20101208_AAAKBH martin_j_Page_181.QC.jpg
9511c7189a93738b3c9e4fa334343601
39130839fe0c6e5c44072631691bbd996d28569c
7958 F20101208_AAAKAT martin_j_Page_170thm.jpg
ad1347e6790e143f83b08d9bef5a83da
bc306a24c1ea46b48dd91415cf4b563898b5784b
38418 F20101208_AAAJWB martin_j_Page_068.QC.jpg
af3e50ebef679abdc2cd8fcea5c02335
7e395d30fea0a0ffcd2e1de56c700e05f525ee66
8826 F20101208_AAAJVN martin_j_Page_055thm.jpg
b8a522f674c6cd2222819f0f53b8bac6
f12df996f2c3f84e8ea57fd1627eec4eba400890
4515 F20101208_AAAJUY martin_j_Page_038thm.jpg
d5bcdc61f22a68f239c3eaedc0da5675
48034af3413fab56c73ab74cd9081c27413ffc9b
53666 F20101208_AAAIRW martin_j_Page_150.pro
6705b9d0d8c3363f98c084ee51d6bbc9
b7d42c4cc447180f4673f6644b5e0e4afb081757
72722 F20101208_AAAISL martin_j_Page_073.jpg
ea4bd1730b27dc9cfe2ed9023490aca1
cfdab59774f57227499d701cc63d1d864e114429
1280 F20101208_AAAKAU martin_j_Page_171thm.jpg
a65fb894a2164a6a7f3fb5e4e2f7f5ce
ad5660dd7c9e9bcc949d13b301b568a03688af81
17836 F20101208_AAAJWC martin_j_Page_070.QC.jpg
e9af867bea240884393d8fcfb67f6095
59c78055b89b255e61c5d44e2e7d0c37420b05d7
7568 F20101208_AAAJVO martin_j_Page_056thm.jpg
379f23cfe0626bd50e62db454249dccd
e493e11267de62f9a918b9488810138e358d6ad8
4107 F20101208_AAAJUZ martin_j_Page_039thm.jpg
bbba8efb10b80b55bff29cf9044d676e
2c235baa5da1b2633bd2cdf8c0d489f412407d3d
F20101208_AAAIRX martin_j_Page_081.tif
402ec84a3207e5bbb81d5d8b4636f0db
276361bfe23d678f95743a93aad0d6201dbc1cae
113920 F20101208_AAAITA martin_j_Page_102.jp2
df44fbbcc04372d9f051523d7d6fb316
98c6174d3f01512e9ec42ea2645095f9ea9e6d0f
92621 F20101208_AAAISM martin_j_Page_029.jp2
abf1b976d8eb743534e31a6110378786
83a12264ac24fd6763c50425b1bd2daed1575ebb
3724 F20101208_AAAKAV martin_j_Page_171.QC.jpg
63c2a1326c1589093c27e7c520c40304
5c1c8a009f6e82bdfabcdaeb2d4a29257a1e5c56
3629 F20101208_AAAJWD martin_j_Page_071thm.jpg
e3be86b489916eeaa405f43f11207309
e6f0a2428d296da2d41753c5bac441566532cf19
33667 F20101208_AAAJVP martin_j_Page_057.QC.jpg
a35fff4e2f3cc07dd03fa16e77949470
6f0040aba0dc14801f5704ccbe7dc4a3b35a8e1c
50864 F20101208_AAAIRY martin_j_Page_127.pro
aafd8d86d9564b0a14e00e187afa2839
16c86993249947c530d2f2fa3be7f4ca9429d0f4
8039 F20101208_AAAITB martin_j_Page_113thm.jpg
7ee8ac80c4ab9e00a17626c7ef8c5801
4614a02e6ae89bb489a674a4cdffddb78058188a
F20101208_AAAISN martin_j_Page_102.tif
e2658fd80ea5bf1f8b9c8334ed3615c3
4735ea689c5e972a8de54ee33f3b06bc0dd24ebe
8732 F20101208_AAAKAW martin_j_Page_172thm.jpg
bbb58bfece832bf909e178a9a4670f27
16672dd80437ff007cd273ed44d1629dd305b9d9
4273 F20101208_AAAJWE martin_j_Page_074thm.jpg
8ed25ff99394de64b300ee125bb78bf0
dfde315bf05613d6d088aa07a988448951bcdfdf
29298 F20101208_AAAJVQ martin_j_Page_059.QC.jpg
9ff3cf0c2ca80379e82e5672655c5780
c5052ecc0b7cd6dcba75dc79659c610ae7560215
2477 F20101208_AAAIRZ martin_j_Page_155.txt
60af9c94a7f4fca9e09b916afb53777a
560323729552ce5d8893c8e5c0866d306e4162d4
1708 F20101208_AAAITC martin_j_Page_028.txt
f50bf8bda4cd10fcdc0f10d298fdb2c8
60c3c7d1dcd840cbe8bf03dead6b1af8a30ae9d8
2039 F20101208_AAAISO martin_j_Page_037.txt
6742609207314782407c7852a5562d63
776fc99fe453d3a607bc1c9e1846c37b0eb04d56
35218 F20101208_AAAKAX martin_j_Page_172.QC.jpg
9295993c62bb1d79f474714a26983e88
a92de01c9893432687c02f6d53f74d1315246658
17083 F20101208_AAAJWF martin_j_Page_074.QC.jpg
f93b645de999b06e5fc0b1b07942f0a5
3b5294d6d5310bece8dd861d7a3182a79deb04c1
8397 F20101208_AAAJVR martin_j_Page_060thm.jpg
c8519c85d6843c6a1f75af4ff0a49051
f33908313b84f424937349dc2d9d8786f6419b80
32044 F20101208_AAAITD martin_j_Page_123.QC.jpg
afc34397e939009e0b8da6ecde41c092
3f7713c6f52ec9e922f422c1d1c0af4e5c4a1793
5097 F20101208_AAAISP martin_j_Page_098.pro
7bed38a0b5b847a8c93bc4e2d1b58781
ec27bc0ea4a3916620aebdc72f74ca010ddea5cb
8831 F20101208_AAAKAY martin_j_Page_173thm.jpg
4a7ea5ba5df8e0ba42476eae47dabc10
836d0f07252b3f816739d66a911c5cb98f7506e8
8892 F20101208_AAAJWG martin_j_Page_075thm.jpg
2eceeafa65e62f551ffead1f61f6c930
b221da2f88ebaafa0759fb6253e99e38be2a60b4
34499 F20101208_AAAJVS martin_j_Page_060.QC.jpg
66d32a67622a642a594c2c9c139a877b
fb85345a4ef8ead8f0b42c1e01dc76b4c8beb905
27202 F20101208_AAAITE martin_j_Page_096.jpg
c698a475612ffeca5e8e390c04577bd2
902c472411bd18eac464fa5f4b5bd09f28b482fc
1196 F20101208_AAAISQ martin_j_Page_161.txt
976603a5ebe1966e2b36ff31d8ebea6a
53c3269a69a8c7605a252d1af8121f15d26481de
34332 F20101208_AAAKAZ martin_j_Page_174.QC.jpg
447347a904fa9514b7ba49440fb52176
e3cb05e677d3790756a5c766cbdff7b2a9eae76a
36405 F20101208_AAAJWH martin_j_Page_075.QC.jpg
2433657deb6b910fa64937610110a193
2969bb56d0a3af37ae2fdbf4cbe9de909e07564b
9041 F20101208_AAAJVT martin_j_Page_061thm.jpg
b7eb69bd25a92b7eede4acf0232aa19c
c291eb57872d0884eb006403fb7d2f4522470e6c
8760 F20101208_AAAITF martin_j_Page_046thm.jpg
ba774c8285efcbfc34beee075da60cff
657ed0d76f452b607a088273cf13ee8b7e865640
108112 F20101208_AAAISR martin_j_Page_125.jp2
3eeaa84aa773aa689be5871eff4a999a
09320c46ed479fa85ae57fed654a9e192efbeaa3
8697 F20101208_AAAJWI martin_j_Page_076thm.jpg
7e49d1ce08d2dc19d5b3b4268a133ccc
1aee0455032252628f42b4deb6bc89fe1b04d0cf
35136 F20101208_AAAJVU martin_j_Page_061.QC.jpg
b8def6a53d6e026efc00c3602833635e
40cd71ede46ecd06750f2b44cb478f4d0c8eabbc
13984 F20101208_AAAITG martin_j_Page_071.QC.jpg
590eaad0c82b463a2859d5f68d1a3a15
2cc9353e0221a0832e0f3d33bced66cc0cdaf664
116303 F20101208_AAAISS martin_j_Page_150.jp2
cc63bc6dab5e26511b175dd3b24096e1
69e1d429a32ba4e4e0e89a29163b19e25c99d34f
35772 F20101208_AAAJWJ martin_j_Page_076.QC.jpg
99841ae9e1d4aa333089dea38caff135
90ba1d2dd1ea0255cfb69eeb9435ecc5e64a3cf5
8642 F20101208_AAAJVV martin_j_Page_062thm.jpg
73bd1ed6cbf002cacbb128aaa739dcd5
e0c4f176f7c021ab40179b82482a6900d534aab7
62896 F20101208_AAAITH martin_j_Page_116.pro
02f9abaea9777a1e144aaee20a5a1ec4
72a3daf962a6f5d4cd040dc9aa4e8d03035fbb44
4329 F20101208_AAAIST martin_j_Page_138thm.jpg
c71f67cf2c615429f49e52d199b35460
5b75e608da1a79adeb7b7c005281c16ffec6f822
7899 F20101208_AAAJWK martin_j_Page_077thm.jpg
379743c98b47c750d36caa7e32a8e139
bed330011683f9d42e08119ae7e2195202d87289
8231 F20101208_AAAJVW martin_j_Page_064thm.jpg
a6845b2319f7a260c81135a44c81596b
3701a84e85bc598491bdb930f381c3787385e611
54355 F20101208_AAAITI martin_j_Page_081.pro
59d598f5166063a7fe6a295fe2d5ef35
37cb738ac7d1af8a8372607b796d510c164b49d0
124492 F20101208_AAAISU martin_j_Page_172.jpg
7580262a8d289ea749ab62e0a4ca0a6c
3873c3b28fc24190b0e92b870fb8e4c30231b970
8177 F20101208_AAAJWL martin_j_Page_078thm.jpg
1f27fff185777631d9e0ce21ecae45d2
c4514d44bac350bc1d2795f42fb5f8f31e631add
F20101208_AAAJVX martin_j_Page_065.QC.jpg
33cfdc78d2b9cf362c0811f5ae2f3b8d
810ba427aa7fc208f7323f4fd16e670a2f2f4e4f
2315 F20101208_AAAITJ martin_j_Page_096thm.jpg
51b6df604b3ebb6f12ea741926ee8e55
33b15461cf44533d64f552fbb60ade22d2a3c0de
37271 F20101208_AAAISV martin_j_Page_052.QC.jpg
b76ff7b4cc0ed1bb96fccf188b0ef3f3
601952934ea7bfd71c0e3cc5d9ae0ab70bdee7e5
33104 F20101208_AAAJXA martin_j_Page_089.QC.jpg
c46ba553214ec55cbe65c51950138c99
39de9eca564b56285e34705419049907ca118c6e
8175 F20101208_AAAJWM martin_j_Page_079thm.jpg
4a753fdd2cec65b56ebdd1aacc56949d
73349baf1e42b6dc6d56696e009195a92ccd91c8
37610 F20101208_AAAJVY martin_j_Page_066.QC.jpg
39d8243450a4ad29f5645ecb1bd7ad4c
dcb6cb1491c1158ebcf59f6944080af9d5988835
F20101208_AAAITK martin_j_Page_109.tif
aa5ea70160c65561eaeafe6302a6cfbc
c668b284863026c4ec04e781eb9dd673e80c23b4
93858 F20101208_AAAISW martin_j_Page_180.jpg
e16778bb971af6aad9091e00dbf3436e
ae7823f64872d21cb9bf118cecc4400f6b02af95
32841 F20101208_AAAJXB martin_j_Page_090.QC.jpg
f51b5e89c68e4790f32633762bb1c0c9
71feae549754defff94b6803179ff51a075b376c
30996 F20101208_AAAJWN martin_j_Page_079.QC.jpg
14e9e5ff91e357f96cec882b305fb56e
99ab367a8025f790644d25fb5b3ffb0ab600ed1e
108533 F20101208_AAAITL martin_j_Page_049.jp2
d3f6a0bfed6666af52fa2298fb7abf3c
442243bd45fb73d89165a97cf835047b37a4dd40
7608 F20101208_AAAJXC martin_j_Page_091thm.jpg
d73b9c9f1e7b649ee8fa370ad7ff7df3
1306e771ebad4b74143c310b73aa81ed23abba57
35522 F20101208_AAAJWO martin_j_Page_080.QC.jpg
d6daee45d461d43b4423227aca1d9396
2cccd649549ec78833bded9f55091b8d94f0b30a
8938 F20101208_AAAJVZ martin_j_Page_067thm.jpg
a272f1670a906c1b5731bec084fce626
078ce6a8df283a35c7a09f41c243148637a9e843
F20101208_AAAIUA martin_j_Page_140.tif
631676f09dd240a6e88dbeca73210684
36cea09134095e1af278e1fc5999eca17a1ff4c7
F20101208_AAAITM martin_j_Page_077.tif
d003b36b713ef02734201b0c6de23d17
48454cfb0e8c320070e6281b588341956655616d
F20101208_AAAISX martin_j_Page_028.tif
70519ded56b7b83c0976a9dfab9489cb
a2ed63d036248e8324a5fa49a8f8da49b099c7b2
29514 F20101208_AAAJXD martin_j_Page_091.QC.jpg
4f5f09e25dc7f3c6817962e8637e1a7b
9b3c86a40a958f51436599505b06009bb8f86449
9210 F20101208_AAAJWP martin_j_Page_081thm.jpg
a7115d37ea1e83b329c9d804934797d6
cf73743abbaf2c5fed5483652155efba1b83a6fe
F20101208_AAAIUB martin_j_Page_104.tif
6d5eb9e799861d7329d4d0ba3af5895f
6d763f726dfbc6fafb8ecf1b005d2b3abaf2bb63
1875 F20101208_AAAITN martin_j_Page_079.txt
7861c82c09c1051dcc1b2234567aa79e
1ec28cf9c44b265939e49e2c00632aa8e2c453cd
1736 F20101208_AAAISY martin_j_Page_029.txt
2e769e16dc4022a1cf9cd005b4fc8c13
143e7607a68f58d48050722a157886c4c5056928
8619 F20101208_AAAJXE martin_j_Page_092thm.jpg
e4c5ee79ea69d23dde6bf26c1a42f3dd
e93888d8ec4c86500b7db1ab7a6e128bb5d9f8be
8562 F20101208_AAAJWQ martin_j_Page_082thm.jpg
f32a74b59e4ca7fe47b40a222c699de9
f98c0975c52afea77817aa229813f59c647d1527
108485 F20101208_AAAIUC martin_j_Page_026.jpg
2a822fe7fb535aadfc3569eb7683ad41
9d4c3b2f8c1ece2fd0c6ee55210a603292af5380
F20101208_AAAITO martin_j_Page_035.tif
5c7b24d0d26c6276aa1b054381799770
06eb7de1f96437394e8b8c61f525ba463eb6c792
F20101208_AAAISZ martin_j_Page_122.tif
390725d8174df46f56b5ff8355ebba9d
d7bdb4d29c4aba807361d10f957ba6f49a352eaf
8820 F20101208_AAAJXF martin_j_Page_093thm.jpg
19cf42fad727d7dfef21e355541ac1d9
fdb8a4731be051b656101204d8fadd23a071ba54
34703 F20101208_AAAJWR martin_j_Page_082.QC.jpg
95a0206a458738f7d5c5967fa2654661
53747ad1d7b6dd77f29bfa2e3b94447c7b446a0a
1968 F20101208_AAAIUD martin_j_Page_090.txt
da70b2c5290a1bbaf1781ba5d29c00fd
eae1ed1903c41b0c1c4e6c41d6aba88736a85cf8
106830 F20101208_AAAITP martin_j_Page_031.jpg
efbf10cd257315b8b24ade09590d4213
f68ff21e2112e3fb2d5fbe572c1892ad803ec91a
9191 F20101208_AAAJXG martin_j_Page_094thm.jpg
21dbc73e382642ac67e306def577917f
eb3bc257d0f11e63f13a385c6a6726349f5150d9
9424 F20101208_AAAJWS martin_j_Page_083thm.jpg
660dc08109200c467daaa28c5fb1e771
1aa0c9cf62f040a1aade040babcc40badd399277
7918 F20101208_AAAIUE martin_j_Page_025thm.jpg
dcc7c8cb2b98d91a4fef5c6feee9e8dc
42aae380e187acb28d25a12fa2cd440c32142149
110064 F20101208_AAAITQ martin_j_Page_163.jpg
ad3b84dbafcb534c4d54e34b6a08dbc4
767dee18c51d184d6af0d5d8033fe10ef90c6cb2
F20101208_AAAJXH martin_j_Page_095thm.jpg
7f799d58f82abd600c09a9c2ab73bb9b
d217796dcf5d4530da372f516d1fcd64ac4534ac
37127 F20101208_AAAJWT martin_j_Page_083.QC.jpg
cf36d3ba5bda471bfa8edf31bd84658b
996bca7646ac0b387225c36fec506834a77dd97c
33175 F20101208_AAAIUF martin_j_Page_131.QC.jpg
d1984d512b0d4291bcd5616a39981a2c
b53d94fe3cc2a955972f7674d06bf649378fa8f1
8520 F20101208_AAAITR martin_j_Page_089thm.jpg
5fda48e609196e5606298624aba66d2f
1e635bd93a786b81cfa8c473fef4ea466533d462
108218 F20101208_AAAJAA martin_j_Page_033.jpg
680c16ea6bd2069e5e98f5a63ee71923
47fb24e76812b8b1d65152111ab46deab9484f2d
37012 F20101208_AAAJXI martin_j_Page_095.QC.jpg
a6f6e8750770230d45d143c5923ea33b
263733326a2bf61b1e1d2af065be5b8fded027bf
8625 F20101208_AAAJWU martin_j_Page_084thm.jpg
c782b64f36f817a81e6cc623dc3f11bc
e491626b5320bc7c163fd73de37679fc7ce61afc
49373 F20101208_AAAIUG martin_j_Page_071.jpg
63ed678c405e25f03c40aedeb2b26b42
86d4a6910725b9056d21c51643bd8195b5badd63
7162 F20101208_AAAITS martin_j_Page_024thm.jpg
728fed536502d6daa91c3fada6be6826
fcbbd65c035ad293ab0e25dc3a07a02d392d71a0
111473 F20101208_AAAJAB martin_j_Page_035.jpg
a9304c490926c4176bb4eb1601b11d9e
cc14c17b2600b38009970226b3e721199e10e3de
9076 F20101208_AAAJXJ martin_j_Page_096.QC.jpg
2816a807b283881644f4f4899a78a19f
76120dd8dd6c1c7462befe44f5cd6ed68a1e1561
35749 F20101208_AAAJWV martin_j_Page_084.QC.jpg
6a48abcb00fe42a6fbf75e2a283d74b1
38c3d6b4ecf9350866718d121f7b43566780b93e
18898 F20101208_AAAIUH martin_j_Page_038.QC.jpg
727ee4c5231afb3229bb47a6a96a3c71
51f1227b5a6c602e9c2b035286c9b91fbb93d22f
8793 F20101208_AAAITT martin_j_Page_052thm.jpg
c61fdd804fda2e574c8f4c9adbd23b0e
4be4893742fb1c2ac77a2c8f6081f52d68493212
107232 F20101208_AAAJAC martin_j_Page_036.jpg
895b2a82dcb253ab849043dd15790b95
01e4814e913d030ebd6685f25c5a90cd220ea0bc
9304 F20101208_AAAJXK martin_j_Page_097thm.jpg
cc624691d66e416fa4120c8ffa2c290e
925171d5ae5d9d005688d3a164feb1902faffd34
38349 F20101208_AAAJWW martin_j_Page_085.QC.jpg
6c5cb2577706e55b3f6cb1eb2765e3fe
8b49b8040969f14aa48b618b78384042c01631e9
2769 F20101208_AAAIUI martin_j_Page_012.txt
0eeb6ce99b9a0df08554584aef1c5790
f9c6e170043f1766af1718d1778f6670b526a5ae
6943 F20101208_AAAITU martin_j_Page_111thm.jpg
3345c7aeb3670414b81b84417ca22880
7f3ce3feb0a77494a6c18cb025d19113ebff5fbe
46762 F20101208_AAAJAD martin_j_Page_039.jpg
7da211ce20aac98c9c3ea1d0623f555e
5dd0792d4a0647662d79d16f37ed4cbaea7d6a8f
40680 F20101208_AAAJXL martin_j_Page_097.QC.jpg
675be9a7ab938b4a8c890c32466fb280
3d7b1d6b24f9fd1dcc87a437d3674f616de94f9a
8566 F20101208_AAAJWX martin_j_Page_086thm.jpg
1f47d871353654f46b19ec266a66e0c8
82557ab59e900247a85bfaeb3bb94a87f5442147
26714 F20101208_AAAIUJ martin_j_Page_180.QC.jpg
54f69e45f5c09c0fce2a9050dddbcd45
f491f51239bb92881835f6a832b9b3f2438ce514
2123 F20101208_AAAITV martin_j_Page_126.txt
d9790a4b8f697a832d0b797071c214e8
6fa78548e2ad4c0dc47eadebeff1a8cadcc7666b
26850 F20101208_AAAJYA martin_j_Page_111.QC.jpg
303fdeb05ea578761285abd509ac7838
12d725eec135fbbb0fccd59a998e73132caa2fc3
3910 F20101208_AAAJXM martin_j_Page_098thm.jpg
08d1324e6fa9f2712619df57c1365747
672aeae0d2500d79c8cb25505770dd2733779a34
35009 F20101208_AAAJWY martin_j_Page_086.QC.jpg
607049b5944eddcaabd9bb474b304d9a
e8b81cdd70bb74893086ac699fd0202d0d7f0bb2
F20101208_AAAIUK martin_j_Page_178.tif
d58049d4801ccf76b691a5762a273aa1
24af202168e863f44154bb9b55c2c8d7d4f22552
3806 F20101208_AAAITW martin_j_Page_072thm.jpg
c39435a2d4b5c30cb2a8249229bf133b
6b07ed975986b4714e18df6b85a4e9d5f1e72b3b
60211 F20101208_AAAJAE martin_j_Page_040.jpg
24b5972fcb959a998a38987f2176916d
85d5d70fabe12d8d4d0d241bbea91db3bea47641
6981 F20101208_AAAJYB martin_j_Page_112thm.jpg
cda25e10e275f72905dbd9809db215d9
d6311ea3c8a7f65bf99546320651585c4720879d
12204 F20101208_AAAJXN martin_j_Page_098.QC.jpg
5fad26a96e07d8000948cae2e9da9219
f48f6b1e79de06e12d1b0cace0d0a2d10249de74
7993 F20101208_AAAJWZ martin_j_Page_088thm.jpg
5a8d4f1b17adb8a16044fe2aca60e216
427fd1e2ee91915d16e973d5d1fa8d8b2ceef964
4680 F20101208_AAAIUL martin_j_Page_140thm.jpg
9a6cad5e71e5a24aa4bbc8bf1b61f690
62d17338c7c9240d242a2b251a7d0efc3a99c28f
12828 F20101208_AAAITX martin_j_Page_138.QC.jpg
d6186be3216bb97e543ef0fd6afc2bc0
661d6303c05669d210996071eefc3d8f70a76659
39533 F20101208_AAAJAF martin_j_Page_041.jpg
c4fa4138f24dfe2a80cc7464705b9b8e
288783052ed2d067a8b0afa5e3423c1cf44a2029
24124 F20101208_AAAJYC martin_j_Page_112.QC.jpg
3c7d15b823b253aaed9f826b09c59847
8d5970b06ff487db8c1827e999ed8ebec6d0d719
5118 F20101208_AAAJXO martin_j_Page_099thm.jpg
aa402fa64b7712968a84279f1410a426
c6240349578626f02f00bc476d7621bde7225ae8
8929 F20101208_AAAIUM martin_j_Page_087thm.jpg
c72b219d8c6363e014b357ecb829bd42
a59d8ee20ba936ee1e0476baf71e937c2ffdd4c5
56958 F20101208_AAAJAG martin_j_Page_042.jpg
d408e3b2cb1155e88d10181823cb8b48
2316802006648303c51f750180e5602724d74a88
95935 F20101208_AAAIVA martin_j_Page_124.jpg
1d425915113988489ea9daf4f74b51c2
3df39391bc3bc4f5ac50a1efc6c677b9b093bfa0
29002 F20101208_AAAJYD martin_j_Page_113.QC.jpg
6874f7fce21aeb896bcbdd984c64b903
8d69793e4d03d6cb199c0d8723698773b3a9e659
19135 F20101208_AAAJXP martin_j_Page_099.QC.jpg
bb9e56c856b8e92521512e9ed046f1de
8a937a068cd420a46721e92bf69fb7795354760b
796 F20101208_AAAIUN martin_j_Page_166.txt
128cbaf56d56c2fc4eab7e3d2f24ef2c
c86b559c6f298d8c41cb8b482e8273407017120a
114704 F20101208_AAAITY martin_j_Page_095.jpg
3bed5c1c5582e8d99e5315f361d21eea
2be89667f68aba06d65a4d185356b57a78a57eb5
108211 F20101208_AAAJAH martin_j_Page_044.jpg
6c90077f7281d4082e41f28861f1411e
f7cbf5ab7bd3d420e7bab67ad89eb005d3eff943
54421 F20101208_AAAIVB martin_j_Page_034.pro
3827ebfb3cd44663b8903eec42bab106
368ab7f9bf12516982cea08a45db56491275df89
8212 F20101208_AAAJYE martin_j_Page_114thm.jpg
573e6c82541e4d425a4d2d09576e599a
f58dd91fd280a2964e50ba210d0d1692e1943c8f
17638 F20101208_AAAJXQ martin_j_Page_101.QC.jpg
7f20bb2911bc13b01146968db319988f
611ed99c16b89a50a4f4726b9bb2e2d0245a57b1
36127 F20101208_AAAIUO martin_j_Page_020.QC.jpg
9431d2e13276156cb78b14b666e04fc5
5ac40c4871d125cf4ffbb33047666cd0beb8e239
F20101208_AAAITZ martin_j_Page_136.tif
7c8bcd1af617a994f4b3c12350668786
0de5b946e2ad0185f96b17c1e4f7f0717b90d7b3
110638 F20101208_AAAJAI martin_j_Page_045.jpg
d54807be516f961a9590751199677889
08534bcec5a02056c5b3d282bd2bdbf0effab3f1
F20101208_AAAIVC martin_j_Page_128.tif
218910271a38117d9764841c744de8a0
8005226dd09b3dc853aeb566dd449fbb08867e31
7967 F20101208_AAAJYF martin_j_Page_115thm.jpg
5db2c4f5c92119fe201db5e1a11803b5
6ea0efcbded8c29da5cc6f3010a7898ee45697ac
8634 F20101208_AAAJXR martin_j_Page_102thm.jpg
ed9027fb960d925a78748473acf2ed77
f062f3412934a097799399d9781454435cb4af1f
92494 F20101208_AAAIUP martin_j_Page_025.jpg
7b09204c6f47e4f9738de7565ab5b270
d03d70e14726723eada47f5774e5d092ccfc07d1
106386 F20101208_AAAJAJ martin_j_Page_046.jpg
fa28a185d1ef6f8413cd2d8914ca2606
f300ed09440d509584b7d395000600127a419a19
115484 F20101208_AAAIVD martin_j_Page_147.jpg
9ca3aa958c0e1fff3a74db9a2dda1c14
59ceac5155fb79d4926d35422a9dfdd3b1320e80
8538 F20101208_AAAJYG martin_j_Page_116thm.jpg
c189feacfb21d2c56e09419dd09a978c
f7b906ed9b69fbacf92c651dc97809898318b8a1
8742 F20101208_AAAJXS martin_j_Page_103thm.jpg
709dc52e8199659fbf6dd76f99f7879b
aa035e193686ea4f52e308f587c6fca5dd7974bb
2095 F20101208_AAAIUQ martin_j_Page_052.txt
080a8588b2f23dce189f96bf887fe98b
3b9d05d517bcb43ed9f18eff27147647e047276f
115267 F20101208_AAAJAK martin_j_Page_050.jpg
bf84c4b4a4636051066ab4b973ecf868
64352930b1858bf516831013834a7f9cd02db985
50429 F20101208_AAAIVE martin_j_Page_151.pro
e3abc479626d2099ed7d175c9725a696
d5ebd6da6d62c672c032b270e02d66b7128781ef
8785 F20101208_AAAJYH martin_j_Page_117thm.jpg
c37064d8761582514c526ff6ba03c0d6
9f1383c6bed5f3864452e852d764d5dad7a932e3
8847 F20101208_AAAJXT martin_j_Page_104thm.jpg
fea930613f870f0bc771becd9287d727
4535357a36c464a1d7e4352e7e83a66a34216689
124964 F20101208_AAAIUR martin_j_Page_065.jp2
9bef7ae9624322d6e273409651a13a74
58e8413d8694cc43d3e3c4b09fe483da61120430
16361 F20101208_AAAJBA martin_j_Page_069.jpg
962f1c0d5875f7e82e4e11d94231cb15
954efe148327ddf099611ad77b961688c498d2a7
109529 F20101208_AAAJAL martin_j_Page_052.jpg
80c92f6e8dab8b84e196fbd062142653
178740003e2f11d352347c8c305958f755afee4f
763430 F20101208_AAAIVF martin_j_Page_073.jp2
37da81e9510fc17f24bff50891f45007
6925bfeb84b938952e6865704a866127e9218c0d
8281 F20101208_AAAJYI martin_j_Page_118thm.jpg
47567e0fd2609cd5e7ae8d1d51578fcb
2b93948b8c524aae1b60c18fcc72336e76ef1552
35689 F20101208_AAAJXU martin_j_Page_104.QC.jpg
119c8bd2ff13f3557f9dbc3cf4da2ffb
9496399a1f159210e172a73fb7dba660272e2b34
14919 F20101208_AAAIUS martin_j_Page_141.QC.jpg
f64e1fdadac37e900039bd39e38f3b9c
c92bde997cfdd7b0ceecb1f58270ab7c5d05c8f5
58177 F20101208_AAAJBB martin_j_Page_072.jpg
8742d8376db40116b893aec2503f20f8
b3effe79de078cf47c65c8caff9d02deb7a143f8
107865 F20101208_AAAJAM martin_j_Page_054.jpg
d84094a1c6f3f96db4e7c47a20542e7c
ff2fab81960d5caff8587483e94a076d60cfb4dd
F20101208_AAAIVG martin_j_Page_162.txt
8a950397e8d44d95ce98b21617dccd43
6319a2aa6100ba9ca3d1f4cd3c0f24acd19c48f5
33616 F20101208_AAAJYJ martin_j_Page_118.QC.jpg
8b4fdc37c7da76f148c5c66fd31e3e4c
0d3eca0ed9ade6bb45e5fe393e0b3f099f3311cf
33170 F20101208_AAAJXV martin_j_Page_105.QC.jpg
e9d1d49c659c69697f794f4dd8e2750a
5c15d5bd38f58651849a8d2fab72f86ceae5f5f4
34175 F20101208_AAAIUT martin_j_Page_103.QC.jpg
8edd0cf7dc1e0f0de838a82f799874aa
9ff0b5166bd5b6e416578c61055a05eb075728ad
60316 F20101208_AAAJBC martin_j_Page_074.jpg
a3d09bd72a81d219bf8453858a35d0f2
1792c8e952031e59cf36c4c537ff61345b539391
103689 F20101208_AAAJAN martin_j_Page_055.jpg
c2c35e8dde4c3a6f51da7a62f4549fb0
07619febcdb312bc25bc611e8d92bb8a8d74f7ba
8125 F20101208_AAAIVH martin_j_Page_057thm.jpg
1e9a10566feb4e4a71209dd89b4adb17
981cf86b93d49ea96464e2635347229c8685b558
7588 F20101208_AAAJYK martin_j_Page_119thm.jpg
ed220e5bd6bc98529d2154cb06e24301
bc94c3e2b828be676a2a99809450f1fc358b214b
28224 F20101208_AAAJXW martin_j_Page_106.QC.jpg
72329b40a793682c4a0f2858dfaa1091
7f014a2ec494c69ad87859340f975d309505877f
1984 F20101208_AAAIUU martin_j_Page_004.txt
a896f7f9bc936f231e739393b773c8a4
1abfcd476175a4a7a4ccb9559cd8f6c6e7784bba
111463 F20101208_AAAJBD martin_j_Page_076.jpg
65927a964efed8faa3c4cbe0a347274e
9d1c304fea1a3c9ff3d42c1c20e7d8bae43c1160
91675 F20101208_AAAJAO martin_j_Page_056.jpg
2f3919bb2bd9e28f015c101a2e2df297
8b4a23c952a3a428cf112ae3e9a733a8f7902c40
34650 F20101208_AAAIVI martin_j_Page_055.QC.jpg
a4b66503d189f33dce903eb23bda248c
91a223ba26f4d7cf13524280df38753bb941dd8a
27221 F20101208_AAAJYL martin_j_Page_119.QC.jpg
f006ee71945596bb860f75e3403ce9c9
74f107c766c0cccfddb5309aebacea9d0da5954f
8276 F20101208_AAAJXX martin_j_Page_108thm.jpg
47710c4e58781fd9c0318aca023e3927
edb4d8c76cacb8ef00dfbe3884b33bb98dfe77a4
2070 F20101208_AAAIUV martin_j_Page_153.txt
cb9526cf265538750a40a9c85680f8f0
2157cb245016b012c87a73befdd93c49a99e194b
102752 F20101208_AAAJBE martin_j_Page_078.jpg
a976ec3dcd2ec6c06ab125793612078a
4e34a00bb660f6a7825f2c44d4339469e992af20
88270 F20101208_AAAJAP martin_j_Page_058.jpg
d3735b4c77722f9e433b3e84e627435a
cf2978c4ce6df1858ae92665d28ad613c4cb9cce
31910 F20101208_AAAIVJ martin_j_Page_023.QC.jpg
f18bb2e07894b89fcd74f6ab7e36cea6
e6447dca552c55b2886df658eae41a9154421784
8609 F20101208_AAAJZA martin_j_Page_132thm.jpg
5b8c9efde99edb1ac98041ed220ad7ca
c02067e967e93f4a97b41dd4c691da3891c70e7f
8693 F20101208_AAAJYM martin_j_Page_120thm.jpg
8978f3f52065c59a2eafd2bd4a9ae617
1f6bf5690047bd1c81c0e4d2bdc145eb001fa467
8659 F20101208_AAAJXY martin_j_Page_109thm.jpg
a43a6b93fba882d0492a64cb64793072
fc691cda38d8b76a888be27fd13f11e31c6ddd2d
34255 F20101208_AAAIUW martin_j_Page_048.QC.jpg
249dbfe73177485e9001b4feaab4a5ba
fe3cdc1ef11a0bee143f05300b820cc47a4336ef
94035 F20101208_AAAJAQ martin_j_Page_059.jpg
9c8dcaa9d58ed835d2a6c4fcd0eb5c2c
fb533fe0ad1077c8f630f3f511936ab06fa3d836
F20101208_AAAIVK martin_j_Page_021.tif
706629e513618b37fecf490c262ba4c5
08b12c94337e438292434d1331bc03b7f9122137
35111 F20101208_AAAJZB martin_j_Page_133.QC.jpg
19e510ea1a3ab9524c6420fe2f2e0cc7
bd49159f8976fd849574ce78a146e7e60108e144
26488 F20101208_AAAJYN martin_j_Page_121.QC.jpg
d56a53c1896aa0a4072e6e6f87bc5d99
3d0081c0f768da14f3c7350d1096ae2576d098aa
35851 F20101208_AAAJXZ martin_j_Page_109.QC.jpg
c14cafeb99fd1967263023aabbc8a798
c21f54775ce8b8675cebfba54b52a919b2786255
112015 F20101208_AAAIUX martin_j_Page_007.pro
29ede9c02cd5b5d9dbb2e9dd67cbcd64
5f91b56d30c0c91713be4d3fd976e615567b1ed6
96425 F20101208_AAAJBF martin_j_Page_079.jpg
8bf44248ff7f953fa4fd73bcf14ffb1e
26120b61da810597012c3a169f339a4e313ef12e
100554 F20101208_AAAJAR martin_j_Page_060.jpg
fc877f5b397a0b1f62955f302d70fd2f
881cfcae26e6d881ec40983263b6af103577da6b
103144 F20101208_AAAIVL martin_j_Page_151.jpg
bd86b9e29ec07ff55318b32cab1e98b2
7f41dcb27b3602c4e3d29d174ec8828e9b987c02
32525 F20101208_AAAJZC martin_j_Page_134.QC.jpg
25134cf0756568935143adaf6d559c0a
21c49d1c2c850f254b55846d62791100782460a7
7371 F20101208_AAAJYO martin_j_Page_122thm.jpg
5a860a33b8323129bad7b413e330a395
dafe11aab4be3a0de79f777e66cfbcbdbbe72e90
82701 F20101208_AAAIUY martin_j_Page_027.jpg
8415c2ed2138e7c5265967c15250b692
3cd24e1b01a7cf8d49ea9df231a4cf028df920f9
108727 F20101208_AAAJBG martin_j_Page_083.jpg
e56cde19b2c82e1139abcd518ef999c9
fa3eeacca24ce7741dd2e09e73eb193abcbef644
40288 F20101208_AAAIWA martin_j_Page_119.pro
c23529caea14964eb4587c9727c34d82
269b5c0e5f93dcaa3be392589dbb321ed23bfbdd
109333 F20101208_AAAJAS martin_j_Page_061.jpg
a24702b242bd73dfddec8e459999511b
6b8ecbe60d7e16e036aa198c89dd6180dd3f8c19
123797 F20101208_AAAIVM martin_j_Page_177.jpg
c09c4fbaa21266f0ccc2db6c89afcfb1
ddccc61f0e0e7a16a9ffa8518df4d4b6949da72d
1806 F20101208_AAAJZD martin_j_Page_135thm.jpg
d7a4d6062af4d315a5504bed0b0580f9
6e2fea9a11545c739fd9b66df77b862ce5bdeaca
30601 F20101208_AAAJYP martin_j_Page_122.QC.jpg
527d2b3d7026f4c19dba72326bb30ea4
178d8103e734d129cb2bf62b7e3e017f44809846
113236 F20101208_AAAJBH martin_j_Page_085.jpg
0f0a86c6f749fda74c5a2ccaaffec312
fbcc03a04d586890ba6d318d0f4b95acd3fb2e00
27765 F20101208_AAAIWB martin_j_Page_024.QC.jpg
cdeb6c09058976bc9fc5114f5f07e90e
71abe8624e2e7189c2dd12be12bebc7cb78ef020
102079 F20101208_AAAJAT martin_j_Page_062.jpg
94efee8a3cac769a4c5a9d9c3fb809ef
72c8bfe3cbc99e767e25a5741d8f4abce0a12989
3865 F20101208_AAAIVN martin_j_Page_161thm.jpg
37d46e2e6440e21d70b2ae9e3fd223c0
381178f64aad2a56d36deb0951f1e02b7aca6044
7976 F20101208_AAAJZE martin_j_Page_136thm.jpg
15697b9dc55cbf171357f93dc70501ba
0b4adebae7a81fb9d977f00fa312540a5f786090
8483 F20101208_AAAJYQ martin_j_Page_124thm.jpg
957833751ce24421ae236099ad4c0fb9
6d72e4a9cf513662265f43fe866a1dc8b552241d
113978 F20101208_AAAIUZ martin_j_Page_160.jp2
b33cc01002536eaf9eefbdda7a85bdc7
97a8b739f9f8f8098693d7edfb7d183bb9ccc55e
87773 F20101208_AAAJBI martin_j_Page_088.jpg
15d8c958e78300196773a1fb4d3b6941
a67704c563c7c2e311a9dd42d895aeef61bb59ea
34745 F20101208_AAAIWC martin_j_Page_132.QC.jpg
d90d1f771d716d5f2146153379cd9afa
285b4ade501533f59e72eaeb46318b79362eb1a4
113249 F20101208_AAAJAU martin_j_Page_063.jpg
bd1b85e13af60ccdb72e0ba2fd9950ea
fa544d2c6fec934526a9d2aff8de3659259ad19a
17558 F20101208_AAAIVO martin_j_Page_069.jp2
0cf639d26320fdcd1bedf4bae74258f5
491ced10969921f30e88316fd7f49b563c8afdda
33273 F20101208_AAAJZF martin_j_Page_136.QC.jpg
e7c4952c888de05a4e0c48a7e65ae0c3
d7cdaaf3baf08f16a95e8871885dc4a36ee2c393
8326 F20101208_AAAJYR martin_j_Page_125thm.jpg
de2a279fbcd64c4b316731996f099490
3de41cd3f37ee1fed2c33860ec6e9dd024a91461
97897 F20101208_AAAJBJ martin_j_Page_089.jpg
558983bdc3ffbe1256efc66ed6b58afa
eca0803201ed8055635195f26d305abec3ba9830
8646 F20101208_AAAIWD martin_j_Page_130thm.jpg
ee29fb6c608e998979a0da07edebb172
7e7b41d7875544e9fa906fbddfb9c57938add963
100879 F20101208_AAAJAV martin_j_Page_064.jpg
d0fbf9439c3c0c3783d9312641197d6d
3e8c3808132fe2ac922b85f4aff662b68603039f
9106 F20101208_AAAIVP martin_j_Page_007thm.jpg
c1a3c9a2e2c99acdf3de7152a7c37112
63b42919e94291423c79a8ab49d1848170808478
6263 F20101208_AAAJZG martin_j_Page_137thm.jpg
6fecd9666c11a84c52904aa9394b2187
de0d6f445e3d34663ab0427da9ea2a8799ed60dd
33308 F20101208_AAAJYS martin_j_Page_125.QC.jpg
78d80eafdff58a0aaea73dfcdd4e482b
76487b75aee75d57933ccdc50539d39270dd5e8b
89657 F20101208_AAAJBK martin_j_Page_091.jpg
f267b07cbf11e2c04c387e4324f6aef7
0c1395845063afab4921923a3c28be29e448252b
106563 F20101208_AAAIWE martin_j_Page_092.jpg
dfce4cb3a8cad89641e441a59920f26e
11feb72da3541e9bd9c51d12ebe2d2f6cab020b8
117390 F20101208_AAAJAW martin_j_Page_065.jpg
b998334119810f5137950c836d593c31
82bfaaa9be943c8632226b5b5e92f58686734278
2619 F20101208_AAAIVQ martin_j_Page_158.txt
6bb0e0e72af15bb4914a02a869667c98
22524823e1f817f18fe76c7eb80ae3767010d534
23765 F20101208_AAAJZH martin_j_Page_137.QC.jpg
2c9ccc0e5bd656460d0db4b1e7f29b64
94cef96b7671e845bf108815ffec5e774dd336fd
35082 F20101208_AAAJYT martin_j_Page_126.QC.jpg
e4c1b19cb0a8240dbdb57072fb4c5040
ea1b214962d2f7dd7ba2333ab2ebf0868ce6e184
112956 F20101208_AAAJBL martin_j_Page_093.jpg
35de20650a3d5573471b37aa652f882d
3519ccd61afcc19fd3b28afb4b20ce0eb33d2e52
109150 F20101208_AAAIWF martin_j_Page_004.jp2
db9f1f028d5e72882915a1e0c194ae5f
ca61cbeef2f8a43a4d6e90a9882a2cca3a013e07
114474 F20101208_AAAJAX martin_j_Page_066.jpg
ffbfef0ff7f8441fc19b176b6e3fcc2b
f3ef1b1b53f7e562966d127ccf66dd0e92b6bfbb
8721 F20101208_AAAIVR martin_j_Page_128thm.jpg
a3c0050fd7c796664bb31095b336816e
05d004f6afd9d7fc97581511cc2d09a3772bff70
120876 F20101208_AAAJCA martin_j_Page_116.jpg
9e66244c1cff4aee18cb29c4645d465c
a7d777b99a607ba049450a7927d19ccdb6908e86
13078 F20101208_AAAJZI martin_j_Page_139.QC.jpg
e9bf4ce1272af08f06c3d7df5462cfb3
77b0a3269b0a2c5ed2fe373320a80f6a7da040e3
8496 F20101208_AAAJYU martin_j_Page_127thm.jpg
a6058454f6e65c778e9e73d4f9524992
ba832944202b49d18e76f1f5af6e22e3c2db2354
111862 F20101208_AAAJBM martin_j_Page_094.jpg
0728158235b82b63f6e008d8821d6ac1
e657e03899c92df7610ab5f4baf2fa4f21224d70
25520 F20101208_AAAIWG martin_j_Page_043.jpg
557525088def52ba2c26819af5ff4df4
746543945eaf126ba036636a0ed446e44670861f
109246 F20101208_AAAJAY martin_j_Page_067.jpg
4721f1a06ad5fc65e9748a7e590aec2b
2509a463312258dd0a589e8fcd6d6d9988574337
F20101208_AAAIVS martin_j_Page_173.tif
e749f7ee937287bddee4ee443de03fa1
ef6e15acbb622cfcd5094aabc8c78aa929031d5f
138471 F20101208_AAAJCB martin_j_Page_117.jpg
dc41f8a8eb7e7d28a8282af2514c87de
963389c7de90145dcf2f854f75946dd0c8fca95c
4667 F20101208_AAAJZJ martin_j_Page_141thm.jpg
4d00a9cc73b28d9e35c3ef6df98cf26c
325fba44511eea7987af273680bdddad627b4339
34512 F20101208_AAAJYV martin_j_Page_127.QC.jpg
414dc66650ab3a80ddfc343f1316d56d
1b212bb10c80aa8b68d54fcee17cb8e0069f4097
142694 F20101208_AAAJBN martin_j_Page_097.jpg
0562f62041d04aa92bf6043e6d8324d8
4404f40ecc8e36c79b49f0a8f74412c56c057fb8
974 F20101208_AAAIWH martin_j_Page_140.txt
a32c616697e4c6a563973239e18f176b
60541ce27478c8e2c69d83fd9d8011ebdb55ca30
116866 F20101208_AAAJAZ martin_j_Page_068.jpg
6615a9a5fbe1179f79cc2cbf954bdc9c
e9ca8246c54e93879af0efcd228c5b45ccc980b8
17664 F20101208_AAAIVT martin_j_Page_042.QC.jpg
ea5a70e332c824a1e42363497902864f
eafb76b2c1d3f30db512426fd1145c20b0512926
F20101208_AAAJCC martin_j_Page_120.jpg
cd161d6b3f8eed31464437aed1687ff2
f17e12c308dfd4239983d5bafcd15c0abf9ae55e
3599 F20101208_AAAJZK martin_j_Page_143thm.jpg
71623a0ff32ea4b9337e94261d0505e2
a50c9fd74526da770d6a3f3cc587680eff404730
35156 F20101208_AAAJYW martin_j_Page_128.QC.jpg
7726ba3aefb0ca417aae287cb246b7a3
d9064c8ba590be8c0823183aedbe8d68091f951e
56715 F20101208_AAAJBO martin_j_Page_099.jpg
0ddf087c85e4a6b50f2ad31bf439b370
d2f23a95b97d95744a09086edc92a7f7295f50d5
2119 F20101208_AAAIWI martin_j_Page_102.txt
c8ca17a591926d5580c39a40314c5e93
c233bf6063ce5989b9bcbe12aacda9153cbda5ad
F20101208_AAAIVU martin_j_Page_070.tif
567f47102118e8a95f998cf8c13304b6
47b3e1b777fbc04e4200b9de82e048e1c1e1f3ae
85015 F20101208_AAAJCD martin_j_Page_121.jpg
be34419bd021c74478a2258246d5ab48
69cdaea927511a1b05bb2cc3543594229293b975
13530 F20101208_AAAJZL martin_j_Page_143.QC.jpg
0990af8f0ad24f8e0ada6cf118de5089
f53d915b19310181eaf14c1f156a3634f1f98da0
8653 F20101208_AAAJYX martin_j_Page_129thm.jpg
c8900a8d7184faf32a8bc52125507291
fb91a97efde155076dfae484148409793226693c
44253 F20101208_AAAJBP martin_j_Page_100.jpg
aa939c2a64cfac8be08fdf46d335d1f0
3e259f710731d6ee1fb34bff795c8c3768219674
F20101208_AAAIWJ martin_j_Page_076.tif
6a88f607f09d6644df5da9157bb17567
a05c4cf4dc51aee198bdfe9fb8e1801a4b5652a5
8641 F20101208_AAAIVV martin_j_Page_033thm.jpg
f4d45f3d883593997ede0a9c275d4fe1
bb9ee9b0e6debe0e42b394587fad96b0580dfb72
85385 F20101208_AAAJCE martin_j_Page_122.jpg
70cd4f7dbd26dee6d8609b9ae41a1758
6d78a7ea46c2f2cdcc9e4e8ce208607889fd55d7
8332 F20101208_AAAJZM martin_j_Page_145thm.jpg
b32238d1d5726eaee48869d1b232251b
9cdc4313afc44581565b0db233eb64936a3a7169
30130 F20101208_AAAJYY martin_j_Page_130.QC.jpg
f654086999a4087be784a7ab5f595b40
c079cb7c82201477949f5ac79633e13e39706042
61956 F20101208_AAAJBQ martin_j_Page_101.jpg
491337a82e23d87b974d922762ca16a6
6b2c96365d306f8ec62f4b9fe570448af4ad5bf7
2278 F20101208_AAAIWK martin_j_Page_160.txt
9e9e3ab88882bf45d9bde89d71859994
3cde619c46b82259239e7dd8fc4d33da173634c4
2184 F20101208_AAAIVW martin_j_Page_035.txt
2fbf6c5ca43833858567fbd40c359439
e4896f24e7afbdce1c35da64f35d9058c71f0dd6
89876 F20101208_AAAJCF martin_j_Page_123.jpg
50236f617d62d7f49c42ba6f44af5c94
39c337d9afd50f893239f0f22aa252cb7c5cee43
8696 F20101208_AAAJZN martin_j_Page_146thm.jpg
81bce633fb11453bee063c1b61986748
e6d2f8ba46de04d2b7c46aa59bb78ee330873fa6
8529 F20101208_AAAJYZ martin_j_Page_131thm.jpg
99c9566d04c4206c99a5109607e4831f
ccbb5ece63a15526774c9bb33c12bd34bba4719c
107521 F20101208_AAAJBR martin_j_Page_102.jpg
b1fbb5a8501c5b242def3b9019069b70
956a5991ec0f8da3834a1934571a8f6d0ad4846d
1995 F20101208_AAAIWL martin_j_Page_125.txt
12036b057122cd734093342b4474d1e4
b0f8f18253bfe52bcc5e5e8dd456e74d2803043f
87786 F20101208_AAAIVX martin_j_Page_165.jpg
5b913a1fd1f06eb6992cc4e8bd015ee6
8847f7ada6bded1a36be49ef3d9843f62b46c94a
8987 F20101208_AAAJZO martin_j_Page_147thm.jpg
9378abc61015dd8566cd0d7b1b1d8ca4
0732b7e30d40e0e1bc78cb7b20d95640b503ab61
17807 F20101208_AAAIXA martin_j_Page_162.QC.jpg
9680753a5c5f89c7f2667a771cbb7622
75450c34d0fdfaa73325986bfb201fdea961d296
98841 F20101208_AAAJBS martin_j_Page_105.jpg
2d5f8067fc0e385f09fec39471821273
077e6a79cc0af024a70cebf3136f0b558d0f3686
979 F20101208_AAAIWM martin_j_Page_003.QC.jpg
f156009369cad2901aef9891edab09ad
8174fbd5915ce686982f89e2963c2920ed542fa8
9101 F20101208_AAAIVY martin_j_Page_168.QC.jpg
1c2defe0b830d80b4d9bf34a082f4b44
7b21844cf8a857163946b914888280d78f8eaeff
99377 F20101208_AAAJCG martin_j_Page_125.jpg
7aed96356510818f378f3620cf984fc0
bc009ae5c710a392c03cc62f301659f5f564f33d
39426 F20101208_AAAJZP martin_j_Page_148.QC.jpg
7f63dc04e8e91b36044e6dc9fb6d0de4
6029d47a62533c36145e221ee730225089c84f11
87656 F20101208_AAAJBT martin_j_Page_106.jpg
c9b8213c0b08d16793cdafd75fdd822f
05fd031e9d76fb5a6565ddb5417102dc53f729c3
22452 F20101208_AAAIWN martin_j_Page_001.jp2
cc9c5729b05b4a9864644e4994c216ae
2b13441928fc70a7a0e383e266b0eb894d7c9d60
1102 F20101208_AAAIVZ martin_j_Page_143.txt
4dfedb931aa7b0e1c8dcf837210d6d51
73fa393a13bd6795d1204eb9b83390120258494e
106836 F20101208_AAAJCH martin_j_Page_126.jpg
6fe4a316863fdb4818bd8006e26c6522
00fc14f73fff478b2a65acd2c32f288b01616466
8427 F20101208_AAAIXB martin_j_Page_011thm.jpg
8cd2257f20e64ad93b651c8889200c73
6df584b55f7a0a47ecd25407886411bcc6b7e5a9
36644 F20101208_AAAJZQ martin_j_Page_149.QC.jpg
25d0639d7dc83040ed8eb21ea3cee1c9
edf7dc31e011d1358df141acc1ae2c69dca1f22b
90119 F20101208_AAAJBU martin_j_Page_107.jpg
3462a0d8e59ea695d2380c7c113f5176
ab7289aca22fa1e593db81f3d36ccc8d837431a6
34721 F20101208_AAAIWO martin_j_Page_129.QC.jpg
9f2359562bed275e35ab3f7deec74c5a
76cb5478640c3152c9261e744f82191a553edc2a
102483 F20101208_AAAJCI martin_j_Page_127.jpg
615b16e6a2b4fe2a80cb948ce62665ad
3618979c286be8b0dd0890ac836b02b21577778c
57237 F20101208_AAAIXC martin_j_Page_164.jp2
32cee904fca4aada5cbfc1b903045e01
0e0b36ec3bb579739aa1d5927f3d569cfe8a9d2e
F20101208_AAAJZR martin_j_Page_150thm.jpg
38b2d5f9c65f1aba68b2fe4b210da022
e39c8658bf79f2b5da2aa16141a8abfbea9f4347
98534 F20101208_AAAJBV martin_j_Page_108.jpg
5f12ea4cfc0912fecefc9a5562090f70
ccdafc2c871300befdc837a40715f6e98c6c297a
1511 F20101208_AAAIWP martin_j_Page_141.txt
4bf6ea0ef53789d82e86b19481fa58d6
1e6870fbca287a3ae00d94d7d97bfe33105135f3
103278 F20101208_AAAJCJ martin_j_Page_129.jpg
6d75ac13542992567fe91e230b2dd6d1
6467f9684dd1c4526a1807a5af9f56063f7e31ed
F20101208_AAAIXD martin_j_Page_101.tif
a3b822f6d35f186acf6343d5013ea720
70147ab81a60ec0df9f44b4c0bd0f3013262cbd2
8253 F20101208_AAAJZS martin_j_Page_152thm.jpg
3882a1b2cca13c0fdc888ab244fc9f20
b3bb3ceccca351791c0f106de64e35fa29295c65
106544 F20101208_AAAJBW martin_j_Page_109.jpg
c7991fd9baf8c4ca38892972d0f094da
2fb765cf430d726b04f9ac440d1dad020a55d61c
54006 F20101208_AAAIWQ martin_j_Page_067.pro
cc2815210d5a5078302a7f2ab863f024
9e9e35bb079bf5d4cdcf5fbacab6429bd435d258
102064 F20101208_AAAJCK martin_j_Page_130.jpg
064354978d27e08839308451547a17db
bd1adfe0075651fa09d0e2f0b6fdb39946d9f070
54742 F20101208_AAAIXE martin_j_Page_020.pro
70d87b007c0c89d9a4bdcb1768c9a0ff
f72bdbe150df7a69cfdd16fff9b08fb8fef30f7f
33238 F20101208_AAAJZT martin_j_Page_152.QC.jpg
f141360a060cf523f82f978db17b6aab
91a7d09c8b77e3c18a5e7c9b310f677b7e37016a
108929 F20101208_AAAJBX martin_j_Page_110.jpg
2ba4d9a74f4d788b149eb4d831865534
5db44d0cf6b7777ac85c525ddb6baa179da02cbc
F20101208_AAAIWR martin_j_Page_083.tif
dbda75f66dd491b235b26ce21f800fad
e6de9c25a6af714bdc964fef0f783b7afcbc4b71
62434 F20101208_AAAJDA martin_j_Page_167.jpg
676d20223b6af36083a692f979785fa9
666837301fe3e1265dad4b5df9297e6bea0c5fef
102888 F20101208_AAAJCL martin_j_Page_131.jpg
cc92dff918a5ae34c48bd4813cf9c0e5
a048d262274e0571520fe7c1f10ba2132e35102f
6661 F20101208_AAAIXF martin_j_Page_069.pro
62f3a63f3000e7e0bd5dc07e5436cb63
f89dd2ddf8305ed297f0289f391fe640f30c11ce
2691 F20101208_AAAJZU martin_j_Page_154thm.jpg
a974804bfd44839ed7c30229aa4a7891
23a2e4c9a9c1e36557857268d4f52f4f5083925c
78384 F20101208_AAAJBY martin_j_Page_112.jpg
db1e8329abd3f5dd8c53bdb84e114365
7df05539909041291666ca52f08e71cc706e6110
113156 F20101208_AAAIWS martin_j_Page_081.jpg
5c575094c781475525e6f577be716d2b
8defb73ec0f59dcc30f4c89735e1346ea6efa50e
82117 F20101208_AAAJDB martin_j_Page_169.jpg
f22aa7e694c84e5f342ece0b25d1f009
b0e2c6b4ca06d291e7085b75c92f858f7c266554
40513 F20101208_AAAJCM martin_j_Page_139.jpg
d7652a6bd3236fabcb6175f833b6cc18
1df3d50c69135e723c19803e37b8cef833cb768c
288873 F20101208_AAAIXG martin_j_Page_168.jp2
8e6298cc0848cde35166becbc459359e
66d41f2e107ec61af9c128786d66d30fdcc9b656
10560 F20101208_AAAJZV martin_j_Page_154.QC.jpg
45222a518367aed4a727cc35ac8a3951
0888185df76dee073d531ba1b06158a021d29d94
97966 F20101208_AAAJBZ martin_j_Page_114.jpg
b9784f2edd61a73e53ceca402e35ba81
531570f3fb7fe5223ef9bc5e9ce126af6d1fc5cf
120621 F20101208_AAAIWT martin_j_Page_178.jpg
e00f083f7a9f37e73c62c4f02c265019
30817f38e65adee4b69d732462a283625001100f
93179 F20101208_AAAJDC martin_j_Page_170.jpg
03bd2bc9814a06861b0f56bd3fb2e9f1
a3d206b03f7b6198b7fa7632ec2fc786d00726dc
47139 F20101208_AAAJCN martin_j_Page_143.jpg
0a832bbb42455d7097a63aa628f20e1a
567b8ff65dc878383e141c93be2b9d2edda03a79
29570 F20101208_AAAIXH martin_j_Page_028.QC.jpg
a12a8c3b942290d878a27947d52c1acd
60d007006589acf9dc3ea3756c573d23acd4805f
6941 F20101208_AAAJZW martin_j_Page_155thm.jpg
494bc3a203b3bfaa76765c8bac5bfaf8
3e7fa137deb74d85ac25e5fd59c09b3fd996e019
7605 F20101208_AAAIWU martin_j_Page_121thm.jpg
a1fb308dae413bbed965ead128818161
0df3bb7b33cb322bb2c2a41226d43ae917e5431d
14099 F20101208_AAAJDD martin_j_Page_171.jpg
81508c19dcee31647b199ea6deda047d
0ce6f238f6371f6b3d4a2f9d59ecf9b075932b3f
102912 F20101208_AAAJCO martin_j_Page_145.jpg
15c87dcea7c05d7c2ebe87f3394aeafb
4571710c97e0ee7c935e3f72266639fc45f434ea
4019 F20101208_AAAIXI martin_j_Page_144thm.jpg
c619949a3822105d1029d534ba64e07b
a9b994609092c7d1e882087561ae8e03cfd839aa
27385 F20101208_AAAJZX martin_j_Page_155.QC.jpg
06a4a3f0b7035a3462fbc5ad6001ac75
6bff9bff3b882585931fec2e5d6e3cd094cce65b
28187 F20101208_AAAIWV martin_j_Page_029.QC.jpg
e0cee6cc7d556f686ac8f909da8bf15e
2f3b48850d657ec172962f0561a1068c99a8989e
121574 F20101208_AAAJDE martin_j_Page_173.jpg
ac70c84a987bcb2e4acc583dd357154d
61d9d4be4d1aa22a3690d286b3fb3c9d779226a3
106260 F20101208_AAAJCP martin_j_Page_146.jpg
55004b30daab08f3f1d8080c8eb51da0
09d904d268ec75ab4099304de8d1a7402c4df77a
103862 F20101208_AAAIXJ martin_j_Page_103.jpg
d972a48696710ff7097e866f3851b38e
fa1981ac068ce1615047cc2911de506676a4c90a
4026 F20101208_AAAJZY martin_j_Page_156thm.jpg
950a8796a69f52f0703c32ffb2ce82b6
4d811a30178cf6e5dd7cc4c9131a783a5a184714
8442 F20101208_AAAIWW martin_j_Page_175thm.jpg
714c4c8fdbf8bffa90e71eafc1f87655
28ff057a87bd0b7ef08d11d02f15476e7e3e0881
121034 F20101208_AAAJDF martin_j_Page_175.jpg
04341f84ce2122ddb5bc23ee6d1e977c
5360a2199cdaa3044cf1b94f064a79d43b34ebc0
119678 F20101208_AAAJCQ martin_j_Page_148.jpg
a5587f76d1eee0f1167c8b02a79f6cab
97ad078d9607645092101bb33eafc3565cc0a3e6
35785 F20101208_AAAIXK martin_j_Page_116.QC.jpg
96bd3b9f685d82c9b225195509806506
a2233a5f66f4af2b3119a81ef72edbf5f917881b
13146 F20101208_AAAJZZ martin_j_Page_156.QC.jpg
3cd88f06ad57a4121fe7c783bcf99962
6bb6db9759807ab149d30e68e52254413eb0272b
F20101208_AAAIWX martin_j_Page_012.tif
ec1585c7ab6c02034ac404c44c8b108e
844ab0dfb459330d64d8b52806c3779e7230e7c8
133567 F20101208_AAAJDG martin_j_Page_176.jpg
9a3b537769cc41a2e3fd951e9634a7c0
5e33710d2737de92d4cc0a596505f8518a564857
108998 F20101208_AAAJCR martin_j_Page_150.jpg
cd877c46d14c3e9395956cd4d373b8c0
a8a8b2d36d0fa92a009d3161e857a0a0ef5c5ff3
F20101208_AAAIXL martin_j_Page_020.tif
ea8cb1172fe69865075a706ca98ebd13
fbaa639ec3974813e046fa85b1ef6f074bd01cc2
49955 F20101208_AAAIWY martin_j_Page_144.jpg
06d29daf5a6f935f1e8a338950fefe7d
5fbbd4501a8d1b6ebbd1aee510cc861d471b08f2
124249 F20101208_AAAIYA martin_j_Page_068.jp2
15fa03cec4974a26f20da60200d95cae
fb1df8fb391e7e3eb0c670328e69f9623dd59a28
106479 F20101208_AAAJCS martin_j_Page_153.jpg
a3a052ea36cbdf1ea5fd9cfe9dd8fd7a
d6571039de26df1cf49782dde29fe064068bda89
13124 F20101208_AAAIXM martin_j_Page_100.QC.jpg
6f1625964b92b83b2d5036d60e547163
cc8b07ddc9cd23210a9d939d2701c0e576d0fe44
44833 F20101208_AAAIWZ martin_j_Page_091.pro
2f8a864bc29b4664e0ac87bf9400205d
7c781437778618aa36bdf284c8eadfd72386f89b
133461 F20101208_AAAJDH martin_j_Page_179.jpg
6340f9650664012b78fa746f4413ea45
0abf2756856e7b6985ed35211b25fc42765d288a
45189 F20101208_AAAIYB martin_j_Page_142.jp2
3441a784114c53584fd3dc05b17dade7
eb8762ea23c09fd1721fb7aa7972b247bb2518ca
31104 F20101208_AAAJCT martin_j_Page_154.jpg
1fb87fa2cf369fb41dc7b9a780835b4e
f89ece3afc6208dbad7c80af2b780bcfb8916552
17464 F20101208_AAAIXN martin_j_Page_156.pro
35c84c9688208aa22b57b257c482ad0a
3935e01ab116f4cb9cc77c80344f915c71f3288f
4809 F20101208_AAAJDI martin_j_Page_002.jp2
6de18ed75cb4100666dd8d52df9c86e9
5db8a8dbdfb61a268aa9d38e3feecc61e606343c
86710 F20101208_AAAIYC martin_j_Page_111.jp2
b10f5594a58de3cd57026aa7e0bf51ff
4365f4c7f4f445af7376ccdf81e7a6f7da0eb8e5
86987 F20101208_AAAJCU martin_j_Page_155.jpg
d5382bcab6e1d6512165b25edbde4b0c
bb4f6159340795f3db2fe30bf0bfff937e5539b7
22451 F20101208_AAAIXO martin_j_Page_142.pro
af8391fef239f8e33f312a93e4fcb77b
acde68044834f9eaca406256aa919b4d212c06f2
4343 F20101208_AAAJDJ martin_j_Page_003.jp2
97a3829982dd9d768711f62507164177
e6ff8d3ecd5e411d230eb062e965e765d8e06f14
51531 F20101208_AAAIYD martin_j_Page_061.pro
c27e0944640a3a7702373592d2509230
106bbcc52ff9f013a95795c4eb67d6a2defe2cba
56438 F20101208_AAAJCV martin_j_Page_158.jpg
64cae725f9f49854114c22b62a253d17
986bf5ad3fe4188cc87f46af6213110c20b6c99d
9112 F20101208_AAAIXP martin_j_Page_065thm.jpg
2fe8e41bc49ad870340e494de2883150
d3f8e03e92878a5dcb9971c2411fe71949935798
88081 F20101208_AAAJDK martin_j_Page_005.jp2
be9c7e4283e77aed1a50071afd196823
8b11ccb51514f43c8812c3b85d31a2320630a036
118062 F20101208_AAAIYE martin_j_Page_136.jpg
0b0049b330129ac09d882ccc60f43727
7a4bc5cda43a3a40230c9174f142deae37e8543e
104660 F20101208_AAAJCW martin_j_Page_160.jpg
677a9ce2b33acb53769b75aa871af2c5
7c5909b07f72dbc054a9cd0ffc92ad732832b02d
35391 F20101208_AAAIXQ martin_j_Page_098.jpg
bc9a10d2374964bec8b15d49839c9cf7
b3a39e5b1bc12fa2b95bc4fdb02653dbd35c581b
1051968 F20101208_AAAJDL martin_j_Page_006.jp2
d8e81b0ea818c6d8c4ef47584da572e2
1071dc13714b0f6e627c39f4a3c15851d48419d6
F20101208_AAAIYF martin_j_Page_005.tif
bc2bfc9c8b27dfe5ce23aaef3052e55c
167672c3f76d2fc277864115f18fa1ef09f4e534
62263 F20101208_AAAJCX martin_j_Page_162.jpg
c1d72a31c512ccb0cd6cf4d7a73931f9
0334ec4dcb1817aeb13a4f594f8955915a1bbf1f
4588 F20101208_AAAIXR martin_j_Page_101thm.jpg
a9ee952b778cfe802d887f3b6e7fbcd0
ef83fee76b3976d094c641ed5730872482c4eb29
115612 F20101208_AAAJEA martin_j_Page_026.jp2
63c8fe65070cec47313162344e815558
2ebdcd8d8ecbcb16053f5a130a09e5077986dde6
1051969 F20101208_AAAJDM martin_j_Page_007.jp2
29b78cb525ac3d43308fd509f3572d5f
4f25526b2831650c72de30c951ee2b28c2924ca6
103021 F20101208_AAAIYG martin_j_Page_090.jpg
ea929df0d935fb6aa5e3a5e2553e147f
ecd1c97cc9ca120d1a08a42f712688d5ffc1718d
60251 F20101208_AAAJCY martin_j_Page_164.jpg
9729782b9625f08146749b22556f2815
e891bc6a6afa14e1d15b026a26651b11d1f35808
61787 F20101208_AAAIXS martin_j_Page_172.pro
890c6ad84a8f4457324367e67e13f3fe
dc825c61bf130415fe1a025c160bd7a66b457840
87157 F20101208_AAAJEB martin_j_Page_027.jp2
6f083722365b9cbaa5793ae74d7deffe
4a949b45ac89c807d8ffeb5296276b0ce70b953b
1051973 F20101208_AAAJDN martin_j_Page_008.jp2
efb0e9dc3dac839c9a4fcc6af611720e
7f0a7092643fc3a2e4af4ade28f47542460082e7
114177 F20101208_AAAIYH martin_j_Page_017.jp2
c8d620ea6d128a729565fef695ea37fe
44697f185ecd501f6446bee0ff54f3f1ce85d9b0
43001 F20101208_AAAJCZ martin_j_Page_166.jpg
6d1eb20db8b7cb6ba0d4d194ea06c73e
dcddb58065070b3d581159a27af1aec026c95d87
581794 F20101208_AAAIXT martin_j_Page_010.jp2
ef281c65a3b4650b1b58dfb412c48bd8
aa0ddcc1011ac1eefe81913003c33594c183b9e9
108757 F20101208_AAAJEC martin_j_Page_030.jp2
dc006dd30965ce9959d4d9ee18b576cb
c12f5e130c18e19fde86fd1402767e5067396683
1051921 F20101208_AAAJDO martin_j_Page_009.jp2
5b292621b2aa4beaf652bcf4cbdc6a55
fbd27d2469edf8adafa952f79a07832b222dd514
706854 F20101208_AAAIYI martin_j_Page_101.jp2
ccc24a9a3ecebdc7768e33f939198f42
5dc32bff09fd406743e14318536003262f6c91d1
30399 F20101208_AAAIXU martin_j_Page_159.jpg
d4fda7cae8763ae3f54f82beab1d9e08
4e9f4061efff57804b71c1d3187d973c9d431f87
115803 F20101208_AAAJED martin_j_Page_031.jp2
4197af92dc384a8b6317069906eac1b2
27d8b92e86d47bfe3f8f37164e5393ffcc36b8f5
1051922 F20101208_AAAJDP martin_j_Page_011.jp2
785dee2cb074fc3274e039418d97c3fc
43bad83d3da411f26d3047f3a3b8527f4898119f
13657 F20101208_AAAIYJ martin_j_Page_140.QC.jpg
1945b91f030d51b150e21e266c430c35
3ee8557f081feb8c88e927b505cf8fa31d3845d1
49410 F20101208_AAAIXV martin_j_Page_105.pro
000753c893311cf7c7266dc54402bef7
20443a0c4daa3cefcc99f4fab5aff01a97fcb850
116437 F20101208_AAAJEE martin_j_Page_033.jp2
564ab06fe050a3270c5f9e5e6a155825
24a1a346256cfbb638bf9301e95e92d618932c2b
1051981 F20101208_AAAJDQ martin_j_Page_012.jp2
4d11e287d0c55d91d3c1735b8c1e9c09
f5ce1f5ab4393c2b71bdcb8d0b1fba81cfb4e700
8849 F20101208_AAAIYK martin_j_Page_054thm.jpg
80b9527c4508189ad092e1e18ec5af23
3734c24260d0c8207113cbe47d4e2b8be577916b
34057 F20101208_AAAIXW martin_j_Page_170.QC.jpg
4184aa6d9c01d37f9d2e74cbab85504c
bbbda7717f336764a0a78b3766abf7743a6d3c2b
116969 F20101208_AAAJEF martin_j_Page_034.jp2
a8c3bbcb058a82ee11bb160eb600b9a2
3195c1daa7e7c86f7e3af460c5f5300c4c53509c
102980 F20101208_AAAJDR martin_j_Page_013.jp2
40f366a552b415cc2b880a2e18197dfb
5ba44019877b0fe5a3c193a7dc7726f5315be385
7949 F20101208_AAAIYL martin_j_Page_123thm.jpg
2ee5cb68d86f85d3576c77b17c26c57f
4b0e3f5911ab093aa5d33f004e734494d0db356c
54133 F20101208_AAAIXX martin_j_Page_118.pro
fb471bdab03a1a7582f433baad165041
7a1f2a68dd4145a169b1dd375f844407e76041f0
113060 F20101208_AAAJEG martin_j_Page_036.jp2
2addbfb77f14844b63aa1cc527ed63d4
96b8c50595b5b3cd37709947657d97e8b29de301
33914 F20101208_AAAIZA martin_j_Page_151.QC.jpg
fedee4913b61e126d7f584163da7e401
e8c186c6d6c6f55e4190ea48e8c2c8764828929d
51189 F20101208_AAAJDS martin_j_Page_014.jp2
d4ed103d3ec5570221ffdb14d776acc6
7960029909ca8703282264f39f23784d7935e4af
48918 F20101208_AAAIYM martin_j_Page_141.jpg
ccc230c3b703f20dec6b14ae507b7a51
8ace711001f5d28dcd4f2f765e6bce6db3347de2
F20101208_AAAIXY martin_j_Page_103.tif
bf3aea5ee36da2613fbe6fa216966718
872e4083e9146e879b841cdee23c8ce2c6f325cd
112907 F20101208_AAAJEH martin_j_Page_037.jp2
dbbb2b54d7a6434030845040b0aa11d7
be134eeb4232119f13a12b0170346f76ab90274c
106580 F20101208_AAAIZB martin_j_Page_053.jpg
4164bd034512ddd1217e826d19a7db8f
a1a1722076d8936031892d5b69b0525b44692b92
112913 F20101208_AAAJDT martin_j_Page_015.jp2
1b280ada2fa5aa6d0486b1eb2d49a615
b8fe25cc64410ed89608d6d14470a62238e0bd89
2017 F20101208_AAAIYN martin_j_Page_055.txt
7072b741512c7bfde13af1a0e098d416
8499931b435d4f7d38fb2a6a29340e1fd6018a5c
26032 F20101208_AAAIXZ martin_j_Page_073.pro
4b1eb2e91bd8f3fcf18cb8d0bceffd3c
f3382a84b5f4504ef1e9d40c67eea7ecacfa904d



PAGE 1

1 POPULATION ECOLOGY AND CONS ERVATION OF THE SNAIL KITE By JULIEN MARTIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Julien Martin

PAGE 3

3 A Mercedes

PAGE 4

4 ACKNOWLEDGMENTS I am greatly indebted to my committee me mbers: Wiley Kitchens, Jim Nichols, Don DeAngelis, Peter Frederick, and Ben Bolker. Wiley Kitchens has been an exceptional supervisor and mentor. His unwavering support and countless hours of expertise helped me grow as an ecologist and contributed to the su ccess of my research. I also ow e special thanks to Jim Nichols who had a major influence on the project. He help ed me think more critically about my research and assisted with many of the statistical anal yses. Don DeAngelis provi ded excellent input and valuable ideas. I am thankful to Peter Fred erick for many useful discussions about water management and bird populations in the Evergl ades. Ben Bolker provided helpful statistical advice. In addition to the input from my committee me mbers, I also greatly benefited from the assistance of other scientists: Jim Hines, Wolf Mooij, Arpat Ozgul, MadanOli, Jeff Hostetler, Hardin Waddle, Vicky Dreitz, Rob Bennetts, Ga ry White, Roger Pradel, Christopher Cattau, Andrea Bowling, Zack Welch, Eric Powers, Christ a Zweig, Sarah Haas, Steve Beissinger, Bret Sandercock, Therese Donovan, Nicholas Schtick zelle, Tom Contreras, Alejandro Paredes, Thomas Cornullier, Maynard Hiss, Brad Stit h, Phil Darby, Steve Daniels and Fred Johnson. Statistical analyses related to capture-mark-reca pture would not have be en possible without the help of Jim Hines. This dissertation could never have been co mpleted without the great vision of the individuals who initiated and de signed the Snail Kite monitori ng program: Rob Bennetts, Wiley Kitchens, Vicky Dreitz, and Jim Nichols. Additionally, in the ear ly stage of my research, Rob Bennetts spent many hours helping me become familiar with kites and the huge Snail Kite database. During my first field season, he repla ced me in the field for several months which

PAGE 5

5 allowed me to attend an important course on cap ture-mark-recapture techniques offered by Gary White and David Anderson. I would like to thank all the fi eld biologists involved with data collection during the last 5 years: Christopher Cattau, Andrea Bowling, Samant ha Musgrave, Sara Stocco, Brian Reichert, Melinda Conners, Danny Huser, Derek Piotrowi cz, Christina Rich, Michaela Speirs, Jamie Dubeirstein, Janell Brush, Zack Welch, Jeff King scott, Freddy Martin, and Paul Pouzergues. I appreciate the help and cooperation of the following biologists and managers who invested time and effort to promoting Snail Ki te recovery: Tylan Dean, Debbie Pierce, Steve Miller, Joe Benedict, Susan Sylvester and Jim Rodgers. Financial support was provided by the U.S. Army Corps of Engineers, the U.S. Fish and Wildlife Service, the St. Johns River Water Mana gement District, the Florida Fish and Wildlife Conservation Commission, and the U.S. Geological Survey. Employees of the Florida Cooperative Fish and Wildlife Research Unit were especially helpful with the administration of this study, particularly Donna Roberts, Franklin Percival, Barbara Fesler and Debra Hughes. Lois Wilcox, Anne Taylor, and El len Main helped with editing. Finally, I would like to thank members of my family and friends who have been a great support throughout these years: Me rcedes, Christophe, Jose, the Martin-Pouzergues family, the McSweeneys, the Glogowskis, the Kitchens, Jacotte, Emilie, Bruno, and Sarah.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ........11 LIST OF FIGURES................................................................................................................ .......12 ABSTRACT....................................................................................................................... ............13 CHAPTER 1 INTRODUCTION..................................................................................................................15 Why Study the Snail Kite?.....................................................................................................15 Background Information on Snail Kites.................................................................................16 Objectives and Outline......................................................................................................... ..18 2 IMPORTANCE OF WELL-DESIGNED MONITORING PROGRAMS FOR THE CONSERVATION OF ENDANGERED SPEC IES: CASE STUDY OF THE SNAIL KITE........................................................................................................................... ............19 Introduction................................................................................................................... ..........19 Methods........................................................................................................................ ..........23 Study Area..................................................................................................................... ..23 Sampling Methods...........................................................................................................23 Marking protocol......................................................................................................23 Population survey protocol.......................................................................................23 Analysis....................................................................................................................... ....24 Superpopulation size estimates of adults.................................................................24 Estimation of population growth rate.......................................................................26 Estimation of the number of young produced..........................................................27 Detection probabilities for num ber of young produced every year.........................27 Count data................................................................................................................28 Average number of kites and gr owth rate based on count data................................28 Detection probabilities for FC and MC....................................................................28 Estimates of Precision and Magnitude of the Difference between Estimates.................29 Results........................................................................................................................ .............30 Population Size and Average Population Growth Rate...................................................30 Average Number of Kites Before and Af ter Decline Based on the Superpopulation Approach......................................................................................................................31 Average Number of Kites Before and After Decline Based on Count Data...................31 Average Number of Kites and Growth Rate Based on Count Data................................31 Number of Young............................................................................................................32 Detection Probabilities for FC and MC...........................................................................32 Discussion..................................................................................................................... ..........32

PAGE 7

7 Population Decline..........................................................................................................32 Problems Associated with Counts and Implications for Recovery Plans........................34 Importance of Monitoring to Diagnose Causes of Decline.............................................37 Conclusion..................................................................................................................... .........38 3 MULTISCALE PATTERNS OF MOVEME NT IN FRAGMENTED LANDSCAPES AND CONSEQUENCES ON DEMOGRAPHY OF THE SNAIL KITE IN FLORIDA......44 Introduction................................................................................................................... ..........44 Hypotheses and Predictions.............................................................................................47 Methods........................................................................................................................ ..........49 Study Area..................................................................................................................... ..49 Criteria for Determining the Regional Impact of the 2001 Drought...............................49 Statistical Models to Estimate Movement and Survival..................................................50 Field Methods for the Study of Movement on a Monthly Scale.....................................51 Statistical Methods to Estimate M ovement on a Monthly Scale Using Radiotelemetry.............................................................................................................51 Estimating monthly movement among regions........................................................51 Estimating monthly movement within regions using radio-telemetry.....................52 Field Methods for the Study of Movement and Survival on an Annual Scale................53 Statistical Methods to Estimate Annual Movement and Survival Using Banding Data........................................................................................................................... ...53 Estimating survival...................................................................................................53 Estimating annual movement probabilitie s among regions using banding data......54 Goodness of fit.........................................................................................................55 Model Selection Procedure..............................................................................................55 Effect of Patch Size and Distance on Movement............................................................56 Effect Size.................................................................................................................... ...56 Estimates of Precision.....................................................................................................56 Results........................................................................................................................ .............57 Monthly Movement Probabilities Among Regions.........................................................57 Effects of patch size and distance............................................................................57 Monthly Movement Probabilities Within Regions..........................................................58 Movement within the Everglades region.................................................................58 Movement within the K region................................................................................59 Comparison Among and Within Regions........................................................................59 Interannual Survival Estimates........................................................................................60 Inter-Annual Movement Among Regions and Drought Effect on Movement................61 Discussion..................................................................................................................... ..........62 Monthly Movement Among Contiguous and Isolated Wetlands....................................62 Patch Size and Distance Between Patc hes as Factors Driving Movement.....................63 Inter-annual Pattern of Movement...................................................................................64 Regional Survival and Resistance of th e Population to Natural Disturbance.................65 Conclusions and Conservation Implications..........................................................................68 4 NATAL LOCATION INFLUENCES MOVE MENT AND SURVIVAL OF THE SNAIL KITE..................................................................................................................... .....75

PAGE 8

8 Introduction................................................................................................................... ..........75 Hypotheses and Predictions.............................................................................................77 Study Area..................................................................................................................... .........78 Material And Methods........................................................................................................... .79 Field Methods..................................................................................................................79 Capture-mark-recapture...........................................................................................79 Data Analysis.................................................................................................................. .79 Multistate modeling..................................................................................................79 Movement.................................................................................................................80 Survival....................................................................................................................82 Model Selection, Goodness of Fit and Program Used....................................................85 Effect Size.................................................................................................................... ...86 Notes Concerning Regional Specific Survival................................................................86 Wetland Conditions.........................................................................................................87 Results........................................................................................................................ .............88 GOF Tests...................................................................................................................... ..88 Movement....................................................................................................................... .88 Comparison of Natal Philopatry a nd Philopatry to Non-Natal Site................................89 Survival....................................................................................................................... .....89 Adults.......................................................................................................................89 Juveniles...................................................................................................................91 Detection Probabilities....................................................................................................91 Discussion..................................................................................................................... ..........91 Effect of Natal Region on Movement.............................................................................91 Influence of Natal Region on Survival............................................................................93 Conclusions and conservation implications............................................................................95 5 EXPLORING THE EFFECTS OF NATU RAL DISTURBANCES AND HABITAT DEGRADATION ON THE VIABIL ITY OF THE SNAIL KITE.......................................102 Introduction................................................................................................................... ........102 Objectives..................................................................................................................... .105 Methods........................................................................................................................ ........106 Study Area.....................................................................................................................106 Life Cycle..................................................................................................................... .106 Hydrological Conditions...............................................................................................107 Data Source and Estimates of Vital Rates.....................................................................108 Survival rates..........................................................................................................108 Fecundity rates.......................................................................................................111 Detection of juveniles.............................................................................................113 Matrix Analyses.............................................................................................................113 1, damping ratio, sensitivity and elastic ity analysis (Objectives 1 and 2)............113 Before versus after effect on 1 and Life Table Response Experiment (Objectives 3 and 4)............................................................................................114 Stochastic population growth ra te (objectives 1, 4, 5 and 6).................................115 Viability of Snail Kites under cu rrent conditions (Objective 1)............................116 Evaluation of Hypothesis 1: Reduction of s after 1998 (Objective 5)..................116

PAGE 9

9 Evaluation of Hypothesis 2: Increase in drying event frequency reduces s (Objective 6).......................................................................................................117 Evaluation of Hypothesis 3: Increase in frequency of drying events and before versus after effect (Objective 7)...........................................................117 Probability of Quasi-Extinction (Objective 1)..............................................................118 Validation..................................................................................................................... .119 Results........................................................................................................................ ...........119 Survival estimates..........................................................................................................119 Probability of Quasi-extinction (Objective 1)...............................................................122 1, sensitivity and Elasticity an alysis (Objectives 1 and 2)...........................................122 Life Table Response Experiment (Objective 3)............................................................122 Stochastic population growth rate s.............................................................................123 Viability of Snail Kites under cu rrent conditions (Objective 1)............................123 Evaluation of Hypothesis 1: Reduction of s after 1998 (Objective 5)..................123 Evaluation of Hypothesis 2: Increase in drying event frequency reduces s (Objective 6).......................................................................................................124 Evaluation of Hypothesis 3: Increase in frequency of drying events and before versus after effect (Objective 7)........................................................................125 Discussion..................................................................................................................... ........125 Snail Kite Viability and Key Vital Rates......................................................................125 Hypothesis 1: Before versus Af ter 1998 Effect (Objective 4))..................................127 Hypothesis 2: Increase in Frequency of Moderate Drying Events (Objective 6)..........128 Hypothesis 3: Increase in Moderate Dryi ng Events and Before versus After 1998 effect (Objective 7)....................................................................................................129 Limits of the Models.....................................................................................................129 Management Implications....................................................................................................132 6 CONCLUSION.....................................................................................................................145 Synthesis of Research Findings............................................................................................145 Monitoring of the Snail Kite..........................................................................................145 New information on movement and demography of the Snail Kite..............................147 Current Status of the Snail Kite and Management Implications...................................148 Perspectives................................................................................................................... .......150 Rescuing the Snail Kite Population from Ex tinctions Risks Associated with Small Populations.................................................................................................................150 Habitat Management Models........................................................................................152 Everkite..................................................................................................................152 Adaptive management............................................................................................152 APPENDIX A SURVEY-SPECIFIC PARAMETER ES TIMATES USED TO COMPUTE ESTIMATES OF SUPERPOPULATION SIZE..................................................................155 B CONFIDENCE INTERVALS, MODEL S ELECTION TABLES AND MOVEMENT ESTIMATES...................................................................................................................... ..156

PAGE 10

10 C SELECTION OF MODELS USED TO ASSESS THE EFFECT OF NATAL LOCATION ON MOVEME NT AND SURVIVAL............................................................163 D CLUSTERING ANALYSIS.................................................................................................165 E ESTIMATION OF Q(T).......................................................................................................169 F DETECTION OF JUVENILES............................................................................................170 LIST OF REFERENCES.............................................................................................................172 BIOGRAPHICAL SKETCH.......................................................................................................181

PAGE 11

11 LIST OF TABLES Table page 2-1 Estimates of Snail Kites annual population growth rates; and 3-year running average of growth rates................................................................................................................ ...39 2-2 Estimates of detection probability of Snail Kites..............................................................40 3-1 Multistate models of monthly movement probabilities of Snail Kites among five major regions based on radio-telemetry data.....................................................................70 3-2 Multistate models of monthly movement probabilities of Snail Kites among wetlands in the E and K region based on radio-telemetry data.........................................................71 3-3 Multistate models of annual apparent survival and movement probabilities....................72 4-1 Multistate models of appa rent survival and annual tran sition probabilities in relation to the place of birth.......................................................................................................... ..97 5-1 Cormack-Jolly-Seber mode ls of apparent survival..........................................................136 A-1 Survey-specific parameter estimates used to compute estimates of superpopulation size for Snail Kites between 2001 and 2005....................................................................155 B-1 Multistate models of monthly movement probabilities of Snail Kites among the 5 major regional patches in Florida based on radio-telemetry data (models with delta AIC > 15)...................................................................................................................... ...157 B-2 Multistate models of monthly movement probabilities among wetlands in the E and K regions based on radio-telemetry da ta (models with delta AIC > 9)...........................158 B-3 Multistate models of annual survival and annual movement probabilities (models with delta AIC > 11)........................................................................................................160 B-4 Annual movement estimates between the 4 major regions used by the Snail Kite (E, O, K and J) during normal and drought years..................................................................162 C-1 Multistate models of appa rent survival and annual tran sition probabilities in relation to the place of birth (models with delta AIC > 12)..........................................................163 D-1 Five hydrological variables used to conduct the hierarch ical clustering analysis...........167

PAGE 12

12 LIST OF FIGURES Figure page 2-1 Map of the wetlands that were sampled to obtain both counts and capture-resighting information of Snail Kite for the estimation of population size........................................41 2-2 Comparison of the estimates of popul ation size of Snail Kites (using the superpopulation approach) with annual counts..................................................................42 2-3 Number of young (i.e., nest lings close to fledging) Sna il Kites marked every year from 1992 to 2005..............................................................................................................43 3-1 Major wetlands used by the Snail Kite in Florida..............................................................73 3-2 Apparent survival between 1992 and 2003 of adult and juvenile Snail Kites...................74 4-1 Major wetland complexes (i.e., regions) used by the Snail Kite in Florida.......................98 4-2 Movement probabilities between na tal region and post-dispersal region..........................99 4-3 Model averaged estimates of natal phil opatry (NPHL) and philopatry to non-natal region (PHLNN)..............................................................................................................100 4-4 Model averaged estimates of region specific and natal re gion specific survival of Snail Kites in four regions...............................................................................................101 5-1 Model averaged estimates of adult an d juvenile survival between 1992 and 2005.........138 5-2 Estimates of probability of quasi-extinction....................................................................139 5-3 Sensitivity (a) and elasticity (b) of 1 to changes in age-specific vital rates...................140 5-4 The difference in age-specific vital rate s between the matrix BEF and matrix AFT......141 5-5 Estimates of stochastic population growth rates ( s) for environmental conditions before and after 1998.......................................................................................................142 5-6 Estimates of stochastic population growth rates ( s) assuming a low frequency (LFMD) and a high frequency of mo derate drying events (HFMD)...............................143 5-7 Estimates of stochastic population growth rates ( s) for environmental conditions before and after 1998.......................................................................................................144 D-1. Agglomerative hierarchical analysis wh ich categorized years from 1992 to 2006 into wet years (blue), moderately dry year s (yellow) and drought years (red).......................168

PAGE 13

13 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy POPULATION ECOLOGY AND CONS ERVATION OF THE SNAIL KITE By Julien Martin May 2007 Chair: Wiley M. Kitchens Major: Wildlife Ecology and Conservation My research was articulated ar ound three primary goals: (1) dete rmine the current status of the Snail Kite ( Rostrhamus sociabilis ) population in Florida; (2) provide information about population ecology relevant to conservation of th e Snail Kite; (3) make recommendations that will help Snail Kite recovery. I found that the Snail Kite population declined dramatically in recent years. Estimates of stochastic population growth rate and probabilities of quasi extin ction suggested that the Snail Kite population in Florida was at high risk of extinction. The sharp de cline observed after 2001 was mostly associated with a multiregional drought that occurred in 2001 and affected movement, survival and reproduction. The occurrence of this disturbance allowed us to evaluate hypotheses related to the effect of droughts on demography and moveme nt of Snail Kites. Only a small proportion of kites escaped a regional drought by moving to refugia (wetlands less affected by drought). Many individuals died after the dr ought. During drought, adu lt survival dropped by 16%, while juvenile survival dropped by 86% (possibly because juveniles were less likely to reach refugia). Although kites exhibit extensive exploratory behavior, particularly among contiguous wetlands, they also show high levels of annual site tenac ity during the breeding season, especially to their place of birth. Fidelity to breeding and natal sites has been given

PAGE 14

14 relatively little attention in the past. However, I found that fide lity to the natal region could have significant effects on movement patterns and survival, and coul d influence the whole dynamics of the kite population. Although the 2001 drought had a considerable effect on su rvival, reproduction, and abundance, our results suggest that the lack of recovery after 2002 was predominantly caused by lack of recruitment. We found evidence that both habitat conversi on (caused by prolonged hydroperiod and increased water depth during the Fall), and the increase in frequency of drying events (during the Spring and Summer), especial ly in Water Conservation Area 3A (WCA3A), could be responsible for the observed reduction in population growth rate. Finally, I present a set of management recommendations to promote Snail Kite recovery.

PAGE 15

15 CHAPTER 1 INTRODUCTION Why Study the Snail Kite? Natural ecosystems provide societies with goods and services worth trillions of dollars annually, and also perform life-s upport services essentia l to the persistence of humanity (Daily et al. 1997). These services include purificati on of air and water, detoxification, decomposition of wastes, and regulation of climates, to list a few (Daily et al. 1997). Yet, most natural ecosystems suffer from escalating destruction ca used by human activities. The multiplication of ecosystem restoration efforts ar ound the world reveals some level of recognition of the problem. The Everglades ecosystems in south Florida are currently the site of one of the largest and most ambitious ecosystem rest oration projects ever undertak en (Mitsch and Gosselink 2000). A major multi-billion dollar restoration project of the Everglades (the Comprehensive Everglades Restoration Plan, CERP) is bei ng implemented by the U.S. Army Corps of Engineers, the U.S. Fish and Wildlife Service, and th e state of Florida to restore th is ecosystem. Restoration of the Everglades is gathering enormous attention bo th nationally and intern ationally (The Economist 2005). If this project succeeds, it is likely to become a model for other large ecosystem restoration projects throughout the world (The Economist 2005). The primary objective of this restoration project is to impr ove the quality of native habita ts and increase diversity and abundance of native plants and animals (RECOVER 2005). The principal idea of the restoration is to act on two primary stressors of the sy stem: hydrology and water quality (RECOVER 2005). Conservation biologists and manage rs have selected a number of indicator species as measures of success of the restoration (RECOVER 2005; Niemi and McDonald 2004). The Snail Kite ( Rostrhamus sociabilis plumbeus ) is one of the indicator species that was selected by the CERP work team (RECOVER 2005).

PAGE 16

16 There are at least six justifi cations to list the Snail Kite as an important indicator of Everglades restoration. First, the Snail Kite is a wetland dependent species. Kites forage and nest preferentially in habitats dominated by plan t communities that CERP strives to restore (RECOVER 2005). Second, Snail Kites respond numeri cally (e.g., change in mortality rates) and behaviorally (e.g., change in movement rates) to changes in hydrological conditions (the primary stressor modified by CERP). The response of ki tes to changes in hydrological conditions is measurable (e.g., estimates of reproductive parame ters, movement, etc.). Third, the Snail Kite population experienced some dramatic decline af ter the drainage and fragmentation of the Everglades, to the point that it was listed as federally endangered in 1967. The recovery of endangered species is also one of the tasks of CERP (RECOVER 2005). F ourth, because of the charismatic nature of this species, the Snail Kite is an indicator that can ease communication with the public. Fifth, the geographi c range of the Snail Kite in Fl orida encompasses most of the wetlands that CERP will attempt to restore (R ECOVER 2005). Sixth, the monitoring of the Snail Kite is one of the few programs in Florida th at is long term (> 15 years) and accounts for detectability and spatial variation (Be nnetts et al. 1999b; Yoccoz et al. 2001). Background Information on Snail Kites The Snail Kite is a medium size raptor which is restricted to the American continent and Cuba. The Snail Kite belongs to the Order of the Falconiformes the Family of the Accipitridae the Subfamily of the Buteoninae and the Genus Rostrhamus (Lerner and Mindell 2005). Three subspecies have been recognized: R. s. plumbeus is found in Florida and Cuba, whereas R. s. sociabilis and R. s. major range from Central to South America. Beissinger (1998) has questioned this classification, which is based on morphometric measurements. The Snail Kite in Florida is restricted to the remaining wetlands that used to constitute the historic Kissimmee-Okeechobee-Everglades wate rshed. The Snail Kite is a wetland dependent

PAGE 17

17 species that feeds almost exclus ively on freshwater Apple Snails Pomacea paludosa (Beissinger 1988). Kites are sensitive to cha nge in hydrological conditions, part ly because snail availability is tightly linked to hydrology (Bei ssinger 1995). The occurrence of droughts in particular reduces snail availability dras tically, and therefore a ffects kites movement and demography (i.e., during drought kites must move or die) (Bennetts and Kitchens 2000). Conversely, aside from some possible short term benefits for juvenile surv ival (Bennetts et al. 2002), prolonged hydroperiod, flooding events or drought suppression may, in the longer term, degrade the vegetation communities that support both kite foraging and ne sting habitats (Kitchens et al. 2002), making the management of kite habitat a complex endeavor. Since 1930, the network of wetlands occupied by kites has changed dramatically. It has been reduced approximately by half of its orig inal size, and has been severely fragmented (Kitchens et al. 2002). In 1967 the Snail Kite fr om Florida and Cuba was first listed as endangered pursuant to the Endangered Species Conservation Act (USFWS 1999). Since its listing as an endangered species th e Snail Kite population in Florida has been monitored using quasi-systematic annual count surveys (Bennetts and Kitchens 1997). In the early seventies counts were less than 200, but increased approximately to 1000 birds in 1995. However, Bennetts et al. (1999b) criticized that monitoring technique, ar guing that these counts were of limited value, if not misleading, because detection probabilities we re not accounted for. Since 1992 a capture-mark-recapture study was initiate d to provide robust estimates of vital rates (i.e., survival, reproduction, move ment) and population size. This protocol was coupled to an extensive radiotelemetry protocol between 1992 and 1995 to provide more precise estimates of movement and survival (Bennetts et al. 1999a).

PAGE 18

18 Dreitz et al. (2002) provided the first estimates of populat ion size that accounted for detection, and their estimates (from 1997 to 2000) were four to five times greater than the recovery target set by the USFW S in 1999. Estimates of Dreitz et al. (2002) during the study period also indicated a fairly stable population. U nder the objectives set by the recovery plan of 1999, these figures were encouraging. Objectives and Outline My research was articulated ar ound three primary goals:(1) determine the current status of the Snail Kite population in Florida; (2) provide information about popula tion ecology relevant to conservation of the Snail Kite; (3) make cl ear recommendations that will help Snail Kite recovery. In Chapter 2, I provided new information related to the status of the Snail Kite in Florida. Based on a recently proposed estimator of abundance (i.e., the superpopulation approach), I presented estimates of population size and popula tion growth rates for th e last 9 years. In Chapter 2 I also emphasized the importance of accounting for major sources of variation when estimating demographic parameters. In Chapter 3, I examined how Snail Kites perceive and move th rough the landscapes of Florida. I also examined the link between m ovement and survival, and investigated the relevance of fragmentation and habitat destruction to Snail Kite conservation. In Chapter 4, I focused on some importan t behavioral components that determine movement and survival of kites. In particul ar, I looked at how fidelity to the natal site influenced movement and survival. At the e nd of the chapter, I emphasized the importance of considering fidelity to the natal site for management. In Chapter 5, I used matrix population mode ls to estimate projected population growth rates and probability of quasi -extinction. I also evaluated competing hypotheses explaining changes in population growth rates. At th e end of the chapter, I provided a set of recommendations for management of Snail Kite habitat. In the final chapter (Chapter 6), I synthesi zed the information presented in chapters 1 through 5, and provided some perspectives for future work on the Snail Kite.

PAGE 19

19 CHAPTER 2 IMPORTANCE OF WELL-DESIGNED MONITORING PROGRAMS FOR THE CONSERVATION OF ENDANGERED SPECIE S: CASE STUDY OF THE SNAIL KITE Introduction Monitoring natural populations is often a necessary step to esta blish the conservation status of species and to help improve management decisions (Yoccoz et al. 2001). However, many monitoring programs do not effec tively address two important components of variation in monitoring data: spatial variation and detectabil ity, which ultimately may limit the utility of monitoring in identifying declines and im proving management (Yoccoz et al. 2001). Detectability refers to the probabi lity that an animal will be detected if it is present in the sampled area (Williams et al. 2002). Many sources of variation may affect detectability (e.g., observer effect, environmental conditions), and mon itoring data that do not ta ke detectability into account will typically lead to bi ased estimates (Williams et al. 2002). Spatial variation is another source of variability of m onitoring data. It results from the inab ility to sample the entire area of interest (i.e., inference is drawn from selected spatial units th at are only a fraction of the area of interest; this is a problem when the areas samp led are not representative of the entire area) (Williams et al. 2002). The principle that monitoring programs should ta ke detectability and spatial variation into consideration is gaining some support among w ildlife biologists (see Williams et al. 2002). However, analyses using uncorrected counts co ntinue to be publishe d in major journals (reviewed in Rosenstock et al 2002; Conn et al. 2004). The co ntinued controversy around the value of uncorrected count-based indexes results partly from the fact that monitoring programs that estimate detectability are of ten more labor intensive (Conn et al. 2004). Some authors have argued that when the focus is on population ch ange rather than populat ion size, uncorrected count-based indexes may be sufficient, but for this latter statement to be correct, detectability

PAGE 20

20 should remain constant over time which is rarely the case when monitoring mobile organisms (Conn et al. 2004). The principle that it is crucial to estimate detection and to account for spatial variation when monitoring animal populations a ppears to take even longer to be accepted by some managers. This is problematic because recovery plans for many endangered species are still based on monitoring programs that ignore these primary sources of variations (e.g., Cape Sable Seaside Sparrow Ammodramus maritimus mirabilis Wood Stork Mycteria americana see USFWS (1999)). We used the monitoring of the Snail Kite ( Rostrhamus sociabilis plumbeus ) in Florida to illustrate the importance of considering detectabi lity and spatial variation. The Snail Kite feeds almost exclusively on freshwater snails and, t hus, is considered a wetland-dependent species (Beissinger 1988). In the United Stat es the Snail Kite is restricted to the remaining wetlands of central and south Florida (Dreitz et al. 2002). Because the availab ility of snails to kites is strongly dependent on hydrol ogical conditions, variations in wate r levels are likely to influence kite behavior and demography (Beissinger and Takekawa 1983). Droughts a ffect kite behavior and demography by reducing snail availability to kites (Beissinge r and Takekawa 1983). The drainage of the Everglad es that began in the early 1930s and was followed by wetland destruction led to the collapse of the kites in Florida (USFWS 1999). In 1967 the Snail Kite was first listed as endangered pursuant to the U. S. Endangered Species Conservation Act (ESA) (USFWS 1999). Three primary qua ntitative recovery criteria we re set by the U.S. Fish and Wildlife Service (USFWS) in 1999 on which to ba se reclassification of the Snail Kite from endangered to threatened (USFWS 1999): (1) the 10-year average for the total population size is estimated as > 650, with a coefficient of variation (C V) less than 20% for the pooled data (USFWS 1999); (2) no annual popu lation estimate is less than 500; and (3) the rate of

PAGE 21

21 increase to be estimated annually or biannually and over the 10 year peri od will be greater than or equal to 1.0, sustained as a 3-year running average (USFW S 1999). These criteria, however, were set in reference to data obtained fr om uncorrected counts (USFWS 1999). Since 1965 several agencies have been c onducting kite surveys throughout th e designated critical kite habitats (reviewed in Bennetts et al. 1999b). One major weakness of most count surveys is that detectability is not considered (Bennetts et al. 1999b). Hereafter we used the term counts to refer to uncorrected counts, which basically co rrespond to the number of animals counted during a survey. These counts represen t an unknown fraction of the targ et population (i.e., detection probability is not taken into account, see Williams et al. 2002). In contrast, the terms estimate of population size and estimate of superpopulatio n size correspond to populat ion parameters of interest that take detectability into consideration. Dreitz et al. (2002) provided the first estimates of populat ion size that accounted for detection, and their estimates (from 1997 to 2000) were four to five times gr eater than the target set by the USFWS in 1999. Estimates of Dreitz et al. (2002) during the study period also indicated a fairly stable populat ion. Under the objectives set by the recovery plan of 1999, these figures were encouraging. However, estimates pres ented in this study indi cate that criteria set by the recovery plan in 1999 need revision and that count surveys of popula tions that occupy large landscapes may be dangerously misleading. Using a recent estimator of superpopulation size, we examined the implications of carefully considering de tection probabilities and spatial va riation when making inference about changes in population size and number of young produced. In this study, the superpopulation consisted of all kites that had a non-zero probabi lity of being detected over the course of the sampling year (Dreitz et al. 2002). The superpop ulation approach employed in our study and in

PAGE 22

22 Dreitz et al. (2002) was based on capture-mark-resighting analys es. This approach allowed for the estimation of superpopulation size that takes in to account dete ctability and spatial variation (Williams et al. 2002). This approach makes us e of capture-mark-resight models such as Cormack-Jolly-Seber models (CJS) (Dreitz et al. 2002). There are six primary steps used in this st udy. 1) We estimated superpopulation size for young and adults separately. 2) Based on these estimates, we examined changes in abundances by estimating population growth rates, and we looked for population decline. 3) We then compared count data with superpopulation estimat es. Since detection probabilities for counts are typically < 1.0, counts will often underestimate th e true population size (Williams et al. 2002). 4) We computed one type of detection probability (denoted P) as the ratio of the number of kites counted over the estimated superpopulation size, and examined how detection varied over time for two types of count surveys. If this detec tion estimate varied substa ntially over time, then count-based indexes could be very misleading. For instance, if detec tion increases over time (e.g., because of additional field personnel), population growth rates derived from counts may suggest that the population is grow ing, or remains stable, while in fact, the tru e population is decreasing. Note that detection probability based on the ratio st atistic (P) is different from detection probability (denoted p ) directly estimated using capture-mark-resight models such as the CJS (Williams et al. 2002). 5) We examined the recovery criteria that the USFWS set for kites to determine whether these criteria were me t or were close to being met based on two types of monitoring data, one that considered de tection and spatial va riation (superpopulation approach), and another that did no t (i.e., count data). 6) Finally, we examined the implications of addressing detectability and spat ial sampling for the conservation of kites and other endangered species.

PAGE 23

23 Methods Study Area The Snail Kite population in Florida has been described as geographica lly isolated (Martin et al. 2006; Dreitz et al. 2002). Kites occupy the remaining wetlands of the KissimmeeOkeechobee-Everglades freshwater watershed. The sampled units we used are identical to the units Dreitz et al. (2002) used and encompassed major kite hab itats (Figure 2-1). Although kites may temporarily emigrate to unsampled areas, it is unlikely that they will not return to the major wetlands included in the survey during some por tion of the sampling period (Dreitz et al. 2002). Sampling Methods Marking protocol Multiple, consecutive surveys of Snail Kites fr om airboats were c onducted during the peak of the breeding season (March through June) throughout the areas sampled from 1992 to 2005. During surveys, workers located nests and bande d young kites when they were ready to fledge (~25 days). A total of 1806 young were marked between 1992 and 2005. Prior to 1995, 134 kites were marked as adults (> 1 year) (Bennetts and Kitchens 2000). Additionally, 76 kites were marked as young prior to 1992. Kites were marked with alpha-numeric bands. Population survey protocol Our protocol for population surveys from 2001 to 2005 followed that described by Dreitz et al. (2002) and was part of the same Florid a Cooperative Fish and Wildlife Research Unit (FCFWRU) kite-monitoring program. Four to si x consecutive surveys from airboats were conducted at 2to 4-week intervals throughout th e designated wetland units from 25 February to 30 June. From 2001 to 2003, the surveys starte d between 1 March and 8 March and ended between 15 June and 19 June. In 2004 and 2005, th e surveys started betw een 25 February and 1 March and ended between 27 June and 30 June. Du ring each survey we inspected every sighted

PAGE 24

24 kite with binoculars and spotting scopes. We categorized each observed individual as: (1) marked if the kite carried a band that could be uniquely identified; (2) unmarked if the sighted kite did not carry an identifiabl e band; or (3) unknown if the bandi ng status of the kite could not be determined. Analysis Superpopulation size estimates of adults We used the superpopulation approach (S chwarz and Arnason 1996) generalized by Schwarz and Stobo (1997) into a robust design framework. This approach allows for movement during secondary occasions (i.e., between surv eys within a year, see Dreitz et al. 2002). Dreitz et al. (2002) were the first to apply this method to the Snail Kite. A ll notations follow Dreitz et al. (2002). The superpopulation approach estimated the total number of kites pr esent in the sampled area for at least some of the surveys during th e sampling period (surveys within a year were denoted i =1, 2,, n ). For any given year (denoted j ), we referred to this estimate as the superpopulation estimate for each year (* j N ). n1 ij j1j i1 N NB (Eq. 2-1) where 1j N is the estimated abundance of the first survey in year j The ij N is estimated as ijij ij ijmu N p (Eq. 2-2) where mij is the number of marked kites and uij the number of unmarked kites at each survey i in year j Kites whose banding status was unknown were excluded from th is analysis. The ij p (sighting probability) is the estimated probability of sighting a kite given that it was present in

PAGE 25

25 survey i of year j Given the constraint, p1j = p2j, abundance can be estimated for all surveys within year j The ij B is the estimate of the number of new k ites entering the sampled area (between survey i and i+1 ) from areas not sampled on each survey 1 ij i,jij ijBNN, i > 0, (Eq. 2-3) where ij(apparent survival) is the estimated probability of not dying and not permanently emigrating to an area not sampled between surveys i and i +1 of year j Given the constraint p1j = p2j ij B can be estimated for i = 1,, ( n-1) As in Dreitz et al. (2002) we used the CJS model implemented in program MARK to estimate ij and ij p (Cooch and White 2005). We analyzed each annual capture -recapture data set separately to obtain estimates of i and i p within a year. We preferred this approach to a multigroup approach (used in Dreitz et al. 2002) because the date fo r the within year surveys did not match exactly from one year to the other. Unlike Dreitz et al. (2002) the number of sampling surveys in our study varied between four and si x (we note that the number of sampling surveys should not affect the superpopulation estimate as long as the sampling periods remain similar). Although we ended our surveys before 30 June in 2001, 2002 and 2003, our data indicates that all surveys conducted after the seco nd week of June did not affect estimates of superpopulation size. Indeed, removing the last survey for year s when surveys ended on approximately 30 June, did not affect the estimate of abundance. This can be explained by the fact that by the end of the sampling season, it is unlikely that many birds will enter the sampled areas for the first time (i.e., most birds have already done so before the last survey).

PAGE 26

26 For model selection among CJS models, we used QAICc, which corresponds to the Akaike information criterion (AIC) corrected for sm all sample size and extrabinomial variation (Burnham and Anderson 2002). We ran four CJS m odels for each year. Models that assumed i and i p remained constant between surveys were de noted with a subscripted dot (.). Conversely, models that allowed i and i p to vary among surveys were denoted with a subscripted t Thus, the four models were denoted: .. p .t p t. p and tt p As recommended by Burnham and Anderson (2002), we used mode l-averaged estimates of and p The purpose of estimating and p with CJS models was primarily to comput e estimates of superpopulation size. Confidence intervals of estimates of super population size were computed with the same parametric bootstrap procedure (500 simulations) descri bed in Dreitz et al. (2002). The assumptions for the superpopulation model are similar to the ones required for the more widely known Jolly-Seber model (Williams et al. 2002). In particular, homogeneity of rates among animals are assumed. The su perpopulation model also assume s that all members of the superpopulation unavailable until t will exhibit similar probability of being available for capture at t + 1. The Goodness of fit test (test 2 + test 3) implemented in program RELEASE, which tests for homogeneity of i and i p and for lack of independence of survival and capture events (Burnham et al. 1987), is also applicable for the superpopulation model (Williams et al. 2002). There is no evidence of heterogene ity or lack of independence for i and i p when probability p is > 0.05 (Cooch and White 2005). Burnham and Ande rson (2002) indicate that model structure is acceptable for extra-binomial factor c < 4, and they suggest to adjust for extrabinomial variation if c > 1. We computed c with program RELEASE (Burnham et al. 1987). Estimation of population growth rate Annual population growth rate ( j ), was estimated as:

PAGE 27

27 1 j j j N N, (Eq. 2-4) We then computed the arithmetic average of all the j over the last 8 years (1998 to 2005) and a 3-year running average (denoted 2 j (j) ). Estimation of the number of young produced We used the superpopulation approach descri bed above to estimate the number of young produced in any given year (her eafter referred as young) for the en tire superpopulation (denoted Yj N ). For this analysis mij and uij (see Eq. 2-2) included exclusivel y kites that were hatched in year j We used this approach only in 2004 and 2005 because we began recording mij and uij for the young in 2004. There was not enough band resight information of young in 2004 and 2005 to estimate ij and ij p that were specific to that particul ar age class. Therefore, we used ijand ij p computed for adults to estimate the number of young produced in 2004 and 2005. Detection probabilities for number of young produced every year Only a proportion of the total number of young produced were detected and marked every year. To estimate the proportion of young marked during each year (i.e., detection probability of young), we used the following estimator (see Williams et al. 2002): Yj Yj YjC P N (Eq. 2-5) where YjPis the detection probabi lity of young in year j, CYj is the number of young observed and marked in year j (hereafter referred as the number of young marked). The YjP differed from survey-specific ij p (directly estimated with CJS models).

PAGE 28

28 Count data We used two types of count survey data that we subsequently compared to estimates of superpopulation size: first count (FC), and maximum count (MC) For FC we used the first FCFWRU annual count survey (total number of birds counted during the first survey) as an indicator of annual abundance. We used the fi rst annual count survey because it was always conducted at the same date (1 March + 1 week). Many agencies, including the Florida Fish and Wildlife Conservation Commission use this type of format for surveys in which a designated study area is sampled annually (typically at the same time of year). The MC was annual count data of the maximum count obtained for any of the FCFWRU surveys w ithin a sampling season. MC and FC included: marked, unmarked and unknown kites. However, because in 1997 unknown birds were not reported, all analyses related to FC and MC data focused on the period 1998 to 2005. Average number of kites and growth rate based on count data We computed the arithmetic average for the two sets of count data ( C). We also used these count data sets to co mpute annual growth rate based on counts (cj). cj were estimated as follow (see also Williams et al. 2002): 1 j cjC Cj, (Eq. 2-6) We then computed the 8-years arithmetic average of all the cj ; as well as the 3-year running average denoted: 2cj(j). Detection probabilities for FC and MC Monitoring based on counts typically assumes de tection probability to equal 1.0; however, in practice, this assumption is rarely met (Williams et al. 2002).

PAGE 29

29 We determined the detection probability of Snail Kites us ing both FC and MC surveys by computing the ratio of the number of kites counted in a given year j ( Cj ) (using either FC or MC), over the estimated superpopulation size for that same year (* j N ): j j jCFC PFC N and j j jCMC PMC N (Eq. 2-7) We emphasize that jPFC and jPMC differed from survey specific ij p (directly estimated with CJS models). We also established, for each type of surv ey (FC and MC), the increase in detection probabilities (in percentage) necessary to obtain an average count of kites ( C) > 650. Estimates of Precision and Magnitude of the Difference between Estimates We used the delta method to compute the vari ances of derived estimates (Williams et al. 2002). 95% confidence intervals (95% CI) of any parameter that was not strictly positive (e.g., estimate of magnitude of the difference, see below) were computed as follows: 95% CI [ ] = + t0.025 ,df SE[ ], where SE[ ] is the estimated standard error of and t0.025 ,df is the upper 97.5 percentile point of the t distribution on df (Burnham and Anderson 2002). As recommended by Burnham and Anderson (2002), for any parameter that is strictly positive (e.g., population size), we used an approximation for computing 95% CI [ ] that is based on a lognormal distribution (p. 259 Burnham and Anderson 2002). The magnitude of the difference between two estimates (MD) was estimated by computing the arit hmetic difference between estimates (Cooch and White 2005).

PAGE 30

30 Results Population Size and Average Population Growth Rate Estimates of ij and ij p were obtained with model averaging of models .. p .t p t. p and tt p for each year (estimates of ij ij p and other survey-specific parameter estimates used to compute estimates of superpopulation size ar e available on-line see Table A-1 in Appendix A). There was no need to adjust for lack of fit of the most general model in 2002, 2003, 2004 and 2005, because c was < 1, and test 2 + test 3 from R ELEASE were all non significant ( p > 0.05). In 2001, the test 2 + te st 3 was significant ( p = 0.02). Therefore, we adjust ed for lack of fit of the most general model in 2001 (c= 2.2). Estimates of superpopulation size (* j N ) from 1997 to 2000 were obtained from Dreitz et al. (2002), whereas estima tes from 2001 to 2005 are the results of the present study. Estimates of superpopulation size between 1997 and 2000 were fairly constant and relatively hi gh (Dreitz et al. 2002; Figure 22). Superpopulation size estimates decreased sharply during the inte rval 2000-2002, but there was an appa rent stabilization, or even slight increase (but note th e 95% CI overlap) in 2004 and 2005. The average superpopulation size for the last 9 years (1997 to 2005) was 2254 (95% CI = 2124 to 2392). Estimates of the 8year average growth rate ba sed on superpopulation estimates was 0.93 (95% CI = 0.84 to 1.03). Estimates of annual growth rate based on supe rpopulation estimates we re > 1, in 1998, 2003 and 2004 (Table 2-1). Estimates of the 8-year averag e growth rate was 1.11 (95% CI = 0.91 to 1.37) based on FC; and 0.99 (95% CI = 0.86 to 1.12) ba sed on MC. Estimates of the 3-year running average growth rate based on superpopulat ion estimates and on MC were > 1 for 0204 only (Table 2-1). Estimates of the 3-year running av erage growth rate based on FC was < 1 for 0002 only (Table 2-1).

PAGE 31

31 Average Number of Kites Before and Aft er Decline Based on the Superpopulation Approach Estimates of superpopulation size suggests thr ee major periods: a pre-decline period (1997 to 2000), a decline period (2000 to 2002) and a post decline period (2002 to 2005) (Figure 2-2). We computed the average number of kites during the pre and post decline periods. Prior to decline (1997 to 2000), the average number of kites based on the superpopulation estimator was 3157 (95% CI = 2909 to 3426). Afte r decline (2002 to 2005) the average number of kites was 1407 (95% CI = 1278 to 1550). There wa s a substantial decrease between before and after decline ( MD = 1750; 95% CI = 1457 to 2041). This re presented a 55% decrease (95% CI = 46% to 67%) when compared with predecline levels. Average Number of Kites Before and After Decline Based on Count Data Average number of kites before decline (1998 to 2000) based on FC data was 397 kites (95% CI = 164 to 959) and 403 (95% CI = 316 to 514) after decline (20 02 to 2005). Therefore, FC data showed a slight increase in kite num bers between the interv als 1998 to 2000 and 2002 to 2005; however, the difference was not biologically significant ( MD = 6; 95% CI = -220 to 208). Based on MC data average number of kites ( C) before decline (1998 to 2000) was 600 (95% CI = 462 to 779) and after declin e (2002 to 2005) was 410 (95% CI = 337 to 499). Although 95% CI intervals of C overlapped, MC data showed a substant ial decrease in kite numbers between the intervals 1998 to 2000 and 2002 to 2005, (MD = 190; 95% CI = 82 to 298). This represented a 32% decrease (95% CI = 20% to 51%). Average Number of Kites and Growth Rate Based on Count Data The average number of kites counted with FC and MC for the last 8 years (1998 to 2005, but excluding 2001), was 401 kites (95% CI = 3 19 to 503) for FC and 491 kites (95% CI = 392 to 616) for MC. The CV was 0.09 for MC and FC.

PAGE 32

32 Estimates of annual growth rate based on FC were < 1 in 2000 and 2002 (Table 2-1). Estimates of annual growth rate based on MC were < 1 from 1999 to 2002 (Table 2-1). Number of Young There was a sharp decline in the number of young marked starting in 1999 (Figure 2-3). The average number of young marked from 1992 to 1998 was 200 (95% CI = 145 to 277), whereas the average number of young marked between 1999 and 2005 was 61 (95% CI = 38 to 96). The difference was substantial ( MD = 139; 95% CI = 75 to 204) This represented a 70% decrease (95% CI= 41% to 100%). The number of young produced in 2004 and 2005 based on the superpopulation approach were Y2004N= 414 and Y2005N= 55. The detection probabilities in 2004 and 2005 were Y2004P= 0.16 and Y2005P= 0.35. Estimates of confidence intervals could not be computed because sample size of resighting of young kites was too small. Detection Probabilities for FC and MC An increase of 63% in detec tion probability was necessary to obtain an average count > 650 for the FC survey; an increase of 33% in detection probability was necessary to reach a similar target based on the MC survey data (Tab le 2-1). Detection estimates increased over the years for both types of surveys (Table 2-1). Discussion Population Decline Our results based on the superpopulation approa ch indicate that the population of Snail Kites in Florida declined sh arply between 2000 and 2002 (Figure 2-2). Although estimates were slightly higher for 2004 and 2005, there was no evid ence of a substantial recovery. The reduction in the estimated average kite abundance before and after decline was s ubstantial (55% reduction

PAGE 33

33 in abundance). The method we used to estimate the superpopulation size of kites was also useful in obtaining the number of young produced per br eeding season. Although this parameter is difficult to estimate in the wild, it is often n eeded to evaluate the viability of threatened populations. For instance, the superpopulation ap proach is an appealing method to compute reliable estimates of fertility rates, which ar e critical to correctly parameterize many types of population viability analyses (Morris and Doak 2002). We only had data to compute estimates of the number of young produced for 2 years (2004 and 2005). We also used these estimates to compute the proportion of young marked during these 2 years (i.e., detection of young). The fact that detection of young varied substantially in 2004 and 2005 suggests that one should be cautious in using the number of young marked as an indicator of the number of young produced. However, we believe detection estimates for these 2 years corresponded to extreme values. We expect ed low detection probability for 2004 because birds bred unusually early, which meant a large proportion of birds fledged before they could be marked. Conversely, in 2005 we inve sted an unprecedented effort in nest searching and marking young, which led to higher detection. Unless de tection declined signi ficantly between the intervals 1992-1998 and 1999-2005, we expect the observed number of young marked to be representative of an importan t decline in the number of young pr oduced. We believe detection is likely to have increased in recent years because we invested more effort in nest-searching activities than in earlier years. An increase in detection implies that th e reduction in the number of young produced in recent years is even more se vere than is apparent in Figure 3-3. Models used to obtain the number of young pr oduced assumed that estimates of ij and ij p for young and adults were similar. Appropr iate sample size of resighting of young kites should be collected

PAGE 34

34 in the future to check the assumption that adult estimates of ij and ij p provide a reasonable approximation to estimate the number of young produced. Problems Associated with Counts and Implications for Recovery Plans Identifying population decline is critical to the process of species conservation. In practice, it is often the documentation of population decline below a critical threshold that leads to the classification of species as endangered under th e ESA. Additionally, identifying a reduction in population size may prevent unsubstantiated down listing. The legal protection offered by the ESA is in many cases essential to the persistence of many species at risk of extinction (Doremus and Pagel 2001). Our results provide a compelling example of the risks associated with setting recovery targets that are based on deficient monitoring programs. Next we explain how some of the current recovery targets pres ented in the Snail Kite recove ry plan (USFWS 1999) could be met (even with a declining population) if monito ring does not account for major sources of error such as detection. One of the major recovery crite ria listed in the plan stat es that the 10-year average population size should be > 650. Even the most recent superpopulation estimates obtained during our study indicated that the actu al Snail Kite population size may be twice this number. This suggests that although the 8-year-a verage counts obtained with FC and MC were all below the recovery target set by th e USFWS (i.e., 650 kites), it is likely that by increasing the search effort (e.g., increase in the number of field personnel), mo re than 650 kites could have been counted during these surveys. In fact, an increase of 33% in detection probability (i.e., the proportion of kites counted from the true population size) during MC counts and 63% during the FC counts would have boosted the average nu mber of kites counted over a period of 8 years to above the 650 target (Table 2-1). In both cases CV was < 0.2 (i.e, < 20%). The second recovery criterion

PAGE 35

35 states that kite numbers should not fall belo w 500 for any given year. This condition would not have been met for the FC count with detec tion increased by 63%, because the count in 1998 would be 460 (but increasing de tection from 0.09 to 0.16 in 1998 would have brought the count for that year to 501). Similarly, this condition w ould not have been met for the MC count with detection increased by 33%, because the count in 2003 would be 463 (but increasing detection from 0.30 to 0.43 in 2003 would have brought the count for that year to 501). The third recovery criterion stipulates that the 3-year running average should not be < 1.0 over a period of 10 years. Out of the five averages that could be computed for the last 8 years of data for the FC count, only one value fell belo w 1.0. Reducing the proportion of birds that were observed in 2001 from 0.19 to 0.11 (see Table 2-1), would have pushed all values of the running average for the FC count above 1.0 (although the lowe r CI of these values may have fallen below 1.0, nothing is mentioned in the recovery plan about parameter uncertainty of growth rate estimates). One can think of scenarios that w ould cause such a reducti on in the proportion of kites counted. For example, dry co nditions could reduce airboat access to wetlands used by kites. Finally, when computing the 8-year average of growth rate for the superpopulation, the FC count and the MC count, we found that the growth rate was < 1.0 for the superpopulation and the MC count but was > 1.0 for the FC count (in all cases lower 95% CI were < 1.0). The fact that the estimate of the average growth based on FC data was > 1.0, even though the population was declining, is most likely due to the increase in de tection probabilities over the years (Table 2-1). This increase in detection probability was also ob served for the MC data, and resulted in an inflated 8-years average growth rate for the MC survey as well. The observed increase in detection probabilities over time could be due to an increase in the number of field personnel in recent years (since 2002).

PAGE 36

36 A particularly disturbing fact regarding count data is th at, despite the drop in kite abundance (55% based on the superpopultion ap proach) the FC count did not indicate a reduction in kite numbers. The MC count, however indicated a reduction in kite numbers. The MC count may be less biased than FC because for every sampling year the maximum count will be closer to the true abundance than any other single count. This is because all counts underestimate true abundance, therefore the maxi mum count should be the closest to the true abundance than any other count. However, since both types of counts ignored detection and did not deal effectively with sampli ng variation they were therefore biased. The FC count (i.e., single annual count) is by far the most common type of count survey. The format of the FC count was very similar to the surveys conducted by the Fl orida Fish and Wildlife Conservation Commission (FFWCC) between 1995 and 2004, ex cept that the FFWCC annual count took place during the midwinter (December to February), and was restricted to fewer wetlands (FFWCC unpublished data). This spatial restriction also increases potential for errors a ssociated with spatial variation. Thus, by simply varying the proportion of kite s observed during counts (i.e., detection) three major recovery criteria in the Snail Kite recovery plan were close to being met based on a monitoring that relied on count s (e.g., FC) in spite of an alarming decrease in estimated population size and reduced reproduction. Our primary purpose was not to set new recove ry criteria for the Snail Kite (although our study strongly suggests that existi ng criteria are in need of revi sions), and we point out that several authors have proposed promising appro aches to set more appropriate criteria (e.g., Gerber and DeMaster 1999; Morris and Doak 2002). Instead, we emphasize the critical importance of designing monitoring programs that address major, common sources of errors,

PAGE 37

37 because reliability of the recovery criteria wi ll strongly depend on the quality of the monitoring data. Importance of Monitoring to Diagnose Causes of Decline Although the identification of popul ation decline is an important step, it is evidently only part of the process of protecting a species. A next st ep should be to diagnose the cause of decline, or alternatively, factors limiting growth. In the case of kites, the drought that occurred in 2001 appears to coincide with the popul ation decline and strongly affected adult and juvenile survival (Martin et al. 2006). However, the drought affected ki te survival only tempor arily (1 to 2 years, see Martin et al. 2006). The lack of evident recovery four years after this natural disturbance suggests that factors affecting reproduction and recruitment ma y prevent growth. The drastic reduction in the number of young kites marked ( 70% decrease), suggests that factors limiting reproduction may deserve more atten tion than they have received in the past. However, rigorous evaluation of the causes of d ecline and factors limiting growth needs to be performed. Hypotheses related to disease, pr edation, food availability and nest substrate should probably be the focus of future investigations (see Peery et al. 2004, Martin et al. unpublished). The multiple competing hypotheses approach (MCH) provides an appealing framework to disentangle the factors that could potentially affect population growth of thr eatened species (Williams et al. 2002; Peery et al. 2004). Ideall y, monitoring programs designed to tease apart ecological hypotheses using MCH, will incorporate both sp atial variation and de tectability. Addressing spatial variation is particular ly important to effectively a ssess hypotheses related to spatial dynamics (Yoccoz et al. 2001). This may be of particular relevance to the management of spatially structured populations of species that occupy large landscapes (Yoccoz et al. 2001).

PAGE 38

38 Conclusion A growing number of ecologists are recognizing the va lue of using design s that incorporate both detectability and spatial sampling because (1 ) they allow for better parameter estimates, and (2) because they favor more effective evalua tion of ecological hypotheses (reviewed in Yoccoz et al. 2001). As illustrated by our results, these sa mpling design issues are extremely relevant to the protection of endangered species. Indeed, ig noring detectability and spatial variation may lead to dangerously inappropriate management decisions (e.g., unsubs tantiated downlisting). Nonetheless, considerable resources continue to be invested in monitoring programs that ignore these sources of variability, and many recovery pl ans continue to rely on these flawed programs. Given the immediate risks of extinction faced by an increasing number of species, it is urgent for managers and conservation biologists to rigorous ly revisit these recovery plans and monitoring programs that do not effectively addre ss spatial sampling and detectability.

PAGE 39

39 Table 2-1. Estimates of Snail Kites annual popula tion growth rates; and 3-year running average of growth rates. Year ( j ) Parametera 1998 1999 2000 2001 2002 2003 2004 Annual rates j (S) 1.14b 0.77b 0.73b 0.69 0.83 1.29 1.05 cj(FC) 1.26 1.57 0.69 1.09 0.77 1.25 1.17 cj(MC) 1.18 0.83 0.85 0.89 0.83 1.16 1.17 Average rates j (j2)(S) 0.88b 0.73b 0.75b 0.94 1.05 cj(j2)(FC) 1.17 1.12 0.85 1.04 1.06 cj(j2)(MC) 0.95 0.85 0.86 0.96 1.05 aParameter explanations: j estimates of annual populat ion growth rate based on superpopulation estimates (S); cj estimates of annual population growth based on first-count surveys (FC) and maximum count survey (MC); j (j2) 3-year running average growth rate based on j (S); cj(j2) 3-year running average growth rate based on cj(FC) and cj(MC). bComputed using data from Dreitz et al. (2002).

PAGE 40

40 Table 2-2. Estimates of detection probability of Snail Kites for first-count surveys (FC) and maximum count surveys (MC) for each year between 1998 and 2005a. Year ( j ) Detection 1998199920002001200220032004 2005 C jPFC 0.090.100.200.190.300.280.27 0.30 399 jjPFC63%*PFC 0.150.160.330.310.490.450.44 0.49 651 jPMC 0.180.190.200.230.300.300.27 0.30 490 jjPMC33%*PMC 0.240.250.270.310.400.400.36 0.40 654 a Estimates were obtained by computing the ratio co unt over superpopulation size for each year (j). The C corresponds to the average number of kites counted using the estimated detection for FC ( jPFC) and for MC ( jPMC) and the detection probabilities that we re increased by 63% for the FC surveys ( jjPFC63%*PFC ) and 33% for the MC surveys ( jjPMC33%*PMC). bDetection probabilities from 1998 to 2000 were com puted using estimates of superpopulation size published in Dreitz et al. (2002). bComputed using data from Dreitz et al. (2002).

PAGE 41

41 Figure 2-1. Map of the wetlands that were sampled to obtain bot h counts and capture-resighting information of Snail Kite for the estima tion of population size. Thick black line delimits areas sampled by the Florida Coopera tive Fish and Wildlife Research Unit.

PAGE 42

42 Figure 2-2. Comparison of the estimates of population size of Snail Kites (using the superpopulation approach) with annual c ounts. Data for three count surveys are plotted in the figure: (1) first count surv ey (FC); (2) maximum count survey. Kite numbers and estimates of population si ze from 1997 to 2000 were obtained from Dreitz et al. (2002), while estimates fro m 2001 to 2005 were results of the present study. Error bars correspond to 95% confidence intervals. The recovery target for Snail Kites (650 birds), set by the USFWS in 1999 is also presented.

PAGE 43

43 Figure 2-3. Number of young (i.e., nestlings close to fledgi ng) Snail Kites marked every year from 1992 to 2005.

PAGE 44

44 CHAPTER 3 MULTISCALE PATTERNS OF MOVEMENT IN FRAGMENTED LANDSCAPES AND CONSEQUENCES ON DEMOGRAPHY OF THE SNAIL KITE IN FLORIDA Introduction Habitat loss and fragmentation are major fact ors affecting populations of many organisms (Holt and Debinski, 2003). One detrimental effect is reduced movement of these organisms (Holt and Debinski 2003; Smith and Hellmann 2002) This may have important population consequences given that movement is a fundament al process driving the dynamics of fragmented populations, as it connects local populations thro ugh emigration and immi gration (Hanski 1999; Clobert et al 2001). To assess how movement influences the dynami cs of spatially-struc tured populations, we need to understand how animals perceive, move through, and learn about the landscapes they occupy (Hanski 2001). We also need to evaluate the relative importanc e of critical factors governing movement processes at a pertinent spatio-temporal scales. Patch size, distance between patches, and patch quality are major fact ors influencing the movement of many animal populations in spatially structured systems (Han ski 1999). Several studies have demonstrated the effect of distance on movement (e.g., Haddad 1999; Hanski 2001). Theoretical models of metapopulation dynamics commonly assume greater emigration from smaller patches (Hanski 2001; Schtickzelle and Baguette 2003), and higher immigration toward la rger habitat patches because of the more frequent encounters of moving animals with patch boundaries ( patch boundary effect ) (Lomolino 1990; Hanski 2001). Fragmentation and habitat reduction reduce pa tch size and increase the linear distance between patches: both alterations are likely to decrease movement (Holt and Debinski 2003). Creating areas unsuitable for foraging or breeding (i.e., matrix ) between or around habitats may also decrease survival (Sch tickzelle and Baguette 2003).

PAGE 45

45 Despite the importance of providing robust quantitative demographic and movement estimates of populations inhabiting fragmented landscapes (Hanski 2001; Williams, Nichols and Conroy 2002), few empirical estimates exist, especially for verteb rates using large landscapes. From 1992 to 2004, we studied an isol ated population of Snail Kites ( Rostrhamus sociabilis ) restricted to Florida. The Snail Kite is a raptor that feeds almost exclusively on freshwater apple snails Pomacea paludosa (Beissinger 1988). The kites restricted diet makes it a wetland-dependent species. Since wetlands in Florida have been severely reduced (Davis and Ogden 1994; Kitchens et al. 2002) since the earl y 1930s, the population is now confined to the remaining fragments of wetlands extending from the southern end to the centre of the state (Figure 3-1). Because the availability of apple snails to kites is re lated to hydrologic conditions, variations in water levels are likely to infl uence Snail Kite behaviour and demography. In particular, snail availability to kites is gr eatly reduced during droughts (Beissinger 1995). Beissinger (1986) and DeAngelis and White (19 94) described the hydrologic environment used by kites as highly spatially-temporally variable In such a variable environment, one might expect kites to show nomadic tendencies (Ben netts and Kitchens 2000). Bennetts and Kitchens (2000) developed a conceptual model of kite movement along a food resource gradient. They hypothesized that when food is scar ce (during drought), kites move to refugia habitats or die. When food is abundant exploratory movements can be done at minimum risk of starvation. During droughts, kites that have previously ex plored wetlands throughout their range are less likely to search randomly for alternative habitats, and thus are less likely to starve. Their model also suggests that when food is superabundant, o ccasional territorial defence may occur for short periods of time, but otherwise kites are typical ly non territorial (Beissi nger 1995). Bennetts and

PAGE 46

46 Kitchens (2000) estimated the average probabil ity of movement among wetland units (Figure 31) to be approximately 0.25 per month, which they associated with a nomadic type of behaviour. However, this probability was obtained without considering the comple xity of the spatial configuration of the system. We attempted to enhance our understanding of how kites perceive and move throughout the landscape by incorporating a de tailed level of spatial complexity into a modelling approach at multiple spatial scales. First, we es timated movement within a group of contiguous wetlands (separated by small physical barriers, easily cros sed by kites: such as a road). The distance between centroids of these contiguous wetlands varied between 16 and 110 km. Second, we estimated movement within a group of wetlands sepa rated by a moderate extent of matrix (< 5 km): moderately isolated wetlands Matrix areas generally cons ist of non-wetland areas (e.g., agricultural or urban areas). The distance betw een centroids of these moderately isolated wetlands varied between 10 and 44 km. Third, we estimated movement among wetlands or groups of wetlands isolated by extensive matrix (> 15 km): isolated wetlands To be consistent with the classification of Bennetts et al. (1999a), we called thes es isolated wetlands: regions Most regions used to be connected through the Kissimmee-Okeechobee-Everglades watershed, and became isolated as a result of habitat re duction (Davis and Ogden 1994; Light and Dineen 1994) (Figure 3-1). The distance between centroids of these regions va ried between 69 and 232 km. We also explored movement at two temporal scales. First, we examined movement at an annual scale. Because of the period of sampling (i.e., peak of breeding season), this informed us about patterns of breeding and natal philopatry of Snail Kites. Second, we examined movement patterns on a monthly scale. This period of samp ling included the entire year (i.e., including

PAGE 47

47 periods outside the breeding season). Thus this study also informed us about movement patterns that were independent of breeding activities (e.g ., exploratory movement). Finally, we discussed the consequences of kite movement on survival. To date, the assumption has been that dur ing a drought, kites move from areas most affected by drought toward areas least affect ed by drought (Beissinger and Takekawa 1983; Bennetts and Kitchens 2000; Takekawa and Bei ssinger 1989); and that the impact of a drought on the kite population will depend on the spatial extent and inte nsity of the drought (Beissinger 1995; Bennetts and Kitchens 2000). However, all hypotheses regarding kite responses to drought are based on count data that do not consider de tection probabilities. Ther efore, these hypotheses have yet to be rigorously test ed and quantified with appropr iate statistical methodologies (Williams et al 2002). Hypotheses and Predictions Prediction 1: Effect of fragmentation on movement We predict that movement will covary positively with connectivity (i.e., amount of matrix between wetlands). Thus, movement among con tiguous wetlands should be greater than among moderately isolated wetlands, and movement among moderately isolated wetlands should be greater than among isolated wetla nds (i.e., regions). Prediction 1 implies that movement within regions will be greater than betw een regions, which could also be explained by a distance effect on movement. However, if movement among contiguous wetlands is greater than among moderately isolated wetlands, the effects of c onnectivity on movement can be separated from the effects of distance (between centroids), since di stances between centroids of contiguous wetlands are greater than that of moderately isolated wetlands in the study area. Prediction 2: Effect of patch configuration on movement

PAGE 48

48 We expect patch size and distance between patch centroids to influence movement. Movement among patches (i.e., wetlands) s hould decrease with distance (Hanski 1999). Emigration should be higher from smaller patc hes (Hanski 2001; Schtickzelle and Baguette 2003), and immigration should be hi gher toward larger patches (Lomolino 1990; Hanski 2001). Prediction 3: Patch configuration affect s juvenile movement more than adult movement Patch size and distance between patches are mo re likely to influence movement of birds that have never dispersed from their natal area (typically young individuals) than birds that are aware of wetlands outside their natal area. Because we expect the number of wetlands visited to increase with time, on average juveniles (< 1 year) should have visited fewer wetlands than adults. Therefore, movement of juveniles should be less influenced by habitat characteristics (e.g., habitat quality) of destina tion sites than movement of adults, whose movement may be partly influenced by their knowledge of the loca tion of multiple wetlands (assuming that kites remember sites they have already visited). This prediction is derived from hypotheses developed by Bennetts and Kitchens (2000) and (Bell 1991), who suggested that many species learn from exploratory movements, and thus modify their movement patterns accord ing to their experience with visited habitats. Thus we expect a stronger relationship between movement and geometric features of the landscapes for juveniles than for adults. Prediction 4: Drought effect on movement a. During a drought, we predict that some bird s will move from areas most affected to areas least affected by drought (e .g., Takekawa and Beissinger 1989). b. Because of their knowledge of alternative wetlands and the paths linking these wetlands, adult birds should be more successful th an juveniles in moving refugia habitats. Prediction 5: Drought effect on survival

PAGE 49

49 a. As predicted by Beissinger (1995) and Bennett s and Kitchens (2000), we expect survival to be lower during drought. b. Survival should be lower in regions most affected by drought. c. Because adults are more likely to move successfully to areas least affected by drought, we expect survival to decrease more for juveniles than for adults. Methods Study Area This study was conducted throughout central and southern Flor ida, encompassing most of the habitats used by the Snail Kite. Thirteen we tlands were sampled (Figure 3-1). Given that kites can cross small physical barriers delimiti ng each wetland (e.g. road) with relative ease (Bennetts 1998), we further aggregated the units into five larger groups of wetlands (regions) (Figure 3-1). We used Bennetts ( 1998) and Bennetts et al. (1999a) definition of a region. Regions were separated from other regions by an extende d matrix (>15 km). Water Conservation Areas (WCAs), Everglades National Park and Big C ypress National Preserve constituted a group of contiguous wetlands and were grouped into one region: the Everglades region (E). The Kissimmee Chain of Lakes region (K) included Lake Tohopekaliga, East Lake Tohopekaliga, Lake Kissimmee, and all the small lakes in the surrounding areas. Wetlands in the K region were isolated by moderate extent of matrix (< 5 km ). Lake Okeechobee (O), St Johns Marsh (J), and Loxahatchee Slough (L), constituted their own re gions. Areas of wetlands and distances between wetlands were estimated using a Geographic In formation System (ArcView GIS 3.2; Xtools, DeLaune 2000). Criteria for Determining the Regi onal Impact of the 2001 Drought We used water-stage data (elevation of water surface measured in feet above the National Geodetic Datum of 1929) recorded daily in each of the major wetland units and made available

PAGE 50

50 by the South Florida Water Management District (http://www.sfwmd.gov/org/ema/dbhydro ) to develop an index of drought impact. We used the data corresponding to the period of study (1992 to 2003). Water stage was averaged by month for th e entire time series. We calculated the mean of the monthly average stages for March through June of each year. This period is especially critical for apple snail breeding and availability to the kites (Darby 1998) and also includes the greater part of the seasonal dry season when water stages are at their annual minimum (i.e., when water levels are most likely to aff ect kite survival and movement). We determined the mean stage for the pe riod of record (1992-2003) for each major wetland unit and determined where drought-year wa ter stage means fell in terms of standard deviations below this value. This method, propos ed by Bennetts (1998), allows for comparisons of drought intensity among wetlands for the period of record. The 2001 drought occurred between January and August (Smith et al 2003). Intensity of dr ought was maximal for the lowest drought score values (DSV). This analys is indicated that regi on E (WCA3B DSV = -2.32; Big Cypress DSV = -2.28; WC A1A DSV = -2.18; WC A3A DSV = -1.92; WC 2B DSV = -1.41; WCA2A DSV = -1.20) and region O (DSV = -2.57) were the mo st-impacted, while region K (Lake Kissimmee DSV =-0.72; Lake Tohopekaliga DSV =-0.84; La ke East Toho DSV = -0.98) was the least affected. Region J wa s also affected (DSV = -1.92). Statistical Models to Esti mate Movement and Survival Multistate capture-recapture models (Hestbeck, Nichols and Malecki 1991; Williams, Nichols and Conroy 2002) were used to estimate apparent survival ( ), movement probabilities ( ) and detection probabilities ( p ) simultaneously. u was defined as the probability for a kite alive in location u (i.e., wetland u ) at time t to survive between time t and t+1 ; and pu was the probability of detecting (sighting) a kite that was alive and associated with wetland u We

PAGE 51

51 defined us as the probability that a kite in wetland u at time t was in wetland s at time t+1 given that it was alive at t+1 Modelled parameters used notation from Senar, Conroy and Borras (2002); time dependency was ( t ) and no time effect was (.). We a ssigned each bird to one of two age classes: juveniles ( juv ), 30 days to 1 year; and adults ( ad ), older than 1 year. Effects embedded in other factors are shown using parentheses. A multiplicative effect is shown by (*) and an additive effect is shown by (+). All computations of the movement and survival probabilities were carried out using program MARK V 4.1 (White and Burnham 1999). Field Methods for the Study of Movement on a Monthly Scale Between 1992 and 1995, 165 adult and 120 juvenile Snail Kites were equipped with radio transmitters with a battery life of approximately 9-18 months (Bennetts and Kitchens 2000). Between 1992 and 1995, aircraft radio-telemetry su rveys were conducted on a weekly basis (two 4-5 hour flights every week) over a large portion of the entire range of the population in Florida. Previous analyses by Bennetts et al (1999a) and Bennetts (1998) found no evidence of radio effects on survival or movement probabilities. Statistical Methods to Estimate Movement on a Monthly Scale Us ing Radiotelemetry Estimating monthly m ovement among regions To estimate monthly movement probabilities ( ) of radio-tagged individuals among regions, we used multistate models. Because monthly survival estimates were beyond the scope of our study, we removed individuals from the anal ysis after they were last observed and fixed survival parameters to 1. For this analysis, we included individuals for which the fate and location could be determined with certainty (i.e ., detection probability equals 1). In addition, birds that temporarily disappeared and then reappeared in the sa mple were censored when they disappeared and were included again when they reappeared (Williams et al 2002). This analysis included six states: the five regi ons described above (E, K, O, L, J) (Figure 3-1), and one state

PAGE 52

52 containing peripheral habitats and matrix area (P all locations outside the sampled areas). To compute the probability of movement out of a patch (wetland or regi on), we summed the transition probabilities out of that patch. To calculate the average monthly probability of movement out of any wetland within a region, we computed the average of the monthly movement probabilities out of every wetland in the region of interest. We tested the effect of patch size ( AR for the surface area of the receiving site, and AD for the surface area of the donor site), distance ( d) region ( r ), age, and time on movement probabilities. The notations for age and time followed the ones common to all analyses. We also tested the effect of year ( year ), given that the radio-teleme try study was conducted between 1992 and 1995. A seasonal effect ( seas ) with respect to three 4-mont hs seasons (January-April, MayAugust, September-December) (Bennetts and Ki tchens 2000); and a breeding season effect ( breed ; breeding season: January-June; non-breed ing season: July-December) were also included. With known fate multistate data (for which the de tection probability is 1), there is currently no appropriate Goodness of Fit test ( GOF ). However, most analyses presented in our study included fairly general models. Estimating monthly movement within regions using radio-telemetry The same method was used for this analysis as for the among-regions analysis. Because two regions comprised several wetland units, we conducted two separate analyses. The analysis for the K region contained four modera tely isolated wetlands (denoted: mw ): Lake Tohopekaliga, Lake East Tohopekaliga, Lake Kissimmee, and a s ite containing all of the small lakes in the surrounding area (Figure 3-1). Analysis for the Everglades region contained five contiguous wetlands (denoted: cw ): WCA3A,WCA3B, Everglades Nation al Park, and Big Cypress (Figure 3-1). We also, aggregated three contiguous wetlands (WCA1,WCA2A an d WCA2B), into one

PAGE 53

53 site, as our data set would not have permitted a seven-site model. Patch size and distance were included as factors in the models of region E only. This analysis was not applicable for region K, because of the site that included all of the small lakes. Field Methods for the Study of Moveme nt and Survival on an Annual Scale We used mark-resighting inform ation collected during the peak of the breeding season (March 1-May 30), for a period of 13 y ears (1992 to 2004). Between 1992 and 2004, 1730 juveniles were marked just before fledging. J uveniles advance to the adult age class at the beginning of the next breeding season (Bennett s et al. 2002). In a ddition, between 1992 and 1995, 134 adults (i.e., older than 1 year) were banded. Bands were uniquely numbered anodised aluminium colour bands. Banded kites were identi fied from a distance, using a spotting scope. Each wetland was surveyed at least once using an airboat. Statistical Methods to Estimate Annual M ovement and Survival Using Banding Data We used a multistate model to estimate annua l movement and survival probabilities. We assigned the location of each bird to four regions (see STUDY AREA). We excluded region L from this analysis to maximize precision, as rela tively few birds were r ecorded in this area. Estimating survival A set of biologically relevant m odels was developed that allowed and p to vary across time, or stay constant for each age class. Because our data set included kites banded as juveniles and as adults, age was modelled both as time si nce marking and as a group effect. We also created models that in cluded drought effect on and p We included a drought effect, which assumed different effects on apparent survival in 2000-2001 and 2001-2002 (denoted: D12). We used this approach because th e drought was likely to affect before and after the 2001 sampling occasion. ND indicated that was constant during the remain ing non-drought years (denoted:

PAGE 54

54 ND ). For juveniles we designed models w ith additive effect of time and region ( t+r ) on but because of the drought few juvenile s were fledged in 2001 (32 juveniles were fledged in K, 3 in J and none in E and O). We thus constructed mode ls with additive effect of time and region on except during the interval 2001-2002, during which was assumed to be similar among regions (denoted: juv ( r + td)). Consequently, during the interval 2001-2002 model juv ( r + td) reflected apparent survival for northern regions (K and J) Because we expect environmental conditions to be more similar among neighbouring regions than among regions that are far apart, we expect survival in regions close to each other, to be simila r. Thus we developed models that assumed similar apparent survival probabi lities in neighbouring regions. Du e to the proximity of regions E and O in the south (separated by 30 km) and K and J in the north (separated by 25 km), (conversely, O and J were separated by 50 km; Figure 3-1), we developed models with a common survival parameter for e ach group of regions (denoted [] E OKJ ; superscripts indicate regions the survival probabi lities pertain to; = indicates that E is the same as O similarly K is the same as J ; indicates that E and O are different from K and J ). Models assuming a different for each region were denoted (( r )). Because the drought intensity was strongest in E, O and J (lowest DSV), and weakest in K (highest DSV), some models assumed similar drought effects on in E, O, and J (denoted [ E=O=J] ( D1-2)); with no drought effect on in K ( K (.)). Estimating annual movement probabilitie s among regions using banding data Our multistate approach using the banding data (described above) provided annual estimates of movement probabilities ( ), among four regions (E, O, K and J). We tested the effect of the drought on move ment between 2000-2001(denoted D1). We also estimated the

PAGE 55

55 probability for a kite to be found in a particular region (u) at year t+1 given that it was present in that same region in year t ( uu) These probability estimates were used to evaluate the level of philopatry at each site. These estimates were obtai ned as one minus the estimated probabilities of moving away from the area. Goodness of fit Previous survival analyses i ndicated a strong age effect on (Bennetts et al 2002). Unfortunately, we are not aware of GOF test accounting for an age effect on for multistate model. However, it is possible to test the fit of adult data sepa rately. We used program U-CARE version 2.02, which tests the fit of the Jolly move (JMV) and Arnason-Schwarz models (AS) (Pradel, Wintrebert and Gimenez 2003). We were only able to test model JMV, which fit the data satisfactorily when testing th e fit of adult data separately ( 2 102 = 104.3, P = 0.42). The fit of the JMV model could not be assessed on juven iles separately (Test M requires > 4 occasions). Thus, as suggested by Senar, et al (2002) we computed a GOF accounting for an age effect (by summing Test 3.SM, Test 2.CT and Test 2.CL, available from program U-CARE, see Choquet et. al 2003), for a site-specific Cormack-Jolly-Seber (CJS) model in lieu of a multistate model. The site-specific CJS model fitted the data satisfactorily ( 2 175 = 152.1, P = 0.89). We concluded that there was no evidence of lack of fit of the multistate model used (i.e., models in Table 3-3 accounted for an age effect on ; Choquet et al. 2003). Model Selection Procedure For each mark-resight analysis, we first devel oped and fitted a set of biologically relevant models that corresponded to our best a priori hypotheses (referred as st arting models). We then developed models whose relevance was linked to the need to eval uate the fit of each of the starting models (Cam, Oro and Jimenez 2004). We used AICc (Burnham and Anderson 2002) as

PAGE 56

56 a criterion to select the model that provided the most parsimonious descrip tion of the variation in the data (i.e. model with the lowest AICc). The value of AICc (the difference between the AICc of a particular model and that of the model with the lowest AICc) was presented in each set of model-selection results. We also used AICc weight ( w ) as a measure of relative support for each model (Burnham and Anderson 2002). We reported only the model whose w was greater than 0.01. Effect of Patch Size an d Distance on Movement Movement probabilities were modelled as lin ear-logistic function of patch size and/or distance (Blums et al 2003). For example, proba bilities of moving from one patch to another in function of distance were modelled as: Logit ( ( d )) = i + d ( d ), where i, d, are the parameters to be estimated. i is the intercept, d is the slope for distance between patch centroids ( d ). Probability of moving was predicted to decrease with increasing distance between patches ( d < 0) (Blums et al 2003). Whenever the 95%CI [d ] estimate did not overlap 0, the relation wa s considered statistically significant. Effect Size To measure the magnitude of the difference be tween estimates we computed estimates of effect size (ES ) as the arithmetic difference betwee n estimates. Whenever the 95%CI [ES ] did not include 0 the difference wa s considered statistically signi ficant (Cooch and White 2005). Estimates of Precision Variances for derived estimates in our study were computed using the delta method (Williams et al 2002). Confidence intervals for estimates that were strictly positive (, ), were

PAGE 57

57 computed using the method proposed by Burnha m et al. (1987) based on the lognormal distribution (Appendix B-1 in Appendix B). Estimates of effect size (not strictly positive), were approximated as follows: 95%CI [ ]= + 1.96 SE[ ]. Results Monthly Movement Probabilities Among Regions Effects of patch size and distance The most parsimonious model (with lowest AICc; Table 3-1.a.), was a model that only included a site-specific effect of movement (()r ). However, the model that assumed movement probabilities to be site-specific for adults, but included a patch-size and a distance-betweenpatches effect plus interaction of these factors for juveniles (()(*)adrjuvARd ), also received some support ( AICc=1.7; Table 3-1.a.). This model had considerably more support than the model that assumed movement pr obabilities to be solely site -specific for adult birds and juveniles (()()adrjuvr ; AICc = 15.2; see also Tabl e B-1.a. in Appendix B). When the analysis is conducte d on juveniles only, the model (*)juvARd is considerably better than ()juvr ( AICc = 14; Table 3-1.b; see also Table B-1.b. in Appendix B), indicating that patch size and distance may be important in determining the movement probabilities of juveniles. Model (*)juvARd indicates that the probability of moving between two locations decreased with distance between these locations ( d = -0.020, 95 %CI =-0.032 to -0.007). Conversely, we could not show a ny relationship between the receiv ing site area and movement with this model ( A R= -0.020, 95 % CI = -0.247 to 0.207). The interaction for this model was positive, but not very strong ( A R*d= 0.002, 95 %CI =0.0001 to 0.003). We also tried a model

PAGE 58

58 with an additive effect of distance an d patch size of the receiving sites, ()().adrjuvARd That model did not reach numerical convergen ce with program MARK when the data set included both juvenile and adult birds; we cons equently ran this model on a data set that only comprised juvenile birds (()juvARd ; Table 3-1b). Although this model was less parsimonious than one that incor porated an interaction effect ( AICc = 3; Table 3-1.b), it was considerably better than the site-specific model ( AICc = 14; see also Table B-1.b. in Appendix B). Model ()juvARd supported the hypothesis of a negativ e relationship between movement probabilities and distance ( d = -0.011, 95 %CI = -0.020 to -0.0030). This model also supported the hypothesis of a positive relationship between movement and size of the receiving sites ( A R= 0.205, 95 %CI = 0.120 to 0.289). The models that included the size effect of the donor patch on juvenile movement ()juvAD received little support ( AICc = 6.5; Table 3-1.b), but the parameter for AD supported the hypothesis that emigration was lower out of larger patches ( A D= -0.191, 95%CI = -0.298 to -0.084). There was no evidence of any patch size or di stance effect on adult movement (Table 3-1.a and Table 3-1.c). Models that included effect s of time, year, or season received no support ( w ~0). Monthly Movement Probabilities Within Regions Movement within the Everglades region The most parsimonious model for this analysis was (*) s eascw ( w ~ 1; Table 3-2.a), which assumed movement probabilities to va ry by season and to be site-specific.

PAGE 59

59 Movement within the K region The most parsimonious model for this analysis assumed movement to vary by season (() s eas ; w = 0.67; Table 3-2.b). Comparison Among and Within Regions The probability that a Snail Kite in any of th e five wetlands in region E moved to another unit in that same region within the next month ( average monthly movement probability among contiguous wetlands), using model (*) s eascw for the Everglades region (Table 3-2.a), was 0.29 (95%CI = 0.24 to 0.35). By contrast, the monthl y movement probabilities from E to the four other regions was only 0.04 (95% CI = 0.03 to 0.05), using model ()r (Table 3-1.a). The same pattern was observed in region K where kites m oved extensively among the moderately isolated wetlands in this region, using model () s eas (Table 3-2.b) we found the average monthly probability = 0.15 (95%CI = 0.13 to 0.17); with only a 0.09 (95%CI = 0.06 to 0.12) monthly movement probability from this region to the four other regions, using model ()r (Table 3-1.a). The probability that kites in any of the five regions moved to another region within the next month (average monthly movement among isolated wetlands), using model ()r (Table 31.a) was 0.10 (95%CI = 0.08 to 0.12). Average monthly movement among contiguous wetlands was significa ntly greater than among moderately isolated wetlands (ES = 0.14, 95%CI = 0.08 to 0.20). Average monthly movement among moderately isolated wetland wa s significantly greater than among isolated wetlands (ES = 0.05, 95%CI = 0.02 to 0.07); and average monthly movement among contiguous wetlands was significantly gr eater than among is olated wetlands (ES = 0.19, 95%CI = 0.13 to 0.25).

PAGE 60

60 Interannual Survival Estimates The most parsimonious model ([][]121()(.)()()(*)(*)EOKJKEOJdjuv ad adad N DDrtprtrD, received overwhelming support from the data ( w = 0.96; Table 3-3). This model had region specific apparent survival for adults, which did not vary over time but differed sign ificantly between drought and non-drought years (Figure 3-2). There was an additive effect of region and time for estimates of apparent survival of juveniles, except for the interval 2001-2002, during which was assumed to be time dependent only. Sighting probabi lities were region and time specific. Movement probabilities were region specific and were a ffected by the drought. Apparent survival estimates for adults kites located in neighbouring regions during non-drought were similar (i.e., E = O and J = K ). During non-drought years E was greater than K (ES = 0.08, 95%CI= 0.03 to 0.13; Figure 3-2). This model also assu med no significant effect of dr ought on adult apparent survival in K (the region with the highest DSV > -1), but assumed a similar effect of drought on adult apparent survival in E, O and J (which all had lower DSV < -1) (see Figure 3-2 for estimates). Average estimates of juvenile apparent survival during nondrought years were higher in southern regions ( E = 0.520, 95%CI = 0.460 to 0.588;O = 0.471, 95%CI = 0.372 to 0.597) than in northern regions ( K = 0.355, 95%CI = 0.233 to 0.541; J = 0.412, 95%CI = 0.295 to 0.575), but confidence intervals ove rlapped. During drought years confidence intervals of region specific juvenile apparent su rvival overlapped widely ( E = 0.07, 95%CI = 0.014 to 0.349;O =0.0647, 95%CI = 0.010 to 0.427; K = 0.054, 95%CI = 0.007 to 0.405; J = 0.058, 95%CI = 0.004 to 0.837). Because no juveniles were marked in 2001 in E and only 4 were

PAGE 61

61 marked in K in 2000, we could not test the hypo thesis of a lower e ffect of the drought on apparent survival of juveniles in K. Given that juvenile apparent survival estimates were not significantly different from one another we averag ed these estimates across regions (Figure 3-2). Estimates of adult apparent survival averag ed across regions remained fairly high and constant over time ( = 0.86; Figure 3-2), but dropped s ubstantially during drought years between 2000 and 2002 (average apparent survival between 2000 and 2002 was = 0.72; Figure 3-2). This represented a relative decrease of 16% in apparent surv ival during the years that were affected by the drought when compared to non-drought years, but the decrease was only significant between 2001 and 2002 (ES = 0.39, 95%CI= 0.24 to 0.53; Figure 3-2). Juvenile apparent survival varied widely over time, but reached a record low between 2000 and 2002 (average between 2000-2002 was = 0.06; Figure 3-2). Juvenile a pparent survival decreased by 86% in 2000 and 2002 (relative decrease) when compared to its average over the non-drought years (average during 1992-1999 and 2002-2003 was = 0.44). Inter-Annual Movement Among Regions and Drought Effect on Movement The most parsimonious model (described a bove; Table 3-3), had site-specific annual transition (movement) probabilities that were constant over time, except during the drought (Table 3-3). This model was substantially be tter supported than the same model without a drought effect ( AICc = 7; Table 3-3). Using the most parsimonious model, we found that during the 2001 drought, movement estimates were higher from the areas with the lowest DSV (i.e., most impacted regions: O and E) toward areas with highest DSV (i.e., least impacted region: K), OK = 0.33 (95%CI = 0.146 to 0.580), EK = 0.030 (95%CI = 0.014 to 0.066), than during non-drought years OK = 0.044 (95%CI = 0.024 to 0.080), EK = 0.015 (95%CI = 0.010

PAGE 62

62 to 0.022). However, the difference was only statis tically significant for birds moving from O to K (ES = 0.28, 95%CI= 0.05 to 0.52). Estimated move ment probabilities toward the mostimpacted region (i.e., E and O) during the dr ought approached 0. This contrasted with nondrought years during which movement probabilities toward E and O were typically much higher than 0 (ranged from 0.02 to 0.16; Table B-4 in Appendix B). Surprisingl y the probability of moving from J to K during the drought appr oached 0, while during non-drought years this probability was JK = 0.06 (95%CI = 0.03 to 0.11). Models including an age effect as well as a drought effect on movement did not reach numerical convergence; however, we did not detect any movement of juvenile bird from the most to the least impacted regions between 2000 and 2001. Models including an age effect on movement but no drought effect were not supported ( w < 0.01; Table B-3 in Appendix B). We used the most parsimonious model to estimat e the probability of staying in each region from one year to another. The probability of staying in E was EE = 0.95 (95%CI = 0.94 to 0.96), the probability of staying in O was OO = 0.76 (95%CI = 0.71 to 0.8 2), the probability of staying in K was KK = 0.72 (95%CI = 0.66 to 0.79), and th e probability of staying in J was JJ = 0.75 (95%CI = 0.69 to 0.82). Discussion Monthly Movement Among Contig uous and Isolated Wetlands We found that kites moved extensively over la rge areas of contiguous wetlands (average monthly movement probability: 0.29). However, our study also showed much less movement among isolated wetlands (average monthly move ment probability: 0.10). As expected average monthly movement probability among moderate ly isolated wetlands was intermediate: 0.15. Differences between these estimates were all stat istically significant. Th ese results agree with

PAGE 63

63 Prediction 1 that loss of connectivity reduces moveme nt of kites. However, as stated in Prediction 1, only by comparing movement among con tiguous wetlands and among moderately isolated wetlands could the effect of connectivity and distance be separated. Indeed, despite the fact that distances between the centroids of contiguous wetlands (E) were greater than between the centroids of moderately isolated wetlands (K), movement among wetlands in E were greater than in K. The results also suggest that seas onality influenced movement within, but not among regions. One possible explanati on is the pronounced wet-dry seas onality resulting in spatiotemporally variable habitat conditions at both the local and regional le vels (Davis and Ogden 1994; Bennetts and Kitchens 2000). The fact that this seasonal pattern was not observed for movements among regions may be due to the high er costs (i.e., mortalit y) associated with moving among regions than when moving within the regions. Patch Size and Distance Between Patc hes as Factors Driving Movement Our modelling approach provi ded supportive evidence that patch size and interpatch distance constitute important factors influencing movement of juveniles at the regional scale. The support for this hypothesis was weak when m ovement was modelled for juveniles and adults simultaneously (Table 3-1.a). However this hypothe sis received substantially more support when juvenile movement was modelled separately (Table 3-1.b. and Table 3-1.c.). Our results are thus consistent with Prediction 2 which predicts that movement probabilities between regions on a monthly scale decrease with distance. The hypothesi s that immigration shoul d be higher toward larger patches because of more freque nt encounters with patch boundaries ( see Prediction 2; Lomolino, 1990; Hanksi, 2001) received some limited support. Indeed, model juv ( AR+ d ) (Table 3-1.b), which supported the hypothesis of a positive rela tionship between movement and the size of the receiving site had a w of 0.17 (Table 3-1.b). Alt hough the model that assumed

PAGE 64

64 higher emigration out of smaller areas for juveniles ( juv ( AD )) was not parsimonious (Table 31.b), examination of the parameter for this model supported this hypothesis. The fact that we only found evidence of a pa tch size and distance effect on the monthly movement probabilities of the juveniles at the regional scale is consistent with Prediction 3 However, we can only infer that juveniles may respond to distance and si ze of the destination site, whereas adults do not (Table 3-1.c.), possibl y indicating that adults are responding to other factors (e.g., habitat quality). Only by includ ing a measure of habitat quality (currently unavailable) in our models could we test the hypot hesis that adult movements are more likely to be determined by the acquired knowledge of the qua lity of the available habitats than by the patch boundary effect. The fact that we found no influence of patch size and distance on monthly movement among contiguous or moderately isolated wetlands can be explained by the fact that movements among these wetlands are so frequent that the effect of patch size effect and distance may be diluted over time (i.e., after a few m onths birds may not search wetlands blindly anymore). If patch size and distance affect moveme nt patterns among patches, one can see how habitat loss and fragmentation may affect disper sal, particularly for juvenile birds. It is particularly likely to increase the search cost when animals move to locate new suitable wetlands. Inter-annual Pattern of Movement Despite relatively high average monthly m ovement probabilities out of regions (e.g., average movement probabilities out of E and K were 0.04 and 0.09, respectiv ely), kites exhibited strong philopatric tendencies to particular regions at an annual scale (e.g., annual estimates of site tenacity for regions E and K were 0.95 and 0.72, respectively).

PAGE 65

65 This extent of site tenacity is surprising given the high environmen tal variability that characterizes the kites range in Florida (Bei ssinger, 1986; DeAngelis and White 1994). Indeed many species that use environments where food resources vary strongly in space and time are often nomadic (e.g., DeAngelis and White 1994). Ho wever movement out of familiar areas may incur important search costs (sta rvation, predation). Kites may also benefit from staying in or returning to familiar regions, as it could cont ribute to maximizing their breeding output and chance of survival (e.g. predation avoidance) (Stamps 2001). In summary, kites movement in this fragmented system varies from site tenacity (between breeding season and at the regi onal scale) to nomadism (with in region on a monthly scale), depending on the spatio-temporal scale of obser vation and hence on the activities of primary relevance at different times a nd places. In particular, one may want to distinguish between breeding (or natal) philopatry and exploratory movements, as the factors governing these processes may be different. Additionally, our results indicate that Snail Kites move substantially less between regions that have b een isolated by human-induced fr agmentation than within these regions. Thus, many kites may have little familiar ity with wetlands located outside their natal region. A regional disturbance coul d therefore have significant de mographic consequences. Kites that are familiar with many landscapes within the populations range may survive a regional drought by moving to other less-aff ected regions, while survival of birds without knowledge of alternative wetlands could be dram atically reduced. The drought th at occurred in Florida in 2001 provided an opportunity to evaluate the effects of this type of natural disturbance on kites. Regional Survival and Resistance of th e Population to Natural Disturbance The analysis of annual movement indicates th at kite movement was affected by the 2001 drought. As expected, a pr oportion of birds moved from the most to the least-impacted regions, which is consistent with Prediction 4.a. (but the drought effect was only significant for kites

PAGE 66

66 moving from O to K). Although m odels including an age effect on annual movement were not supported (possibly because of low sample size), no juveniles that had fledged one year prior to the drought were found to have moved toward re fugia (i.e., only adult birds were observed moving to region K in 2001). This latter obser vation is not based on any robust estimation procedure and therefore should be interpreted with caution. Howeve r, it is worth pointing it out as it supports Prediction 4.b which states that because adul ts are more familiar with the surrounding landscapes they are more likely to reach refugia habitats than juveniles. Despite the fact that a propor tion of kites moved from the most to the least-impacted regions, most birds did not appear to successfully reach refugia habitats and overall, this regional drought had a substantial demogra phic effect on the population (Figur e 3-2), which is consistent with Prediction 5.a The survival analysis conducted over th e last 13 years, at the scale of the whole population, also indicates that appare nt survival varied among regions. During nondrought, adult survival was lower in northern regions (K and J) than in southern regions (E and O), possibly because of lower apple snail av ailability in the northern regions (Cattau unpublished). Juvenile apparent survival was also lower in northern regions than in southern regions during non-drought years, but differences were not statistically significant. Our results supported Prediction 5.b which predicted that survival shoul d be lower in areas most impacted by the drought than in areas least impacted. Adult apparent survival in regions E, O and J (lowest DSV), decreased significantly du ring the drought, while survival in K (highest DSV) did not decrease (Figure 3-2). Prediction 5.b. could not be tested for juveniles because of low sample size. When averaging survival over all the region s apparent survival of adults decreased by 16% during the drought while juvenile apparent su rvival dropped by more than 86% during the drought (Figure 3-2). Thus, the drought had a larger effect on juvenile appa rent survival than on

PAGE 67

67 adult apparent survival, which is consistent with Prediction 5.c. Interestingly, adult apparent survival only decreased signifi cantly between 2001 and 2002, while juvenile apparent survival had already decreased significan tly between 2000 and 2001, indicati ng that juveniles were also more susceptible to early effects of the drought. A d eclining trend in juvenile apparent survival is also evident in Figure 3-2. However, we had no good a priori reason to expect th is trend. It could be due to stochastic variation or unrecognised variations in wetland conditions. Additionally, we should note that out of 65 juveniles equipped in 2003 with radio transmitters, 36 were observed alive between March and May 2004 (M artin et al. unpublished data). Th erefore, juvenile survival between 2003 and 2004 rebounded since the drought to at least 0.55 (detection probability was not accounted for this estimate). The dry-down effects of the drought began in mid January 2001; most of the birds that fledged during the previous breeding season (from the 2000 cohort) were approximately 9 months. Because juveniles are somewhat profic ient at capturing snails after only 2 months (Beissinger 1988), by 9 months thes e birds should be equally effi cient at capturing and extracting snails. Field observations of kite interactions i ndicate no dominance of adu lts over juveniles that are 4 months or older (Martin et al unpublished data). The only major difference in foraging abilities between young and older birds, that we are aware of, woul d be their respective familiarity with the landscapes. Adults would potentially have explored more wetlands than juveniles (Bennetts et al 2002). This may thus explain the weak er effect of the drought on adults (see Prediction 5.c ). We note that the survival estimates presented in this study are apparent survival estimates, indicating that the complement of these estimates includes both mortality and permanent emigration from the study system. Thus, lower survival during drought could be due to both

PAGE 68

68 permanent movement out of the system and lower true survival due to th e drought. It is possible that some kites moved temporarily to periphera l habitats (typically hi ghly disturbed habitats: agricultural areas, large canal) during drought. Although these habitats will typically retain more water than major kite habitats during drought, they are unlikely to be suitable for breeding activity; thus, when conditions improve, most bird s should move back to major wetlands. Hence, because the Snail Kite population in Florida is as sumed to be an isolated population (Bennetts et al. 1999a) and because the geographic scope of our study encompasses the major kite habitats, it is unlikely that many kites remained outside th e sampled areas for three consecutive sampling seasons after the drought. Even if substantial te mporary emigration in to unsampled areas occurred during drought it would not have biased survival if it was followed by movement back into the study system when conditions improved. Conclusions and Conservation Implications Reducing habitat fragmentation has now b ecome almost a rubber-stamp recommendation for maintaining populations of many species of terrestrial mammals, insects, and even birds with reduced dispersal abilities. Howe ver, the benefits may be less obvious when dealing with species able to cover several hundred ki lometres in one day and whose da ily dispersal abilities exceed the distance separating patches that have been isolated through fragme ntation. As suggested by previous theoretical studies (e.g., Doak, Marino and Kareiva 1992) we found that considering scale issues was critical to unde rstanding movement of kites in fragmented landscapes. The case study of the Snail Kite in Fl orida also provides an exampl e of how fragmentation could indirectly affect the persisten ce of species with great disper sal abilities. As suggested by Bennetts and Kitchens (2000) and Bell (1991), e xploratory behaviours may be important for many animals to resist periodic low food avai lability events (such as droughts). Thus, if fragmentation reduces expl oratory movements of kites, it could also reduce resistance of the kite

PAGE 69

69 population to disturbance events. Further work to support this hypothesis may be particularly critical to conserve th is endangered species, but may also be relevant to other avian nomads (e.g., waterbirds in Australia, see Roshier et al 2001).

PAGE 70

70 Table 3-1. Multistate models (with survival and detec tion probabilities equa l to 1) of monthly movement probabilities ( ) of adult ( ad ) and juvenile ( juv ) Snail Kites among the five major regions (E, O, K, L, J) and P (per ipheral and matrix ar eas), based on radiotelemetry data. These models evaluate the effect of patch size distance and regional identity alone on movement probabilities. Model AICc w K a-Movement among regions of juvenile and adult modelled simultaneously (r) 0 0.69 30 ad (r) juv(AR*d) 1.7 0.30 43 ad (r) juv(AD) 8.2 0.01 40 b-Movement among region s modelled using data from juvenile only juv(AR*d) 0 0.79 13 juv(AR+d) 3.0 0.17 12 juv(AD) 6.5 0.03 10 juv(AR) 8.8 0.01 11 c-Movement among regions modelled using data from adult only ad(r) 0 1.00 30 Notes: AICc is the Akaikes Information Criterion. AICc for the ith model is computed as AICcimin (AICc). w refers to AICc weight. K refers to the number of parameters. Only models with w > 0.01 are presented (see Table B-1 in Appendix B, for models with w <0.01). r: region (includes 6 states: E, O, K, J L and P (peripheral and matrix); AR: Area of the receiving site; AD: Area of the donor site; d: distance.

PAGE 71

71 Table 3-2. Multistate (with survival and detecti on probabilities equal to 1) models of monthly movement probabilities ( ) of adult ( ad ) and juvenile ( juv ) Snail Kites among wetlands in the E and K region based on radi o-telemetry data. These models evaluate the effect of patch size, distance, se ason, wetland identity alone on movement probabilities. Model AICc w K a-Movement within the E region of adult and juvenile Snail Kites ( seas*cw ) 0 1 20 b-Movement within the K region of adult and juvenile Snail Kites ( seas ) 0 0.67 3 ad ( seas ) juv ( seas ) 2.6 0.18 6 (.) 5.4 0.04 1 ad (.) juv (.) 6.5 0.03 2 ( mw ) 6.5 0.03 12 ( breed ) 7 0.02 2 ( years*seas ) 8 0.01 10 Notes: Only models with w > 0.01 are presented (see Table B-2 in Appendix B, for models with w < 0.01). cw: contiguous wetland; mw: moderately isolated wetland; seas: season; breed: breeding season. For other notations see Table 3-1.

PAGE 72

72 Table 3-3. Multistate models of annual apparent survival (), sighting ( p ), and movement probabilities ( ) of adults ( ad ) and juveniles ( juv ) Snail Kites based on banding data. The drought effect on during 2000-2002 was denoted D1-2. The drought effect on in 2001 was denoted D1. Constant during non-drought years (1992-2000 and 20022004) was denoted ND Because all models included in Table 3 had region and time dependent sighting probabilities ( p ( r*t )), Table 3-3 only includes model structures for and Model AICc w K [][]121()(.)()()(*)EOKJKEOJdjuv ad adad N DDrtrD 0 0.96 85 [][]12()(.)()()()EOKJKEOJdjuv ad adad N DDrtr 7 0.03 78 Notes: Only models with w > 0.01 are presented (see Table B-4 in Appendix B, for models with w < 0.01). t: time (years); r+td: additive effect of region and time on except during 2001-2002,during which was time dependent only; .: is constant during 19922004. Superscript indicate region specific ; =: regions have identical ; : regions have different For other notations see Table 3-1.

PAGE 73

73 Figure 3-1. Major wetlands used by the Snail Kite in Florida. Regions: Kissimmee Chain of Lakes (K), Everglades (E), Lake Okeec hobee (O), Saint Johns Marsh (J), and Loxahatchee Slough (L). Modera tely isolated wetlands incl uded in K are: East Lake Tohopekaliga (1), Lake Tohopekaliga (2), La ke Kissimmee (3), as well as the small lakes coloured in grey with in the rectangle. Contiguous wetlands included in E: Water Conservation Areas 1A (4), 2A (5), 2B (6), 3A (7), 3B (8), Everglades National Park (9), and Big Cypress National Preserve (10). The grey colouring of the wetlands indicates the area of the wetlands that were included in this study. The thick contour lines delimit regions that include several wetlands. The dotted line indicates the historic Kissimmee-Okeechobee-Evergl ades watershed which constituted a network of well connected wetlands (Dav is and Ogden 1994; Light and Dineen 1994).

PAGE 74

74 Figure 3-2. Apparent survival ( ) between 1992 and 2003 of adult and juvenile Snail Kites, obtained using the most parsimonious model in Table 3. Error bars correspond to 95% confidence intervals. During non-drought years (1992-2000 and 2002-2003), of adults were similar in E and O; and in K and J. During drought (2000-2002), of adults were similar in E, O and J, but different in K. For readability, only in E and K are presented for adults. of juveniles were averag ed across regions. Arrow indicates the beginning of the drought that started in January 2001. Estimates between 1992 and 1999 were consistent with Bennetts et al (2002).

PAGE 75

75 CHAPTER 4 NATAL LOCATION INFLUENCES MOVEMENT AND SURVIVAL OF THE SNAIL KITE Introduction In heterogeneous environments, the place of birth (or natal location) is likely to be critical to the growth, survival, and future reproduction of organisms. Indeed, resource availability and other factors (e.g., parasitism) that can affect vital rates (e.g., survival and reproduction) often vary in space and time (Latham and Poulin 2003; Pettorelli et al. 2003). Moreover, it has been shown that environmental conditi ons during prenatal and early pos t-natal stages can have long term consequences on key life history traits (e.g., lifespan and age of first reproduction) (Metcalfe and Monaghan 2001). Natal location, by shaping the habitat preferences of many animals, may also affect their movement a nd settling decisions during the course of their lifetimes (Stamps 2001; Davis and Stamps 2004). Thus, natal location may influence the ecological dynamics of wild popul ations through its effects on movement, habitat selection and survival (Stamps 2001; Blums et al. 2003; Davis and Stamps 2004). Despite the importance of better understand ing how natal location affects movement decisions and survival, rigorous analyses to esti mate movement and survival in relation to the place of birth are lacking (Blums et al. 2003). Most past studies used ad hoc methods to estimate site fidelity (e.g., return rates, revi ewed in Doherty et al. 2002). These ad hoc measures can be severely biased, because they are also functions of resighting and survival probabilities (Doherty et al. 2002). Modern analytical techniques, su ch as multistate models provide more robust estimates of movement and survival (Dohert y et al. 2002). Although a few studies have now successfully used multistate models to estimate site fidelity (typically to breeding sites) (Hestbeck et al. 1991; Lindberg et al. 1998), only a handful of studies have used these models to estimate philopatry to the natal site (Lindberg et al. 1998). In fact, we ar e not aware of any study

PAGE 76

76 that used robust estimators of m ovement to compare fidelity to th e natal site (at the adult stage) with fidelity to non-natal sites. Nonetheless, the fact that some animals may exhibit a particular attraction toward their natal site at the adult stage could be critic al to the dynamics of many wild populations. Indeed, attraction by in dividuals to the natal site c ould influence habitat selection, patterns of patch occupancy, reproduction, and survival (see Schjrring 2002). We studied patterns and consequences of m ovement related to the natal site in a geographically isolated and spatially st ructured population of Snail Kites ( Rostrhamus sociabilis ) in Florida. The Snail Kite is a highly specializ ed raptor that feeds almost exclusively on freshwater snails (Beissinger 1988) Because of this food specializ ation, Snail Kites are confined to the remaining wetlands in Central and Sout h Florida (Takekawa and Beissinger 1989). The environment occupied by the Snail Kite in Flor ida is highly variable sp atially and temporally (Beissinger 1986). Animals such as Snail Kites that occupy habitats whose food abundance is unpredictable in space and time ar e generally expected to be no madic (Bennetts and Kitchens 2000; Wiens 1976). By modeling movement rates among wetlands used by this bird, Bennetts and Kitchens (2000) measured the extent of nomad ism. However, Martin et al. (2006) found that despite frequent exploratory movement, Snail Ki tes showed strong site tenacity on an annual scale during the breeding seas on. Bennetts and Kitchens (2000 ) hypothesized that exploratory movement may familiarize kites with their landscapes. It has been recognized that exploring unfamiliar habitats is energetically costly (Sch jrring 2002). However, if the exploration phase occurs at times when search costs are minimal (i .e., when food is most abundant), the familiarity with alternative locations gained by birds duri ng exploration may greatly outweigh exploration costs by reducing the risk of mortality during a subsequent disturbance, such as a regional

PAGE 77

77 drought, which dramatically decrease snail availa bility to kites (Beissinger 1995; Bennetts and Kitchens 2000; Martin et al. 2006). On the other hand, benefits of staying in or returning to a familiar area range from increased competitive ability to enhanced predat or avoidance (Stamps 2001). Thus staying in or returning to a familiar habitat may also, in some circumstances, increase an animals probability of surviving and/or of breeding successfully. Although Martin et al. ( 2006) found that kites exhibit site fidelity (i.e., phil opatry) to certain regions during the breeding season, these authors did not specifically examine the level of site tenacity relative to the natal site. Nonetheless, distinguishing between site fideli ty specific to the natal site as opposed to site fidelity to nonnatal sites, and examining consequences on su rvival, may be essential to disentangle the ecological dynamics of many vertebrates. Hypotheses and Predictions Prediction 1 If kites prefer their pl ace of birth when compared to any post-dispersal sites they may have explored in the course of their lifetimes, we expect: (1) movement from post-dispersal sites toward birds natal site to be greater than movement from postdispersal sites toward non-natal sites; (2) we expect greater movement toward th e natal site than away from the natal site; (3) finally, we predict that philopatr y to the natal site should be gr eater than philopatry to non-natal sites. Here, we define philopatry as the probability for a kite to be found in a particular region at year t+1 given that it was present in that same region in year t (Martin et al. 2006). Thus, philopatry to the natal site (or natal philopatry) corresponds to the probability for a kite to be found in its natal region at year t+1 given that it was present in its natal region in year t Prediction 2

PAGE 78

78 Wetland conditions affect surviv al (Beissinger 1995; Bennett s and Kitchens 2000; Martin et al. 2006) and movement (Takekawa and Beissi nger 1989; Bennetts and K itchens 2000; Martin et al. 2006) of Snail Kites. During a regional dr ought a proportion of Snai l Kites is likely to move to the least disturbed areas (Takekawa and Beissinger 1989; Benne tts and Kitchens 2000; Martin et al. 2006). Consequently, we expect the level of natal ph ilopatry to vary according to wetland conditions. Kites should be less philopatric to their natal site wh en it is affected by a drought. Conversely, during a drought, natal philopatry of birds whos e natal areas are located in refugia habitats (areas least affected by droughts) should be great er than during non-drought years. Because of their greater familiarity with the paths linking post-dispersal habitats to their natal habitat, birds hatched in a refugia habitat should also have higher pr obabilities of moving to that refugia during a drought. Prediction 3 If natal location influences adul t survival, we expect adult survival to vary substantially among groups of kites that were hatched in different regions. Prediction 4 Because we expect kites to be more likely to stay in or return to their natal location (see Prediction 1 ), we predict that during a drought, kites whose natal site is located in refugia habitats should be less impacted by a drought (i.e, their survival should decr ease less) than birds whose natal site is located out side of a refugia habitat. Study Area We sampled four major wetland complexes that encompass a very large proportion of all the landscapes used by the Snail Kite in Florid a (Figure 4-1). Most of these wetland complexes (hereafter referred to as regions) consist of several wetlands that were separated by small physical barriers (e.g., road, levee or limited exte nt of non-wetland areas) easily crossed by Snail

PAGE 79

79 Kites (Bennetts 1998). On the other hand, each regi on was isolated from the others by extensive areas mostly unsuitable for breeding or foraging (i .e., matrix), and not easily crossed by kites. We used the same four primary regions as Benne tts et al. (1999a): Everglades E; Kissimmee Chain of Lakes K; Lake Okeechobee L; Saint Johns J (Figure 4-1). Material And Methods Field Methods Capture-mark-recapture The Snail Kite population in Florida has b een monitored since 1992 using capture-markrecapture methods (Bennetts et al. 1999a). Beca use our study focused on movement and survival related to the natal region of Snail Kites, we only included birds bande d as juveniles (before fledging, at approximately 30 days), whose nata l regions we knew (sample size: 1722 birds). Those birds were captured and marked directly at the nest during the peak of the breeding season between 1992 and 2004. Birds were then resighted during annual surveys (which also took place during the peak of breeding seas on: March through June ). Each region was surveyed at least once using an airboat, and bands were identified using a spotting scope. Data Analysis Multistate modeling Multistate models (Hestbeck et al. 1991; Williams et al. 2002) simultaneously estimate transition probabilities among geographic states sighting probabilities p, and apparent survival probabilities (hereafter simply referred to as survival). We defined QS as the probability that an animal in region Q at time t is in region S at time t+1 given that it is alive at t+1 In this example, regions Q and S are geographic states. We defined pQ as the probability that an individual alive in region Q in year t is sighted (Williams et al., 2002). We

PAGE 80

80 defined Q as the probability of surviving (and not permanently emigrating from the study system) over the interval [ t t+1 ] for a kite alive in year t in region Q (Williams et al., 2002). Modeled parameters used notation from Martin et al. (2006); time dependency was t and no time effect was .. We assigned each bird to on e of two age classes: j uveniles (denoted J), 30 days to one year in age; and adults (denoted AD), older th an one year. Effects embedded in parameters (i.e., and p) were shown using parentheses. A multiplicative effect was denoted *. Additional effects included: (1) a drought effect; (2 ) an age effect denoted age; (3) a natal region effect denoted nr; and (4) a regional effect de noted r. The regional effect r allowed parameters of interest (i.e., or p) to vary among regions. Regions included both natal and non-natal regions. We us ed multistate models with four groups (each group included birds hatched in the same natal region: E, L, K or J); and four geographic states (each state corresponded to a region: E, L, K or J) to estimate movement and survival of Snail Kites. It is im portant to recognize the distinction between groups and geographic states. A particular kite can belo ng to only one group (whi ch corresponds to its natal region). In contrast, a k ite can move from one geographic state to another. A geographic state corresponds to a region occupied by a kite at some point in time. Movement First, we used a classic parameterization of multistate models which ignored any effects of natal regions (e.g., Hestbeck et al. 1991, Blums et al. 2003, Martin et al. 2006). For instance, model (r) allowed transitions to vary among regions (i.e., geographic states) but assumed no natal region effect. We used an alternative parameteri zation (hereafter referred as parameterization NOA) that allowed us to de velop models that estimated three types of transition probabilities for each group of kite s hatched in the same region (each group was

PAGE 81

81 exclusively comprised of birds from the same na tal region). The first two types of probabilities were transitions between the natal region (denoted N, for natal) and post-dispersal regions (denoted O, for other). Post-dispersal regi on O potentially included any of the four major regions used by kites except for their natal region. The notation NO corresponded to transition probability from region N to O, while ON corresponded to transition probability from region O to N. In addition to ON and NO, we were able to estimate transition OA: transition from region O (occupied at time t ) toward any post-dispersal region A (at time t+1 ). A included any regions except for the natal region and the post-dispersal region occupied at t while in region O. Next, we provide an example of the constraints that were used to estimate ON, NO, and OA. For example, for all kites that were hatched in region E, we developed models that assumed: (1) EEEELEKEJ ENO (subscripts indicate the natal region; in this example all transitions probabilitie s pertain to kites that were hatched in region E); (2) EEEEEELKLJKLKJJLJKEOA; and (3) EEELEKEJE EON. We used similar constraints to model movement of kites hatche d in the other three regions. These constraints were imposed in order to compar e models that formalised our a priori hypotheses (see HYPOTHESES AND PREDICTIONS). We developed four different types of models using the parameterization NOA: (1) [NOONOA]nr (which assumes NO, ON and OA to be different and to vary among natal regions); (2) [NOOAON]nr; (3) [NOONOA]nr; and (4) [NOONOA]nr. These four types of models were developed to evaluate whether NO, ON and OA differed substantially within each natal region. In addition, we developed a model that included an interaction between a natal regi on effect (nr) and a region effect (r) on movement. This model was denoted r*nr Finally, because a drought occurred between

PAGE 82

82 January and August 2001, we evaluated mode ls that included a drought effect on between 2000 and 2001 (see also Martin et al. 2006). This effect was denoted Dm. We developed a model which assumed that NO, ON and OA differed among groups of birds hatched in different natal regions but also differed duri ng drought and non-drought y ears. This model was denoted [NOONOA]mnr*D. We also considered a model that included an interaction between nr, r and Dm. This model was denoted mr*nr*D Based on model averaged estimates (see MODEL SELECTION) of movement parameters of all models described above we derived natal philopatry (denoted NN; the probability for a kite to be found in region N at time t+1 given that it was present in N at time t ); and philopatry to nonnatal site (denoted OO; the probability for a kite to be found in region O at time t+1 given that it was pr esent in O at time t ; in other words OO is the probability that a kite located in a particular non-natal region at time t will be found in that same non-natal region at time t+1 ). These estimates of philopatry were obtained as one minus the estimated rates of moving away from the area of interest, and the associated variance of these derived estimates was computed using the delta method (see also Hestbeck et al. 1991; Williams et al. 2002). Survival We developed models that allowed and p to vary across time or remain constant for each age class. Given that environmental conditi ons are similar among certain groups of regions as a result of spatial proximity, similar topograph y, similar latitude or management (see Bennetts 1998), we developed models that assumed similar apparent survival probabilities across groups of regions. Thus, we developed models based on proximity (see Figure 41) that evaluated the similarity of survival in the northern regions K and J. Even though the boundaries of E

PAGE 83

83 and L are closer together than L and J, th e distance between the centers of E and L is greater than the distance between the centers of L and J (Figure 4-1). Therefore, we considered models allowing survival in L to be different from K J and E. An analysis of hydrological data by Bennetts (1998) showed that wetland conditions in E and L were the most positively correlated among all the regions. We therefore developed a model that assumed apparent survival to be similar in these two regions. To illustrate our notation system we chose a model with a common survival paramete r for two groups of natal regions: nr[ELKJ] The superscript nr indicates that is natal region specific. In m odels with the superscript nr, survival was allowed to vary among natal regions (i .e., groups) but it was c onstrained to be equal among regions occupied (i.e., geographic states ). The superscripts on the right of model nr[ELKJ] indicate regions to which the survival probabilities pertain (= indicates that nrE is the same as nrL; similarly nrK is the same as nrJ ; indicates that nrE and nrL are different from nrK and nrJ ). We also constructed models in which adult survival varied among regions; we referred to thes e survival rates as r egion specific survival. In models that estimated region specific survival, survival parameters were allowed to vary among regions occupied (i.e., geog raphic states), but we re constrained to be equal among natal regions (i.e., groups). Region specific surv ival estimates were obtained using the same parameterization as Martin et al (2006) and were denoted by a superscript r instead of nr (e.g., r[ELKJ]). In age structured models, adult survival was denoted AD and juvenile survival was denoted J We also created models th at included a drought effect on As in Martin et al. (2006), the drought effect on survival was modeled as a two years effect and was denoted D. Conversely, a no drought effect on was denoted ND. We used this approach

PAGE 84

84 because the drought was likely to affect before and after the 2001 sampling occasion. We used subscripts on the right of to indicate whether the mode l structure of a particular component of pertained to drought years (subscript d reflected survival during the interval 2000-2002) or non-drought years (subscript nd reflected survival dur ing the interval 1992-2000 and 2002-2004). For example, the first part of model [ELKJ]K[ELJ] AD, AD,AD,rrr(.)(ND)(D)d ndd indicates that duri ng non-drought years E ADr L ADr K ADr and J ADr differed substantially while remaining constant over time. The second part of this model (i.e., K AD,r(ND)d) indicates that during the period that corresponded to droug ht years (2000-2002) there was no drought effect on K ADr. The third part of the model (i.e., [ELJ] AD,r(D)d) indicates that during drought years there was an effect of the drought on E ADr L ADr and J ADr. The symbol between E, L and J indicates that E ADr L ADr and J ADr were different during the drought. Because the drought intensity wa s strongest in E, L and J, and weakest in K (see WETLAND CONDITI ONS), some models assume d similar drought effects on r in E, L, and J (e.g., [ELJ] AD,r(D)d) with no drought effect on r in K (e.g., K AD,r(ND)d) (see also Martin et al. 2006) The absence of subscript nd or d indicated that the survival term reflected the entire pe riod of study (1992-2004). For example model AD(r*nr) assumed a multiplicative effect between r and nr on adult survival for the entire period of study (1992-2004). This model did not assume any dr ought effect on adult survival. Model AD(r*nr*D), on the other hand, assumed a multiplica tive effect between r, nr and D.

PAGE 85

85 Model Selection, Goodness of Fit and Program Used We developed a set of models that corresponded to our best a priori hypotheses. Next we used Akaike information criterion adjusted for small sample size (AICc) (Burnham and Anderson, 2002) as a criterion to select the model that provided the most parsimonious description of the variation in th e data (i.e., model with the lowest AICc). Models with a value of AICc (the difference between the AICc of a particular model a nd that of the model with the lowest AICc) less than two were considered to receive a substantial le vel of empirical support (Burnham and Anderson 2002). We al so used the AICc weight ( w ) as a measure of relative support for each model (Burnham and Anderson, 2002). Values of w range from 0 to 1 (with 0 indicating no support; and 1 indi cating maximum support). The sum of weights of all models including a particular effect was denoted ( wt ). As recommended by Burnham and Anderson (2002), we used model averaging to comput e estimates of movement and survival. All computations of the movement and survival ra tes were carried out using program MARK V 4.1 (White and Burnham 1999). We used the sin link function available from program MARK. Estimates of standard error ( SE) and 95% confidence intervals (95%CI) were obtained directly from program MARK. We used the delta method to compute estimates of precision for derived parameters (Burnham and Anderson 2002). All the c onfidence intervals for the derived estimates in our study were approximated as follows: 95%CI [ ]= + 1.96* SE[ ]. Goodness of fit (GOF) tests of multistate models were computed using program U-CARE version V2.22 (Pradel et al. 2003; Choquet et al. 2005). Program U-CARE test the fit of the fully time dependent Jolly Move model (JMV) to the data. To our knowledge there is currently no rigorous test that directly asse sses the fit of the Arnason-Schwarz (AS) model that accounts for an age effect on survival ((age*r*t)(r*t)p(r*t)). However, program U-CARE tests the fit

PAGE 86

86 of the Jolly-move model (JMV) th at accounts for an age effect on (hereafter referred as model JMVA) (Choquet et al. 2005). This test is also generally valid for the AS model that accounts for an age effect on survival (Cooch and White 2006). Indeed, in practice the JMV model is unlikely to show signifi cantly better fit to the data than the AS model (Cooch and White 2006). The JMV model differs from the AS model in that it allows the capture probability for time t+1 to depend on the state at periods t and t+1 whereas the AS model only allows the encounter probability to depend on the current st ate and time. However, the dependence of the capture probability at time t+1 on the state at periods t and t+1 is unlikely to be often observed in practice (Cooch and White 2006). The test for the JMVA model requires the summation of the component WBWA, 3G.Sm, M.ITEC, M.LTEC (see Choquet et al. 2005, for a detailed description of the procedure). If the test is significant (i .e., P < 0.05), Choquet et al. (2005) recommend a correction for overdispersion. Effect Size We estimated the magnitude of the differen ce between two estimates of movement (or survival) by computing the arithmetic difference between these estimates (hereafter referred as effect size denoted ES). Whenever the 95%CI[ES] did not overlap 0, we considered the difference to be statistically significant (Cooch and White 2006) We note that estimates of 95%CI[ES] take into consideration covariances between estimated movement and survival rates. Therefore, differences between estimates ma y be statistically si gnificant based on 95%CI[ES], even though 95%CI of the estimates to be compared overlap. Notes Concerning Regional Specific Survival We note that the parameterizations used in the present study to estimate region specific survival were similar to the ones described in Mar tin et al. (2006). Therefore estimates of region

PAGE 87

87 specific survival should theore tically not differ substantially between the two studies. Any differences between these estimates should primarily be due to the fact that the data set used in the present study included exclusiv ely kites whose natal locations were know n. In contrast, all models assessing the influen ce of natal location on move ment and survival used parameterizations that differed s ubstantially from any models pres ented in Martin et al. (2006). We emphasize that comparing estimates of regio n specific survival (parameterizations used by Hestbeck et al. 1991; Martin et al. 2006 and other authors) wi th estimates of natal region specific survival (parameterization specific to th is paper), was necessary to fully evaluate the importance of natal location on regional survival of adult birds. It was also important to make these comparisons using a common data set that only included individual s whose natal locations were known (as opposed to using the data set anal yzed by Martin et al. 2006). Model averaged estimates of region specific survival were obtained by model averaging estimates from models that assumed survival to be region specific wh ereas model averaged estimates of natal region specific survival were obtained by model averagi ng estimates of models that assumed survival to be natal region specific. Wetland Conditions Martin et al. (2006) determined wetland c onditions during drought us ing hydrological data. Their results suggest that E, L and J were the regions most aff ected by a drought that occurred between January and August 2001 and K was the least affected. Martin et al. (2007a) and Bennetts (1998) found high levels of spatio-t emporal variation in wetland conditions. Not surprisingly they found a positive co rrelation in water levels am ong wetlands that were located nearby. The coefficient of correlation decreased as distance increased. Martin et al. (2007a.) also found that during most multiregional drought s that affected region E (e.g., 1981, 1985, 1989,

PAGE 88

88 1990, 1991, 1992 and 2001) at least one of the wetland s within region K was wetter (based on hydrological indicators) than any other wetland wi thin the region E (Martin et al. 2007a). Results GOF Tests The GOF test indicated that model JMVA fit the data satisfactorily ( 2 137 = 82, P > 0.99). This test is also generally valid for the most general model in our model set (i.e., (age*r*t)(r*t)p(r*t)). Therefore, there was no need to correct for overdispersion. Movement The two most parsimonious models included the component [NOONOA]nr ( wt = 0.54, Table 1). These models assumed moveme nt probabilities between N and O (NO); between O and N (ON); and between O and A ( OA) to be different for each natal region. Model averaged movement estimates ON were greater than either NO or OA. This was true for all natal regions (Figure 4-2). Differences between ON and OA were statistically significant for regions E (ES= 0.45, 95%CI = 0.27 to 0.62), K (ES= 0.14, 95%CI = 0.04 to 0.25), and J (ES= 0.21, 95%CI = 0.05 to 0.38), but not for region L (ES= 0.05, 95%CI = -0.09 to 0.18). Differences between ON and NO were statistically significant for regions E (ES= 0.46, 95%CI = 0.29 to 0.63), K (ES= 0.10, 95%CI = 0.01 to 0.19), and J (ES= 0.17, 95%CI = 0.02 to 0.33), but not for region L ( ES= 0.09, 95%CI = 0.02 to 0.19). Models [NOOAON]mnr*D received less support from the data than model [NOOAON]nr ( AICc = 14.2; see also Appendi x C). Similarly models [NOOAON]mnr*D received less support from the data than model [NOOAON]nr (difference in AICc between

PAGE 89

89 these two models was 6.8, Tabl e 1). Therefore the hypothesis of a drought effect on movement received little support fr om our data. Models [NOONOA]nr, [NOONOA]nr, [NOONOA]nr, r t and received considerably le ss support based on AICc weights ( w ~ 0) (see also Appendix C). Models [NOONOA]nr*age, nr*r and mnr*r*D did not reach numerical converg ence when optimizing the likelihood. Comparison of Natal Philopatry and Philopatry to Non-Natal Site Estimates of natal philopatry were greater than estimates of philopatr y to non-natal regions (Figure 4-3). The differences between these esti mates were statistically significant for regions E (ES= 0.49, 95%CI = 0.31 to 0.66) and L (ES= 0.16, 95%CI = 0.02 to 0.31), but not for regions K ( ES= 0.01, 95%CI = -0.12 to 0.13), and J (ES= 0.10, 95%CI = -0.07 to 0.26). Survival Adults Models that measured region specific survival received mo re support than models that measured natal region specific survival ( AICc > 8.4, Table 1). Models that assumed region specific survival to be similar for regions K a nd J in the north and E and L in the south during non-drought years (e.g., [ELJK] AD,r(.)nd) were among the most parsimonious models ( wt = 0.74, Table 1). Models that assumed no drought effect on region K were well supported ( wt ~ 1, Table 1), as were models that assumed si milar drought effect for regions E and L ( wt ~ 0.99, Table 1). Figure 4-4 shows a substantial decreas e in survival in regions E and L. On the other hand, the effect of the drought on region J was not clear. Indeed, the AICc was only 1.7 between the most parsimonious model (i.e.,[ELJK][KJ][EL] AD,AD,AD,rrr(.)(ND)(D)nddd), which assumed no drought effect on J and the second most parsimonious model which assumed a

PAGE 90

90 drought effect on J (i.e.,[ELJK]K[ELJ] AD, AD,AD,rrr(.)(ND)(D)d ndd, Table 1). Models AD(r*nr), AD(r*nr*D), [ELJK] ADr(t), [ELJK][ELJK] AD,AD,rr(.)(ND)ndd, [ELJK][ELJK] AD,AD,rr(.)(ND)ndd received almost no support based on AICc weights ( w ~ 0) (see also Appendix C). Among models that estimated natal region specific survival, model [ELJK][KJ][EL] AD,AD,AD,nrnrnr(.)(ND)(D)nddd was the most parsimonious, and model [ELJK][KJ][EL] AD,AD,AD,nrnrnr(.)(ND)(D)nddd was the second most parsimonious (difference in AICc between these two models was 2.5, see Table 1). As pointed out earlier, models that assume d region specific survival were better supported than models that assumed natal regi on specific survival. However, model averaged survival estimates for these two types of models were similar ( ES 0.06, Figure 4-4), except for region J (ES = 0.14, but the difference was not statis tically significant 95 %CI = -0.02 to 0.30, Figure 4-4). Estimates of natal region specific survival vari ed significantly among groups of kites hatched in different regions (Figure 4-4). During non-drought periods natal region specific survival estimates were not signi ficantly greater in E than in L (ES= 0.031, 95%CI = 0 to 0.06, Figure 4-4). Nat al region specific survival estimat es were greater in L than in K ( ES= 0.08, 95%CI = 0.04 to 0.12, Figure 4-4); and were greater in L than in J ( ES= 0.08, 95%CI = 0.04 to 0.12, Figure 4-4). Overall, during non-drought years, natal region specific survival estimates were significantly great er in the southern regions (E and L) than in the northern regi ons (K and J) (ES= 0.11, 95%CI = 0.05 to 0.16, Figure 4-4). During the time interval when the drought most severely im pacted adult survival (2001-2002), natal region specific survival in the southern regions (E and L) was significantly lower than in northern

PAGE 91

91 region K (ES= 0.35, 95%CI = 0.23 to 0.48, Figure 4-4); it was also lower than in northern region J ( ES= 0.35, 95%CI = 0.19 to 0.52, Figure 4-4). Juveniles Models that assumed juvenile surv ival to be time dependent (i.e., J(t)) received the most support ( wt ~ 1, Table 4-1). Models J(nr*t), J(nr*D) and J(.) received almost no support ( w ~ 0, see also Appendix C). We do not present juvenile survival estimates because they were not the focus of the current analysis. Furthermore these estimates were almost identical to the ones published in Martin et al. (2006). Detection Probabilities Models that assumed a multiplicative effect of region and time on detection (i.e., p(r*t)) received the most support ( wt ~1, Table 4-1). Models p(t); p (age*t); p(age*r*t); p(r), p(.) received almost no support ( wt ~ 0, see also Appendix C). Discussion Effect of Natal Region on Movement This study, which focuses on movement and surv ival related to the place of birth, shows that Snail Kites, in addition to exhibiting some hi gh level of site tenacity to most regions during the breeding season (Martin et al 2006), also exhibit a particular attraction for their natal region. Indeed, using multistate models with four ge ographic states we found that estimates of movement from post-dispersal regions toward bird s natal region were greater than movement from post-dispersal regions toward non-natal regi ons (differences were st atistically significant for regions E, K and J, Figure 4-2). We al so found that estimates of natal philopatry were greater than estimates of philopatry to non-natal regions (differe nces were statisti cally significant

PAGE 92

92 for regions E and L, Figure 43). Finally, we found that estima tes of movement were greater toward the natal region than away from the natal region (differences were statistically significant for regions E, J and K). These findings provide evidence supporting Prediction 1 The extent of affinity to the natal region appeared to depend on the region (Figure 4-2 and Figure 43). Interestingly, our resu lts show that it may be worthwhile to compute simultaneously the three estimates of natal attraction described in this st udy. The fact that the e ffect of natal location on movement was not statistically significant for a ll regions when using all estimators may be due to lack of statistical power (e .g., due to small sample size). Alte rnatively, one should note that even if movements toward natal sites are substant ially greater than toward non-natal sites, when movements from post-dispersal sites toward nonnatal sites are small, estimates of natal philopatry may not differ from estimates of phil opatry to non-natal site s (Figure 4-2 and Figure 4-3). Fidelity to the natal region may benefit kite s in numerous ways including reducing the risks associated with movement. Movement of ten incurs costs (Baker and Rao 2004), it is energetically costly (Schjrring 2002) and it can increase the risks of predation (Stamps 2001). Furthermore, familiarity with a particular region may potentially increase the fitness of kites if performance of activities related to foraging or reproduction can be improved through training and learning of habitat feat ures specific to a partic ular region (Stamps 2001). On the other hand, exploring habitats away fr om the natal region may be critical to the survival of most kites, given the unpredictabil ity of the system used by kites (Beissinger 1986; Bennetts and Kitchens 2000; Ma rtin et al. 2006). A combinat ion of exploratory movement during the non-breeding peri od when search costs are relatively low, and a natal-philopatric type

PAGE 93

93 of behavior when birds begin re productive activities, is an appe aling hypothesis to explain the movement of kites in Florida, but this hypothesis remains to be tested. Models that included a drought effect on movement were not well supported based on AICc, and therefore, did not support Prediction 2 This contrasts with a previous study by Martin et al. (2006) that indicated ev idence of a drought effect on move ment. The discrepancy between the two studies may be due to a difference in samp le size. In the present study we used only data on kites for which the natal lo cation was known, which significantly reduced our sample size, and therefore, our ability to detect drought e ffect on movement. Given previous findings that most droughts affect regions differe ntially, and given the fact that the probability of staying or returning to a particular region is greatly dependent on the region of birth, we can expect great influences of the natal region on kite survival. Next we examine effects of the natal region on kite survival. Influence of Natal Region on Survival We found that estimates of adu lt survival varied substantiall y among groups of kites that were hatched in different regions (see natal region specific survival Figure 4-4). This result is consistent with Prediction 3 When comparing models that assumed natal region specific survival with models that assumed region specific survival, region specific models received more support from the data. However, estimates of region specific and n atal region specific survival were very similar (Figure 4-4). Thus, our results suggest that kites experience survival rates that are characteristic of the region occ upied (i.e., geographic state) when survival is measured, but because kites have a tendency to stay in or return to their na tal region more than to any other regions, as a consequen ce adult survival is ultimately influenced by the natal region. Interestingly, during most years, adult surviv al rates of birds hatched in the southern regions (i.e., E, L) were hi gher than for birds hatched in th e northern regions (i.e., K and

PAGE 94

94 J regions). The difference in survival rates be tween birds hatched in various regions may be explained by varying quality among habitats. For inst ance, snail availa bility to kites appear to be lower in the northern regions during non-drought conditions (Cattau unpublished data, also reviewed in Mart in et al. 2006). During the drought that occurred in 2001, adu lt birds hatched in E and L suffered high mortality rates, while birds hatched in K were substantially less affected (Figure 4-4). This finding is consistent with Prediction 4 Martin et al. (2006) found that during a drought kites moved from regions most affected by the drough t toward regions least impacted by the drought. In the current study, we did not find any eviden ce of a drought effect on movement (possibly because of low sample size). Nonetheless, we f ound that kites exhibit a pa rticular attraction to their natal region (Figure 4-2 and Figure 4-3). Therefore, at any one time, kites are likely to be found in their natal region. Thus many of the birds hatched in K may have survived the drought by staying in region K, or by moving back to natal regi on K (which as reflected by the hydrological indicator was th e region least impacted by the 2001 drought). Indeed, we expect birds that were hatched in K to be more likely to reach that region than birds hatched in other regions because of their potential ly greater experience with the pa th linking region K to their post-dispersal regions. The impact of the dr ought on J remains unclear. On one hand, model averaged estimates of natal region specific surv ival estimates did not i ndicate any substantial decrease during the drought for region J (Fig ure 4-4). On the other hand region specific survival indicated a decrease during the drought a lthough not as great as in regions E and L and this decrease was not statisti cally significant (Figure 4-4). Interestingly, during most multiregional droughts for which kite distribution data were available (1981-1982, 1985 and 2000-2001), many kites ap peared to use region K as a refugia

PAGE 95

95 habitat (Martin et al. 2006; Takekawa and Be issinger 1989), suggesting that this region is generally less impacted than the major southe rn regions (i.e., E a nd L). This is also confirmed by hydrological data for the last 30 ye ars (Martin et al. 2007a). Indeed, during most multiregional droughts which affected region E (e.g., 1981, 1985, 1989, 1990, 1991, 1992 and 2001), at least one of the wetlands within region K was wetter (based on hydrological indicators) than any other wetland within the region E (Martin et al. 2007a). Thus, birds hatched in region K may have lower survival and reproduction rates for many years, but may be more resistant to multiregional disturbance events, which are believed to be a major cause of mortality among Snail Kite s (Beissinger 1995; Bennetts and Kitchens 2000; Martin et al. 2006). Conclusions and conservation implications Most ecologists would probably recognize that the effect of natal location on movement and survival may have important conservation imp lications for species which are notorious for their high degree of natal phil opatry (e.g. albatross species, Be kkum et al. 2006). In contrast, conservation implications may be less obvious wh en dealing with vertebrate species which exhibit more subtle patterns of natal philopatr y. The Snail Kite which has been described as a nomad (e.g. Bennetts and Kitchens 2000), may be one of such species. Indeed, we found that natal region is critical in influencing movement and survival of Snail Kites in Florida and that large variations in these vita l rates may occur among habitats, in part because of temporal variation in habitat conditions. Thus, one should be cautious wh en evaluating the conservation value of habitats (see also Holt and Gomulkiewicz 2004; Jonze et al. 2004). In particular, this study and that of Martin et al. (2006) show that regions in whic h survival is low for many years may be critical during disturbance events, such as droughts, by serving as refuges during drought and possibly by providing the enti re population with a pool of indi viduals (i.e., kites hatched in

PAGE 96

96 refugia habitats) with greater ability to resist such disturbance (see also Holt and Gomulkiewicz 2004). Aside from its practical implications for c onservation, our study highlights the importance of considering natal location as a potentially important factor affecting the ecological dynamics of spatially structured populations of animals, which, like Snail Kites, inhabit heterogeneous environments and use experience to base their settling decisions.

PAGE 97

97 Table 4-1. Multistate models of apparent survival (AD: survival of adults; J: survival of juveniles) and annual tr ansition probabilities ( ) among the four major wetland complexes used by Snail Kites in Florid a between 1992 to 2004. Factors incorporated in the models included: age, region, natal region; and a drought effect on movement and survival. Model AICc w K DEV [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 0.0 0.38 76 2591.2 [ELJK]K[ELJ][NOONOA] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 1.7 0.16 76 2592.9 [ELJK][KJ][EL][NOOAON] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 2.0 0.14 72 2601.7 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 2.0 0.14 77 2591.1 [ELJK]K[ELJ][NOOAON] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 3.7 0.06 72 2603.3 [ELJK]K[ELJ][NOONOA] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 3.8 0.06 77 2592.8 [ELJK]K[ELJ][NOONOA] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 5.1 0.03 79 2589.9 [ELJK]K[ELJ][NOONOA] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 7.1 0.01 80 2589.8 [ELJK]K[ELJ][NOONOA] AD,J AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)d ndd 8.1 0.01 82 2586.6 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 8.4 0.01 77 2597.4 [ELJK][KJ][EL][NOOAON] J AD,AD,AD,mrrr(.)(ND)(D)(t)nr*Dp(r*t)nddd 8.8 0.00 80 2591.5 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 10.9 0.00 76 2602.1 Notes: AICc: Akaike information criterion. AICc for the ith model is computed as AICci min (AICc). w: AICc weight. K: number of parameters. DEV: deviance as given by program MARK nr: natal region; r: region; t: time; .: no time effect; age: age effect; : multiplicative effect. Supe rscripts on the right of indicate the natal region survival pertain to (e.g., E: survival in region E; there were 4 regions: Ever glades E; Lake Okeechobee L; St Johns J; Kissimmee K). Superscripts on the left of indicate whether survival is na tal-region-specific (denoted nr ) or simply regionspecific (denoted r). : regions have different ; =: regions have similar (e.g., [ELJ] survival rates are similar in E and L but di fferent in J) Subscript nd: survival term reflects non-drought years (1992-2000 and 20022004). Subscript d: survival term reflects drought years (2000-2002). ND: no drought effect on during the interval 2000-2002. D: drought effect on during the interval 2000-2002. S uperscripts on the right of indicate the direction of movement between two regions (NO: transition from N to O; and OA transition from O to A). N: natal region; O is a post-dispersal region (O includes all regions except for the natal region); A is a post-dispersal region ( A includes all regions except for the natal re gion and the post-dispersal region O occupied in the previous year); : regions have different transitio ns probabilities (e.g., [NOONOA] NO, ON and OA are different). Dm: drought effect on between 2000 and 2001. p: sighting probability. Only models with AICc < 11 are presented (see supplementary materials for models with AICc > 11).

PAGE 98

98 Figure 4-1. Major wetland complexes (i.e., re gions) used by the Snail Kite in Florida. Kissimmee Chain of Lakes (K), Everglades (E), Lake Okeechobee (L), and Saint Johns Marsh (J).

PAGE 99

99 Figure 4-2. Movement probabilities ( ) between natal region N; post-dispersal region O (potentially includes al l regions except for the natal region); a nd post-dispersal region A (potentially includes all regions except for the natal re gion and the post-dispersal region O occupied in the year prior to moving to A) for Snail Kites hatched in four regions. Model averaged estimate s of movement between N and O ( NO ); between O and N (ON); and between O and A ( OA ) are presented. Error bars: 95%CI. Asterisks indicate that differences between ON and OA were statistically si gnificant (i.e., 95%CI[ ES] do not include 0).

PAGE 100

100 Figure 4-3. Model averaged estimates of natal philopatry (NPHL) and ph ilopatry to non-natal region (PHLNN). Error bars: 95%CI. Asteri sks indicate that differences between estimates of NPHL and PHLNN were statistically significant (i.e., 95%CI[ES] do not include 0).

PAGE 101

101 Figure 4-4. Model averaged estimates of region specific and natal region specific survival ( ) of Snail Kites in four regions. Reg ion specific survival is denoted RSSND during non-drought, and RSSD during drough t (time interval 2001-2002). Natal region specific survival is denoted NRSSND during non-drought, and NRSSD during drought (time interval 2001-2002). Error bars: 95%CI.

PAGE 102

102 CHAPTER 5 EXPLORING THE EFFECTS OF NATU RAL DISTURBANCES AND HABITAT DEGRADATION ON THE VIABILITY OF THE SNAIL KITE Introduction Natural disturbances and habitat degr adation have major influences on the dynamics of many wild populations. Natural dist urbances are environmental events that generally induce temporary cha nges in the system state (R eeves, Benda et al. 1995). Habitat degradation disturban ces are typically caused by human activities and often lead to more permanent changes in the system state (Reeves, Benda et al. 1995). Natural disturbances are often perceived as cat astrophes (e.g., flood, droughts) that impose large numerical responses on wild populati ons (Casagrandi and Gatto 2002). However, natural disturbances may also play a role in organizing ecosystems (DeAngelis and White 1994). In the long term, natural disturbances ma y be crucial to the persistence of wild populations (DeAngelis and White 1994). In some instances, degradation of natural habitats by human activities may have altered the organizing role of disturbances, and as a consequence, may have changed the shor t and long term response of populations to these naturally occurring even ts. Because most wildlife stud ies are conducted in the short term and at relatively small spatial scales, mo st studies investigating the effects of natural disturbances on natural populations have focu sed on the negative eff ects of disturbances (White and Pickett 1985). Studies that clearly show beneficial effects of disturbances remain scarce. A troubling fact is that s uppressing the natural effects of natural disturbances can actually cause the degrada tion of natural habitats with long term consequences for populations inhabiting thes e systems (e.g., Reeves et al. 1995). Better understanding the role of natural disturban ces on populations, both in the short term and in the long term, is an importa nt challenge in order to uncov er the ecologica l dynamics of

PAGE 103

103 many natural systems in a changing world. This ch allenge is also of pa rticular interest to management and conservation because mana gement actions that benefit certain threatened species in the short term, may be detrimental to the long term restoration of the systems. We examined three hypotheses related to the effect of natural disturbances and habitat degradations on Snail Kite dynamics and viability. The Snail Kite is a wetland specialist whose diet consists almost exclus ively of freshwater snails (Beissinger 1988). Vital rates of Snail Kites are finely tuned to the hydrology of its habitat. Drying events reduce availability of snails to kites, and therefore can dramatically increase mortality (Beissinger and Takekawa 1983; Takekawa and Beissinger 1989; Martin et al. 2006). The impact of drying events on kites depends on th e intensity, spatial ex tent and duration of these disturbances (Kitchens et al. 2002; Mooij et al. 2007). Drying events often occur naturally, but in some instances can be induced by management (e.g., managed draw downs). When natural disturbanc es occur adult Snail Kites are better able than juvenile birds to resist disturbances by moving to refugia habitats (M artin et al. 2006). As pointed out by Kitchens et al. (2002), droughts are na tural disturbances th at are an organizing force of Florida ecosystems, and suppressi ng them may cause habitat conversion. In addition, prolonged hydroperiods and excessive flooding intensity and frequency may also shift the vegetation communities toward communities less suitable for foraging and nesting activities of Snail K ites (Kitchens et al. 2002; Da rby et al. 2005). During the period 1993 to 2005 most primary kite habitats experienced a high frequency of flooding events and unusually prolonged hydroperiods during the Fall, and a single extensive drought was reported between 1993 and 2005 (Mar tin et al. 2007a). The increase in

PAGE 104

104 flooding frequency resulted from an increase in precipitation but also from the regulation of pumping stations, partic ularly in WCA3 and Lake Okeechobee. In this study we consider habitat conversion caused by wa ter management as a type of habitat degradation. Several authors have argued that these changes in hydropatterns have degraded kites habitat (Kitche ns et al. 2002). Martin et al (2007b) noted that the number of juveniles appeared to decrease dram atically after 1998. The 2001 drought had a large effect on kite survival and the population decr eased substantially af ter 2001; however, the kite population has not recove red since (Martin et al. 2007b ). Adult survival does not appear to have been reduced substantiall y, except during the drought that occurred in 2001 (Martin et al. 2006). Survival of juvenile s appears to have d ecreased since 1999, but the pattern is not absolutely clear and could be due to stochastic va riation (Martin et al. 2006). Based on these observations, we defi ned two periods: a pre-1998 period (which included 1998) and a post-1998 period (w hich did not include 1998). The pre-1998 period reflected a pre-declin e environment (i.e., before the number of young produced decreased, see Martin et al. 2007b). Based on pa st research on Snail Kites (e.g., Kitchens et al. 2002; Darby et al. 2005; Mooij et al. 2007; and Martin et al. 2007a), we considered three primary hypotheses to explain lack of r ecovery of the kite popu lation after the 2001 drought. First, we considered the habitat de gradation hypothesis, wh ich assumes that an increase in frequency of flooding events ( during the period August to January, hereafter referred as the Fall) has led to a deterioration of foraging a nd nesting habitats of Snail Kites ( Hypothesis 1 ). Second, we considered the hypothesi s that an increase in moderate drying events during the Spring-Summer (i.e ., during the breeding season, period March to June) is limiting recovery ( Hypothesis 2 ). Indeed, Martin et al (2007a) found that after

PAGE 105

105 1998 there was an increase in moderate dr ying events in WCA3A during the SpringSummer (which includes the peak of the bree ding season). As explained in the section Hydrological Conditions moderate drying events are more moderate disturbances than droughts. Finally, we considered Hypothesis 3 which assumes that both habitat degradation and an increase in moderate drying events are limiting population growth rate of kites. It is important to recognize that analys es associated with the examination of Hypothesis 1 2 and 3 are exploratory. The goal of our exercise was not to demonstrate the effect of specific factors (e.g., habitat degradation) on po pulation growth rate. Instead, our objective was to measure the potential impact of selected factor s in the context of each of the three hypotheses. For example, in the context of Hypothesis 1 our question was, what would be the effect of habitat degradation on popul ation growth rate of kites if in fact Hypothesis 1 was true? We applied the same reasoning for Hypotheses 2 and 3 We selected these three hypotheses because they are the most relevant to current management of kite habitats and are al so among the hypotheses best supported by the existing literature. Objectives Beissinger (1995) used matrix population mode ls to examine the viability of kites in relation to drought frequency. In this pa per we used a similar matrix population approach to address seven primary objectives. Objective 1 : based on the most current estimates of vital rates we provided several measures of Snail Kite viability under the current state of the system (deterministic population growth rate, stochastic population growth rate, and probability of quasi-extinction). Objective 2 : we determined the key vital rates that governed the popul ation growth of the kite popu lation (based on sensitivity

PAGE 106

106 analyses). Objective 3 : We compared deterministic population growth rates (1) before and after 1998. Objective 4 : Based on deterministic one way Life Table Response Experiment (LTRE, Caswell 2001), we estimated how differences in vital rates between the pre and post-1998 environments have cont ributed to changes in population growth rates. Objective 5 : We estimated stochastic population growth rates for pre and post-1998 environment and varied the frequency of drought This analysis allowed us to evaluate the relative importance of an increase in drought frequency as opposed to the before versus after 1998 effect on population growth rate (e.g., due to habi tat conversion under Hypothesis 1 ). Objective 6 : we examined the hypothesis that the primary factor for kite decline was related to an increase in m oderate drying events and not to habitat degradation ( Hypothesis 2 ). Objective 7 : we evaluated a hypothesis that assumed that both an increase in moderate drying events freq uency and the before versus after effect (e.g., due to habitat conversi on) were responsible for l ack of recovery of kites ( Hypothesis 3 ). In the final section of our pres entation we discuss the conservation implications of our findings. Methods Study Area Our study encompassed the entire population of Snail Kites located in Florida; see Dreitz et al. (2002) for a deta iled map of the sampled areas. Life Cycle All population matrix models in this pa per assumed a pre-breeding census and considered only the female part of the population (Caswell 2001). We assumed a 50:50 sex ratio (Beissinger 1995).

PAGE 107

107 As in Beissinger (1995) we considered three stage classes: juven iles (35 days to < 1 year old); subadults (1 to < 2 years old); adults (> 2 years). However, because we assumed a pre-breeding census, the youngest bi rds during the census are the individuals that were produced during the previous breed ing event, and these individuals will be subadults by the time of the birth pulse (Morris and Doak 2002). If subadults survive (survival probability Ps) to the next census they will b ecome adults. If adults survive (survival probability: Pa), they will remain adults. To cont ribute to age class 1 at the next census a reproductive female will have to generate a number of juvenile females (referred herein as fecundity of females: ma; ms, is the fecundity of subadult females), and these juvenile females will have to survive until th e next census (with survival probability of juvenile: Pj) (Morris and Doak 2002). Thus, as explai ned in Caswell (2001) adult fertility ( Fa) is: Fa = Pj x ma (Eq. 5.1) and subadult fertility ( Fs) is: Fs = Pj x ms (Eq. 5.2) Hydrological Conditions We performed an agglomerative hierarch ical clustering analysis in order to establish three groups of years characterized by contrasted hydrological conditions. The three categories of years were: wet year s (1993 to 1998, 2003 and 2005), moderately dry years (1999 to 2000, 2002 and 2004), and drought years (1992 and 2001; [see appendix D for details]). Droughts were characterized by the greatest intensity (e.g., lowest water levels during the dry season), the longest dur ation and the greatest spatial extent (e.g., proportion of wetlands affected by dry conditions).

PAGE 108

108 All but one hydrological variab le (spatial extent) were measured in Everglades Water Conservation Area 3A (WCA3A). Ther e were two primary reasons for focusing on hydrological conditions in WCA3A; (1) WCA3A has been identified as the single most important wetland unit currently occupied by k ites (Martin et al. 2007a ); (2) hydrological conditions in WCA3A are highly correlated w ith conditions in all the WCAs, Everglades National Park and Lake Okeechobee which, by far constitute the largest extent of wetlands occupied by kite s (Martin et al. 2007a). Data Source and Estimates of Vital Rates Survival rates Because apparent survival was estimated and modeled for the entire Florida population of Snail Kites, data from all regions were combined into a single analysis. Cormack-Jolly-Seber model (CJS) (Lebreton et al. 1992) were used to produce estimates of (probability of surviving and not permanently emigrating to an area not sampled between year t and t+1 ) and estimates of p (probability of sighti ng an individual in year t given that it was present and alive in th e sampling area). Modeled parameters used notation from Lebreton et al (1992); time dependency was (t) and no time effect was (.) We assigned each bird to one of two age cla sses: juveniles (denoted by subscript J), which were 30 days to 1 year old; and adults (denoted by subscript AD), older than 1 year. Effects embedded in other factors are shown using parentheses. A multiplicative effect is shown by (*) and an a dditive effect is shown by (+). Because previous survival analyses (Be nnetts, Kitchens & Dreitz, 2002) indicated an age effect on we computed a Goodness Of Fit test (GOF) accounting for an age effect by summing Test 3.SM, Test 2.CT a nd Test 2.CL (available from program U-

PAGE 109

109 CARE), yet the model still fit the data poorly ( 2 120 = 197.18, P < 0.0001) indicating excess variation (Choquet et al ., 2003). We thus calculated a variance inflation factor, which shows the amount of extra-binomial variation present in the data (Burnham & Anderson, 1998). We computed this variance inflation factor as = (Test 3.SM + Test 2.CT + Test 2.CL) / (df 3.SM + df 2.CT + df 2.CL), and obtained 1.64 (Choquet et al 2003). This value was well below 3, above wh ich the model structure is frequently considered to be inadequate (Burnham & Anderson, 1998). We used the value in the program MARK to estimate variance of and p and for model-selection procedures (Burnham & Anderson, 1998; White & Burnham, 1999). A set of biologically relevant m odels was developed that allowed and p to vary across time (denoted t ), or stay constant (denoted ) for each age class (i.e., adults: subscript AD, and juveniles: subscript J). Age was modeled both as time since marking and as a group effect. We also crea ted models that incl uded drought effect on and p We included three types of drought effects. D1: was assumed to be constant except during the drought period 00-02, ( during 00-01, was assumed to be different from 01-02, see also Martin et al. 2006). D2 : same as D2 except that 92-93 was also considered a drought period (see section Hydrological Conditions ). Under D2, the drought effect on survival was assumed to be the same in 92-93 and 01-02. D3: same as D2 except that the drought effect in 92 -93 was assumed to be different from 01-02 (see also Martin et al. 2006 for additional information on drought effect on Snail Kite survival). In the section Hydrological Conditions we identified three categories of years: wet, moderately dry and drought. Hence, we designed models with the effectWMD1: was associated with 3 years categories, wet (93 to 98), moderately dry (99 to 01, 02-

PAGE 110

110 03, 04-05), and drought (92-93, 01-02, but in thes e models the drought effect on survival is assumed to be the same in 92-93 and 01-02). We also constructed models with the effect: WMD2 (same as WMD1 except that th e drought effect on survival was assumed different in 92-93 a nd 01-02). D1: was constant except during the drought period 00-02, (during 00-01, was assumed to be different from 01-02). In addition, we included some trend mode ls (see Cooch and White 2005). In these models, survival probabilities were modeled as linear-logistic func tion of a covariate: TCOV (where COV is a covariate). TAVGJD was a trend model with annual mean water levels for the period January to December (of year t ) as a covariate. TAVGMJ was a trend model with annual mean water levels for the period March to June (of year t ) as a covariate. TMWL was a trend model with mi nimum annual water levels as a covariate. TSPAT was a trend model with spatial extent of drying (i.e., propor tion of regions with index of dryness less than 1) event as a cova riate. TTIME was a trend model with time as a covariate. Some trend models included a drought effect (i.e ., survival was modeled as a function of a covariate for most years except during drought y ears of 2000-2002). These models were denoted J(D4/TCOV) (for juvenile survival) and AD(D5/TCOV) (for adult survival). Because in 1992 most juvenile s were fledged outside the WCAs (i.e., in areas not affected by the 1992 drying event, see Hydrological conditions), we built models that were similar to J(D4/TCOV), but did not include the interval 92-93 as part of the trend (i.e., the survival estimate for this time period was estimated independently from the trend). These models were denoted J(D6/TCOV). We also designed models similar to AD(D5/TCOV), but which assumed identical effects of the drought on adult survival during the period 1992-1993 and 20012002 (in these models adult survival

PAGE 111

111 during the interval 2000-2001 was not included in the trend, and was assumed to be different than during the intervals 19921993 and 2001-2002), these models were denoted A D(D7/TCOV). For each trend model we made a priori predictions about the direction of the relationship (i.e., positive or negative relationship). Because past research suggests that survival should be greater during wet y ears and lower during dry years, we predicted a positive relationship between survival and the slope of TAVGJD, TAVGMJ, and TMWL. Based on work from Bennetts and Kitc hens (2000), we predicted that survival would decrease as the spatial extent of drying events incr eased. If juvenile and adult survival declined over time the slope of TTIME should be negative for juveniles and adults. We used QAICc (see Chapter 3 and 4) as a criterion to select the model that provided the most parsimonious de scription of the variation in the data (i.e., model with the lowest QAICc). Survival rates were obt ained by model averaging (see Chapter 4). Fecundity rates In order to be able to construct population projection models, we need to be able to estimate ma (fecundity of adults) and ms (fecundity of subadults): ,() () () () ()fj a fs faNt mt Nt Nt qt (Eq. 5.3) Where ,()faNt is the estimated number of adult females and ,()fjNt the estimated number of juvenile females in the population. () qt) is the estimated ratio of fecundity rates: () a s m qt m (Eq. 5.4) Hence,

PAGE 112

112 () a sm m qt (Eq. 5.5) In 2004 and 2005 we directly estimated the number of juveniles in the population (see Martin et al. 2007b for methodology). The nu mber of juvenile females was obtained by dividing these superpopulation estimates for juveniles by two (assumi ng a sex ratio of 50:50). For the period 1992 to 2003 we did not ha ve estimates of the number of juveniles, but counts were available. For these years ,()fjNt can be obtained using: ,() () ()fj fjCt Nt t (Eq. 5.6) Where ,()fjCt is a count of the number of juvenile females at time t and () t is the fraction of juveniles counted from the overall population of juveniles at time t (i.e., hereafter referred as detection of j uveniles (see Martin et al. 2007b). ,,,()()()fafSUPfsNtNtNt (Eq. 5.7) Where ,()fSUPNt is the estimate of superpopulation size of females at time t (Estimates of ,()fSUPNt are obtained by dividing superpop ulation size estimates by two, assuming a 50:50 sex ratio; estimates of super population size are availa ble in (Martin et al. 2007)). ,()fsNt in Eq. 5.7, corresponds to the num ber of subadult females at time t which was computed as: ,,()()()j fsfjNtNtPt (Eq. 5.8) We computed () qt based on estimates obtained fr om Bennetts (1998). We obtained () qt= 4.2 (see Appendix E for more details).Varying q(t) from 4 to 8.5 had little effect on population growth rates and ot her relevant measures.

PAGE 113

113 Detection of juveniles Estimates of detection of juveniles ( () t ) were needed to compute fecundity rates. We used a range of values of () t to evaluate the robustness of our results (see electronic supplementary ma terials for details on how () t was derived). First, we assumed that () t was constant over time and was equa l to 0.16 (this estimate was the lowest of our empirical estimates and proba bly underestimated the true detection for most years, Appendix F). Second, we assumed that () t was constant over time but was equal to 0.70 (this estimate probably great ly overestimated the true detection, we picked this value because it was twice the value of our greatest estimate of detection, see Appendix F for more details). Third, we assumed that () t varied from 0.16 to 0.35 (see Appendix F for additional details). Matrix Analyses 1, damping ratio, sensitivity and elastici ty analysis (Objectives 1 and 2) All matrix population models were based on the demographic life cycle described earlier. We constructed a matrix based on the mo st current estimates of vital rates; this included 9 years of data from (1997 to 2005). He reafter, we refer to this matrix as the FULL1 matrix. Survival rates in the FULL1 matrix were computed as the average survival rates for the period 1997-2005. Fertili ties were estimated as the product of the average fecundity and the average survival ra te for juveniles. We computed the dominant eigenvalue ( 1) for this matrix, which corresponds to the asymptotic population growth rate (Caswell 2001). We followed procedures described by Caswell (2001) to compute sensitivities and elasticities. Sensitivity m easures how absolute ch anges in individual

PAGE 114

114 vital rates influence 1 (Caswell 2001). Elastic ity measures the propor tional change in 1 resulting from a proportional change in i ndividual vital rates (Morris and Doak 2002; Caswell 2001). Elasticities are more read ily comparable among vital rates than sensitivities (Horvitz and Schemske 1995). The damping ratio was comput ed by dividing the largest ei genvalue by the absolute number of the second largest eigenvalue. Th is ratio is indicativ e of the rate of convergence to the stable ag e structure (Caswell 2001). Before versus after effect on 1 and Life Table Response Experiment (Objectives 3 and 4) We applied a deterministic, fixed design one-way Life Table Response Experiment described by Caswell (2001), to assess how differences in vital rates before and after 1998 contributed to changes in 1. Matrix BEF summarized the environment before 1998. Survival rates included in BEF were co mputed as the averag e annual survival for the interval 1993 to 1998, excluding estimates for 1992 identified by Martin et al (2007a) as a drought year. Fertilities were computed as the product of the average juvenile survival rates (for 1993 to 1998) and the aver age fecundity for the interval 1997 to 1998. Matrix AFT summarized the environment af ter 1998. Survival and fertility rates for this matrix included estimates from 1999 to 2005, excluding estimates of survival and fertilities affected by the 2001 drought and th e 2004 drying event (both drying events had large effects on juvenile survival). Therefor e, estimates of juven ile survival did not include estimates for the interval 2000 to 2002. Estimates of fecundity did not include estimates for 2001 and 2004. For this analysis we used matrix BEF as the reference matrix, and AFT matrix as the treatment level (Caswell 2001).

PAGE 115

115 Stochastic population growth ra te (objectives 1, 4, 5 and 6) We computed the stochastic population growth rate ( s) and 95%CI of s, by using the simulation method described by Caswell (2001, p.396). 1 01 logT s t tr T and an approximate 95%CI: () log1.96sVr T where T is the simulation time, ((1)) log (())f t fNt r Nt where Nf( t ) is the population size of females at t V(r) is the variance of rt. T was set to 10,000 time steps (i.e., 10, 000 years). The initial population size and population structure were set based on the obs erved values in 2005 (Martin et al. 2007b). The simulations assumed independent and identic ally distributed sequences (iid): at each time step t a matrix (which correspo nded to a particular environmental state) was drawn from a fixed distribution (Caswell 2001). For all the stochastic analysis (i.e., s and probability of quasi-extinction (PQE)) we incorp orated the variance associated with each vital rate (which included both process and samp ling variance), as well as the within year correlation among vital rates using the appr oach described by Morris and Doak (2002: 284-285). The correlation matrix was computed based on fecundity and survival rates for the period 1997 to 2004 (estimates of fecund ity were available for the period 1997 to 2005 but estimates of survival were only available for the period 1992 to 2004). As suggested by Morris and Doak (2002), we used the Stretched-Beta distribution to simulate random variation in fecundity a nd fertility rates, whereas we used the Beta distribution to simulate random variation in survival rates. The variance for each parameter (e.g., survival and fecundities) was computed as: 2() var() 1in These

PAGE 116

116 estimates of variance included both sampling and process variance. In instances where only one estimate was available (e.g. dr ought year of 2001), we used the sampling variance. Variances for derived parameters were computed using the delta method (Williams et al. 2002). We estimated s for 11 sets of environmen tal conditions (denoted by Conditions (x) with x = 1 to 11 ) that addressed Objectives 1, 5, 6 and 7 : Viability of Snail Kites under curren t conditions (Objective 1) Condition (1) : we computed s for the time series 1997 to 2005. For this simulation we could have used matrix FULL1. However, because the 2001 drought was considered an extreme event, we foll owed the suggestion of Morris and Doak (2002, p. 272) and used three matrices for this simulation: (a) matrix FULL2 (means and variances were estimated for the period 1997 to 2005, but excluded data for the period 2000-2001 which corres ponded to the drought); (b) matrix DRO1 (based on the data for 2000); (c) matrix DRO2 (based on the data for 2001). These matrices were drawn at random with the observed frequencies for the period 1997 to 2005: frequency for FULL2 was: 7/ 9 and frequencies for DRO1 and DRO2 were 1/9. Evaluation of Hypothesis 1: Reduction of s after 1998 (Objective 5) Conditions (2) and (3) : we computed s for the environmental state that corresponded to the period before 1998. This simulation was based on two matrices. (a) One matrix summarized th e conditions before 1998 (i.e., means and variances were obtained for the period 1993 to 1998). (b) One matrix summarized the drought conditions DRO (see section Hydrological conditions ). We estimated two values of s: one assuming a drought frequenc y of 0.111 (i.e., drought occurred on average every 9 years: Condition (2) ) and another assuming a drought frequency of 0.25 (i.e., one drought every four years: Condition (3) ). Conditions (4) : same as Condition (2) but for the environmental state that corresponded to the period after 1998. Conditions (5) : same as Condition (3) but for the environmental state that corresponded to the period after 1998. Hypothesis 1 assumes that habitat degrad ation has caused a reduction in s after 1998. The logical predicti on for this hypothesis is Prediction 1: s under Condition (2) > s under Condition (4) ; and s under Condition (3) > s under Condition (5).

PAGE 117

117 Evaluation of Hypothesis 2: Increase in drying event frequency reduces s (Objective 6) Hypothesis 2 assumes that an increase in m oderate drying events frequency has caused a reduction in s after 1998. In the section Hydrological conditions we explained that the frequency in moderate drying ev ent was greater after 1998 than before 1998. Condition (6) : we computed s by simulating an environment with three environmental states: WET1, MOD and DRO (see Hydrological conditions ). This simulation ignored the effect of habitat degradation. The matrix summarizing wet years was based on means and average of all years that were identified as wet years for the period 1992 to 2005 (i.e., for ma and ms: 1993 to 1998, 2003 and 2005; for Pa and Py 1993 to 1998 and 2003). The matrix summarizing moderately dry years was based on means and averages of all ye ars that were identified as dry years for the period 1992 to 2005 (i.e., for ma and ms: 1999 to 2000, 2002 and 2004; for Pa and Py 1999, 2002 and 2004). The matrix summ arizing drought years is described in Hydrological conditions. For condition (6) we applied observed frequencies for each environmental state for the first se ven years of the study (i.e., 1992 to 1998; we refer to this environment as Low Fr equency of Moderate Drying event: LFMD). During this time period: 1 drought year was obser ved (frequency: 1/7); no moderately dry years were observed (f requency: 0/7); and 6 wet years were observed (frequency: 6/7). Condition (7), same as cond ition (6), except that we us ed frequencies observed for the last 7 years (i.e., 1999 to 2005, Hi gh Frequency of Moderate Drying event (HFMD), 1 drought: 1/7; 5 moderately dry years: 4/7; and tw o wet years: 2/7). Under Hypothesis 2 we predict Prediction 2 : s under Condition (6) > s under Condition (7). Evaluation of Hypothesis 3: Increase in frequency of drying events and before versus after effect (Objective 7) Conditions (8), we computed s by simulating an environment with three environmental states: WET2, MOD and DRO. The matrix summarizing the wet years was estimated based on means and vari ances of parameters estimated prior to 1998. Ideally, we would also have comput ed matrices for MOD and DRO based on parameters estimated prior to 1998. Unfort unately, there were no data available for moderately dry years prior to 1998, and in Hydrological conditions we explained why we did not use parameters estimated from 1992. Thus, matrices MOD and DRO were based on data collected after 1998, and were the same matrices used for Condition (6). For condition (8) we app lied the observed frequency for each environmental state for the first 7 year s of the study (LFMD, see Condition (6)).

PAGE 118

118 Condition (9), same as cond ition (8), except that we us ed frequencies observed for the last 7 years (HFMD: 1 drought: 1/7; 5 moderately dr y years: 4/7; and two wet years: 2/7). Condition (10) same as condition (8) excep t that the matrix summarizing the wet years was based on parameters estimated af ter 1998. We refer to this matrix as WET3. For Condition (10) we applied the observed frequency for each environmental state for the first 7 years of the study (LFMD, see Conditions (6 and 8)). Condition (11) same as Condition (10) but we applied the observed frequency for each environmental state for the last 7 y ears of the study (HFMD, see Conditions (7 and 9)). Hypothesis 3 assumes that both an increase in moderate drying events and the before versus after effects have contributed to a decrease in s after 1998. Under this hypothesis we predict Prediction 3a : s under Condition (8) > s under Condition (9) ; and s under Condition (10) > s under Condition (11). We also predict Prediction 3b : s under Condition (8) > s under Condition (10) ; and s under Condition (9) > s under Condition (11). Probability of Quasi-Extinction (Objective 1) We calculated the probability of reaching a quasi-extinction state for a set time. We chose a quasi-extinction threshold of 50 fe males. This number appeared to be a reasonable compromise to deal with concerns about demographic st ochasticity, yet would be low enough to make the population at this level immediately imperiled (Morris and Doak 2002). We estimated the PQE at a projec ted time of 150 years. We calculated the cumulative distribution for quasi-extincti on time by simulation (100 runs) by computing the proportion of trajectories that fell belo w 50 kites during and until a projected time of 150 years. We repeated this procedure 100 ti mes. From the distribution of PQE at each time step we determined the lower and upper 95% percentiles for each analysis. We used these latter values as estimates of 95%CI of the estimate of PQE. We used the mean of

PAGE 119

119 the distribution as an estimate of PQE. The initial population size a nd age structure used in the simulation were based on average es timates of population size obtained in 2005 (Martin et al. 2006). We generated the same stochastic environmental conditions as for the computation of s. Validation We repeated all the analyses to estimate 1, s and PQE assuming (1) constant detection of juveniles () t = 0.16; (2) detection varying between () t = 0.16 and 0.35; and (3) constant dete ction of juveniles () t = 0.70 (see Detection of Juveniles and Appendix F). Analyses assuming constant () t = 0.16 resulted in the most optimistic scenarios (i.e., lowest PQE, greatest s); whereas, analyses assuming constant () t = 0.70 resulted in the most pessimistic scenarios. Analyses that assumed that () t varied between 0.16 to 0.35 were probably the most reasonable (see Appendix F for more details). All matrix analyses were conducte d using MATLAB (Mathworks 2005). Results Survival estimates The most parsimonious modelADJ(WMD1)(t)p(t), was marginally better supported by the data than the next 4 most parsimonious models ( w = 0.234; Table 5-1). The QAICc among the five most parsimonious models were less than 1, which indicate that these models received very similar support from the data. Models that assumed a time effect on juvenile survival a nd detection probabilities were well supported by the data ( wt = 1; Table 5-1). Model averaged estimates indi cated that adult surv ival was significantly

PAGE 120

120 lower during the periods 92-93 and 00-02 (Fi gure 5-1). However adult survival was exceptionally low during the interval 01-02 (F igure 5-1; see also Martin et al. 2006). Models that assumed a wet-moderately drydrought effect on survival (e.g., WMD1) received some support, in fact the most parsimonious model was:ADJ(WMD1)(t)p(t). However, the differences in adult survival (model averaged estimates) during these three year-categories were not subs tantially different (Figure 5-1). Although models assuming a WMD1 or WMD2 effect on juvenile survival were not well supported by the data based on QAICc (when compared to the time depe ndent model), we found that there was a substantial differences in juvenile survival during wet, moderately dry and drought years. As expected Juvenile survival was highest during wet years ( J = 0.557; 95%CI = 0.498 to 0.617), lowest during the 2001 drought ( J = 0.074; 95%CI = 0.012 to 0.339), and intermediate during moderately dry years ( J = 0.255; 95%CI = 0.194 to 0.316) (see also Figure 5-1). Interestingly, j uvenile survival in 92-93 was high even though this period was categorized as a drought period based on th e Hydrological Analysis. This is probably explained by the fact that, most birds were fledged in areas located outside of the WCAs: Lake Okeechobee, St Johns Marsh and KCL, all these areas were not affected dryer than average in 1992 (see Hydrological Conditions, Table 7-1 and 7-2). The trend models received almost no support based on QAICc weights (wt ~ 0, Table 51). However, the slope parameters for mo st trend models indicated a significant relationship between the covari ates and survival of juve niles. The slope parameter TAVJD from model ADJAVGJD(D2)(D6/T)p(t) (Table 5-1) indicated a positive relationship between average water levels for the peri od January to December and survival of

PAGE 121

121 juveniles (TAVJD .= 2.13; 95%CI = 1.22 to 3.05). The slope parameter TAVMJ from model ADJAVGMJ(D2)(D6/T)p(t) (Table 5-1) indicated a positive relationship between average water levels for the period March to June and survival of juveniles (TAVMJ .= 1.23; 95%CI = 0.475 to 1.98). The slope parameter TMWL from model ADJMWL(D2)(D6/T)p(t) (Table 5-1) indicated a positive relationship between annual minimum water levels and survival of juveniles (TMWL .= 1.93; 95%CI = 1.37 to 2.48, see Figure 5-1). The slope parameter TDUR from model ADJDUR(D2)(D6/T)p(t) (Table 5-1) indicated a negative relations hip between duration of drying event and survival of juveniles (TDUR .= -0.04; 95%CI = -0.06 to -0.03). The slope parameter TSPAT from model ADJSPAT(D2)(D6/T)p(t) (Table 5-1) indicated a ne gative relationship between spatial extent of drought a nd survival of juveniles (TSPAT .= -6.95; 95%CI = -10.75 to 3.16). The slope parameter TTIME from model ADJTIME(D2)(D6/T)p(t) (Table 5-1) indicated a negative relations hip between time and survival of juveniles (slope for juveniles:TTIME .= -0.14; 95%CI = -0.22 to -0.06). Th is estimate suggested that juvenile survival decreased significantly over time. In contrast the slope parameter from TTIME from model ADJTIMETIME(D7/T)(D6/T)p(t), did not support the hypothesis of a significant decrease of adult surviv al over time (slope for adults:TTIME .= 0.04; 95%CI = 0.09 to 0.16).. Model averaged estimates for adul t and juvenile survival are presented in Figure 5-1.

PAGE 122

122 Probability of Quasi-extinction (Objective 1) PQE increased dramatically as detection of juveniles increased (Figure 5-2). PQE at 50 years was 0.65 (95%CI = 0.59 to 0.70) when () t = 0.16, and it was 0.975 (95%CI = 0.96 to 1) when () t = 0.16 to 0.35. 1, sensitivity and Elasticity an alysis (Objectives 1 and 2) The deterministic lambda: 1 for the last nine years were 0.92 (with () t = 0.16), 0.89 (with () t = 0.16 to 0.35), and 0.85 (with () t = 0.70). The damping ratios were: 13.7 (for () t =0.16), 19.2 ( () t = 0.16 to 0.35) and 56.7(for () t = 0.70). These values were large and indicated rapid convergence to the stable populat ion structure following perturbations (Caswell 2001). Sensitivity analysis of matrix FULL, indicated that 1 was most sensitive to changes in adult survival and a dult fertility (note that sensitivities were slightly higher for changes in adult survival, Fi gure 5-3.a). Sensitivity of 1 to changes in other vital rates ( Fs and Ps) was considerably lower (Figure 5-3a). Elasticity analyses indicated that 1 was most sensitive to proportional changes in Pa (Figure 5-3.b) Elasticity of 1 to changes in Fa and Ps were considerably lowe r (< 0.1). Elasticity of 1 to change in Fs was almost 0, and this was true independently of the detection of juveniles assumed (Figure 53.b). Life Table Response Ex periment (Objective 3) The deterministic lambda: 1 for matrix BEF was 1.2 (with () t = 0.16), 1.13 (with () t = 0.16 to 0.35), and 0.95 (with () t = 0.70). The deterministic lambda: 1 for

PAGE 123

123 matrix AFT was 0.97 (with () t = 0.16), 0.92 (with () t = 0.16 to 0.35), and 0.89 (with () t = 0.70). There was a large difference in Fa between matrix BEF and matrix AFT when detection of juveniles was assumed to be 0. 16 or varied from 0.16 to 0.35 (Figure 5-4). The reduction in Fa was considerably lower when detection of juveniles was assumed to be large ( () t = 0.70). Reduction in other vital ra tes was less than 0.05 (Figure 5-4). When () t = 0.16 and () t = 0.16 to 0.35, was assumed Fa contributed the most to the reduction in 1 between BEF and AFT (contributi on > 90%), whereas Pa, Ps and Fs contributed less than 4% each (Figure 5-4). When () t = 0.70 was assumed, Fa still contributed the most (84%), but the contribution of Pa was also greater (14%). In all cases, there were close corre spondences between the sum of LTRE contributions and observed differences in 1 (differences < 0.22%), which suggest that the first order approximation used was sufficient (Caswell 2001). Stochastic population growth rate s Viability of Snail Kites under curren t conditions (Objective 1) Condition (1): When assuming () t = 0.16, s for the last 9 years was 0.940 (95%CI = 0.936 to 0.944). When assuming () t = 0.16 to 0.35, s for the last 9 years was 0.9033 (95%CI = 0.8998 to 0.9069). When assuming () t = 0.7, s for the last 9 years was 0.853 (95%CI = 0.8509 to 0.8561). Evaluation of Hypothesis 1: Reduction of s after 1998 (Objective 5) Conditions (2) to (5): There was an add itive effect of drought frequency and the before versus after 1998 effect. The additive effect was due to the fact that we assumed

PAGE 124

124 the drought effect to be the same in the before and after 1998 environment. Increasing drought frequency from 0.11 to 0. 25 caused a large decrease in s (Figure 5-5). There was evidence of a large reduction in s when comparing the before versus after 1998 environments. s estimated from vital rates obtai ned prior to 1998 was substantially greater than s based on vital rates ob tained after 1998. This was true no matter what detection of juveniles was assumed. However, the magnitude of the difference decreased as detection of juveniles increased (Figur e 5-5). The magnitude of the difference in s due to the change in conditions before and af ter 1998 was greater than the effect of an increase in drought frequency (from 0.11 to 0.25) when () t = 0.16 and () t = 0.16 to 0.35 was assumed. When () t = 0.70 was assumed, the reduction in s due to an increase in drought frequency was greater than the effect of the change in conditions before and after 1998 (Figure 5-5). Evaluation of Hypothesis 2: Increase in drying event frequency reduces s (Objective 6) Conditions (6) and (7): s was greater when we simulated environmental states at the frequency observed during the first seve n years of our study (LFMD, Figure 5-6). This pattern was apparent no matter what de tection of juveniles was assumed (Figure 56). However, the magnitude of the difference was greater when detection of juveniles was assumed to be high (Figure 5-6). The reduction in s, was due to the fact that the frequency of moderate drying events in creased during the last seven years (or alternatively, the frequency of wet years de creased). The frequency of droughts remained the same under these two environments, thus droughts did not contri bute to the reduction in s when comparing these two environments.

PAGE 125

125 Evaluation of Hypothesis 3: Increase in frequency of drying events and before versus after effect (Objective 7) Conditions (8) to (11): s was greater when we simulated environmental states at the frequency observed for the first seven y ears (HFMD) than for the last seven years (LFMD). This pattern was observed no matter what detection of j uveniles was assumed. However, the magnitude of the difference decreased as () t increased. In addition, in most cases s was greater when the simulations were based on parameters estimated with data collected prior to 1998. This was true when the frequency of the wet, dry and drought mimicked the frequency observed for the first seven years (LFMD, Figure 5-7). The magnitude of the difference was considerab ly smaller when the frequency of wet, dry and drought mimicked the frequency observed for the last seven years (HFMD, Figure 57). The magnitude of the difference also decreased considerably as () t increased over time (Figure 5-7). The principal explanation fo r the stronger effect of the before versus after 1998 effect under the LFMD scenario is related to the higher frequency of wet years under the LFMD. Indeed, only the wet matrix differed when comparing the before and after 1998 environment. Therefore, the difference in s caused by the difference in wet years was amplified as the frequency of wet years increased. Discussion Snail Kite Viability and Key Vital Rates Our results indicate that if the enviro nmental conditions observed between 1997 and 2005 are representative of future conditi ons there is a high likelihood of a rapid extinction of the Snail Ki te population in Florida. This finding addressed Objective 1 The deterministic population growth rate for the FULL1 matrix was low ( 1 = 0.919), even when the detection probability for juveniles was assumed to be low (i.e., () t =

PAGE 126

126 0.16 to 0.35). When incorporating variation and correlation among vital rates we obtained values of s that were slightly greater than 1. The probability of quasi-extinction (PQE) was high even when assuming a low detection probability for juveniles (Figure 5-2). PQE increased dramatically when detection probabilities () t = 0.70 were assumed. Although one could infer the time to extinction base d on Figure 5-1, several authors have warned against interpreting time to extinction litera lly (e.g., Reed et al. [2002]). Instead, our analyses should be viewed as supportive eviden ce that Snail Kites ar e facing high risks of extinction. Caswell (2001) suggests that popu lation projection models are particularly useful for evaluating the importance of de mographic or environmental factors in influencing population dynamics. This latter idea motivated th e sensitivity analyses as well as the exploratory analyses discussed in the remaining part of our presentation. We determined the sensitivity of 1 to absolute changes (i.e., sensitivities) and relative changes (elasticities) in vita l rates. These analyses addressed Objective 2 Elasticity of 1 to changes in adult survival was by far the greatest (Figure 5-3), which indicates that a proportional ch ange in adult survival woul d cause the largest proportional change on 1. This pattern is typical of the patterns observed for other long-lived species (Stahl and Oli 2006). Regarding sensitivity patterns, 1 was most sensitive to absolute changes in adult survival. Desp ite the greater elasticity of 1 to changes in adult survival, we found that adult fertility was part icularly critical in influencing 1 (Figure 5-3). Asymptotic population growth rate based on vi tal rates estimated for the period prior to 1998 was greater than 1 estimated from vital rates obtained for the period 1999 to 2005. The difference in 1 was substantial even after rem oving the effects of the 2001 and 2004 drying events. We performed a LTRE analysis that indicated that the observed reduction

PAGE 127

127 in 1 for the two periods was largely due to a re duction in adult fertilities between these two periods (Figure 5-4). This finding held true for the th ree levels of detection of juveniles that we considered. Thus, even t hough the dramatic reduction in adult survival during the 2001 drought c ontributed to a large decrease in population size (Martin et al. 2006; Martin et al. 2007b), other factors associated with a reduc tion in adult fe rtilities are responsible for the reduc tion in population growth. Next, we examined the three primary hypot heses explaining reduc tion in Snail Kite population growth during the most recent years. Hypothesis 1: Before versus After 1998 Effect (Objective 4)) Habitat conversion in the wetlands of sout h Florida, resulting from frequent and prolonged floods, has concerned many ecologi sts (Gunderson and Loftus 1994; Bennetts et al. 1998; Kitchens et al. 2002). Habitat conversion is believed to reduce apple snail densities (Darby et al. 2005) a nd could also reduce the availa bility of suitable nesting habitats for Snail Kites (Bennetts et al. 1998). Although other factors may have contributed to the decrease in population gr owth rate after 1998, habitat conversion, especially in the WCAs, appears to be we ll documented and should receive particular attention. As explained earlier, 1 was greater in the pre than in the post 1998 environment even when removing the effects of the drying events of 2001 and 2004. The same pattern was observed for s (Figure 5-5). The magnitude of the difference decreased when detection of juveniles was assumed to be hi gh. But in all instances the difference was substantial, indicating that some changes in the environment may have caused a decrease in population growth. This result supported Prediction 1 We were able to examine the relative effects of drought frequency and the before versus after 1998 effect on s.

PAGE 128

128 Thus, if we are ready to assume that the re duction in population growth was entirely due to habitat conversion, our stocha stic analyses suggest that th e effect of habitat conversion was greater (except when () t = 0.70) than a substantia l increase in drought frequency (from 1 drought every nine year s to 1 drought every 4 years) (Figure 5-5). This prospect is particularly alarming given the slow dynamic of habitat conversion. Under this scenario, the positive effects of adequate habitat restoration plans on kite population growth would only be perceptible at the tim escale of multiple years (i.e., multiple years are needed to convert vegetation communities back to suitable kite habitats). However, as explained in the next section, it is unlikely that habitat c onversion was solely responsible for the observed reduction in population growth. Hypothesis 2: Increase in Frequency of Moderate Drying Events (Objective 6) In the previous section we considered the effects of habitat co nversion and increase in drought frequency on kite population growth. Under Hypothesis 2 we ignored the effect of habitat degradation, but still considered the eff ect of drought frequency. In addition to droughts with a large spatial and temporal extent (e.g ., 2001 drought), we also considered drying events with a more moderate spatial and temporal extent. In fact, based on hydrological criteria deve loped by (Bennetts 1998) thes e moderate drying events would not have been considered as a threat to kites persistence because it was believed that kites could read ily escape drying conditions by m oving to less affected wetlands. Interestingly, we found that even if we ignor ed the effect of habitat degradation, an increase in moderate dryi ng event frequency could reduce s substantially. We found that the increase in moderate drying events freque ncy observed during the last 7 years of our study could explain some of the observed variation in s between the pre and post 1998

PAGE 129

129 environments (Figure 5-6). This result suppor ted Prediction 2. Ther e are several reasons that can explain the impact of moderate dr ying events on kite popul ation growth. First, even though a proportion of adults may be ab le to move to wetlands less affected by a drying event, fledglings, on the other hand, are less likely to be aware of refugia habitats and therefore are more likely to die during a drying event (Martin et al 2006). This hypothesis is consistent with th e fact that adult survival is remarkably stable (except during extensive droughts). However, juvenile survival is strongly affected by drying events, even when they are moderate (Figure 5-1). Hypothesis 3: Increase in Moderate Dryi ng Events and Before versus After 1998 effect (Objective 7) We believe that both habitat conversion a nd an increase in fre quency of moderate drying events are highly re levant to our study system. Therefore, we examined s under Hypothesis 3 which considered both factors. Alt hough we only had data to estimate vital rates during wet years in the pre and post 1998 environment, we found evidence of an effect of both factors on s (Figure 5-7). This result supports Prediction 3a and 3b If we had used the 1992 data to construct the drought matrix for the pre 1998 environment, the differences in s due to the before versus after 1998 effect would have appeared to be even greater. It is in fact possible that long term habita t degradation has reduced the resistance of kite to drought s; however, given the uncertain ty associated with 1992 (see Hydrological Conditions ) we preferred not to incl ude 1992 into our analyses. Limits of the Models First, because estimates of detection of juveniles were only available for 2004 and 2005, we had to find ways to account for detection for the period 1997 to 2003 (see Methods ). Although, estimates of detection of ju veniles were not dir ectly estimated 1997

PAGE 130

130 to 2003, it is probably safe to a ssume that the true values of () t were in fact between 0.16 and 0.70. Therefore, by includin g analyses that assumed constant () t= 0.16 and () t = 0.70 we evaluated the robustness of our analyses to large changes in () t. An alternative to compute fecundity rates would have been to derive this rate from: the proportion of birds attempting to breed (i ), the proportion of successful breeding attempts (Si), the number of breeding attempts per year (i ), the number of juveniles per successful nest (Yi), and the sex ratio (assumed to be 50:50) (Bennetts 1998). The reason we did not use this approach is that we currently do not have good estimates of i and i under varying environm ental conditions. Second, we ignored density dependence becau se at this point there is no evidence of density dependence in this population of Sn ail Kite (i.e., correla ting vital rates with population size do not indicate any relation reflecting dens ity dependence, Martin unpublished data). As pointed out by Beissinger (1995) the setting of an upper boundary would have been arbitrary, and therefor e, should be avoide d. Ignoring density dependence in our simulation models allowed population size to grow to levels unlikely to occur in Florida (Beissinger 1995). Thus, we expect estimates of population growth to be positively biased and PQE to be negativel y biased (i.e., underestimated). Although we did not directly include the e ffect of demographic stochastic ity in our PVAs, we followed Morris and Doaks (2002) s uggestion and picked a quasi-e xtinction threshold of 50, which obviates the need to dir ectly simulate demographic st ochasticity (Morris and Doak 2002). On the other hand, concerns about geneti c stochasticity were not included. At this point the Snail Kite populati on is assumed to be demographically isolated. However,

PAGE 131

131 additional studies are needed to evaluate the genetic structure of th e Snail Kite population in Florida, as well as possible gene flow s with other populations (e.g., Snail Kite population in Cuba). If the Snail Kite population in Florida is completely isolated, genetic stochasticity could further in crease the risk of extinction of the Florida population. Another shortcoming of our st udy is that we ignored tempor al correlation. For instance, some ecologists have hypothesized that foragi ng conditions for kites should be lower than average one year af ter a drought (Beissinger 1995). Th is is certainly a reasonable assumption. We did not include this effect fo r two reasons. First, fecundity and survival were lower than average for the year afte r the 2001 drought, but this could be because 2002 was a moderately dry year. In fact, if we compare vital ra tes of 2002 with vital rates after 1998, the rates are not lo wer than average. So includi ng a lag effect based on our data would not have addressed concerns a bout lag effect. Thus, rather than imposing arbitrary reduction in vital rates we preferre d to ignore this effect If lag effects have major influences on kites, our estimates of stochastic population gr owth rates would be overestimated. Another limitation of our approach is that we did not account for the spatial structure of the kite population. Taking spatia l structure into consid eration would add an important element of realism. We believe th at extending our work to spatially explicit models would be an important next step (Mooij and DeAngelis 2003). Nonetheless, we think that our approach is a useful one b ecause it is based on robust parameter estimates (e.g., survival rates were based on Cormack -Jolly-Seber model) and provides a good starting point for further m odeling efforts. Moreover, even though the Snail Kite population is spatially structured, there is enough movement among habitats for the

PAGE 132

132 population to be considered a single populat ion as opposed to a strict metapopulation (Bennetts and Kitchens 2000; Martin et al. 2006). Finally, our work focused on a limited number of factors. Other dist urbances or catastrophes such as diseases, fire, etc. were ignored. All these factors are likely to increase PQE. Management Implications Our results were based on recent demogr aphic information on Snail Kites in Florida. They were consistent with the hypot hesized critical impor tance of considering drought frequency when evaluating kites pe rsistence (Bei ssinger 1995). However, our results also emphasized the potential importan ce of more moderate drying events that appear to primarily affect reproduction a nd juvenile survival. These moderate drying events had not previously been considered cr itical to kite persis tence (Bennetts 1998). However, under the current system adult fertility appears to be crucial, and factors likely to have large effects on adult reproduction a nd juvenile survival should receive more attention. Thus, our result s suggest that water management plans should focus on reducing the frequency of mode rate drying events. As discus sed earlier, the effect of habitat conversion on kite dynamics should also receive attention, as it has the potential to have a large impact on kite population growth. As suggested by recent studies, the repeated flooding of the wetla nds during the Fall in south Florida is likely to be responsible for habitat conve rsion deemed detrimental to kites (Gunderson and Loftus 1994; Bennetts et al. 1998; Kitc hens et al. 2002). Therefor e, water management plans aimed at improving kite persis tence should also focus on re ducing water levels in the WCAs during the Fall. Reducing flood durat ion and frequency should help restore vegetation communities favorable to foraging kite habitats. We note that the positive effect of this restoration effo rt on kite population growth is likely to only be perceptible

PAGE 133

133 after multiple years of implementation. In c ontrast, during the spring-summer, managers should attempt to reduce the frequency of moderate drying events. Nevertheless, as discussed earlier droughts may play a critical organizing role in the system occupied by kites, therefore managers should probably not attempt to systematically prevent drying events to occur. Ideally, when a drought occurs, managers should accommodate kites by storing some water in refugi a habitats contiguous or near the wetlands affected by the drought (see Martin et al. 2006 and Martin et al. 2007c). Some managers are concerned that lowering water levels during the Fall a nd raising water levels during the Spring and Summer would reduce the amplitude of water le vel fluctuations to a degree that could be detrimental to the system. In fact it has b een proposed to decrease the highs during the Fall, but also decrease the lows during th e spring-summer. Although this management plan could contribute to restor ing suitable foraging habitat for some time during the year, the frequency of moderate drying events a nd possibly drought would also be increased, which would further reduce reproduction and survival. This management strategy would likely increase the probability of extinction. An alternative to this management strategy would be to maintain water levels at higher levels during the spring-summer (i.e., > than during the years that were identified as dry years), but occasionally allow the area to dry down to lower levels than during most moderate drying event. We believe that there is a minimum threshold for each wetland unit below which kites have to leave or die (Mooij et al. 2007). Habitat conversion may in fact have raised this threshold in WCA3A by shifting suitable foraging kite habitats to higher elevations, whic h are more prone to frequent and prolonged drying periods (Mooij et al. 2007). On ce water levels fall below this threshold, it probably does not matter any more from the kite perspective whether

PAGE 134

134 water levels are 1 cm below the threshold or 20 cm below the threshold; kites will have to move or die in both cases. Th erefore, management plans that favor infrequent moderate drying events (e.g. 1 moderate drying event every 7 years), yet allow occasional droughts (e.g., water levels that fall 20 cm below the threshold but occur lets say every 7 years), may be better for kites than management pl ans that would favor more frequent drying events (water levels falls 5 cm below the th reshold but occur with frequency: 6/7 years) even if no drought ever occurred. Indeed, wh en assuming a high frequency of moderate drying events (drought frequency: 0/7; moderate ly dry years frequency: 6/7; wet years: 1/7; with () t = 0.16 to 0.35) s was 0.92; whereas when assuming a low frequency of moderate drying events and droughts (drought frequency: 1/7; moderately dry years frequency: 1/7; wet years: 5/7; with () t = 0.16 to 0.35) s was 0.95. However, the issue is complicated by the fact that lower water le vels may also be more likely to be followed by longer drying events. Drought duration is known to be cr itical to the survival and recruitment of snails (Darby 1998), and thus will adversely affect kites (Mooij et al. 2007). Therefore, managers should also cons ider this factor when setting their management actions. In summary, current management practices of kites habitats appear to be detrimental in two ways: (1) they promot e habitat degradation by favoring prolonged hydroperiod and by increasing flooding fre quency during the period September to February; (2) they increase the propensity of drying events occurring during the dry season (April to June). More generally, our results s upport the argument that it is critical to carefully consider the organizing role of natural di sturbances when modeling the dynamics of

PAGE 135

135 populations of organisms. Indeed, focusing exclusively on their c atastrophic effects without considering their organizing role in the ecosystems may be misleading to both our understanding of the ecological dynamics of the system and to management (Reeves et al. 1995).

PAGE 136

136 Table 5-1. Cormack-Jolly-Seber mode ls (CJS) of apparent survival (AD: survival of adults; J: survival of juveniles) and annual transition probabilities () among the four major wetland complexes used by snail k ites in Florida between 1992 to 2004. Factor s incorporated in the models included: age, region, natal region; and a drought ef fect on movement and survival. Modelbis QAICc AICc w K ADJ(WMD1)(t)p(t) 5597.5420 0.234 31 ADJ(D1)(t)p(t) 5597.5460 0.233 31 ADJ(D2)(t)p(t) 5597.750.21 0.211 31 ADJ(WMD2)(t)p(t) 5598.2860.74 0.161 32 ADJ(D3)(t)p(t) 5598.3670.82 0.155 32 p(t)(age*t) 5604.6667.12 0.007 41 p(t)(aget) 5610.43912.9 0.004 28 ADJAVGJD(D2)(D6/T)p(t) 5615.35917.82 0.0000321 ADJDUR(D2)(D6/T)p(t) 5616.19318.65 0.0000221 ADJAVGJD(D3)(D6/T)p(t) 5616.2218.68 0.0000222 ADJMWL(D2)(D6/T)p(t) 5618.5921.05 0.0000121 ADJ(WMD1)WMD1)p(t) 5622.89425.35 0 22 ADJ(.)(t)p(t) 5623.01925.48 0 29 ADJAVGJD(D2)(D4/T)p(t) 5625.34927.81 0 20 ADJMWL(D2)(T)p(t) 5265.4727.92 0 20 ADJAVGJD(D3)(D4/T)p(t) 5626.50228.96 0 21 Notes: AICc: Akaike information criterion. AICc for the ith model is computed as AICci min (AICc). w: AICc weight. K: number of parameters.: apparent survival. p: detec tion probability.t: time; .: no tim e effect; age: age effect; *: multiplicative effect; +: additive effect. WMD1: is associated with 3 years categorie s, wet (93 to 98), moderately dry (99 to 01, 02-03, 04-05), and drought (92-93, 01-02, but in these models the drought effect on survival is assumed to be the same in 92-93 and 01-02). WMD2: same as WMD1 except that the drought effect on survival is assumed different in 92-93 and 01-02. D1: is constant except during the drought period 00-02, ( during 00-01, is assumed different from 01-02). D2: same as D2 except that 92-93 is also considered a drought the drought effect on surv ival is assumed to be the same in 92-93 and 01-02. D3: same as D2 except that the dr ought effect in 92-93 is assumed to be different from 01-02. TAVGJD: trend model with annual mean water levels for the period Jan-Dec as a covariate. TAVGMJ: trend model with annual mean water levels for the period Mar-Jun as a covariate. TAVGJD: trend model with annual mean water levels for the period Jan-Dec as a covariate. TSPAT: trend model with spatial extent of drying event as a covariate. TDUR: trend model with drying event duration as a covariate. TMWL: trend model with annual minimum water levels as a covariate. TTIME: trend model with time as a covariate. See text for additional details.

PAGE 137

137 Table 5-1. (continued) CJS mode ls of apparent survival Model QAICc AICc K ADJAVGJDAVGJD(D7/T)(D4/T)p(t) 5626.7629.21 21 ADJ(WMD2)WMD2)p(t) 5626.912 29.37 23 ADJAVGJDAVGJD(D5/T)(D4/T)p(t) 5627.374 29.83 20 ADJAVGJD(D1)(D4/T)p(t) 5627.504 29.96 20 ADJTIME(D2)(D6/T)p(t) 5635.9338.39 21 ADJSPAT(D2)(D6/T)p(t) 5636.6839.14 21 ADJDUR(D2)(D4/T)p(t) 5637.064 39.52 20 ADJMWL(D2)(D4/T)p(t) 5637.214 39.67 20 ADJTIMETIME(D7/T)(D6/T)p(t) 5637.6440.1 22 ADJAVGMJ(D2)(D6/T)p(t) 5639.2541.71 21 ADJAVGJDAVGJD(T)(T)]p(t)[ 5641.983 44.44 17 ADJAVGJDAVGJD(T)(T)p(t) 5642.593 45.05 18 ADJAVGMJ(D2)(D4/T)p(t) 5644.897 47.35 20 ADJSPAT(D2)(D4/T)p(t) 5647.766 50.22 20 ADJ(D1)(D1)p(t) 5649.877 52.33 20 ADJ(WMD1)(WMD2)p(t) 5652.142 54.6 22 ADJDURDUR(T)(T)]p(t)[ 5657.333 59.79 17 ADJMWL(D2)(T)p(t) 5657.839 60.3 19 ADJMWLMWL(T)(T)]p(t)[ 5663.667 66.12 17 ADJMWLMWL(T)(T)p(t) 5664.292 66.75 18 ADJAVGMJAVGMJ(T)(T)p(t) 5670.601 73.06 18 ADJAVGMJAVGMJ(T)(T)]p(t)[ 5672.816 75.27 17 ADJSPAT(D2)(T)p(t) 5680.487 82.94 19 ADJ(D1)(.)p(t) 5682.229 84.69 18 ADJ(.)(.)p(t) 5729.546 132 16

PAGE 138

138 Year 92-93 93-94 94-95 95-96 96-97 97-98 98-99 99-00 00-01 01-02 02-03 03-04 04-05 Survival 0.0 0.2 0.4 0.6 0.8 1.0 Adults Juveniles Figure 5-1. Model averaged estimates of a dult and juvenile survival between 1992 and 2005. The colors correspond to the hydrological conditions: red indicate drought years, yellow indicates moderate ly dry years and blue indicate wet years (see Appendix D for details about how the categorization was established).

PAGE 139

139 Figure 5-2. Estimates of probability of quasi -extinction (quasi-extinction threshold was fixed at 50 females) assuming three se parate detections of juveniles: () t = 0.16, () t = 0.7 and () t = 0.16 to 0.35. Error bars correspond to 95% confidence intervals.

PAGE 140

140 Figure 5-3. Sensitivity (a) and elasticity (b) of 1 to changes in age-specific vital rates (F corresponds to fertility rates, and P to survival, subscrip ts indicate the age classes, subscirpt s indicates the suba dults and subscript a refers to the adults)..

PAGE 141

141 Figure 5-4. (Negative bars) The difference in age-specific vital rates (F, corresponds to fertility rates and P, to survival, subscrip ts indicate the age cl asses, a: adults, s: subadults) between the matrix BEF and matrix AFT. (Positive bars) The contributions (in %) of those differen ces to the effect of changes in the environmnent on 1. Differences and contribution were computed for three detection of juveniles ( () t = 0.16, () t = 0.16 to 0.35 and () t = 0.7).

PAGE 142

142 Environment BeforeAfter Stochastic population growth rate 0.75 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 = 0.16, F = 0.11 = 0.16, F = 0.25 = 0.16-0.35, F = 0.11 = 0.16-0.35, F = 0.25 = 0.70, F = 0.11 = 0.70, F = 0.25 Figure 5-5. Estimates of stocha stic population growth rates ( s) for environmental conditions before and after 1998. For each environment (i.e., before and after), we simulated two frequency of droughts: F = 0.25 (one drought every 4 years) and F= 0.11(one drought every 9 years). Estimates of s were computed for three detections of juveniles (() t = 0.16, () t = 0.16 to 0.35 and () t = 0.7). Error bars correspond to 95% confidence intervals.

PAGE 143

143 Environment LFMDHFMD Stochastic population growth rate 0.80 0.85 0.90 0.95 1.00 1.05 = 0.16 = 0.16-0.35 = 0.70 Figure 5-6. Estimates of stocha stic population growth rates ( s) assuming a low frequency of moderate drying events (LFMD) and a high frequency of moderate drying events (HFMD). LFMD corresponded to the frequency of wet, dry and drought years that were observed during the period 1992 to 1998. HFMD corresponded to the frequency of wet, dry and drought years that were observed during the period 1999 to 2005. Estimates of s were computed for three detection of juveniles ( () t = 0.16, () t = 0.16 to 0.35 and () t = 0.7).. Error bars correspond to 95% confidence intervals.

PAGE 144

144 Environment BeforeAfter Stochastic population growth rate 0.80 0.85 0.90 0.95 1.00 1.05 1.10 = 0.16, LFMD = 0.16, HFMD = 0.16-0.35, LFMD = 0.16-0.35, HFMD = 0.70, LFMD = 0.70, HFMD Figure 5-7. Estimates of stocha stic population growth rates ( s) for environmental conditions before and after 1998. For each environment (i.e., before and after), we simulated environments with two frequencies of wet, dry and droughts: LFMD (frequencies of wet, dry, droughts observed during 1992-1998); and HFMD (frequencies of wet, dry, droughts observed during 1999-2005). Estimates of s were computed for three detections of juveniles ( () t = 0.16, () t = 0.16 to 0.35 and () t = 0.7). Error bars correspond to 95% confidence intervals.

PAGE 145

145 CHAPTER 6 CONCLUSION In this final chapter I synthesized the inform ation presented in the preceding chapters and provided a framework for understanding the populat ion ecology of the Snail Kite and developed a vision for its recovery. Synthesis of Research Findings Monitoring of the Snail Kite During the last decade a grow ing number of wildlife studies have emerged which estimate detectability. Additionally, st rong statements about the importa nce of estimating detectability have been made decades ago (Pollock et al. 1990, Bennetts et al. 1999b), yet, Rosenstock et al. (2002) showed that out of 224 papers published in major journals, 95% of bird monitoring studies that they reviewed did not account for detectability. More over, analyses of uncorrected counts continue to be published in major journa ls (reviewed in Conn et al. 2004). As indicated by Conn et al. (2004), the controversy about the va lue of uncorrected coun ts persists because protocols to estimate detectability are typically mo re labor intensive and expensive to implement. Furthermore, studies that address both detectability and spatial vari ation are even less frequent (Yoccoz et al. 2001). As indicate d in Chapter 2, the principle that detectabili ty and spatial variation should carefully be considered in mon itoring programs appears to take even longer to reach managers. As a consequence, recovery pl ans for many endangered species are still based on count surveys that do not account for either so urces of variability. This is very troubling because, as shown in Chapter 2, primary recove ry criteria based on uncorrected counts could have realistically been met, while in fact the sp ecies is declining precip itously. Studies such as those from Cassey and Ussher ( 1999) nicely demonstr ate that uncorrected counts can largely underestimate abundance of endangered species; however, these authors did not examine the

PAGE 146

146 biases associated with using count based indices as measures of population change. Yet, this is exactly where the controversy cont inues to be the most intense. Chapter 2 provides a clear and strong message to conservation biologists and ma nagers to urge them to revise monitoring programs and recovery plans that igno re primary sources of variability. Analyses in Chapter 2 are based on a superpopul ation model that is simple to implement and is well suited to deal effectively with both detectability and spatial variation when estimating population size. This is also a valuable met hod to estimate fecundity and fertility rates (parameters that are often diffi cult to estimate appropr iately). Bennetts (1998) pointed out that fecundity (i.e., number of young females produced per adult female) is one of the most important reproductive rates to understand Snail Kite dynamics (see also Caughley 1977). Bennetts (1998) proposed to derive this rate fr om the proportion of birds attempti ng to breed, the proportion of breeding attempts that are successful and the nu mber of breeding attempts per year, the number of young produced per successful nest and the sex ratio of the young produced. This approach is prohibitively expensive (see Bennetts 1998). A cheaper alternative presented in Chapters 2 and 5 allows for a more direct estimation of fecundity ra tes. Fertility rates can then be derived from the approach described in Chapter 5 or from th e anonymous reproduction equation described by Caswell (p.173, 2001). Accounting for detectability and spatial vari ation is also important when estimating survival and movement rates (e.g., Williams et al. 2002). Bennetts and Kitchens (1997) implemented a very nice study design to estimate movement and survival rates while accounting for these sources of variation. In Chapters 3 and 4 we expanded some of their analyses (e.g., Bennetts and Kitchens 2000, Bennett s et al. 1999a) to take scale i ssues into consideration. We

PAGE 147

147 found that a careful cons ideration of scale both in space (wetlands versus regions) and time (month versus year) was necessary to better understand kite dynamics. Thus, the research presented in this volume pr ovided compelling evidence that it is critical to consider scale, detectability and spatial va riation when estimating movement and demographic parameters of Snail Kites. Although monitoring programs that account for detectability and spatial variations are more costly (Conn, Bailey et al. 2004), in the case of the Snail Kite this investment is clearly necessary. New information on movement and demography of the Snail Kite Snail Kites have been desc ribed as nomads (Beissinger 1983, Bennetts and Kitchens 2000). Bennetts and Kitchens (2000) estimated the average probability of movement among wetland units to be approximately 0.25 per month. However, when we incorporated additional levels of complexity of the spatial configuratio n of the system, we discovered a different pattern. In Chapter 3, we found that kites moved extensively among contiguous wetlands, but significantly less among isolated wetlands. Our an alyses indicated that kite movement was affected by fragmentation. Results in Chapter 3 also suggest that young birds may be even more affected by fragmentation than adults. Familiarity with the landscapes is particularly important during regional droughts (Bennett s and Kitchens 2000). In Chapters 3 and 4 we found that only a small proportion of kites escaped a regional drought by moving to refugia (wetlands less affected by drought). Many individuals died after the dr ought. During the 2001 drought adult survival dropped by 16%, while juvenile su rvival dropped by 86% (possibly because juveniles were less likely to reach refugia). Although kites exhibit ex tensive exploratory beha vior, they also show high levels of annual site tenacity during the bree ding season, particularly to their place of birth (see Chapter 4). Fidelity to breedi ng and natal sites has been given re latively little attention in the past. However, as discussed in Chapters 3 and 4, site fidelity could ha ve a large influence on

PAGE 148

148 movement patterns and survival, and could influen ce the entire dynamics of the kite population. In Chapter 3 and 4 we found that survival vari ed significantly among regi ons. During most years, survival was higher in the southe rn regions than in the north. In contrast, during the 2001 drought survival of kites occupying th e northern regions was higher. This confirmed hypotheses from Takekawa and Beissinger (1989) and Bennetts and Kitchens ( 1997), which stated that the Kissimmee Chain of Lakes may serve as a refuge ha bitat. It also emphasized the importance of considering spatiotemporal patte rns of variation in habitat conditions when assessing the conservation values of particular wetlands (see Chapters 3 and 4). Current Status of the Snail Kite and Management Implications Population size estimates for the period 1997 to 2000 presented by Dr eitz et al. (2002) were four to five times greater than the recovery target set by the USFWS, and survival estimates for the period 1992 to 1999 presented by Bennetts et al. (2002) remained high for both adults and juveniles. Although neither Bennett s et al. (2002) nor Dr eitz et al. (2002) s uggested that Snail Kites were recovering, their results were interpre ted by some managers as evidence that the Snail Kite population in Florida was heading toward rec overy. However, results presented in Chapters 2 and 5 refute the idea of a Snail Kite rec overy. In fact, recent s uperpopulation estimates (Chapter 2) indicated that the Snail Kite popula tion declined dramatically in recent years. Estimates of stochastic populati on growth rate and probabilities of quasi extinction presented in Chapter 5 suggest that the Snail Kite is still at high risk of extinction. The sharp decline observed after 2001 was mostly associated with a multiregional drought that occurred in 2001 and affected both survival and reproduction (see Ch apters 2, 3, 4, and 5). Howeve r, the lack of recovery after 2002 suggests that other factors are limiting the growth of Snail Kites in Florida. The Life Table Response Experiment presented in Chapter 5, sugge sts that a reduction in r ecruitment is probably responsible for the lack of recovery. In Ch apter 5, we found evidence that both habitat

PAGE 149

149 conversion (caused by prolonged hyd roperiod and increased water depth during the Fall), and increase in frequency of drying events (duri ng the Spring and Summer) are responsible for the observed reduction in population growth rate (see Chapter 2). Chapters 3 and 4 indicate that fragmentation, movement and site fi delity should also be considered when managing the network of wetla nds occupied by kites. Although fragmentation is certainly not the cause of th e recent decline in kite abund ance, Chapter 3 provides evidence that further fragmentation and habitat reduction would reduce the resistance of kites to drought and therefore increase probability of extinctions. The fact that k ites respond to fragmentation and to the geographic features of the landscapes (s ee Chapter 3), leads to the logical recommendation that managers should manage the wetlands occupied by kites as a network and not as independent units (see also Takekawa and Be issinger 1989 and Bennetts and Kitchens 1997, and Kitchens et al. 2002). Given the paradigm that the persistence of good natural habitats requires occasional drying events (see Chapter 5, Benne tts et al. 1998, and Kitchens et al. 2002). Restoration projects that involve wholesale dry downs of an entire region (e.g., restoration of Lake Tohopekaliga) (Welch 2004) may want to c onsider the option of co nserving water in at least some local patches within the region to be a ffected to serve as a refuge for Snail Kites. The draw downs of local patches s hould occur sequentially, allowing a sufficient recovery period for previously dried areas to retu rn to a productive level. Moreo ver, the pattern of drying and inundation should optimally attempt to mimic as closely as possible the hydrology of the Everglades under a more natural landscape (Fennema et al 1994). Finally, findings presented in Chapters 3 a nd 4 emphasized the importance of considering spatiotemporal patterns of variation in habitat conditions when assessing the conservation values of particular wetlands. For instance, we found that during most years kite survival was higher in

PAGE 150

150 the southern than in the nor thern regions. However, during drought kite survival in the Kissimmee Chain of Lakes was less impacted by droughts than the southern regions. Therefore, in order to make a correct assessment of the cons ervation values of habitats one should consider the frequency of droughts in these wetlands. Perspectives The population monitoring program of the Snail K ite that has been in place for more than 16 years offers many additional research questions related to conservation, management and population ecology. In this section I present a non-exhaustive list of research directions and ideas that I consider reasonable prioriti es for conservation and management. Rescuing the Snail Kite Population from Ex tinctions Risks Associated with Small Populations Recent estimates indicate that the Snail Kite population size includes less than 2000 individuals (Chapter 2) Projections presented in Chapte r 5 suggest that the population is expected to experience further decrease. Although current estima tes of abundance do not suggest that Allee effects and demographi c stochasticity are sources of im mediate concern (at least in the short term), genetic stochasticity on the other hand may constitute an immediate problem for the Snail Kite population. As explained in Chapter 1, there are other populati ons of Snail Kites in Cuba and in Central and South Am erica. It is not clear at this point if all these populations constitute different Evolutionary Significant Units ( sensu Fraser and Bernatchez 2001)). Some authors have noted differences in morphometr ics measurements among these populations (Sykes et al. 1995). However, genetic and ecological exchangeability among these populations have not been carefully considered (Beissi nger 1988). Nevertheless, establishing the level of gene flow (or genetic divergence) among these pop ulations could be of great cons ervation value. If gene flow between the Snail Kite populati on in Florida and other populations (e.g., Cuba, Central America)

PAGE 151

151 is low (i.e., less than one migrant per generation) genetic stochasticity (e.g., due to inbreeding depression) could be a serious co ncern for the viability of the Sn ail Kite. Given that population viability analyses presented in Chapter 5 suggest that the Snail Kite population in Florida may reach extremely low numbers of individuals in a relatively short period of time, management considerations to prevent gene tic extinction or increase geneti c diversity of the Snail Kite population in Florida may need to be consid ered soon. Captive breeding and supplemental translocations are among the conservation optio ns that deal with these problems. Although captive breeding is a popular rec overy technique recommended by management agencies, it is often impractical for many vertebrate populati ons (Allendorf and Luikart 2007). In the case of the Snail Kite supplemental translocation (i.e ., introduction of a few individuals from source populations in Central America (or other populations) into the population in Florida), appears more promising. However, supplemental tran slocation is risky wh en populations are not genetically and ecological exchangeable (Allendo rf and Luikart 2007). With these concerns in mind, I recommend (1) additional studies about the genetic structure of Snail Kite in Florida (e.g., effective population size, level of heterozygozity etc.); (2) st udies to establis h the level of genetic exchangeability (e.g., reciprocal m onophyly based on mitochondrial DNA); studies to determine the level of gene flow among popul ations; (4) studies to determine ecological exchangeability (e.g., study of courtship behavior etc.) with populations that could be potential candidates for translocation programs. Evidently these programs will not be substitutes for the need to restore Snail Kite habitats. In the next section I provide some ideas on how to improve habitat management techniques.

PAGE 152

152 Habitat Management Models Everkite In 2002, Mooij et al. published an article that described a de tailed individual based model of the Snail Kite population in Fl orida: Everkite. Everkite is a comprehensive spatially explicit model that includes a large number of parameters (e.g., reproduction, survival, movement etc), that are linked to hydrological c onditions. This model is a valuab le tool to explore questions related to the population ecology of kites, but can also be ap plied to address management questions, in fact it could be incorporated into an adaptive management framework (see below). More specifically, Everkite can be used to asse ss the effect of water management scenarios on Snail Kites. At this point many of the biological parameters included in Everkite are based on count data. Updating this model with movement a nd survival estimates presented in Chapters 2, 3 and 4 would probably improve th is models predictive performa nce and value for management. In addition, establishing a clea r mechanistic link between hydrol ogy, vegetation, Apple Snails and Snail Kites would also be useful. Adaptive management Adaptive management is a powerful framewor k to manage ecological systems (Dorazio and Johnson 2003, Williams et al. 2002). Kitchens et al. (2002) recognized the value of adaptive management to help Snail Kite recovery. In the following section I propose to use adaptive management to address a specific problem relate d to management of kite habitat; but this framework can be used to resolve many other mana gement issues related to Snail Kite recovery. As explained in Chapter 5, re storing suitable nesting and fora ging kite habitats in WCA3A is key to kite recovery. This restoration effort can be achieve d by acting on hydrology. I believe that adaptive management is a met hod of choice to reach this goal.

PAGE 153

153 Nichols (2001) notes that adap tive management require five pr erequisites: (1) specification of an objective function; (2) a model set includi ng predictive models associated to competing hypotheses about how the system will respond to management actions; (3) prior probabilities associated to each predictive model; (4) a fini te set of management actions; (5) a monitoring program to keep track of the system state a nd other variables releva nt to the objective. Once these prerequisites have been met, Nichols (2001) suggests the following four iterative steps: (1) identify th e state of the system at time t ; (2) probabilities associated to each predictive model are updated based on informa tion obtained about the system state at time t ; (3) the management action which is expected to pro duce the optimal return (in term of the initial objective function) is selected; (4) after this ma nagement action is implemented we returns to step 1. We can apply the approach described by Nic hols to the implementation of optimal water regimes in WCA3A that will maximize the persistence of kites. A logical objective for this management problem could be to minimize the pr obability of quasi-extinction of Snail Kite. As explained in Chapter 5, the primar y factors that we identified as most critical to kite recovery are: drying events intensity, dur ation and frequency, and flooding events intensity, duration and frequency. We need to develop models that wi ll predict how kite will respond to management actions (e.g., given intensity, duration and freque ncy of drying and flooding events). Each model can be assigned a probability based on its cr edibility (Kendall 2001; Nichols 2001). We should then derive the management action that maximize the return (i.e., minimi ze probability of quasiextinction). After implementation of the selected mana gement action, we s hould reevaluate the predictive models and give more weight to the models whose predictions agreed the most with

PAGE 154

154 the observed changes in the system-state (Ke ndall 2001; Nichols 2001). A nother iteration of the process can then be started. This adaptive management approach has not yet been rigorously a pplied to Snail Kite habitat management. Unforeseen difficulties may of course arise during its implementation (e.g., agreement over the finite set of management actions). However, if these difficulties can be surmounted, this approach could prove itself to be a very efficient way to manage kite habitats (Nichols and Williams 2006).

PAGE 155

155 APPENDIX A SURVEY-SPECIFIC PARAMETER ESTIMATES USED TO COMPUTE ESTIMATES OF SUPERPOPULATION SIZE Table A-1. Survey-specific parameter estimate s used to compute estimates of superpopulation size for Snail Kites between 2001 and 2005. Year Survey ij ijSE() ij p ijSE(p) ij N ij B Yij N Yij B 2001 1 0.71 0.08 0.39 0.07 944 2 0.70 0.07 0.39 0.07 1135 467 3 0.69 0.07 0.39 0.07 1190 398 4 0.70 0.07 0.39 0.07 827 4 5 0.71 0.07 0.39 0.07 487 0 6 0.39 0.08 559 215 2002 1 0.91 0.10 0.33 0.09 1117 2 0.91 0.15 0.33 0.09 1174 163 3 0.91 0.10 0.32 0.09 1192 126 4 0.32 0.07 870 0 2003 1 0.98 0.10 0.29 0.09 902 2 0.92 0.15 0.29 0.09 854 0 3 0.98 0.11 0.28 0.09 697 0 4 0.27 0.07 943 260 2004 1 0.89 0.07 0.27 0.05 1155 2 0.87 0.07 0.27 0.05 928 0 186 3 0.86 0.08 0.25 0.05 975 169 350 189 4 0.86 0.08 0.26 0.05 737 0 151 0 5 0.87 0.08 0.26 0.05 716 82 123 0 6 0.26 0.05 716 92 148 41 2005 1 0.88 0.07 0.29 0.06 1371 2 0.93 0.05 0.29 0.06 1401 195 3 0.90 0.06 0.26 0.06 1254 0 4 0.93 0.05 0.29 0.05 953 0 14 5 0.89 0.05 0.36 0.09 827 0 31 18 6 0.28 0.05 722 0 50 23 aParameter definitions: ij, apparent survival probability; ij p detection probability; ij N abundance of adults; Yij N abundance of young of the year; ij B number of new adult individuals in the sampled populations; Yij B number of new young individuals in the sampled population.

PAGE 156

156 APPENDIX B CONFIDENCE INTERVALS, MODEL S ELECTION TABLES AND MOVEMENT ESTIMATES Appendix B-1. Method from Burnham et al (1987) to compute confidence intervals on parameter When a parameter is assumed to be lognormally distributed, lower and upper bounds L and U for an approximate (1)%CI[ ], are computed as: L = / C ; and U = C ; where 2 /2expln(1[()]) Czcv cv( ) is coefficient of variation of ; z0.025 = 1.96 (for 95%CI, = 0.05). This approach provides a better ap proximation than normal based 95%CI [ ]= + 1.96* SE[ ], when parameter is strictly positive and/or cv( ) > 0.1 (Burnham and Anderson 2002; Burnham et al 1987).

PAGE 157

157 Table B-1. Multistate models of monthly movement probabilities ( ) of adult ( ad ) and juvenile ( juv ) snail kites among the 5 major regiona l patches in Florida based on radiotelemetry data. These models evaluate the effect of patch size distance, regional patch identity alone on movement probabilit ies. The models presented below are not supported by the data but were developed to evaluate the fit of the models presented in Table 3-1. Model AICc K a-Movement amon g re g ions of j uveniles and adults modelled simultaneousl y ad(r) juv(r) 15.2 60 ad(AR*d) juv(AR*d) 16.5 26 ad(AR+d) juv(AR+d) 21.8 25 (AR+d) 22.4 12 (AR*d) 24 13 (r*season) 29.2 90 (AR) 48 11 (r*years) 96.9 120 ad(AD) juv(AD) 103 19 (AD) 104 11 ad(d) juv(d) 118.6 22 (d) 127.1 11 ad (r*seas) juv(r*seas) 134.1 180 ad (seas) juv(seas) 173.4 6 ad (years) juv(years) 184.3 8 (.) 203.2 1 (seas) 186.7 3 (seas*years) 189.2 10 (years) 203.7 4 (t) 215.3 40 (breed) 205.1 2 ad(t) juv(t) 222.1 78 ad(r) juv(d) 933 36 ad(r) juv(AR) 1081 29 ad(AR) juv(AR) 69300 15 b-Juveniles only juv(r) 14 30 c-Adults only ad(AR+d) 13.6 12 ad(AR*d) 15.4 13 Notes: AICc is the Akaikes Information Crite rion adjusted for small sample size. AICc for the ith model is computed as AICcimin (AICc). w refers to AICc weight. K refers to the number of parameters. Only models with w < 0.01 are presented. t: month; r: region; AR: area of the receiving site; AD: area of the donor site; d: distance; seas: seasonal; breed: breeding season. Models that never reached numerical convergence were not presented.

PAGE 158

158 Table B-2. Multistate models (wit h survival and detection probabil ities equal to 1) of monthly movement probabilities ( ) of adult ( ad ) and juvenile ( juv ) snail kites among wetlands in the E and K regions based on ra dio-telemetry data. The models presented below are not supported by the data but were developed to evaluate the fit of the models presented in Table 3-2. Model AICc K Movement within the E region (cw) 12.4 20 ad(cw) juv(cw) 21.5 38 ad(cw) juv(AR*d) 25.8 22 ad(cw) juv(AD) 39.2 21 ad(cw) juv(AR+d) 40.3 21 ad(cw) juv(AR) 42.4 21 ad(cw) juv(d) 49.3 20 (cw*years) 58.6 72 (AR*d) 60.6 3 (seas*region) 65 120 (AD) 82.4 2 ad(AD) juv(AD) 85 4 (seas*year) 99.3 10 (AR+d) 102 3 (seas) 102.3 3 ad(AR+d) juv(AR+d) 103.2 6 ad(AR) juv(AR) 103.2 4 (AR) 104.2 2 (.) 106.6 1 (d) 107.7 2 ad(.) juv(.) 108.3 2 (breed) 108.5 2 (breed*year) 108.8 7 (years) 109.7 4 ad(d) juv(d) 111.1 4 (t) 111.8 34 ad(years) juv(years) 783.7 7 ad(t) juv(t) 799.6 48 ad(cw*years) juv(cw*years) 1606.1 95

PAGE 159

159 Table B-2 (cont.). Model AICc K Movement within the K region (years*breed) 9.2 7 ad(years) juv(years) 10.1 7 ad(mw) juv(mw) 20.4 24 (t) 29.2 37 Notes: AICc is the Akaikes Information Criterion. AICc for the ith model is computed as AICcimin (AICc). w refers to AICc weight. K refers to the number of parameters. Only models with w < 0.01 are presented. cw: contiguous wetland; mw: moderately isolated wetland; seas: season; breed: breeding season. Models that never reached numerical convergence were not presented.

PAGE 160

160 Table B-3. Multistate mode ls of annual survival (), sighting probabilities ( p ), and annual movement probabilities ( ) of adults ( ad ) and juveniles ( juv ) snail kites based on banding data. The drought eff ect during 2000-2002 was denoted D1-2. Constant during non-drought years (1992-200 0 and 2002-2004) was denoted ND The models presented below are not supported by the data but were developed to evaluate the fit of the models presented in Table 3-3. Model AICc K [][]12()(.)()()(*)()EOKJKEOJdjuv ad adad N DDrtprtr 11 80 [][]12()(.)()()(*)()EOKJKEOJjuv ad adad N DDtprtr 14 76 [][]12()(.)()()()()EOKJKEOJdjuv ad adad N DDrtadrjuvr 15.6 88 [][]12()(.)()()(*)()EOKJKEOJjuv ad adad N DDtprtr 16 77 []12(*)(.)()()(*)()KEOJjuv adad adrNDDtprtr 16.3 82 [][]12()(.)()()(*)()EOKJKEOJjuv ad adad N DDtprtr 16.6 79 [][]12()()()(*)()EOKJEOJKjuv adad N DDtprtr 16.7 78 [][]12()(.)()()(*)()EOKJKEOJjuv ad adad N DDtprtr 16.8 79 [][][]12()(.)()()(*)()EOKJKEOJEOKJad juv adad N DDtprtr 17.2 88 [][]12()()()(*)()EOKJEOJKdjuv adad N DDrtprtr 17.8 81 12(*)(*)()(*)()juv adadrNDrDtprtr 18.7 84 [][]12()()()(*)()EOKJEOJKjuv adad N DDtprtr 19.2 81 [][]121()(.)()()(*)(*)EOKJKEOJjuv ad adad N DDtprtrD 19.6 88 [][]12()(.)()()(*)()EOKJKEOJdjuv ad adad N DDrtprtr 19.7 77 [][]12()()()(*)()EOKJEOJKdjuv adad N DDrtprtr 19.8 75 ()()(*)()djuv adrtrtprtr 22.9 86 []121(*)(.)()()(*)(*)KEOJjuv adad adrNDDtprtrD 22.9 95 [][]12()(.)()()(*)()()EOKJKEOJadjuvjuv ad adad N DDtprtrr 24.2 87 [](.)()(*)()EOKJdjuv adrtprtr 36.8 75 []12(*)(.)()()(*)()KEOJdjuv adad adrNDDrtprtr 42.4 75 []12(*)(.)()()(*)()KEOJdjuv adad adrNDDrtprtr 43.5 80 12(*)(*)()(*)()()dadjuvjuv adadrNDrDrtprtrr 60.7 81

PAGE 161

161 Table B-3 (cont.). Model AICc K [][]12()(.)()()()()EOKJKEOJdjuv ad adad N DDrtprtr 108.6 44 [][]12()(.)()()()()EOKJKEOJdjuv ad adad N DDrtptr 145.8 40 [][]12()(.)()()(*)()EOKJKEOJjuv ad adad N DDtprtt 171.1 76 [][]12()(.)()()(*)(.)EOKJKEOJjuv ad adadNDDtprt 192.9 65 [][]12()(.)()()()()EOKJKEOJjuv ad adad N DDtprr 226.4 32 [][]12()(.)()()(.)()EOKJKEOJjuv ad adad N DDtpr 263.4 29 Notes: AICc is the Akaikes Information Cr iterion adjusted for small sample size. AICc for the ith model is computed as AICcimin (AICc). w refers to AICc weight. K refers to the number of parameters. Only models with w < 0.01 are presented. t: time (years); r+td: additive effect of region and time on except during 2001-2002,during which was time dependent only; .: is constant during 1992-2004. Superscript indicate region specific ; =: regions have identical ; : regions have different

PAGE 162

162 Table B-4. Annual movement estimates () between the 4 major regions used by the snail kite (E, O, K and J) during normal and drought years estimated with model [][]121()(.)()()(*)(*)EOKJKEOJdjuv ad adad N DDrtprtrD. MOVEMENT FROM TO SE[ ] 95%CI Year E O 0.017 0.004 0.010-0.026 Normal E O 0 0 0 Drought E K 0.015 0.003 0.010-0.022 Normal E K 0.03 0.012 0.014-0.066 Drought E J 0.016 0.003 0.010-0.025 Normal E J 0.027 0.017 0.008-0.091 Drought O E 0.156 0.024 0.115-0.209 Normal O E 0 0 0 Drought O K 0.044 0.014 0.024-0.080 Normal O K 0.33 0.117 0.146-0.580 Drought O J 0.039 0.013 0.076 Normal O J 0.018 0.096 0-1 Drought K E 0.092 0.021 0.058-0.133 Normal K E 0 0 0 Drought K O 0.1 0.025 0.063-0.161 Normal K O 0 0 0 Drought K J 0.084 0.02 0.052-0.133 Normal K J 0.14 0.111 0.026-0.497 Drought J E 0.155 0.033 0.010-0.231 Normal J E 0 0 0 Drought J O 0.028 0.017 0.009-0.090 Normal J O 0 0 0 Drought J K 0.062 0.02 0.033-0.115 Normal J K 0.000 0.000 0 Drought

PAGE 163

163 APPENDIX C SELECTION OF MODELS USED TO ASSE SS THE EFFECT OF NATAL LOCATION ON MOVEMENT AND SURVIVAL Table C-1. Multistate models of apparent survival (AD: survival of adults; J: survival of juveniles) and annual tr ansition probabilities ( ) among the four major wetland complexes used by snail kites in Florida between 1992 to 2004. Factors incorporated in the models included: age, region, nata l region; and a drought effect on movement and survival. Model AICc K DEV [ELJK][ELJK][NOONOA] J AD,AD,rr(.)(D)(t)nrp(r*t)ndd 12.26 84 2586.5 [ELJK][KJ][EL][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 12.32 72 2612.0 [ELJK][K][ELJ][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 12.63 80 2595.3 [ELJK][K][ELJ][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 13.56 76 2604.8 [ELJK][K][ELJ][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 14.15 82 2592.6 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,mrrr(.)(ND)(D)(t)nr*Dp(r*t)nddd 14.17 88 2579.8 [ELJK][K][ELJ][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 15.07 79 2599.9 [ELJK][K][ELJ][NOOAON] J AD,AD,AD,nrnrnr(.)(ND)(D)(t)nrp(r*t)nddd 15.19 77 2604.3 [ELJK][ELJK][NOOAON] J AD,AD,nrnr(.)(D)(t)nrp(r*t)ndd 17.72 84 2591.9 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(nr*t)nrp(r*t)nddd 21.2 103 2554.6 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(age*r*t)nddd 24.53 119 2523.2 [ELJK][ELJK][NOOAON] J AD,AD,rr(.)(D)(t)nrp(r*t)ndd 29.48 75 2622.8 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 33.08 72 2632.7 [ELJK][ELKJ][NOONOA] J AD,AD,rr(.)(ND)(t)nrp(r*t)ndd 46.74 74 2642.2 [ELJK][NOONOA] ADJr(t)(t)nrp(r*t) 49.4 116 2554.6 [ELJK][ELJK][NOONOA] J AD,AD,rr(.)(ND)(t)nrp(r*t)ndd 49.7 76 2640.9 [ELJK][ELKJ][NOONOA] J AD,AD,rr(.)(ND)(nr*D)nrp(r*t)ndd 51.63 73 2649.2 [ELJK][ELKJ][NOONOA] J AD,AD,rr(.)(ND)(t)nrp(r*t)ndd 59.84 73 2657.4 [NOONOA] ADJ(r*nr*D)(t)nrp(r*t) 61.2 119 2559.8 [NOONOA] ADJ(r*nr)(t)nrp(r*t) 68.6 88 2634.2 Table continues next page

PAGE 164

164 Table C-1 (cont.) Model AICc K DEV [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(.)nrp(r*t)nddd 82.81 65 2697.2 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(age*t)nddd 91.54 50 2737.3 [ELJK][KJ][EL] J AD,AD,AD,rrr(.)(ND)(D)(t)rp(r*t)nddd 102.05 76 2693.2 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(t)nddd 102.49 40 2768.9 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 112.14 72 2711.8 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r*t)nddd 119.47 68 2727.6 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(r)nddd 161.3 32 2844.2 [ELJK][KJ][EL][NOONOA] J AD,AD,AD,rrr(.)(ND)(D)(t)nrp(.)nddd 188.26 29 2877.3 [ELJK][KJ][EL] J AD,AD,AD,rrr(.)(ND)(D)(t)tp(r*t)nddd 239.47 76 2830.7 [ELJK][KJ][EL] J AD,AD,AD,rrr(.)(ND)(D)(t).p(r*t)nddd 253.09 65 2867.5 (age*r*t)r*tp(r*t) 337.12 275 2472.5 Notes: Only models with AICc > 11 (see Table 4-1 for models with AICc < 11). AICc weights are not presented in digital appendix S1 because w ~ 0 for all models. For other notations see Table 4-1.

PAGE 165

165 APPENDIX D CLUSTERING ANALYSIS Cluster analyses are numerical methods th at allow the identification of groups in multivariate data (Venables & Ripley 2002). We conducted an agglomerative hierarchical clustering analysis on a m x n matrix (Everitt 2005). The rows ( m ) corresponded to the years (form 1992 to 2006), and the columns ( n ) to hydrological variables. To avoid the dependence on the choice of m easurements units, we standardized the measurements based on recommendations from Kaufman and Rousseeuw (1990): igg ig g x m z s where mg is the arithmetic mean of variable g, xig is the ith observation of variable g ( i = 1 to n ), and sg is the mean absolute deviation : 121 ... g ggggnggsxmxmxm n We selected 5 biologically rele vant hydrological variables: minimum water levels for the entire year. average water levels for the period March to June (which is more biol ogically relevant to kites than average water levels for the entire year, see Martin et al. (2006) and Mooij et al (2007)). proportion of wetlands affected by dry conditi ons (< 1 SD below the annual minimum for the period March-June, see Bennetts and Kitchens 1997 and Martin et al. 2006). duration of drying event (i.e., number of days for which water levels fell below 1 SD below the mean annual minimum, see Bennetts and Kitchens 1997). duration of moderate dry conditions (i.e., numbe r of days for which water levels fell below the mean annual minimum). The results of the agglomerative hierarchic al analysis are summarized in Figure D-1. From this analysis we could identify 3 groups of years:

PAGE 166

166 Group 1 (G1): 1992 and 2001 Group 2 (G2): 1999, 2000, 2002, 2004 and 2006 Group 3 (G3): 1993, 1994, 1995, 1996, 1997, 1998, 2003 and 2005 This grouping was consistent with grouping base d on visualization of Table D-1. Figure D1 shows that G1 included the driest years, G2 included years with more moderate dry conditions, G3 included years with the wettest conditionsers with other wet years (F igure D-1). Therefore based on Table D-1 and Figure D-1 we identified 3 categories: 1-Drought years: years included in G1. Note from Table 7-4 that 2001 was a more intense drought with a greater spatial a nd temporal extent than 1992. 2-Moderately dry years: years included in G2. 3-Wet years: years included in G3.

PAGE 167

167 Table D-1. Five hydrological vari ables used to conduct the hierarch ical agglomerative clustering analysis. (1) Annual minimum water levels at station 3A65 (MWL); (2) average water levels for the period March to June (WL (Mar-Jun)) at station 3A65; (3) spatial extent of wetlands affected by dry cond itions; (4) duration of drying event (i.e., number of days for which water levels fell below 1 SD below the mean annual minimum, see Bennetts and Kitchens 1997) (D uration II); (5) durati on of moderate dry conditions (i.e., number of days for which water levels fell below the mean annual minimum) (Duration I). Year MinWL WL (Mar-Jun)Duration IDuration II Spat. Ext. 1992 7.92 8.6054.0014.0054% 1993 9.13 9.680.000.009% 1994 8.93 9.360.000.000 1995 9.09 9.700.000.000 1996 8.95 9.310.000.008% 1997 8.89 9.300.000.000 1998 8.49 9.5317.000.000 1999 8.46 8.9439.000.0017% 2000 8.31 8.8619.000.000 2001 7.83 8.35107.0030.0072% 2002 8.37 9.0030.000.000 2003 8.90 9.310.000.000 2004 8.25 8.8345.000.008% 2005 8.62 9.160.000.000 2006 8.36 8.8648.000.0025%

PAGE 168

168 Figure D-1. Agglomerative hierar chical analysis which categori zed years from 1992 to 2006 into wet years (blue), moderately dry year s (yellow) and drought years (red).

PAGE 169

169 APPENDIX E ESTIMATION OF Q(T) Estimates of () qt can be obtained based on es timates from Bennetts (1998). ()aaaaa s ssssmSY qt mSY Where Si corresponds to nest success (subscripts i ndicate the age class to which the parameter pertains to, adults: a and subadults: s ); i is the number of breed ing attempts per year; i is the proportion of birds attempting to breed; Yi is the number of young per successful nest; and the sex ratio is assumed to be 50:50 (see Beissinger 1995). Beissinger (1995) assumed no difference in nest success and number of young pe r successful nest. Based on a radiotelemetry study, Bennetts (1998) estimated (1995)a = 1 (i.e., 100% of the adult Snail Kites attempted to breed in 1995 which was a wet year), and (1995)s = 0.33. Based on his radiotelemetry study, Bennetts estimated (1995)a = 1.4, but did not get an estimate for s Therefore we set s = 1 based on Beissinger (1995), which noted most subadults nest later in the season. Thus, based on Bennetts estimates 0.311.41.9 (1995)4.24 0.30.3311.9 q In contrast, based on Beissinger (1995), () qt= 8 during lag years, 8.8 during wet years, and 1 during drought (however this latter estimate was based on less reliable data). We used () qt based on Bennetts estimates because these estimates we re more recent and based on a more robust methodology. Varying q(t) from 4 to 8.8 had little effect on population growth rates and other relevant measures.

PAGE 170

170 APPENDIX F DETECTION OF JUVENILES Estimates of detection of juveniles () t were available for 2004 and 2005 only (Martin et al. 2007b). In contrast, estimates of detection of adults were available for the period 1997 to 2005. Martin et al. (2007b) found that detection for the adults increased linearly over time between 1997 and 2005. They suggest that this incr ease in detection was the result of an increase in search effort. Thus, we fitted a linear regression to their data (Table 2, MC data); the dependent variable was detection of adults ()at and the independent variable was year ( t ). We found a significant rela tionship (F = 24.63, df = 1 and 6, P = 0.0025, R2 = 0.80): ()at = 38.1158 + 0.0192 x t We adjusted this equation for detection of j uveniles so that detection in 2005 would equal 0.35, which was the estimate obtained by Martin et al. (2007b). We obtained the following equation: () t = -38.146 + 0.0192 x t We computed annual detec tion of juveniles using the latter equation, except for 2004 for which an estimate of detection was available ( () t in 2004 was 0.16). Martin et al. (2007b) emphasized that this estimate was unusually low, probably because of logistical problems that occurred in 2004. Thus, detection for this analysis varied between 0.16 and 0.35. Martin et al. (2007b) cau tioned about these estimates of detection for juveniles because of scattered data. Therefore, in order to assess the robu stness of our analyses we also repeated our analys es based on a conservative ov erestimate of juveniles, () t = 0.70, which was basically twice the estimate obtained in 2005. When detection of juveniles was considered to be 0.70, we assumed a constant detection for all years. We also conducted our

PAGE 171

171 analyses assuming a constant detection probability of () t = 0.16. It is probably safe to assume that the average detection of juveniles during our study peri od was between 0.16 and 0.70.

PAGE 172

172 LIST OF REFERENCES Allendorf, F. W., and G. Luikart. 2007. C onservation and the genetics of populations. Blackwell Publishing, Oxford. Baker, M. B., and S. Rao. 2004. Incremental costs and benefits shape natal dispersal: theory and example with Hemilepistus reaumuri Ecology 85:1039-1051. Bakker, V. J., and D. H. Van Vuren. 2004. Gap-crossing decision by the red squirrel, a forest-dependent small ma mmal. Conservation Biology 18:689-697. Beissinger, S. R., and J. E. Takekawa. 1983. Habitat use by and disp ersal of Snail Kites in Florida during drought conditio ns. Florida Field Naturalist 11:89-106. Beissinger, S. R. 1986. Dem ography, environmental uncerta inty, and the evolution of mate desertion in the Snail Kite. Ecology 67:1445-1459. Beissinger, S. R. 1988. The Snail Kite. Pages 148-165 in R. S. Palmer, editor. Handbook of north american birds. Yale University Press, New Haven, Connecticut. Beissinger, S. R. 1990. Alternative foods of a diet specialist, the Snail Kite. The Auk 107:327-333. Beissinger, S. R. 1995. Modelling extinction in periodic environments: Everglades water levels and Snail Kite population vi ability. Ecological Applications 5:618-631. Bekkum, M. V., P. M. Sagar, J.-C. Stahl, and G. K. Chambers. 2006. Natal philopatry does not lead to population genetic differen tiation in Buller's al batross. Molecular Ecology 15:73-79. Bennetts, R. E. 1998. The demography and move ments of Snail Kites in Florida. PhD thesis. University of Florida, Gainesville. Bennetts, R. E., W. M. Kitche ns, and D. L. DeAngelis. 1998. Recovery of the Snail Kite in Florida: Beyond a reductionist paradigm. in K. G. Wadworth, editor. Transaction of the 63rd North Ameri can Wildlife and Natural Resources Conference, Wildlife Management Institute. Bennetts, R. E., V. J. Dreitz, W. M. Kitche ns, J. E. Hines, and J. D. Nichols. 1999a. Annual survival of Snail Kites in Flor ida: radio telemetry versus captureresighting data. The Auk 116:435-447. Bennetts, R. E., W. A. Link, J. R. Sauer, a nd P. W. Sykes, Jr. 1999b. Factors influencing counts in an annual survey of Snail Kites in Florida. Auk 116:316-323. Bennetts, R. E., and W. M. Kitchens. 2000. F actors influencing movement probabilities of a nomadic food specialist: proximate fo raging benefits or ultimate gains from exploration? Oikos 91:459-467.

PAGE 173

173 Bennetts, R. E., W. M. Kitche ns, and V. J. Dreitz. 2002. Influence of an extreme highwater event on survival, reproduction, and di stribution of Snail kites in Florida. Wetlands 22:366-373. Blums, P., J. D. Nichols, J. E. Hines, M. S. Lindberg, and A. Mednis. 2003. Estimating natal dispersal movement rates of female European ducks with multistate modelling. Journal of Animal Ecology 72:1027-1042. Bowman, J., N. Cappuccino, and L. Fahrig 2002. Patch size and population density: the effect of immigration be havior. Conservation Ecology 6:(1): 9. [online] URL: http://www.consecol.org/vol6/iss1/art9. Brownie, C., J. E. Hines, J. D. Nichols, K. H. Pollock, and J. B. Hestbeck. 1993. Capturerecaptures studies for multiple strata including non-Markovian transitions. Biometrics 49:1173-1187. Bullock, J. M., R. E. Kenward, and R. S. Hails, editors. 2002. Dispersal Ecology. Blackwell, Oxford. Burnham, K. P., D. R. Anderson, G. C. Wh ite, C. Brownie, and K. H. Pollock. 1987. Design and analysis of methods for fish survival experiments based on releaserecapture. American Fisheries Society Monograph 5:1-437. Burnham, K. P., and D. R. Anderson. 1998. Model Selection and In ference: A practical Information-theoretic Approach. Springer-Verlag, New York. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: A practical information-theoretic approach second edition. Springer-Verlag, New York. Cam, E., D. Oro, R. Pradel, and J. Ji menez. 2004. Assessment of hypotheses about dispersal in a long-lived seabird usin g multistate capture-recapture models. Journal of Animal Ecology 73:723-736. Caswell, H. 2001. Matrix population models: co nstruction, analysis, and interpretation. Sinauer Associates, Sunderland. Casagrandi, R., and M. Gatto. 2002. Habitat de struction, environmental catastrophes, and metapopulation extinction. Pages 127-140 in Theoretical Population Biology. Cassey, P., and G. T. Ussher. 1999. Estim ating abundance of tuatara. Biological Conservation 88:361-366. Caughley, G. 1977. Analysis of vertebrate populations. John Wiley and Sons, New York. Choquet, R., A. M. Reboulet, R. Pradel, O. Gimenez, and J.-D. Lebreton. 2003. User's manual for U-Care, Utilities-CApture-REcapture, V2.02, CEFE/CNRS. Montpellier (ftep://ftp.cefe .cnrs-mop.fr/biom/U-CARE).

PAGE 174

174 Choquet, R., A. M. Reboulet, O. Gimenez, J. -D. Lebreton, and R. Pradel. 2005. User's manual for U-Care, Utilities-CApture-REcapture, V2.22, CEFE/CNRS. Montpellier (ftep://ftp.cefe .cnrs-mop.fr/biom/U-CARE). Clobert, J., A. A. Dhont, E. Danchin, a nd J. D. Nichols, editors. 2001. Population dynamics consequences of dispersal in local populations a nd in metapopulations. Oxford University Press, Oxford. Conn, P. B., L. L. Bailey, and J. R. Saue r. 2004. Indexes as surrogates to abundance for low-abundance species. Pages 59-74 in W. L. Thompson, editor. Sampling rare or elusive species. Island Press, Washington, D.C. Cooch, E. G., and G. C. White. 2005. MARK an introductory guide. Available online from: http://phidot.o rg/software/docs/book/ Daily, G. C., S. Alexander, P. E. Ehrlich, L. Goulder, J. Lubchenco, et al. 1997. Ecosystem services: benefits supplied to human societies by natural ecosystems. Issues in ecology 1. Darby, P. C. 1998. Florida apple snail (Pomacea paludosa) life history in the context of a hydrologically fluctuating environment. Pages 154 in Wildlife Ecology and Conservation. University of Florida, Gainesville. Darby, P. C., L. B. Karunaratne, and R. E. Bennetts. 2005. The influence of hydrology and associated habitat structure on spatia l and temporal patterns of apple snail abundance and recruitment. University of West Florida/ US Geological Survey, Pensacola, FL, USA. Technical report. Davis, J. M., and J. A. Stamps. 2004. The eff ect of natal experience on habitat preference. Trends in Ecology and Evolution 19:411-416. Davis, S., and J. Ogden, editors. 1994a. Evergl ades, The Ecosystem and its restoration. St Lucie Press, Boca Raton. Davis, S., and J. Ogden. 1994b. Toward Ecosystem Restoration. Pages 769-796 in S. Davis and J. Ogden, editors. Everglades The Ecosystem and its restoration. St Lucie Press, Boca Raton. DeAngelis, D. L., and P. S. White. 1994. Ec osystems as products of spatially and temporally varying driving forces, ecological processe s and landscapes: A theoretical perspective. in J. C. Ogden and S. M. Davis, editors. Everglades: The Ecosystems and Its Restoration. St Lucie Press, Delray Beach, Fla. DeLaune, M. G. 2000. XTools ArcView Extension (Version 9/15/03) URL: http://www.odf.state.or.us/divisions/management/State_forests/XTools.asp.

PAGE 175

175 Doherty, P. F., J. D. Nichols, J. Tautin, J. F. Voelzer, G. W. Smith, et al. 2002. Sources of variation in breeding-gr ound fidelity of mallards ( Anas platyrhyncos ). Behavioral Ecology 13:543-550. Dorazio, R. M., and F. A. Johnson. 2003. Ba yesian inference and decision theory-a framework for decision making in natu ral resource management. Ecological Applications 13:556-563. Doremus, H., and J. E. Pagel. 2001. Why listing may be forever: Perspectives on delisting under the U.S. Endangered Species Act. Conservation Biology 15:12581268. Dreitz, V. J., J. D. Nichols, J. E. Hines, R. E. Bennetts, W. M. Kitchens, et al. 2002. The use of resighting data to es timate the rate of population gr owth of the Snail Kite in Florida. Journal of Applied Statistics 29:609-623. Economist. 2005. Water, bird and man-The Everglades. The Economist (U.S. Edition) 377. Everitt, B.S. 2005. An R and S-PLUS companion to multivariate analysis, Springer, London. Fennema, R. J., C. J. Neidraueuer, R. A. J ohnson, T. K. MacVicar, and W. A. Perkins. 1994. A computer model to simulate na tural Everglades hydrology. Pages 249289 in S. Davis and J. Ogden, editors. Everglades: The Ecosystem and its restoration. St. Lucie Press, Delray Beach. Fraser, D. J., and L. Bernatchez. 2001. Adap tive evolutionary conservation: toward a unified concept for defining cons ervation units. Molecular Ecology 10:27412752. Gerber, L. R., and D. P. DeMaster. 1999. A quantitative approach to endangered species act classification of long lived vertebra tes: application to the North Pacific Humbpack Whale. Conservation Biology 13:1203-1214. Gunderson, L. H., and W. F. Loftus. 1994. Vege tation of the Everglades: determinants of community. Pages 769-796 in J. C. Ogden, editor. Everglades, The Ecosystem and its restoration. St Lucie Press, Delray Beach. Haddad, N. M. 1999. Corridor and distan ce effects on interpatch movements: A landscape experiment with butterf lies. Ecological Applications 9:612-622. Hanski, I. 1999. Metapopulation Ecology. Oxford University Press, Oxford. Hanski, I. 2001. Population dynamics consequen ces of dispersal in local populations and in metapopulations. in J. Clobert, A. A. Dhont, E. Danchin, and J. D. Nichols, editors. Dispersal. Oxford University Press, Oxford.

PAGE 176

176 Harris, R. J., and J. M. Reed. 2002. Behavior al barriers to non-mig ratory movements of birds. Annales zoologici Fennici 39:275-290. Hestbeck, J. B., J. D. Nichols, and R. A. Malecki. 1991. Estimates of movement and site fidelity using mark-resight data of wintering Canada Geese. Ecology 72:523-533. Hill, J. K., C. D. Thomas, and O. T. Le wis. 1996. Effects of habitat patch size and isolation on dispersal by Hesperia comma butterflies: implications for metapopulation structure. J ournal of Animal Ecology 65:725-735. Holt, R. D., and D. M. Debinski. 2003. Reflections on Landscapes Experiments and Ecological Theory: Tools for the Study of Habitat Fragmentation. in G. A. Bradshaw and P. A. Marquet, editors. How Landscape Change. Springer-Verlag, Berlin Heidelberg. Holt, R. D., and R. Gomulkiewicz. 2004. Conservation implications of niche conservatism and evolution in heter ogeneous environments. Pages 244-264 in Evolutionary conservation biology. Camb ridge University Press, Cambridge. Holyoak, M. 2000. Habitat patch arrange ment and metapopulation persistence of predators and prey. American Naturalist 156:378-389. Horvitz, C. C., and D. W. Schemske. 1995. Spatiotemporal variation in demographic transitions of a tropical understory herb: projection matrix analysis. Ecological Monogaphs 65:155-192. Howell. 1932. Florida Bird Life. Coward-McMann, New York. Johnson, C. J., K. L. Parker, C. H. D ouglas, and M. P. Gillingham. 2002. Movement parameters of ungulates and scale-specific responses to the environment. Journal of Animal Ecology 71:225-235. Jonze, N., C. Wilcox, and H. P. Possingha m. 2004. Habitat Selection and Population Regulation in Temporally Fluctuating Environments. American Naturalist 164:E104-E114. Kendall, W. L. 2001. Using models to facilitate complex decisions. in T. M. Shenk and A. B. Franklin, editors. Modelling in natural resource management: valid development, interpretation and appli cation. Island Press, Washington D.C. Kitchens, W. M., R. E. Bennetts, and D. L. DeAngelis. 2002. Linkages between the Snail Kite population and the wetland dynamics in a highly fragmented South Florida Hydroscape. Pages 183-203 in K. G. Porter, editor. The Everglades, Florida Bay, and Coral Reefs of the Florida Keys. CRC Press, Boca Raton. Latham, A. D. M., and R. Poulin. 2003. Spatio temporal heterogeneity in recruitment of larval parasites to shore crab intermediate hosts: the influence of shorebird definitive hosts. Canadian Journal of Zoology 81:1282-1291.

PAGE 177

177 Lebreton, J.-D., K. P. Burnham, J. Clobe rt, and D. R. Anderson. 1992. Modelling survival and testing biological hypothes is using marked animals: a unified approach with case studies. Ecological Monogaphs 62:67-118. Lerner, H. R. L., and D. P. Mindell. 2005. Phylogeny of eagles, Old World vultures and other Accipitridae based on nuclear and mitochondrial DNA. Pages 327-346 in Molecular Phylogenetics and Evolution. Lindberg, M. S., J. S. Sedinger, D. V. Derksen, and R. F. Rockwell. 1998. Natal and breeding philopatry in a black brant, Branta bernicla nigricans metapopulation. Ecology 79:1893-1904. Lomolino, M. V. 1990. The target area hypot hesis: the influen ce of island area immigration rates of non volant mammals. Oikos 57:297-300. Martin, J., J. D. Nichols, W. M. Kitchens, and J. E. Hines. 2006. Multiscale patterns of movement in fragmented landscapes a nd consequences on demography of the Snail Kite in Florida. Journal of Animal Ecology 75:527-539. Martin, J., W. M. Kitchens, C. Cattau, A. Bowling, S. Stocco, et al. 2007a. Demography and conservation of the Snail Kite in Fl orida, annual report for US Fish and Wildlife Service. Florida Fish and W ildlife Cooperative Research Unit and University of Florida, Gainesville. Martin, J., W. M. Kitchens, and J. E. Hines. 2007b. Importance of well designed monitoring programs for the conservation of endangered species: Case study of the Snail Kite. Conservation Biology. Martin, J., W. M. Kitchens, and J. E. Hine s. 2007c. Natal location influences movement and survival of a spatially structure populat ion of Snail Kite in Florida. Oecologia Mathworks. 2005. MATLAB: the language of technical computi ng, using MATLAB. Version 7.1. in The Mathworks, Natick, Massachussets, USA. Metcalfe, N. B., and P. Monaghan. 2001. Co mpensation for a bad start: grow now, pay later? Trends in Ecology and Evolution 16:254-260. Mitsch, W. J., and J. G. Gosselink. 2000. We tlands, Third edition. John Wiley and Sons, New York. Mooij, W. M., R. E. Bennetts, W. M. Kitche ns, and D. L. DeAngelis. 2002. Exploring the effect of drought extent a nd interval on the Florida Sn ail Kite: inte rplay between spatial and temporal scales. Ecological Modelling 149:25-39. Mooij, W. M., and D. L. DeAngelis. 2003. Un certainty in spatially explicit animal dispersal models. Pages 794-805 in Ecological Applications.

PAGE 178

178 Mooij, W. M., J. Martin, W. M. Kitchens and D. L. DeAngelis 2007. Exploring the temporal effects of seasonal water ava ilability on the Snail Kite of Florida. in I. Storch, editor. Temporal explicitn ess in landscape ecology: the missing dimension. Springer, New York. Morris, W. F., and D. F. Doak. 2002. Quantitative conservation biology. Theory and practice of population viability analysis Sinauer, Sunderland, Massachusetts. Naeem, S., F. S. Chapin III, R. Costanza, P. E. Ehrlich, F. B. Golley, et al. 1998. Biodiversity and Ecosystem Functioning: Maintaining Natural Life Support Processes. in Issues in ecology. Nichols, J. D. 2001. Using models in the c onduct of science and management of natural resources. Pages 11-34 in T. M. Shenk and A. B. Franklin, editors. Modelling in natural resource management: development, interpretation and application. Island Press, Washington D.C. Nichols, J. D. and B. K. Williams. 2006. M onitoring for conservation. Trends in Ecology and Evolution Niemi, G. J., and M. E. McDonald. 2004. A pplication of ecologica l indicators. Annual Review of Ecology and Systematics 35:89-111. Peery, Z. M., S. R. Beissinger, S. H. Newm an, E. B. Burkett, and T. D. Williams. 2004. Applying the declining population pa radigm: diagnosing causes of poor reproduction in the Marbled Murrelet. Conservation Biology 18:1088-1098. Pettorelli, N., S. Dray, J. M. Gaillard, D. Chessel, P. Duncan, et al. 2003. Spatial variation in springtime food resources infl uences the winter body mass of roe deer fawns. Oecologia 137:363-369. Pollock, K. H., J. D. Nichols, C. Brownie, a nd J. E. Hines. 1990. St atistical inference for capture-recapture experime nts. Wildlife Monographs 107. Porter, J. W., and K. G. Porter, editors. 2002. The Everglades, Florida Bay, and Coral Reefs of the Florida Keys. CRC Press, Boca Raton. Pradel, R., C. M. A. Wintrebert, and O. Gimenez. 2003. A proposal for a Goodness-of-Fit test to the Arnason-Schwarz multisite capture-recapture model. Biometrics 59:4353. Prato, T. 2004. Accounting for uncertainty in making species protection decisions. Pages 806-814 in Conservation Biology. RECOVER. 2005. The RECOVER team's reco mmandations for interim targets for the Comprehensive Everglades Restoration Project. c/o U.S. Army Corps of Engineers and South Florida Water Mana gement District, West Palm Beach.

PAGE 179

179 Reed, J. M., S. C. Mills, J. B. Dunning, E. S. Menges, K. S. McKelvey, et al. 2002. Emerging issues in population viabi lity analysis. Cons ervation Biology 16:7-19. Reeves, G. H., L. E. Benda, K. M. Burnett, P. A. Bisson, and J. R. Sedell. 1995. A disturbance-based ecosystem approach to maintaining and re storing freshwater habitats of ESU of Anadromous Salmoni ds in the Pacific Northwest. Pages 334349 in American Fisheries Society Symposium. Ruckelshaus, M. H., H. Cynthia, and P. Kareiva. 1997. Assessing the data requirements of spatially explicit dispersa l models. Conservation Biology 11:1298-1306. Schjrring, S. 2002. The evolution of informed natal dispersal: inherent versus acquired information. Evolutionary Ecology Research 4:227-238. Schtickzelle, N., and M. Baguette. 2003. Behavioral responses to habitat patch boundaries restrict dispersal and generate emigration-patch area relationships in fragmented landscapes. Journal of Animal Ecology 72:533-545. Serrano, D., and J. L. Tella. 2003. Dispersal within a spatially stru ctured population of lesser kestrels: the role of spatial isola tion and conspecific at traction. Journal of Animal Ecology 72:400-410. Smith, J. N. M., and J. J. Hellmann. 2 002. Population persistence in fragmented landscapes. Trends in Ecology and Evolution 17:397-399. Smith, S. M., D. E. Gawlik, R. Rutchey, G. E. Crozier, and S. Gray. 2003. Assessing drought-related ecological risk in th e Florida Everglades. Journal of Environmental Management 68:355-366. Spendelow, J. A., J. D. Nichols, I. C. T. Nisbet, H. Hays, G. D. Cormons, et al. 1995. Estimating annual survival and movement rates of adults within a metapopulation of Roseate Terns. Ecology 76:2415-2428. Stahl, J. T., M. K. Oli. 2006. Relative im portance of avian life history variables to population growth rate. Ecological Modelling 198:28-39. Stamps, J. A. 2001. Habitat selection by disp ersers: integrating proximate and ultimate approaches. Pages 230-242 in J. Clobert, E. Danchin, A. A. Dhont, and J. D. Nichols, editors. Dispersal. Oxford University Press, Oxford. Sykes, P. W., J. A. Rodgers, and R. E. Bennetts. 1995. Snail Kite. Pages 1-31 in A. Pool and F. Gill, editors. The birds of North America., The Academy of Natural Sciences, Philadelphia, a nd The American Ornithologists' Union, Washington D.C. Takekawa, J. E., and S. R. Beissinger. 1989. Cyclic drought, dispersal and conservation of the Snail Kite in Florida: lessons in critical habitat. Conservation Biology 3:302-311.

PAGE 180

180 Taylor, B. L., P. R. Wade, U. Ramakrishna n, M. Gilpin, and H. R. Akcakaya. 2002. Incorporating uncertainty in Population viability analyses for the purpose of classifying species by risk. Pages 147-168 in S. R. Beissinger and D. R. McCullough, editors. Population viability analysis. The University of Chicago Press, Chicago, Illinois. USFWS. 1999. South Florida multi-species recovery plan. http://ecos.fws.gov/docs/r ecovery_plans/1999/990518.pdf Waples, R. S. 2002. Definition and estima tion of effective population size in the conservation of endangered species. Pages 147-168 in S. R. Beissinger and D. R. McCullough, editors. Population viability analysis. The university of chicago press, Chicago. Welch, Z. C. 2004. Littoral vegetation of Lake Tohopekaliga: Community description prior to a large-scale fisheries habitat-e nhancement project. MS thesis. University of Florida, Gainesville. White, G. C., and K. P. Burnham. 1999. Pr ogram MARK: Survival rate estimation from both live and dead en counters. Bird Study 46 (Suppl.):S120-S139. White, P. S., and S. T. A. Pickett. 1985. Natural disturbance and patch dynamics: and introduction. in P. S. White, editor. The ecology of natural disturbance and patch dynamics. Academic Press, Orlando, Florida. Wiens, J. A. 1976. Population responses to patchy environments. Annual Review of Ecology and Systematics 7:81-120. Wilcove, D. S., and L. Y. Chen. 1998. Ma nagement costs of endangered species. Conservation Biology 12:1405-1407. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations, Academic Press edition, San Diego, California. Yoccoz, N. G., J. D. Nichols, and T. Boulin ier. 2001. Monitoring of biological diversity in space and time. Trends in Ecology and Evolution 16:446-453.

PAGE 181

BIOGRAPHICAL SKETCH Contents of this page removed per demand of copyright holder, Dr. Julien Martin. No other changes in document. 181


Permanent Link: http://ufdc.ufl.edu/UFE0019653/00001

Material Information

Title: Population Ecology and Conservation of the Snail Kite
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0019653:00001

Permanent Link: http://ufdc.ufl.edu/UFE0019653/00001

Material Information

Title: Population Ecology and Conservation of the Snail Kite
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0019653:00001


This item has the following downloads:


Full Text





POPULATION ECOLOGY AND CONSERVATION OF THE SNAIL KITE


By

JULIEN MARTIN

















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2007




























2007 Julien Martin


































A Mercedes









ACKNOWLEDGMENTS

I am greatly indebted to my committee members: Wiley Kitchens, Jim Nichols, Don

DeAngelis, Peter Frederick, and Ben Bolker. Wiley Kitchens has been an exceptional supervisor

and mentor. His unwavering support and countless hours of expertise helped me grow as an

ecologist and contributed to the success of my research. I also owe special thanks to Jim Nichols

who had a major influence on the project. He helped me think more critically about my research

and assisted with many of the statistical analyses. Don DeAngelis provided excellent input and

valuable ideas. I am thankful to Peter Frederick for many useful discussions about water

management and bird populations in the Everglades. Ben Bolker provided helpful statistical

advice.

In addition to the input from my committee members, I also greatly benefited from the

assistance of other scientists: Jim Hines, Wolf Mooij, Arpat Ozgul, MadanOli, Jeff Hostetler,

Hardin Waddle, Vicky Dreitz, Rob Bennetts, Gary White, Roger Pradel, Christopher Cattau,

Andrea Bowling, Zack Welch, Eric Powers, Christa Zweig, Sarah Haas, Steve Beissinger, Bret

Sandercock, Therese Donovan, Nicholas Schtickzelle, Tom Contreras, Alejandro Paredes,

Thomas Comullier, Maynard Hiss, Brad Stith, Phil Darby, Steve Daniels and Fred Johnson.

Statistical analyses related to capture-mark-recapture would not have been possible without the

help of Jim Hines.

This dissertation could never have been completed without the great vision of the

individuals who initiated and designed the Snail Kite monitoring program: Rob Bennetts, Wiley

Kitchens, Vicky Dreitz, and Jim Nichols. Additionally, in the early stage of my research, Rob

Bennetts spent many hours helping me become familiar with kites and the huge Snail Kite

database. During my first field season, he replaced me in the field for several months which









allowed me to attend an important course on capture-mark-recapture techniques offered by Gary

White and David Anderson.

I would like to thank all the field biologists involved with data collection during the last 5

years: Christopher Cattau, Andrea Bowling, Samantha Musgrave, Sara Stocco, Brian Reichert,

Melinda Conners, Danny Huser, Derek Piotrowicz, Christina Rich, Michaela Speirs, Jamie

Dubeirstein, Janell Brush, Zack Welch, Jeff Kingscott, Freddy Martin, and Paul Pouzergues.

I appreciate the help and cooperation of the following biologists and managers who

invested time and effort to promoting Snail Kite recovery: Tylan Dean, Debbie Pierce, Steve

Miller, Joe Benedict, Susan Sylvester and Jim Rodgers.

Financial support was provided by the U.S. Army Corps of Engineers, the U.S. Fish and

Wildlife Service, the St. Johns River Water Management District, the Florida Fish and Wildlife

Conservation Commission, and the U.S. Geological Survey. Employees of the Florida

Cooperative Fish and Wildlife Research Unit were especially helpful with the administration of

this study, particularly Donna Roberts, Franklin Percival, Barbara Fesler and Debra Hughes.

Lois Wilcox, Anne Taylor, and Ellen Main helped with editing.

Finally, I would like to thank members of my family and friends who have been a great

support throughout these years: Mercedes, Christophe, Jose, the Martin-Pouzergues family, the

McSweeneys, the Glogowskis, the Kitchens, Jacotte, Emilie, Bruno, and Sarah.









TABLE OF CONTENTS

page

A CK N O W LED G M EN T S ................................................................. ........... ............. .....

L IS T O F T A B L E S ......................................................................................... 1 1

LIST OF FIGURES .................................. .. .... ..... ................. 12

ABSTRAC T ................................................... ............... 13

CHAPTER

1 INTRODUCTION ............... ................. ........... ......................... .... 15

W hy Study the Snail Kite? ........................................ .......... .............15
B background Inform ation on Snail K ites..................................................................... ...... 16
Objectives and Outline ..................................... .. .... ...... .. ............18

2 IMPORTANCE OF WELL-DESIGNED MONITORING PROGRAMS FOR THE
CONSERVATION OF ENDANGERED SPECIES: CASE STUDY OF THE SNAIL
K I T E ....................................................................................................................................... 1 9

In tro d u ctio n ................... ...................1...................9..........
M eth o d s ...........................................................................2 3
S tu d y A re a ................................................................................................................. 2 3
Sampling M methods ................................. ................................ ........ 23
M parking protocol ..................................................................... ......... 23
P population survey protocol................................................................................. 23
A analysis ....................................................... .................. ................. 24
Superpopulation size estimates of adults ........................................ .....24
Estimation of population growth rate ..................................... 26
Estimation of the number of young produced ................................ ............. 27
Detection probabilities for number of young produced every year .........................27
C ount data ....................................... ............................. 28
Average number of kites and growth rate based on count data .............................28
Detection probabilities for FC and M C ................................ ...... ................... 28
Estimates of Precision and Magnitude of the Difference between Estimates ..............29
Results ........... .......... ....... ................. ........ ............................ 30
Population Size and Average Population Growth Rate ..............................................30
Average Number of Kites Before and After Decline Based on the Superpopulation
A p p ro ach ................... ..... ........ ....... ... .............. ......................... ... 3
Average Number of Kites Before and After Decline Based on Count Data ...................31
Average Number of Kites and Growth Rate Based on Count Data .............................31
N um b er of Y young ............................................................... ....................... 32
Detection Probabilities for FC and MC ...................................... ............... 32
Discussion ..................... .............................. ..................32


6









Population D decline ................. .. .. ...... .... .... ............. .. .... ............... 32
Problems Associated with Counts and Implications for Recovery Plans......................34
Importance of Monitoring to Diagnose Causes of Decline............................................37
C o n c lu sio n ................... .......................................................... ................ 3 8

3 MULTISCALE PATTERNS OF MOVEMENT IN FRAGMENTED LANDSCAPES
AND CONSEQUENCES ON DEMOGRAPHY OF THE SNAIL KITE IN FLORIDA......44

In tro du ctio n ..................................44...........................
H ypotheses and Predictions.................................................. ............................... 47
M eth o d s ...........................................................................4 9
Study Area ............................................ .....................49
Criteria for Determining the Regional Impact of the 2001 Drought.............................49
Statistical Models to Estimate Movement and Survival.............................................50
Field Methods for the Study of Movement on a Monthly Scale............................ .. 51
Statistical Methods to Estimate Movement on a Monthly Scale Using
R adiotelem etry ......................................................................... ... ..... 5 1
Estimating monthly movement among regions......................................51
Estimating monthly movement within regions using radio-telemetry ...................52
Field Methods for the Study of Movement and Survival on an Annual Scale ..............53
Statistical Methods to Estimate Annual Movement and Survival Using Banding
D ata ................. ... ......................................................... .. ........... .....53
E stim ating survival ......................... ..... ...................... .... ...... ... ............. 53
Estimating annual movement probabilities among regions using banding data......54
G oodness of fit .................................................................. .................... ...... ... 55
M odel Selection P rocedure........................................... .............................................55
Effect of Patch Size and Distance on Movement...............................................56
E ffe c t S iz e ................................................................................................................. 5 6
Estim ates of Precision ................... ............... .... ..... ............. 56
Results .................................... ........ ........................... 57
M monthly M movement Probabilities Among Regions.................................... ................ 57
Effects of patch size and distance .................................... .......... ........ ...............57
M monthly M movement Probabilities W within Regions............................... ............... 58
M movement within the Everglades region ...................................... ............... 58
M ovem ent w within the K region ........................................ .......................... 59
Comparison Among and W within Regions.................................... ........................ 59
Interannual Survival Estimates............................ ....... ...............60
Inter-Annual Movement Among Regions and Drought Effect on Movement ..............61
D discussion .................... .. ...... .. .. ............ .. ........... .......... 62
Monthly Movement Among Contiguous and Isolated Wetlands.............................62
Patch Size and Distance Between Patches as Factors Driving Movement ...................63
Inter-annual Pattern of M ovem ent..................................... ..................................... 64
Regional Survival and Resistance of the Population to Natural Disturbance .................65
Conclusions and Conservation Im plications ........................................ ....... ............... 68

4 NATAL LOCATION INFLUENCES MOVEMENT AND SURVIVAL OF THE
SN A IL K ITE ............................................. 75


7









Introduction ................................... ...........................................75
H ypotheses and Predictions.................................................. ............................... 77
Study Area ........................................ 78
M material A nd M methods ......... ... ....... ............................................................. 79
Field M methods ................. ............................................................................79
C apture-m ark-recaptu re ..................................................................................... 79
D ata A n a ly sis ............................................................................................................. 7 9
M ultistate m odeling.............................................. ................79
M o v e m e n t........................................................................................................... 8 0
Survival ........................ .............. .... .......................82
Model Selection, Goodness of Fit and Program Used ..................................... 85
Effect Size ...................... .................. ......... ... ..... 86
Notes Concerning Regional Specific Survival ..................................... ............ ...86
W etland C conditions ...................................... .......... .............................. 87
R results ....................................... .... ................. 88
G O F T ests ................................................................................... ... 88
M ov em ent ......................... ........... ............................................................. ............... 8 8
Comparison of Natal Philopatry and Philopatry to Non-Natal Site ................................89
S u rv iv a l ................... ...................8...................9..........
A d u lts ...............................................................................8 9
Ju v e n ile s ...................................................................................................... 9 1
Detection Probabilities .................................. .. .. .. ...... .. ............91
D discussion .................................... .... .......................................... 9 1
Effect of Natal Region on M movement ........................................................ ............... 91
Influence of N atal Region on Survival ................................ ......................... ........ 93
Conclusions and conservation implications......................................... 95

5 EXPLORING THE EFFECTS OF NATURAL DISTURBANCES AND HABITAT
DEGRADATION ON THE VIABILITY OF THE SNAIL KITE ............. ... ...............102

In tro d u ctio n ................... ...................1.............................2
O b j e c tiv e s ................................................................................................................ 1 0 5
M eth o d s ...........................................................................10 6
S tu dy A rea ................................................................................ 10 6
L ife C y c le ........................................................................................................1 0 6
H ydrological C conditions ....................................................................107
Data Source and Estimates of Vital Rates ...........................................108
Survival rates............................................. 108
F ecu n dity rates ........................................................................................1 1
D election of juveniles ............................................. ............... ................ 113
M atrix A analyses ..................... .... ......................... ........ ... ..........................113
1A, damping ratio, sensitivity and elasticity analysis (Objectives 1 and 2) ..........13
Before versus after effect on Xi and Life Table Response Experiment
(O objectives 3 and 4) .................. ............ .. ....... .... ...... .... ...... .. ..... 114
Stochastic population growth rate (objectives 1, 4, 5 and 6) ................................. 115
Viability of Snail Kites under current conditions (Objective 1) ............................116
Evaluation of Hypothesis 1: Reduction of Xs after 1998 (Objective 5).............. 116


8









Evaluation of Hypothesis 2: Increase in drying event frequency reduces Xs
(Objective 6) .................................... ............................... ........117
Evaluation of Hypothesis 3: Increase in frequency of drying events and
"before versus after" effect (Objective 7)................................ .................... 117
Probability of Quasi-Extinction (Objective 1) ............. ...................... ....................118
V alidation ..................................... .................. ............... ........... 119
R results ................... ............................................................................. 119
Su rviv al estim ates .................. ................................................................. 119
Probability of Quasi-extinction (Objective 1) ...................... ................ .................. 122
Xi, sensitivity and Elasticity analysis (Objectives 1 and 2)............... ............ .....122
Life Table Response Experiment (Objective 3) ...................................................122
Stochastic population growth rate s ......................... .. ..... ...............123
Viability of Snail Kites under current conditions (Objective 1) ............................123
Evaluation of Hypothesis 1: Reduction of Xs after 1998 (Objective 5)..................123
Evaluation of Hypothesis 2: Increase in drying event frequency reduces Xs
(O bj ectiv e 6) ................................................................................................ 124
Evaluation of Hypothesis 3: Increase in frequency of drying events and "before
versus after" effect (O objective 7)...................................... ....... ............... 125
D iscu ssion ...................................... ......................................................... 12 5
Snail Kite Viability and Key Vital Rates ........................... ............... 125
Hypothesis 1: "Before versus After 1998" Effect (Objective 4))...............................127
Hypothesis 2: Increase in Frequency of Moderate Drying Events (Objective 6)..........128
Hypothesis 3: Increase in Moderate Drying Events and "Before versus After 1998"
effect (O objective 7) .......................................... .............. .. ...... .... 129
Lim its of the M models ................................................... ........ .. ........... 129
M anagem ent Im plications ........................................................................... ................... 132

6 CONCLUSION................ ....... ........ ............. ............... .. 145

Synthesis of Research Findings .......................................................................... 145
M monitoring of the Snail K ite...................... ........... ...... .......................................145
New information on movement and demography of the Snail Kite.............................147
Current Status of the Snail Kite and Management Implications ................................148
P e rsp e c tiv e s ........................................... ......... ..... ..................... ........ .................... 1 5 0
Rescuing the Snail Kite Population from Extinctions Risks Associated with Small
P o p u latio n s .............................................. .. ................ ................ 1 5 0
H habitat M anagem ent M odels ............................................... ............................. 152
E v e rk ite ............................................................................................................ 1 5 2
A daptive m anagem ent...................... .... ...................... ................. ............... 152

APPENDIX

A SURVEY-SPECIFIC PARAMETER ESTIMATES USED TO COMPUTE
ESTIMATES OF SUPERPOPULATION SIZE ...........................................................155

B CONFIDENCE INTERVALS, MODEL SELECTION TABLES AND MOVEMENT
E S T IM A T E S ............................................................................ 15 6









C SELECTION OF MODELS USED TO ASSESS THE EFFECT OF NATAL
LOCATION ON MOVEMENT AND SURVIVAL ...................................... ...............163

D C L U STE R IN G A N A L Y SIS ...................................................................... ..................... 165

E E ST IM A T IO N O F Q (T )............................................................................. .................... 169

F D E TE C TIO N O F JU V EN ILE S......................................... .............................................170

L IST O F R E F E R E N C E S ..................................................................................... ..................172

B IO G R A PH IC A L SK E T C H ......................................................................... .. ...................... 181










































10









LIST OF TABLES


Table page

2-1 Estimates of Snail Kites annual population growth rates; and 3-year running average
of grow th rates. ..................................................................................................................39

2-2 Estimates of detection probability of Snail Kites ................................... .................40

3-1 Multistate models of monthly movement probabilities of Snail Kites among five
major regions based on radio-telemetry data. ......................................... ...............70

3-2 Multistate models of monthly movement probabilities of Snail Kites among wetlands
in the E and K region based on radio-telemetry data ......................................................71

3-3 Multistate models of annual apparent survival and movement probabilities ..................72

4-1 Multistate models of apparent survival and annual transition probabilities in relation
to the place of birth .................................... .................................. ........... 97

5-1 Cormack-Jolly-Seber models of apparent survival............................................136

A-i Survey-specific parameter estimates used to compute estimates of superpopulation
size for Snail Kites between 2001 and 2005 .............................................. ............... 155

B-l Multistate models of monthly movement probabilities of Snail Kites among the 5
major regional patches in Florida based on radio-telemetry data (models with delta
A IC > 15)........................................................ ............................... 157

B-2 Multistate models of monthly movement probabilities among wetlands in the E and
K regions based on radio-telemetry data (models with delta AIC > 9). ........................158

B-3 Multistate models of annual survival and annual movement probabilities (models
w ith delta A IC > 11) .............. ........................................... .................... 160

B-4 Annual movement estimates between the 4 major regions used by the Snail Kite (E,
O, K and J) during norm al and drought years............................................................... 162

C-l Multistate models of apparent survival and annual transition probabilities in relation
to the place of birth (models with delta AIC > 12) .......................................................163

D-l Five hydrological variables used to conduct the hierarchical clustering analysis .........167









LIST OF FIGURES


Figure page

2-1 Map of the wetlands that were sampled to obtain both counts and capture-resighting
information of Snail Kite for the estimation of population size .....................................41

2-2 Comparison of the estimates of population size of Snail Kites (using the
superpopulation approach) with annual counts....................................... ............... 42

2-3 Number of young (i.e., nestlings close to fledging) Snail Kites marked every year
from 19 92 to 2 0 0 5 ..................................................... ................ 4 3

3-1 Major wetlands used by the Snail Kite in Florida...................... .. ....................73

3-2 Apparent survival between 1992 and 2003 of adult and juvenile Snail Kites...................74

4-1 Major wetland complexes (i.e., regions) used by the Snail Kite in Florida..................98

4-2 Movement probabilities between natal region and post-dispersal region........................99

4-3 Model averaged estimates of natal philopatry (NPHL) and philopatry to non-natal
region (PHLNN) .................. ......... ............ .......... ............ 100

4-4 Model averaged estimates of "region specific" and "natal region specific" survival of
Snail K ites in four region s ....................................................................... ..................10 1

5-1 Model averaged estimates of adult and juvenile survival between 1992 and 2005.........138

5-2 Estim ates of probability of quasi-extinction................................................................. 139

5-3 Sensitivity (a) and elasticity (b) of Xi to changes in age-specific vital rates .................140

5-4 The difference in age-specific vital rates between the matrix BEF and matrix AFT......141

5-5 Estimates of stochastic population growth rates (s) for environmental conditions
before and after 1998 ................................... .. ....... .. ..............142

5-6 Estimates of stochastic population growth rates (ks) assuming a low frequency
(LFMD) and a high frequency of moderate drying events (HFMD).............................143

5-7 Estimates of stochastic population growth rates (s) for environmental conditions
before and after 1998 ................................... .. ....... .. ..............144

D-1. Agglomerative hierarchical analysis which categorized years from 1992 to 2006 into
wet years (blue), moderately dry years (yellow) and drought years (red).....................168









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

POPULATION ECOLOGY AND CONSERVATION OF THE SNAIL KITE

By

Julien Martin

May 2007

Chair: Wiley M. Kitchens
Major: Wildlife Ecology and Conservation

My research was articulated around three primary goals: (1) determine the current status of

the Snail Kite (Rostrhamus sociabilis) population in Florida; (2) provide information about

population ecology relevant to conservation of the Snail Kite; (3) make recommendations that

will help Snail Kite recovery.

I found that the Snail Kite population declined dramatically in recent years. Estimates of

stochastic population growth rate and probabilities of quasi extinction suggested that the Snail

Kite population in Florida was at high risk of extinction. The sharp decline observed after 2001

was mostly associated with a multiregional drought that occurred in 2001 and affected

movement, survival and reproduction. The occurrence of this disturbance allowed us to evaluate

hypotheses related to the effect of droughts on demography and movement of Snail Kites. Only a

small proportion of kites escaped a regional drought by moving to refugia (wetlands less affected

by drought). Many individuals died after the drought. During drought, adult survival dropped by

16%, while juvenile survival dropped by 86% (possibly because juveniles were less likely to

reach refugia). Although kites exhibit extensive exploratory behavior, particularly among

contiguous wetlands, they also show high levels of annual site tenacity during the breeding

season, especially to their place of birth. Fidelity to breeding and natal sites has been given









relatively little attention in the past. However, I found that fidelity to the natal region could have

significant effects on movement patterns and survival, and could influence the whole dynamics

of the kite population.

Although the 2001 drought had a considerable effect on survival, reproduction, and

abundance, our results suggest that the lack of recovery after 2002 was predominantly caused by

lack of recruitment. We found evidence that both habitat conversion (caused by prolonged

hydroperiod and increased water depth during the Fall), and the increase in frequency of drying

events (during the Spring and Summer), especially in Water Conservation Area 3A (WCA3A),

could be responsible for the observed reduction in population growth rate. Finally, I present a set

of management recommendations to promote Snail Kite recovery.









CHAPTER 1
INTRODUCTION

Why Study the Snail Kite?

Natural ecosystems provide societies with goods and services worth trillions of dollars

annually, and also perform "life-support services" essential to the persistence of humanity (Daily

et al. 1997). These services include purification of air and water, detoxification, decomposition

of wastes, and regulation of climates, to list a few (Daily et al. 1997). Yet, most natural

ecosystems suffer from escalating destruction caused by human activities. The multiplication of

ecosystem restoration efforts around the world reveals some level of recognition of the problem.

The Everglades ecosystems in south Florida are currently the site of one of the largest and

most ambitious ecosystem restoration projects ever undertaken (Mitsch and Gosselink 2000). A

major multi-billion dollar restoration project of the Everglades (the Comprehensive Everglades

Restoration Plan, CERP) is being implemented by the U.S. Army Corps of Engineers, the U.S.

Fish and Wildlife Service, and the state of Florida to restore this ecosystem. Restoration of the

Everglades is gathering enormous attention both nationally and internationally (The Economist

2005). If this project succeeds, it is likely to become a model for other large ecosystem

restoration projects throughout the world (The Economist 2005). The primary objective of this

restoration project is to improve the quality of native habitats and increase diversity and

abundance of native plants and animals (RECOVER 2005). The principal idea of the restoration

is to act on two primary stressors of the system: hydrology and water quality (RECOVER 2005).

Conservation biologists and managers have selected a number of indicator species as measures

of success of the restoration (RECOVER 2005; Niemi and McDonald 2004). The Snail Kite

(Rostrhamus sociabilisplumbeus) is one of the indicator species that was selected by the CERP

work team (RECOVER 2005).









There are at least six justifications to list the Snail Kite as an important indicator of

Everglades restoration. First, the Snail Kite is a wetland dependent species. Kites forage and nest

preferentially in habitats dominated by plant communities that CERP strives to restore

(RECOVER 2005). Second, Snail Kites respond numerically (e.g., change in mortality rates) and

behaviorally (e.g., change in movement rates) to changes in hydrological conditions (the primary

stressor modified by CERP). The response of kites to changes in hydrological conditions is

measurable (e.g., estimates of reproductive parameters, movement, etc.). Third, the Snail Kite

population experienced some dramatic decline after the drainage and fragmentation of the

Everglades, to the point that it was listed as federally endangered in 1967. The recovery of

endangered species is also one of the tasks of CERP (RECOVER 2005). Fourth, because of the

charismatic nature of this species, the Snail Kite is an indicator that can ease communication

with the public. Fifth, the geographic range of the Snail Kite in Florida encompasses most of the

wetlands that CERP will attempt to restore (RECOVER 2005). Sixth, the monitoring of the Snail

Kite is one of the few programs in Florida that is long term (> 15 years) and accounts for

detectability and spatial variation (Bennetts et al. 1999b; Yoccoz et al. 2001).

Background Information on Snail Kites

The Snail Kite is a medium size raptor which is restricted to the American continent and

Cuba. The Snail Kite belongs to the Order of the Falconiformes, the Family of the Accipitridae,

the Subfamily of the Buteoninae and the Genus Rostrhamus (Lerner and Mindell 2005). Three

subspecies have been recognized: R. s. plumbeus is found in Florida and Cuba, whereas R. s.

sociabilis, and R. s. major range from Central to South America. Beissinger (1998) has

questioned this classification, which is based on morphometric measurements.

The Snail Kite in Florida is restricted to the remaining wetlands that used to constitute the

historic Kissimmee-Okeechobee-Everglades watershed. The Snail Kite is a wetland dependent









species that feeds almost exclusively on freshwater Apple Snails Pomaceapaludosa (Beissinger

1988). Kites are sensitive to change in hydrological conditions, partly because snail availability

is tightly linked to hydrology (Beissinger 1995). The occurrence of droughts in particular reduces

snail availability drastically, and therefore affects kites movement and demography (i.e., during

drought kites must move or die) (Bennetts and Kitchens 2000). Conversely, aside from some

possible short term benefits for juvenile survival (Bennetts et al. 2002), prolonged hydroperiod,

flooding events or drought suppression may, in the longer term, degrade the vegetation

communities that support both kite foraging and nesting habitats (Kitchens et al. 2002), making

the management of kite habitat a complex endeavor.

Since 1930, the network of wetlands occupied by kites has changed dramatically. It has

been reduced approximately by half of its original size, and has been severely fragmented

(Kitchens et al. 2002). In 1967 the Snail Kite from Florida and Cuba was first listed as

endangered pursuant to the Endangered Species Conservation Act (USFWS 1999).

Since its listing as an endangered species the Snail Kite population in Florida has been

monitored using quasi-systematic annual count surveys (Bennetts and Kitchens 1997). In the

early seventies counts were less than 200, but increased approximately to 1000 birds in 1995.

However, Bennetts et al. (1999b) criticized that monitoring technique, arguing that these counts

were of limited value, if not misleading, because detection probabilities were not accounted for.

Since 1992 a capture-mark-recapture study was initiated to provide robust estimates of vital rates

(i.e., survival, reproduction, movement) and population size. This protocol was coupled to an

extensive radiotelemetry protocol between 1992 and 1995 to provide more precise estimates of

movement and survival (Bennetts et al. 1999a).









Dreitz et al. (2002) provided the first estimates of population size that accounted for

detection, and their estimates (from 1997 to 2000) were four to five times greater than the

recovery target set by the USFWS in 1999. Estimates of Dreitz et al. (2002) during the study

period also indicated a fairly stable population. Under the objectives set by the recovery plan of

1999, these figures were encouraging.

Objectives and Outline

My research was articulated around three primary goals:(1) determine the current status of

the Snail Kite population in Florida; (2) provide information about population ecology relevant

to conservation of the Snail Kite; (3) make clear recommendations that will help Snail Kite

recovery.

* In Chapter 2, I provided new information related to the status of the Snail Kite in Florida.
Based on a recently proposed estimator of abundance (i.e., the superpopulation approach),
I presented estimates of population size and population growth rates for the last 9 years. In
Chapter 2 I also emphasized the importance of accounting for major sources of variation
when estimating demographic parameters.

* In Chapter 3, I examined how Snail Kites perceive and move through the landscapes of
Florida. I also examined the link between movement and survival, and investigated the
relevance of fragmentation and habitat destruction to Snail Kite conservation.

* In Chapter 4, I focused on some important behavioral components that determine
movement and survival of kites. In particular, I looked at how fidelity to the natal site
influenced movement and survival. At the end of the chapter, I emphasized the importance
of considering fidelity to the natal site for management.

* In Chapter 5, I used matrix population models to estimate projected population growth
rates and probability of quasi-extinction. I also evaluated competing hypotheses explaining
changes in population growth rates. At the end of the chapter, I provided a set of
recommendations for management of Snail Kite habitat.

* In the final chapter (Chapter 6), I synthesized the information presented in chapters 1
through 5, and provided some perspectives for future work on the Snail Kite.









CHAPTER 2
IMPORTANCE OF WELL-DESIGNED MONITORING PROGRAMS FOR THE
CONSERVATION OF ENDANGERED SPECIES: CASE STUDY OF THE SNAIL KITE

Introduction

Monitoring natural populations is often a necessary step to establish the conservation status

of species and to help improve management decisions (Yoccoz et al. 2001). However, many

monitoring programs do not effectively address two important components of variation in

monitoring data: spatial variation and detectability, which ultimately may limit the utility of

monitoring in identifying declines and improving management (Yoccoz et al. 2001).

Detectability refers to the probability that an animal will be detected if it is present in the

sampled area (Williams et al. 2002). Many sources of variation may affect detectability (e.g.,

observer effect, environmental conditions), and monitoring data that do not take detectability into

account will typically lead to biased estimates (Williams et al. 2002). Spatial variation is another

source of variability of monitoring data. It results from the inability to sample the entire area of

interest (i.e., inference is drawn from selected spatial units that are only a fraction of the area of

interest; this is a problem when the areas sampled are not representative of the entire area)

(Williams et al. 2002).

The principle that monitoring programs should take detectability and spatial variation into

consideration is gaining some support among wildlife biologists (see Williams et al. 2002).

However, analyses using uncorrected counts continue to be published in major journals

(reviewed in Rosenstock et al. 2002; Conn et al. 2004). The continued controversy around the

value of uncorrected count-based indexes results partly from the fact that monitoring programs

that estimate detectability are often more labor intensive (Conn et al. 2004). Some authors have

argued that when the focus is on population change rather than population size, uncorrected

count-based indexes may be sufficient, but for this latter statement to be correct, detectability









should remain constant over time, which is rarely the case when monitoring mobile organisms

(Conn et al. 2004). The principle that it is crucial to estimate detection and to account for spatial

variation when monitoring animal populations appears to take even longer to be accepted by

some managers. This is problematic because recovery plans for many endangered species are

still based on monitoring programs that ignore these primary sources of variations (e.g., Cape

Sable Seaside Sparrow Ammodramus maritimus mirabilis, Wood Stork Mycteria americana, see

USFWS (1999)).

We used the monitoring of the Snail Kite (Rostrhamus sociabilis plumbeus) in Florida to

illustrate the importance of considering detectability and spatial variation. The Snail Kite feeds

almost exclusively on freshwater snails and, thus, is considered a wetland-dependent species

(Beissinger 1988). In the United States the Snail Kite is restricted to the remaining wetlands of

central and south Florida (Dreitz et al. 2002). Because the availability of snails to kites is

strongly dependent on hydrological conditions, variations in water levels are likely to influence

kite behavior and demography (Beissinger and Takekawa 1983). Droughts affect kite behavior

and demography by reducing snail availability to kites (Beissinger and Takekawa 1983).

The drainage of the Everglades that began in the early 1930s and was followed by wetland

destruction led to the collapse of the kites in Florida (USFWS 1999). In 1967 the Snail Kite was

first listed as endangered pursuant to the U.S. Endangered Species Conservation Act (ESA)

(USFWS 1999). Three primary quantitative recovery criteria were set by the U.S. Fish and

Wildlife Service (USFWS) in 1999 on which to base reclassification of the Snail Kite from

endangered to threatened (USFWS 1999): (1) "the 10-year average for the total population size is

estimated as > 650, with a coefficient of variation (CV) less than 20% for the pooled data"

(USFWS 1999); (2) "no annual population estimate is less than 500"; and (3) "the rate of









increase to be estimated annually or biannually and over the 10 year period will be greater than

or equal to 1.0, sustained as a 3-year running average" (USFWS 1999). These criteria, however,

were set in reference to data obtained from uncorrected counts (USFWS 1999). Since 1965

several agencies have been conducting kite surveys throughout the designated critical kite

habitats (reviewed in Bennetts et al. 1999b). One major weakness of most count surveys is that

detectability is not considered (Bennetts et al. 1999b). Hereafter we used the term "counts" to

refer to uncorrected counts, which basically correspond to the number of animals counted during

a survey. These counts represent an unknown fraction of the target population (i.e., detection

probability is not taken into account, see Williams et al. 2002). In contrast, the terms "estimate of

population size" and "estimate of superpopulation size" correspond to population parameters of

interest that take detectability into consideration.

Dreitz et al. (2002) provided the first estimates of population size that accounted for

detection, and their estimates (from 1997 to 2000) were four to five times greater than the target

set by the USFWS in 1999. Estimates of Dreitz et al. (2002) during the study period also

indicated a fairly stable population. Under the objectives set by the recovery plan of 1999, these

figures were encouraging. However, estimates presented in this study indicate that criteria set by

the recovery plan in 1999 need revision and that count surveys of populations that occupy large

landscapes may be dangerously misleading.

Using a recent estimator of superpopulation size, we examined the implications of

carefully considering detection probabilities and spatial variation when making inference about

changes in population size and number of young produced. In this study, the superpopulation

consisted of all kites that had a non-zero probability of being detected over the course of the

sampling year (Dreitz et al. 2002). The superpopulation approach employed in our study and in









Dreitz et al. (2002) was based on capture-mark-resighting analyses. This approach allowed for

the estimation of superpopulation size that takes into account detectability and spatial variation

(Williams et al. 2002). This approach makes use of capture-mark-resight models such as

Cormack-Jolly-Seber models (CJS) (Dreitz et al. 2002).

There are six primary steps used in this study. 1) We estimated superpopulation size for

young and adults separately. 2) Based on these estimates, we examined changes in abundances

by estimating population growth rates, and we looked for population decline. 3) We then

compared count data with superpopulation estimates. Since detection probabilities for counts are

typically < 1.0, counts will often underestimate the "true" population size (Williams et al. 2002).

4) We computed one type of detection probability (denoted P) as the ratio of the number of kites

counted over the estimated superpopulation size, and examined how detection varied over time

for two types of count surveys. If this detection estimate varied substantially over time, then

count-based indexes could be very misleading. For instance, if detection increases over time

(e.g., because of additional field personnel), population growth rates derived from counts may

suggest that the population is growing, or remains stable, while in fact, the "true" population is

decreasing. Note that detection probability based on the ratio statistic (P) is different from

detection probability (denotedp) directly estimated using capture-mark-resight models such as

the CJS (Williams et al. 2002). 5) We examined the recovery criteria that the USFWS set for

kites to determine whether these criteria were met or were close to being met based on two types

of monitoring data, one that considered detection and spatial variation (superpopulation

approach), and another that did not (i.e., count data). 6) Finally, we examined the implications of

addressing detectability and spatial sampling for the conservation of kites and other endangered

species.









Methods


Study Area

The Snail Kite population in Florida has been described as geographically isolated (Martin

et al. 2006; Dreitz et al. 2002). Kites occupy the remaining wetlands of the Kissimmee-

Okeechobee-Everglades freshwater watershed. The sampled units we used are identical to the

units Dreitz et al. (2002) used and encompassed major kite habitats (Figure 2-1). Although kites

may temporarily emigrate to unsampled areas, it is unlikely that they will not return to the major

wetlands included in the survey during some portion of the sampling period (Dreitz et al. 2002).

Sampling Methods

Marking protocol

Multiple, consecutive surveys of Snail Kites from airboats were conducted during the peak

of the breeding season (March through June) throughout the areas sampled from 1992 to 2005.

During surveys, workers located nests and banded young kites when they were ready to fledge

(-25 days). A total of 1806 young were marked between 1992 and 2005. Prior to 1995, 134 kites

were marked as adults (> 1 year) (Bennetts and Kitchens 2000). Additionally, 76 kites were

marked as young prior to 1992. Kites were marked with alpha-numeric bands.

Population survey protocol

Our protocol for population surveys from 2001 to 2005 followed that described by Dreitz

et al. (2002) and was part of the same Florida Cooperative Fish and Wildlife Research Unit

(FCFWRU) kite-monitoring program. Four to six consecutive surveys from airboats were

conducted at 2- to 4-week intervals throughout the designated wetland units from 25 February to

30 June. From 2001 to 2003, the surveys started between 1 March and 8 March and ended

between 15 June and 19 June. In 2004 and 2005, the surveys started between 25 February and 1

March and ended between 27 June and 30 June. During each survey we inspected every sighted









kite with binoculars and spotting scopes. We categorized each observed individual as: (1)

marked if the kite carried a band that could be uniquely identified; (2) unmarked if the sighted

kite did not carry an identifiable band; or (3) unknown if the banding status of the kite could not

be determined.

Analysis

Superpopulation size estimates of adults

We used the superpopulation approach (Schwarz and Arnason 1996) generalized by

Schwarz and Stobo (1997) into a robust design framework. This approach allows for movement

during secondary occasions (i.e., between surveys within a year, see Dreitz et al. 2002). Dreitz et

al. (2002) were the first to apply this method to the Snail Kite. All notations follow Dreitz et al.

(2002). The superpopulation approach estimated the total number of kites present in the sampled

area for at least some of the surveys during the sampling period (surveys within a year were

denoted i=1, 2,..., n). For any given year (denotedj), we referred to this estimate as the


superpopulation estimate for each year (Nj).


Nj = Nl + Z B+ (Eq. 2-1)
i=i

where Nij is the estimated abundance of the first survey in year. The Nij is estimated as

Smj +u,
Nij -u, (Eq. 2-2)
Pj

where m,j is the number of marked kites and u, the number of unmarked kites at each survey i in

year. Kites whose banding status was unknown were excluded from this analysis. The

p,, (sighting probability) is the estimated probability of sighting a kite given that it was present in









survey i of year. Given the constraint, pl = p2,, abundance can be estimated for all surveys

within year.

The Bi is the estimate of the number of new kites entering the sampled area (between

survey i and i+1) from areas not sampled on each survey

By = Ni+i,j Njy, i> O, (Eq. 2-3)


where ,j (apparent survival) is the estimated probability of not dying and not permanently

emigrating to an area not sampled between surveys i and i+1 of year. Given the constraint pi =

p2j B# can be estimated for i= 1,..., (n-1).

As in Dreitz et al. (2002) we used the CJS model implemented in program MARK to

estimate )., and p, (Cooch and White 2005). We analyzed each annual capture-recapture data

set separately to obtain estimates of ji and pi within a year. We preferred this approach to a

multigroup approach (used in Dreitz et al. 2002) because the date for the within year surveys did

not match exactly from one year to the other. Unlike Dreitz et al. (2002) the number of sampling

surveys in our study varied between four and six (we note that the number of sampling surveys

should not affect the superpopulation estimate as long as the sampling periods remain similar).

Although we ended our surveys before 30 June in 2001, 2002 and 2003, our data indicates that

all surveys conducted after the second week of June did not affect estimates of superpopulation

size. Indeed, removing the last survey for years when surveys ended on approximately 30 June,

did not affect the estimate of abundance. This can be explained by the fact that by the end of the

sampling season, it is unlikely that many birds will enter the sampled areas for the first time (i.e.,

most birds have already done so before the last survey).









For model selection among CJS models, we used QAICc, which corresponds to the Akaike

information criterion (AIC) corrected for small sample size and extrabinomial variation

(Burnham and Anderson 2002). We ran four CJS models for each year. Models that assumed (i

and pi remained constant between surveys were denoted with a subscripted dot (.). Conversely,

models that allowed i and pi to vary among surveys were denoted with a subscripted t. Thus,

the four models were denoted: p .pt, ,tP., and tpt. As recommended by Burnham and

Anderson (2002), we used model-averaged estimates of 4 and p. The purpose of estimating 4

and p with CJS models was primarily to compute estimates of superpopulation size. Confidence

intervals of estimates of superpopulation size were computed with the same parametric bootstrap

procedure (500 simulations) described in Dreitz et al. (2002).

The assumptions for the superpopulation model are similar to the ones required for the

more widely known Jolly-Seber model (Williams et al. 2002). In particular, homogeneity of rates

among animals are assumed. The superpopulation model also assumes that all members of the

superpopulation unavailable until t will exhibit similar probability of being available for capture

at t + 1. The Goodness of fit test (test 2 + test 3) implemented in program RELEASE, which tests

for homogeneity of i and pi and for lack of independence of survival and capture events

(Burnham et al. 1987), is also applicable for the superpopulation model (Williams et al. 2002).

There is no evidence of heterogeneity or lack of independence for i and pi when probability p

is > 0.05 (Cooch and White 2005). Burnham and Anderson (2002) indicate that model structure

is acceptable for extra-binomial factor c < 4, and they suggest to adjust for extrabinomial

variation if c > 1. We computed c with program RELEASE (Burnham et al. 1987).

Estimation of population growth rate

Annual population growth rate ( j), was estimated as:










N j+l
j = (Eq. 2-4)
Nj

We then computed the arithmetic average of all the Xj over the last 8 years (1998 to 2005) and a


3-year running average (denoted Xj-(j+2)).

Estimation of the number of young produced

We used the superpopulation approach described above to estimate the number of young

produced in any given year (hereafter referred as young) for the entire superpopulation (denoted


Nyj). For this analysis m, and u, (see Eq. 2-2) included exclusively kites that were hatched in

year. We used this approach only in 2004 and 2005 because we began recording m, and u, for

the young in 2004. There was not enough band resight information of young in 2004 and 2005 to

estimate #j and py that were specific to that particular age class. Therefore, we used ()j and pi

computed for adults to estimate the number of young produced in 2004 and 2005.

Detection probabilities for number of young produced every year

Only a proportion of the total number of young produced were detected and marked every

year. To estimate the proportion of young marked during each year (i.e., detection probability of

young), we used the following estimator (see Williams et al. 2002):

P9y i- j (Eq. 2-5)
NYj

where Pyj is the detection probability of young in year, Cy, is the number of young observed

and marked in year (hereafter referred as the number of young marked). The PYj differed from

survey-specific pi (directly estimated with CJS models).









Count data

We used two types of count survey data that we subsequently compared to estimates of

superpopulation size: first count (FC), and maximum count (MC). For FC we used the first

FCFWRU annual count survey (total number of birds counted during the first survey) as an

indicator of annual abundance. We used the first annual count survey because it was always

conducted at the same date (1 March + 1 week). Many agencies, including the Florida Fish and

Wildlife Conservation Commission use this type of format for surveys in which a designated

study area is sampled annually (typically at the same time of year). The MC was annual count

data of the maximum count obtained for any of the FCFWRU surveys within a sampling season.

MC and FC included: marked, unmarked and unknown kites. However, because in 1997

unknown birds were not reported, all analyses related to FC and MC data focused on the period

1998 to 2005.

Average number of kites and growth rate based on count data

We computed the arithmetic average for the two sets of count data (C).

We also used these count data sets to compute annual growth rate based on counts ( ,j). 1

were estimated as follow (see also Williams et al. 2002):


j C, (Eq. 2-6)


We then computed the 8-years arithmetic average of all the A2c; as well as the 3-year


running average denoted: Xcj-(j+2).

Detection probabilities for FC and MC

Monitoring based on counts typically assumes detection probability to equal 1.0; however,

in practice, this assumption is rarely met (Williams et al. 2002).









We determined the detection probability of Snail Kites using both FC and MC surveys by

computing the ratio of the number of kites counted in a given year (Cj) (using either FC or

MC), over the estimated superpopulation size for that same year (Nj):

C (FC) C (MC)
Pj (FC) = J and Pj (MC) = J (Eq. 2-7)
Nj Nj

We emphasize that Pj (FC) and Pj (MC) differed from survey specific p, (directly

estimated with CJS models).

We also established, for each type of survey (FC and MC), the increase in detection

probabilities (in percentage) necessary to obtain an average count of kites (C) > 650.

Estimates of Precision and Magnitude of the Difference between Estimates

We used the delta method to compute the variances of derived estimates (Williams et al.

2002). 95% confidence intervals (95% CI) of any parameter 0 that was not strictly positive (e.g.,

estimate of magnitude of the difference, see below) were computed as follows: 95% CI [6 ] = 0

+ tO.025,df SE [ ], where SE [ ] is the estimated standard error of 0 and to.025,df is the upper

97.5 percentile point of the t distribution on df (Burnham and Anderson 2002). As recommended

by Burnham and Anderson (2002), for any parameter 0 that is strictly positive (e.g., population

size), we used an approximation for computing 95% CI [6 ] that is based on a lognormal

distribution (p. 259 Bumham and Anderson 2002). The magnitude of the difference between two

estimates (MD) was estimated by computing the arithmetic difference between estimates

(Cooch and White 2005).









Results

Population Size and Average Population Growth Rate

Estimates of ,v. and p. were obtained with model averaging of models .p .Pt, CtP.,


and ,tpt for each year (estimates of p, p P and other survey-specific parameter estimates used

to compute estimates of superpopulation size are available on-line see Table A-i in Appendix

A). There was no need to adjust for lack of fit of the most general model in 2002, 2003, 2004 and

2005, because c was < 1, and test 2 + test 3 from RELEASE were all non significant (p > 0.05).

In 2001, the test 2 + test 3 was significant (p = 0.02). Therefore, we adjusted for lack of fit of the

most general model in 2001 (c = 2.2). Estimates of superpopulation size (Nj) from 1997 to

2000 were obtained from Dreitz et al. (2002), whereas estimates from 2001 to 2005 are the

results of the present study. Estimates of superpopulation size between 1997 and 2000 were

fairly constant and relatively high (Dreitz et al. 2002; Figure 2-2). Superpopulation size estimates

decreased sharply during the interval 2000-2002, but there was an apparent stabilization, or even

slight increase (but note the 95% CI overlap) in 2004 and 2005. The average superpopulation

size for the last 9 years (1997 to 2005) was 2254 (95% CI = 2124 to 2392). Estimates of the 8-

year average growth rate based on superpopulation estimates was 0.93 (95% CI = 0.84 to 1.03).

Estimates of annual growth rate based on superpopulation estimates were > 1, in 1998, 2003 and

2004 (Table 2-1). Estimates of the 8-year average growth rate was 1.11 (95% CI = 0.91 to 1.37)

based on FC; and 0.99 (95% CI = 0.86 to 1.12) based on MC. Estimates of the 3-year running

average growth rate based on superpopulation estimates and on MC were > 1 for 202-04 only

(Table 2-1). Estimates of the 3-year running average growth rate based on FC was < 1 for oo0-02

only (Table 2-1).









Average Number of Kites Before and After Decline Based on the Superpopulation
Approach

Estimates of superpopulation size suggests three major periods: a pre-decline period (1997

to 2000), a decline period (2000 to 2002) and a post decline period (2002 to 2005) (Figure 2-2).

We computed the average number of kites during the pre and post decline periods.

Prior to decline (1997 to 2000), the average number of kites based on the superpopulation

estimator was 3157 (95% CI = 2909 to 3426). After decline (2002 to 2005) the average number

of kites was 1407 (95% CI = 1278 to 1550). There was a substantial decrease between before and

after decline (MD = 1750; 95% CI = 1457 to 2041). This represented a 55% decrease (95% CI=

46% to 67%) when compared with predecline levels.

Average Number of Kites Before and After Decline Based on Count Data

Average number of kites before decline (1998 to 2000) based on FC data was 397 kites

(95% CI = 164 to 959) and 403 (95% CI = 316 to 514) after decline (2002 to 2005). Therefore,

FC data showed a slight increase in kite numbers between the intervals 1998 to 2000 and 2002 to

2005; however, the difference was not biologically significant (MD = 6; 95% CI = -220 to 208).

Based on MC data average number of kites (C) before decline (1998 to 2000) was 600 (95% CI

= 462 to 779) and after decline (2002 to 2005) was 410 (95% CI = 337 to 499). Although 95%

CI intervals of C overlapped, MC data showed a substantial decrease in kite numbers between

the intervals 1998 to 2000 and 2002 to 2005, (MD = 190; 95% CI = 82 to 298). This represented

a 32% decrease (95% CI = 20% to 51%).

Average Number of Kites and Growth Rate Based on Count Data

The average number of kites counted with FC and MC for the last 8 years (1998 to 2005,

but excluding 2001), was 401 kites (95% CI = 319 to 503) for FC and 491 kites (95% CI = 392

to 616) for MC. The CV was 0.09 for MC and FC.









Estimates of annual growth rate based on FC were < 1 in 2000 and 2002 (Table 2-1).

Estimates of annual growth rate based on MC were < 1 from 1999 to 2002 (Table 2-1).

Number of Young

There was a sharp decline in the number of young marked starting in 1999 (Figure 2-3).

The average number of young marked from 1992 to 1998 was 200 (95% CI = 145 to 277),

whereas the average number of young marked between 1999 and 2005 was 61 (95% CI = 38 to

96). The difference was substantial (MD = 139; 95% CI = 75 to 204). This represented a 70%

decrease (95% CI= 41% to 100%).

The number of young produced in 2004 and 2005 based on the superpopulation approach

,* *
were NY2004 = 414 and N2005 = 55. The detection probabilities in 2004 and 2005 were PY2004 =

0.16 and PY2005 = 0.35. Estimates of confidence intervals could not be computed because sample

size of resighting of young kites was too small.

Detection Probabilities for FC and MC

An increase of 63% in detection probability was necessary to obtain an average count >

650 for the FC survey; an increase of 33% in detection probability was necessary to reach a

similar target based on the MC survey data (Table 2-1). Detection estimates increased over the

years for both types of surveys (Table 2-1).

Discussion

Population Decline

Our results based on the superpopulation approach indicate that the population of Snail

Kites in Florida declined sharply between 2000 and 2002 (Figure 2-2). Although estimates were

slightly higher for 2004 and 2005, there was no evidence of a substantial recovery. The reduction

in the estimated average kite abundance before and after decline was substantial (55% reduction









in abundance). The method we used to estimate the superpopulation size of kites was also useful

in obtaining the number of young produced per breeding season. Although this parameter is

difficult to estimate in the wild, it is often needed to evaluate the viability of threatened

populations. For instance, the superpopulation approach is an appealing method to compute

reliable estimates of fertility rates, which are critical to correctly parameterize many types of

population viability analyses (Morris and Doak 2002).

We only had data to compute estimates of the number of young produced for 2 years (2004

and 2005). We also used these estimates to compute the proportion of young marked during

these 2 years (i.e., detection of young). The fact that detection of young varied substantially in

2004 and 2005 suggests that one should be cautious in using the number of young marked as an

indicator of the number of young produced. However, we believe detection estimates for these 2

years corresponded to extreme values. We expected low detection probability for 2004 because

birds bred unusually early, which meant a large proportion of birds fledged before they could be

marked. Conversely, in 2005 we invested an unprecedented effort in nest searching and marking

young, which led to higher detection. Unless detection declined significantly between the

intervals 1992-1998 and 1999-2005, we expect the observed number of young marked to be

representative of an important decline in the number of young produced. We believe detection is

likely to have increased in recent years because we invested more effort in nest-searching

activities than in earlier years. An increase in detection implies that the reduction in the number

of young produced in recent years is even more severe than is apparent in Figure 3-3. Models

used to obtain the number of young produced assumed that estimates of (.j and p, for young

and adults were similar. Appropriate sample size of resighting of young kites should be collected









in the future to check the assumption that adult estimates of #ij and p# provide a reasonable

approximation to estimate the number of young produced.

Problems Associated with Counts and Implications for Recovery Plans

Identifying population decline is critical to the process of species conservation. In practice,

it is often the documentation of population decline below a critical threshold that leads to the

classification of species as endangered under the ESA. Additionally, identifying a reduction in

population size may prevent unsubstantiated downlisting. The legal protection offered by the

ESA is in many cases essential to the persistence of many species at risk of extinction (Doremus

and Pagel 2001). Our results provide a compelling example of the risks associated with setting

recovery targets that are based on deficient monitoring programs. Next we explain how some of

the current recovery targets presented in the Snail Kite recovery plan (USFWS 1999) could be

met (even with a declining population) if monitoring does not account for major sources of error

such as detection.

One of the major recovery criteria listed in the plan states that the 10-year average

population size should be > 650. Even the most recent superpopulation estimates obtained during

our study indicated that the actual Snail Kite population size may be twice this number. This

suggests that although the 8-year-average counts obtained with FC and MC were all below the

recovery target set by the USFWS (i.e., 650 kites), it is likely that by increasing the search effort

(e.g., increase in the number of field personnel), more than 650 kites could have been counted

during these surveys. In fact, an increase of 33% in detection probability (i.e., the proportion of

kites counted from the "true" population size) during MC counts and 63% during the FC counts

would have boosted the average number of kites counted over a period of 8 years to above the

650 target (Table 2-1). In both cases CV was < 0.2 (i.e, < 20%). The second recovery criterion









states that kite numbers should not fall below 500 for any given year. This condition would not

have been met for the FC count with detection increased by 63%, because the count in 1998

would be 460 (but increasing detection from 0.09 to 0.16 in 1998 would have brought the count

for that year to 501). Similarly, this condition would not have been met for the MC count with

detection increased by 33%, because the count in 2003 would be 463 (but increasing detection

from 0.30 to 0.43 in 2003 would have brought the count for that year to 501).

The third recovery criterion stipulates that the 3-year running average should not be < 1.0

over a period of 10 years. Out of the five averages that could be computed for the last 8 years of

data for the FC count, only one value fell below 1.0. Reducing the proportion of birds that were

observed in 2001 from 0.19 to 0.11 (see Table 2-1), would have pushed all values of the running

average for the FC count above 1.0 (although the lower CI of these values may have fallen below

1.0, nothing is mentioned in the recovery plan about parameter uncertainty of growth rate

estimates). One can think of scenarios that would cause such a reduction in the proportion of

kites counted. For example, dry conditions could reduce airboat access to wetlands used by kites.

Finally, when computing the 8-year average of growth rate for the superpopulation, the FC

count and the MC count, we found that the growth rate was < 1.0 for the superpopulation and the

MC count but was > 1.0 for the FC count (in all cases lower 95% CI were < 1.0). The fact that

the estimate of the average growth based on FC data was > 1.0, even though the population was

declining, is most likely due to the increase in detection probabilities over the years (Table 2-1).

This increase in detection probability was also observed for the MC data, and resulted in an

inflated 8-years average growth rate for the MC survey as well. The observed increase in

detection probabilities over time could be due to an increase in the number of field personnel in

recent years (since 2002).









A particularly disturbing fact regarding count data is that, despite the drop in kite

abundance (55% based on the superpopultion approach) the FC count did not indicate a

reduction in kite numbers. The MC count, however, indicated a reduction in kite numbers. The

MC count may be less biased than FC because for every sampling year the maximum count will

be closer to the true abundance than any other single count. This is because all counts

underestimate true abundance, therefore the maximum count should be the closest to the true

abundance than any other count. However, since both types of counts ignored detection and did

not deal effectively with sampling variation they were therefore biased. The FC count (i.e., single

annual count) is by far the most common type of count survey. The format of the FC count was

very similar to the surveys conducted by the Florida Fish and Wildlife Conservation Commission

(FFWCC) between 1995 and 2004, except that the FFWCC annual count took place during the

midwinter (December to February), and was restricted to fewer wetlands (FFWCC unpublished

data). This spatial restriction also increases potential for errors associated with spatial variation.

Thus, by simply varying the proportion of kites observed during counts (i.e., detection)

three major recovery criteria in the Snail Kite recovery plan were close to being met based on a

monitoring that relied on counts (e.g., FC) in spite of an alarming decrease in estimated

population size and reduced reproduction.

Our primary purpose was not to set new recovery criteria for the Snail Kite (although our

study strongly suggests that existing criteria are in need of revisions), and we point out that

several authors have proposed promising approaches to set more appropriate criteria (e.g.,

Gerber and DeMaster 1999; Morris and Doak 2002). Instead, we emphasize the critical

importance of designing monitoring programs that address major, common sources of errors,









because reliability of the recovery criteria will strongly depend on the quality of the monitoring

data.

Importance of Monitoring to Diagnose Causes of Decline

Although the identification of population decline is an important step, it is evidently only

part of the process of protecting a species. A next step should be to diagnose the cause of decline,

or alternatively, factors limiting growth. In the case of kites, the drought that occurred in 2001

appears to coincide with the population decline and strongly affected adult and juvenile survival

(Martin et al. 2006). However, the drought affected kite survival only temporarily (1 to 2 years,

see Martin et al. 2006). The lack of evident recovery four years after this natural disturbance

suggests that factors affecting reproduction and recruitment may prevent growth. The drastic

reduction in the number of young kites marked (70% decrease), suggests that factors limiting

reproduction may deserve more attention than they have received in the past. However, rigorous

evaluation of the causes of decline and factors limiting growth needs to be performed.

Hypotheses related to disease, predation, food availability and nest substrate should probably be

the focus of future investigations (see Peery et al. 2004, Martin et al. unpublished). The multiple

competing hypotheses approach (MCH) provides an appealing framework to disentangle the

factors that could potentially affect population growth of threatened species (Williams et al.

2002; Peery et al. 2004). Ideally, monitoring programs designed to tease apart ecological

hypotheses using MCH, will incorporate both spatial variation and detectability. Addressing

spatial variation is particularly important to effectively assess hypotheses related to spatial

dynamics (Yoccoz et al. 2001). This may be of particular relevance to the management of

spatially structured populations of species that occupy large landscapes (Yoccoz et al. 2001).









Conclusion

A growing number of ecologists are recognizing the value of using designs that incorporate

both detectability and spatial sampling because (1) they allow for better parameter estimates, and

(2) because they favor more effective evaluation of ecological hypotheses (reviewed in Yoccoz

et al. 2001). As illustrated by our results, these sampling design issues are extremely relevant to

the protection of endangered species. Indeed, ignoring detectability and spatial variation may

lead to dangerously inappropriate management decisions (e.g., unsubstantiated downlisting).

Nonetheless, considerable resources continue to be invested in monitoring programs that ignore

these sources of variability, and many recovery plans continue to rely on these flawed programs.

Given the immediate risks of extinction faced by an increasing number of species, it is urgent for

managers and conservation biologists to rigorously revisit these recovery plans and monitoring

programs that do not effectively address spatial sampling and detectability.










Table 2-1. Estimates of Snail Kites annual population growth rates; and 3-year running average of
growth rates.


Parametera
Annual rates

Xj (S)

Xcj (FC)

Xcj (MC)
Average rates

Xj-(j+2)(S)

Xcj-(j+2) (FC)

Xcj-(j+2) (MC)


Year (/)
1998 1999 2000 2001


1.14b

1.26

1.18


0.88b

1.17

0.95


0.77b

1.57

0.83


0.73 b

1.12


0.73 b

0.69

0.85


0.75b

0.85


2002 2003 2004


0.69

1.09

0.89


0.94

1.04


0.85 0.86 0.96 1.05


1.05

1.17

1.17


aParameter explanations: ij, estimates of annual population growth rate based on

superpopulation estimates (S); 2cj, estimates of annual population growth based on

first-count surveys (FC) and maximum count survey (MC); Xj-(j+2), 3-year running

average growth rate based on Xj (S); Xcj-(j+2), 3-year running average growth rate

based on icj (FC) and icj (MC).
bComputed using data from Dreitz et al. (2002).









Table 2-2. Estimates of detection probability of Snail Kites for first-count surveys (FC) and
maximum count surveys (MC) for each year between 1998 and 2005'.
Year (j)
Detection 1998 1999 2000 2001 2002 2003 2004 2005 C
Pj(FC) 0.09 0.10 0.20 0.19 0.30 0.28 0.27 0.30 399
Pj(FC)+ 63% *Pj (FC) 0.15 0.16 0.33 0.31 0.49 0.45 0.44 0.49 651
Pj (MC) 0.18 0.19 0.20 0.23 0.30 0.30 0.27 0.30 490
Pj (MC)+33%*Pj (MC) 0.24 0.25 0.27 0.31 0.40 0.40 0.36 0.40 654
a Estimates were obtained by computing the ratio count over superpopulation size for each year (j). The
C corresponds to the average number of kites counted using the estimated detection for FC (Pj (FC))
and for MC (Pj (MC)) and the detection probabilities that were increased by 63% for the FC surveys
(Pj (FC) + 63% Pj (FC) ) and 33% for the MC surveys (Pj (MC) + 33% Pj (MC)).
bDetection probabilities from 1998 to 2000 were computed using estimates of superpopulation size
published in Dreitz et al. (2002).
bComputed using data from Dreitz et al. (2002).










































Figure 2-1. Map of the wetlands that were sampled to obtain both counts and capture-resighting
information of Snail Kite for the estimation of population size. Thick black line
delimits areas sampled by the Florida Cooperative Fish and Wildlife Research Unit.










5000
*-- Superpopulation estimates
---- Recovery threshold
S4000 -- MC count
S. ... o FC count

3000
CU
4-

Z- 2000 -


S1000 I


0- .. -- - -----------
0I I I I I
1996 1998 2000 2002 2004 2006
Year

Figure 2-2. Comparison of the estimates of population size of Snail Kites (using the
superpopulation approach) with annual counts. Data for three count surveys are
plotted in the figure: (1) first count survey (FC); (2) maximum count survey. Kite
numbers and estimates of population size from 1997 to 2000 were obtained from
Dreitz et al. (2002), while estimates from 2001 to 2005 were results of the present
study. Error bars correspond to 95% confidence intervals. The recovery target for
Snail Kites (650 birds), set by the USFWS in 1999 is also presented.










350

300

c 250
0
200
0
Q 150-
E
= 100 -
z

50 -

0 1
1990 1992 1994 1996 1998 2000 2002 2004 2006

Year

Figure 2-3. Number of young (i.e., nestlings close to fledging) Snail Kites marked every year
from 1992 to 2005.









CHAPTER 3
MULTISCALE PATTERNS OF MOVEMENT IN FRAGMENTED LANDSCAPES AND
CONSEQUENCES ON DEMOGRAPHY OF THE SNAIL KITE IN FLORIDA

Introduction

Habitat loss and fragmentation are major factors affecting populations of many organisms

(Holt and Debinski, 2003). One detrimental effect is reduced movement of these organisms (Holt

and Debinski 2003; Smith and Hellmann 2002). This may have important population

consequences given that movement is a fundamental process driving the dynamics of fragmented

populations, as it connects local populations through emigration and immigration (Hanski 1999;

Clobert et al. 2001).

To assess how movement influences the dynamics of spatially-structured populations, we

need to understand how animals perceive, move through, and learn about the landscapes they

occupy (Hanski 2001). We also need to evaluate the relative importance of critical factors

governing movement processes at a pertinent spatio-temporal scales. Patch size, distance

between patches, and patch quality are major factors influencing the movement of many animal

populations in spatially structured systems (Hanski 1999). Several studies have demonstrated the

effect of distance on movement (e.g., Haddad 1999; Hanski 2001). Theoretical models of

metapopulation dynamics commonly assume greater emigration from smaller patches (Hanski

2001; Schtickzelle and Baguette 2003), and higher immigration toward larger habitat patches

because of the more frequent encounters of moving animals with patch boundaries (patch

boundary effect) (Lomolino 1990; Hanski 2001).

Fragmentation and habitat reduction reduce patch size and increase the linear distance

between patches: both alterations are likely to decrease movement (Holt and Debinski 2003).

Creating areas unsuitable for foraging or breeding (i.e., matrix) between or around habitats may

also decrease survival (Schtickzelle and Baguette 2003).









Despite the importance of providing robust quantitative demographic and movement

estimates of populations inhabiting fragmented landscapes (Hanski 2001; Williams, Nichols and

Conroy 2002), few empirical estimates exist, especially for vertebrates using large landscapes.

From 1992 to 2004, we studied an isolated population of Snail Kites (Rostrhamus

sociabilis) restricted to Florida. The Snail Kite is a raptor that feeds almost exclusively on

freshwater apple snails Pomaceapaludosa (Beissinger 1988). The kite's restricted diet makes it a

wetland-dependent species. Since wetlands in Florida have been severely reduced (Davis and

Ogden 1994; Kitchens et al. 2002) since the early 1930s, the population is now confined to the

remaining fragments of wetlands extending from the southern end to the centre of the state

(Figure 3-1).

Because the availability of apple snails to kites is related to hydrologic conditions,

variations in water levels are likely to influence Snail Kite behaviour and demography. In

particular, snail availability to kites is greatly reduced during droughts (Beissinger 1995).

Beissinger (1986) and DeAngelis and White (1994) described the hydrologic environment used

by kites as highly spatially-temporally variable. In such a variable environment, one might

expect kites to show nomadic tendencies (Bennetts and Kitchens 2000). Bennetts and Kitchens

(2000) developed a conceptual model of kite movement along a food resource gradient. They

hypothesized that when food is scarce (during drought), kites move to refugia habitats or die.

When food is abundant exploratory movements can be done at minimum risk of starvation.

During droughts, kites that have previously explored wetlands throughout their range are less

likely to search randomly for alternative habitats, and thus are less likely to starve. Their model

also suggests that when food is superabundant, occasional territorial defence may occur for short

periods of time, but otherwise kites are typically non territorial (Beissinger 1995). Bennetts and









Kitchens (2000) estimated the average probability of movement among wetland units (Figure 3-

1) to be approximately 0.25 per month, which they associated with a nomadic type of behaviour.

However, this probability was obtained without considering the complexity of the spatial

configuration of the system.

We attempted to enhance our understanding of how kites perceive and move throughout

the landscape by incorporating a detailed level of spatial complexity into a modelling approach at

multiple spatial scales. First, we estimated movement within a group of contiguous wetlands

(separated by small physical barriers, easily crossed by kites: such as a road). The distance

between centroids of these contiguous wetlands varied between 16 and 110 km. Second, we

estimated movement within a group of wetlands separated by a moderate extent of matrix (< 5

km): moderately isolated wetlands. Matrix areas generally consist of non-wetland areas (e.g.,

agricultural or urban areas). The distance between centroids of these moderately isolated

wetlands varied between 10 and 44 km. Third, we estimated movement among wetlands or

groups of wetlands isolated by extensive matrix (> 15 km): isolated wetlands. To be consistent

with the classification ofBennetts et al. (1999a), we called theses isolated wetlands: regions.

Most regions used to be connected through the Kissimmee-Okeechobee-Everglades watershed,

and became isolated as a result of habitat reduction (Davis and Ogden 1994; Light and Dineen

1994) (Figure 3-1). The distance between centroids of these regions varied between 69 and 232

km.

We also explored movement at two temporal scales. First, we examined movement at an

annual scale. Because of the period of sampling (i.e., peak of breeding season), this informed us

about patterns of breeding and natal philopatry of Snail Kites. Second, we examined movement

patterns on a monthly scale. This period of sampling included the entire year (i.e., including









periods outside the breeding season). Thus this study also informed us about movement patterns

that were independent of breeding activities (e.g., exploratory movement). Finally, we discussed

the consequences of kite movement on survival.

To date, the assumption has been that during a drought, kites move from areas most

affected by drought toward areas least affected by drought (Beissinger and Takekawa 1983;

Bennetts and Kitchens 2000; Takekawa and Beissinger 1989); and that the impact of a drought

on the kite population will depend on the spatial extent and intensity of the drought (Beissinger

1995; Bennetts and Kitchens 2000). However, all hypotheses regarding kite responses to drought

are based on count data that do not consider detection probabilities. Therefore, these hypotheses

have yet to be rigorously tested and quantified with appropriate statistical methodologies

(Williams et al. 2002).

Hypotheses and Predictions

Prediction 1: Effect of fragmentation on movement

We predict that movement will covary positively with connectivity (i.e., amount of matrix

between wetlands). Thus, movement among contiguous wetlands should be greater than among

moderately isolated wetlands, and movement among moderately isolated wetlands should be

greater than among isolated wetlands (i.e., regions). Prediction 1 implies that movement within

regions will be greater than between regions, which could also be explained by a distance effect

on movement. However, if movement among contiguous wetlands is greater than among

moderately isolated wetlands, the effects of connectivity on movement can be separated from the

effects of distance (between centroids), since distances between centroids of contiguous wetlands

are greater than that of moderately isolated wetlands in the study area.

Prediction 2: Effect ofpatch configuration on movement









We expect patch size and distance between patch centroids to influence movement.

Movement among patches (i.e., wetlands) should decrease with distance (Hanski 1999).

Emigration should be higher from smaller patches (Hanski 2001; Schtickzelle and Baguette

2003), and immigration should be higher toward larger patches (Lomolino 1990; Hanski 2001).

Prediction 3. Patch configuration affects juvenile movement more than adult movement

Patch size and distance between patches are more likely to influence movement of birds

that have never dispersed from their natal area (typically young individuals), than birds that are

aware of wetlands outside their natal area. Because we expect the number of wetlands visited to

increase with time, on average juveniles (<1 year) should have visited fewer wetlands than

adults. Therefore, movement of juveniles should be less influenced by habitat characteristics

(e.g., habitat quality) of destination sites than movement of adults, whose movement may be

partly influenced by their knowledge of the location of multiple wetlands (assuming that kites

remember sites they have already visited). This prediction is derived from hypotheses developed

by Bennetts and Kitchens (2000) and (Bell 1991), who suggested that many species learn from

exploratory movements, and thus modify their movement patterns according to their experience

with visited habitats. Thus we expect a stronger relationship between movement and geometric

features of the landscapes for juveniles than for adults.

Prediction 4: Drought effect on movement

a. During a drought, we predict that some birds will move from areas most affected to

areas least affected by drought (e.g., Takekawa and Beissinger 1989).

b. Because of their knowledge of alternative wetlands and the paths linking these wetlands,

adult birds should be more successful than juveniles in moving refugia habitats.

Prediction 5: Drought effect on survival









a. As predicted by Beissinger (1995) and Bennetts and Kitchens (2000), we expect survival

to be lower during drought.

b. Survival should be lower in regions most affected by drought.

c. Because adults are more likely to move successfully to areas least affected by drought,

we expect survival to decrease more for juveniles than for adults.

Methods

Study Area

This study was conducted throughout central and southern Florida, encompassing most of

the habitats used by the Snail Kite. Thirteen wetlands were sampled (Figure 3-1). Given that

kites can cross small physical barriers delimiting each wetland (e.g. road) with relative ease

(Bennetts 1998), we further aggregated the units into five larger groups of wetlands (regions)

(Figure 3-1). We used Bennetts (1998) and Bennetts et al. (1999a) definition of a region. Regions

were separated from other regions by an extended matrix (>15 km). Water Conservation Areas

(WCAs), Everglades National Park and Big Cypress National Preserve constituted a group of

contiguous wetlands and were grouped into one region: the Everglades region (E). The

Kissimmee Chain of Lakes region (K) included Lake Tohopekaliga, East Lake Tohopekaliga,

Lake Kissimmee, and all the small lakes in the surrounding areas. Wetlands in the K region were

isolated by moderate extent of matrix (< 5 km). Lake Okeechobee (0), St Johns Marsh (J), and

Loxahatchee Slough (L), constituted their own regions. Areas of wetlands and distances between

wetlands were estimated using a Geographic Information System (ArcView GIS 3.2; Xtools,

DeLaune 2000).

Criteria for Determining the Regional Impact of the 2001 Drought

We used water-stage data (elevation of water surface measured in feet above the National

Geodetic Datum of 1929) recorded daily in each of the major wetland units and made available









by the South Florida Water Management District (http://www.sfwmd.gov/org/ema/dbhydro) to

develop an index of drought impact. We used the data corresponding to the period of study (1992

to 2003). Water stage was averaged by month for the entire time series. We calculated the mean

of the monthly average stages for March through June of each year. This period is especially

critical for apple snail breeding and availability to the kites (Darby 1998) and also includes the

greater part of the seasonal dry season when water stages are at their annual minimum (i.e., when

water levels are most likely to affect kite survival and movement).

We determined the mean stage for the period of record (1992-2003) for each major

wetland unit and determined where drought-year water stage means fell in terms of standard

deviations below this value. This method, proposed by Bennetts (1998), allows for comparisons

of drought intensity among wetlands for the period of record. The 2001 drought occurred

between January and August (Smith et al. 2003). Intensity of drought was maximal for the

lowest drought score values (DSV). This analysis indicated that region E (WCA3B DSV = -2.32;

Big Cypress DSV = -2.28; WCA1A DSV = -2.18; WCA3A DSV = -1.92; WC2B DSV = -1.41;

WCA2A DSV = -1.20) and region O (DSV = -2.57) were the most-impacted, while region K

(Lake Kissimmee DSV =-0.72; Lake Tohopekaliga DSV =-0.84; Lake East Toho DSV = -0.98)

was the least affected. Region J was also affected (DSV = -1.92).

Statistical Models to Estimate Movement and Survival

Multistate capture-recapture models (Hestbeck, Nichols and Malecki 1991; Williams,

Nichols and Conroy 2002) were used to estimate apparent survival (q), movement probabilities

(q) and detection probabilities (p) simultaneously. 0 was defined as the probability for a kite

alive in location u (i.e., wetland u) at time t to survive between time t and t+ ; andp" was the

probability of detecting (sighting) a kite that was alive and associated with wetland u. We









defined q "" as the probability that a kite in wetland u at time t was in wetland s at time t+ 1,

given that it was alive at t+ 1. Modelled parameters used notation from Senar, Conroy and Borras

(2002); time dependency was (t) and no time effect was (.). We assigned each bird to one of two

age classes: juveniles (juv), 30 days to 1 year; and adults (ad), older than 1 year. Effects

embedded in other factors are shown using parentheses. A multiplicative effect is shown by (*)

and an additive effect is shown by (+). All computations of the movement and survival

probabilities were carried out using program MARK V 4.1 (White and Burnham 1999).

Field Methods for the Study of Movement on a Monthly Scale

Between 1992 and 1995, 165 adult and 120 juvenile Snail Kites were equipped with radio

transmitters with a battery life of approximately 9-18 months (Bennetts and Kitchens 2000).

Between 1992 and 1995, aircraft radio-telemetry surveys were conducted on a weekly basis (two

4-5 hour flights every week) over a large portion of the entire range of the population in Florida.

Previous analyses by Bennetts et al. (1999a) and Bennetts (1998) found no evidence of radio

effects on survival or movement probabilities.

Statistical Methods to Estimate Movement on a Monthly Scale Using Radiotelemetry

Estimating monthly movement among regions

To estimate monthly movement probabilities (q) of radio-tagged individuals among

regions, we used multistate models. Because monthly survival estimates were beyond the scope

of our study, we removed individuals from the analysis after they were last observed and fixed

survival parameters to 1. For this analysis, we included individuals for which the fate and

location could be determined with certainty (i.e., detection probability equals 1). In addition,

birds that temporarily disappeared and then reappeared in the sample were censored when they

disappeared and were included again when they reappeared (Williams et al. 2002). This analysis

included six states: the five regions described above (E, K, O, L, J) (Figure 3-1), and one state









containing peripheral habitats and matrix area (P, all locations outside the sampled areas). To

compute the probability of movement out of a patch (wetland or region), we summed the

transition probabilities out of that patch. To calculate the average monthly probability of

movement out of any wetland within a region, we computed the average of the monthly

movement probabilities out of every wetland in the region of interest.

We tested the effect of patch size (AR for the surface area of the receiving site, and AD for

the surface area of the donor site), distance (d), region (r), age, and time on movement

probabilities. The notations for age and time followed the ones common to all analyses. We also

tested the effect of year (year), given that the radio-telemetry study was conducted between 1992

and 1995. A seasonal effect (seas) with respect to three 4-months seasons (January-April, May-

August, September-December) (Bennetts and Kitchens 2000); and a breeding season effect

(breed; breeding season: January-June; non-breeding season: July-December) were also

included.

With known fate multistate data (for which the detection probability is 1), there is currently

no appropriate Goodness of Fit test (GOF). However, most analyses presented in our study

included fairly general models.

Estimating monthly movement within regions using radio-telemetry

The same method was used for this analysis as for the among-regions analysis. Because

two regions comprised several wetland units, we conducted two separate analyses. The analysis

for the K region contained four moderately isolated wetlands (denoted: mw): Lake Tohopekaliga,

Lake East Tohopekaliga, Lake Kissimmee, and a site containing all of the small lakes in the

surrounding area (Figure 3-1). Analysis for the Everglades region contained five contiguous

wetlands (denoted: cw): WCA3A,WCA3B, Everglades National Park, and Big Cypress (Figure

3-1). We also, aggregated three contiguous wetlands (WCA1,WCA2A and WCA2B), into one









site, as our data set would not have permitted a seven-site model. Patch size and distance were

included as factors in the models of region E only. This analysis was not applicable for region K,

because of the site that included all of the small lakes.

Field Methods for the Study of Movement and Survival on an Annual Scale

We used mark-resighting information collected during the peak of the breeding season

(March 1-May 30), for a period of 13 years (1992 to 2004). Between 1992 and 2004, 1730

juveniles were marked just before fledging. Juveniles advance to the adult age class at the

beginning of the next breeding season (Bennetts et al. 2002). In addition, between 1992 and

1995, 134 adults (i.e., older than 1 year) were banded. Bands were uniquely numbered anodised

aluminium colour bands. Banded kites were identified from a distance, using a spotting scope.

Each wetland was surveyed at least once using an airboat.

Statistical Methods to Estimate Annual Movement and Survival Using Banding Data

We used a multistate model to estimate annual movement and survival probabilities. We

assigned the location of each bird to four regions (see STUDY AREA). We excluded region L

from this analysis to maximize precision, as relatively few birds were recorded in this area.


Estimating survival

A set of biologically relevant models was developed that allowed q andp to vary across

time, or stay constant for each age class. Because our data set included kites banded as juveniles

and as adults, age was modelled both as time since marking and as a group effect. We also

created models that included drought effect on q andp. We included a drought effect, which

assumed different effects on apparent survival in 2000-2001 and 2001-2002 (denoted: D1-2). We

used this approach because the drought was likely to affect q before and after the 2001 sampling

occasion. ND indicated that q was constant during the remaining non-drought years (denoted:









ND). For juveniles we designed models with additive effect of time and region (t+r) on q, but

because of the drought few juveniles were fledged in 2001 (32 juveniles were fledged in K, 3 in J

and none in E and O). We thus constructed models with additive effect of time and region on 0,

except during the interval 2001-2002, during which q was assumed to be similar among regions

(denoted: 4juv(r+td)). Consequently, during the interval 2001-2002 model 0juv(r+td) reflected

apparent survival for northern regions (K and J). Because we expect environmental conditions to

be more similar among neighboring regions than among regions that are far apart, we expect

survival in regions close to each other, to be similar. Thus we developed models that assumed

similar apparent survival probabilities in neighboring regions. Due to the proximity of regions E

and O in the south (separated by 30 km) and K and J in the north (separated by 25 km),

(conversely, O and J were separated by 50 km; Figure 3-1), we developed models with a

common survival parameter for each group of regions (denoted [E=O K=J]; superscripts

indicate regions the survival probabilities pertain to; "=" indicates that 0E is the same as 0o,

similarly OK is the same as J ; indicates that OE and 0o are different from OK and iJ).

Models assuming a different q for each region were denoted ((r)).

Because the drought intensity was strongest in E, O and J (lowest DSV), and weakest in K

(highest DSV), some models assumed similar drought effects on q in E, O, and J (denoted

#E=0=J] (D1-2)); with no drought effect on q in K (K (.)).

Estimating annual movement probabilities among regions using banding data

Our multistate approach using the banding data (described above) provided annual

estimates of movement probabilities (q), among four regions (E, O, K and J). We tested the

effect of the drought on movement between 2000-2001(denoted D1). We also estimated the









probability for a kite to be found in a particular region (u) at year t+ 1, given that it was present in

that same region in year t (/"). These probability estimates were used to evaluate the level of

philopatry at each site. These estimates were obtained as one minus the estimated probabilities of

moving away from the area.

Goodness of fit

Previous survival analyses indicated a strong age effect on q (Bennetts et al. 2002).

Unfortunately, we are not aware of GOF test accounting for an age effect on q for multistate

model. However, it is possible to test the fit of adult data separately. We used program U-CARE

version 2.02, which tests the fit of the "Jolly move" (JMV) and Arnason-Schwarz models (AS)

(Pradel, Wintrebert and Gimenez 2003). We were only able to test model JMV, which fit the

data satisfactorily when testing the fit of adult data separately (X2 102 = 104.3, P = 0.42). The fit

of the JMV model could not be assessed on juveniles separately (Test M requires > 4 occasions).

Thus, as suggested by Senar, et al. (2002) we computed a GOF accounting for an age effect (by

summing Test 3.SM, Test 2.CT and Test 2.CL, available from program U-CARE, see Choquet

et. al. 2003), for a site-specific Cormack-Jolly-Seber (CJS) model in lieu of a multistate model.

The site-specific CJS model fitted the data satisfactorily (X2 175 = 152.1, P = 0.89). We concluded

that there was no evidence of lack of fit of the multistate model used (i.e., models in Table 3-3

accounted for an age effect on q; Choquet et al. 2003).

Model Selection Procedure

For each mark-resight analysis, we first developed and fitted a set of biologically relevant

models that corresponded to our best apriori hypotheses (referred as starting models). We then

developed models whose relevance was linked to the need to evaluate the fit of each of the

starting models (Cam, Oro and Jimenez 2004). We used AICc (Burnham and Anderson 2002) as









a criterion to select the model that provided the most parsimonious description of the variation in

the data (i.e. model with the lowest AICc). The value of AAICc (the difference between the AICc

of a particular model and that of the model with the lowest AICc) was presented in each set of

model-selection results. We also used AICc weight (w) as a measure of relative support for each

model (Burnham and Anderson 2002). We reported only the model whose w was greater than

0.01.

Effect of Patch Size and Distance on Movement

Movement probabilities were modelled as linear-logistic function of patch size and/or

distance (Blums et al. 2003). For example, probabilities of moving from one patch to another in

function of distance were modelled as:

Logit (V (d)) = f, + d (d),

where fl, fld, are the parameters to be estimated. f, is the intercept, fld is the slope for

distance between patch centroids (d). Probability of moving was predicted to decrease with

increasing distance between patches (fd < 0) (Blums et al. 2003). Whenever the 95%CI [/fd

estimate did not overlap 0, the relation was considered statistically significant.

Effect Size

To measure the magnitude of the difference between estimates we computed estimates of

"effect size" (ES) as the arithmetic difference between estimates. Whenever the 95%CI [ES ]

did not include 0 the difference was considered statistically significant (Cooch and White 2005).

Estimates of Precision

Variances for derived estimates in our study were computed using the delta method

(Williams et al. 2002). Confidence intervals for estimates that were strictly positive (0, w), were









computed using the method proposed by Burnham et al. (1987) based on the lognormal

distribution (Appendix B-l in Appendix B).

Estimates of effect size (not strictly positive), were approximated as follows: 95%CI [ ]=

0 1.96 SE [ ].

Results

Monthly Movement Probabilities Among Regions

Effects of patch size and distance

The most parsimonious model (with lowest AICc; Table 3-1.a.), was a model that only

included a site-specific effect of movement (/(r)). However, the model that assumed movement

probabilities to be site-specific for adults, but included a patch-size and a distance-between-

patches effect plus interaction of these factors for juveniles (y/ad(r) y/juv(AR d)), also received

some support (AAICc=1.7; Table 3-1.a.). This model had considerably more support than the

model that assumed movement probabilities to be solely site-specific for adult birds and

juveniles (y/ad(r) q/juv(r) ; AAICc = 15.2; see also Table B-l.a. in Appendix B).

When the analysis is conducted on juveniles only, the model y/juv(AR d) is considerably

better than yijuv(r) (AAICc = 14; Table 3-1.b; see also Table B-l.b. in Appendix B), indicating

that patch size and distance may be important in determining the movement probabilities of

juveniles. Model y/juv(AR d) indicates that the probability of moving between two locations


decreased with distance between these locations (/d = -0.020, 95 %CI =-0.032 to -0.007).

Conversely, we could not show any relationship between the receiving site area and movement

with this model ( pAR = -0.020, 95 % CI = -0.247 to 0.207). The interaction for this model was


positive, but not very strong (P AR*d = 0.002, 95 %CI =0.0001 to 0.003). We also tried a model









with an additive effect of distance and patch size of the receiving sites, yiad(r) yijuv(AR + d).

That model did not reach numerical convergence with program MARK when the data set

included both juvenile and adult birds; we consequently ran this model on a data set that only

comprised juvenile birds ( yjuv(AR + d); Table 3-1b). Although this model was less

parsimonious than one that incorporated an interaction effect (AAICc = 3; Table 3-1.b), it was

considerably better than the site-specific model (AAICc = 14; see also Table B-l.b. in Appendix

B). Model yijuv(AR + d) supported the hypothesis of a negative relationship between movement


probabilities and distance (/d = -0.011, 95 %CI = -0.020 to -0.0030). This model also

supported the hypothesis of a positive relationship between movement and size of the receiving

sites (fOAR = 0.205, 95 %CI = 0.120 to 0.289). The models that included the size effect of the

donor patch on juvenile movement yijuv(AD) received little support (AAICc = 6.5; Table 3-1.b),

but the fl parameter for AD supported the hypothesis that emigration was lower out of larger

patches (/AD = -0.191, 95%CI = -0.298 to -0.084).

There was no evidence of any patch size or distance effect on adult movement (Table 3-1.a

and Table 3-1.c). Models that included effects of time, year, or season received no support

(w~0).

Monthly Movement Probabilities Within Regions

Movement within the Everglades region

The most parsimonious model for this analysis was y/(seas *cw) (w ~ 1; Table 3-2.a),

which assumed movement probabilities to vary by season and to be site-specific.









Movement within the K region

The most parsimonious model for this analysis assumed movement to vary by season

(yV(seas); w = 0.67; Table 3-2.b).

Comparison Among and Within Regions

The probability that a Snail Kite in any of the five wetlands in region E moved to another

unit in that same region within the next month (average monthly movement probability among

contiguous wetlands), using model y(seas cw) for the Everglades region (Table 3-2.a), was

0.29 (95%CI = 0.24 to 0.35). By contrast, the monthly movement probabilities from E to the four

other regions was only 0.04 (95%CI = 0.03 to 0.05), using model V/(r) (Table 3-1.a). The same

pattern was observed in region K where kites moved extensively among the moderately isolated

wetlands in this region, using model yi(seas) (Table 3-2.b) we found the average monthly

probability f = 0.15 (95%CI = 0.13 to 0.17); with only a 0.09 (95%CI = 0.06 to 0.12) monthly

movement probability from this region to the four other regions, using model V/(r) (Table 3-1.a).

The probability that kites in any of the five regions moved to another region within the

next month (average monthly movement among isolated wetlands), using model V/(r) (Table 3-

1.a) was 0.10 (95%CI = 0.08 to 0.12).

Average monthly movement among contiguous wetlands was significantly greater than

among moderately isolated wetlands (ES = 0.14, 95%CI = 0.08 to 0.20). Average monthly

movement among moderately isolated wetland was significantly greater than among isolated

wetlands (ES = 0.05, 95%CI = 0.02 to 0.07); and average monthly movement among

contiguous wetlands was significantly greater than among isolated wetlands (ES = 0.19, 95%CI

= 0.13 to 0.25).









Interannual Survival Estimates

The most parsimonious model

(4K[EO=KJ] (ND) (.)E0d (DI12) juv(r + td)p(r t) (r D), received overwhelming support

from the data (w = 0.96; Table 3-3). This model had region specific apparent survival for adults,

which did not vary over time but differed significantly between drought and non-drought years

(Figure 3-2). There was an additive effect of region and time for estimates of apparent survival

of juveniles, except for the interval 2001-2002, during which q was assumed to be time

dependent only. Sighting probabilities were region and time specific. Movement probabilities

were region specific and were affected by the drought. Apparent survival estimates for adults

^E ^0 ^J
kites located in neighboring regions during non-drought were similar (i.e., 0 = 0 and =

"K -E -K
K ). During non-drought years E was greater than E (ES = 0.08, 95%CI= 0.03 to 0.13;

Figure 3-2). This model also assumed no significant effect of drought on adult apparent survival

in K (the region with the highest DSV > -1), but assumed a similar effect of drought on adult

apparent survival in E, O and J (which all had lower DSV < -1) (see Figure 3-2 for estimates).

Average estimates of juvenile apparent survival during non-drought years were higher in

southern regions ( = 0.520, 95%CI = 0.460 to 0.588; o = 0.471, 95%CI = 0.372 to 0.597)

-K -J
than in northern regions (0 = 0.355, 95%CI = 0.233 to 0.541; 0 = 0.412, 95%CI = 0.295 to

0.575), but confidence intervals overlapped. During drought years confidence intervals of region

-E
specific juvenile apparent survival overlapped widely ( = 0.07, 95%CI = 0.014 to


0.349; =0.0647, 95%CI = 0.010 to 0.427; = 0.054, 95%CI = 0.007 to 0.405; = 0.058,

95%CI = 0.004 to 0.837). Because no juveniles were marked in 2001 in E and only 4 were









marked in K in 2000, we could not test the hypothesis of a lower effect of the drought on

apparent survival of juveniles in K. Given that juvenile apparent survival estimates were not

significantly different from one another we averaged these estimates across regions (Figure 3-2).

Estimates of adult apparent survival averaged across regions remained fairly high and

constant over time ( = 0.86; Figure 3-2), but dropped substantially during drought years

between 2000 and 2002 (average apparent survival between 2000 and 2002 was 0 = 0.72; Figure

3-2). This represented a relative decrease of 16% in apparent survival during the years that were

affected by the drought when compared to non-drought years, but the decrease was only

significant between 2001 and 2002 (ES = 0.39, 95%CI= 0.24 to 0.53; Figure 3-2). Juvenile

apparent survival varied widely over time, but reached a record low between 2000 and 2002

(average 0 between 2000-2002 was = 0.06; Figure 3-2). Juvenile apparent survival decreased

by 86% in 2000 and 2002 (relative decrease) when compared to its average over the non-drought

years (average q during 1992-1999 and 2002-2003 was = 0.44).

Inter-Annual Movement Among Regions and Drought Effect on Movement

The most parsimonious model (described above; Table 3-3), had site-specific annual

transition (movement) probabilities that were constant over time, except during the drought

(Table 3-3). This model was substantially better supported than the same model without a

drought effect (AAICc = 7; Table 3-3). Using the most parsimonious model, we found that

during the 2001 drought, movement estimates were higher from the areas with the lowest DSV

(i.e., most impacted regions: O and E) toward areas with highest DSV (i.e., least impacted

region: K), qOK= 0.33 (95%CI = 0.146 to 0.580), K= 0.030 (95%C = 0.014 to 0.066), than

during non-drought years OK = 0.044 (95%CI = 0.024 to 0.080), / = 0.015 (95%CI = 0.010









to 0.022). However, the difference was only statistically significant for birds moving from O to

K (ES = 0.28, 95%CI= 0.05 to 0.52). Estimated movement probabilities toward the most-

impacted region (i.e., E and 0) during the drought approached 0. This contrasted with non-

drought years during which movement probabilities toward E and O were typically much higher

than 0 (ranged from 0.02 to 0.16; Table B-4 in Appendix B). Surprisingly the probability of

moving from J to K during the drought approached 0, while during non-drought years this

probability was JK = 0.06 (95%CI = 0.03 to 0.11).

Models including an age effect as well as a drought effect on movement did not reach

numerical convergence; however, we did not detect any movement of juvenile bird from the

most to the least impacted regions between 2000 and 2001. Models including an age effect on

movement but no drought effect were not supported (w < 0.01; Table B-3 in Appendix B).

We used the most parsimonious model to estimate the probability of staying in each region

from one year to another. The probability of staying in E was /EE = 0.95 (95%CI = 0.94 to

0.96), the probability of staying in O was too = 0.76 (95%CI = 0.71 to 0.82), the probability of

staying in K was K = 0.72 (95%CI = 0.66 to 0.79), and the probability of staying in J was rJJ

= 0.75 (95%CI = 0.69 to 0.82).

Discussion

Monthly Movement Among Contiguous and Isolated Wetlands

We found that kites moved extensively over large areas of contiguous wetlands (average

monthly movement probability: 0.29). However, our study also showed much less movement

among isolated wetlands (average monthly movement probability: 0.10). As expected average

monthly movement probability among moderately isolated wetlands was intermediate: 0.15.

Differences between these estimates were all statistically significant. These results agree with









Prediction 1, that loss of connectivity reduces movement of kites. However, as stated in

Prediction 1, only by comparing movement among contiguous wetlands and among moderately

isolated wetlands could the effect of connectivity and distance be separated. Indeed, despite the

fact that distances between the centroids of contiguous wetlands (E) were greater than between

the centroids of moderately isolated wetlands (K), movement among wetlands in E were greater

than in K. The results also suggest that seasonality influenced movement within, but not among

regions. One possible explanation is the pronounced wet-dry seasonality resulting in spatio-

temporally variable habitat conditions at both the local and regional levels (Davis and Ogden

1994; Bennetts and Kitchens 2000). The fact that this seasonal pattern was not observed for

movements among regions may be due to the higher costs (i.e., mortality) associated with

moving among regions than when moving within the regions.

Patch Size and Distance Between Patches as Factors Driving Movement

Our modelling approach provided supportive evidence that patch size and interpatch

distance constitute important factors influencing movement of juveniles at the regional scale.

The support for this hypothesis was weak when movement was modelled for juveniles and adults

simultaneously (Table 3-1.a). However this hypothesis received substantially more support when

juvenile movement was modelled separately (Table 3-1.b. and Table 3-1.c.). Our results are thus

consistent with Prediction 2, which predicts that movement probabilities between regions on a

monthly scale decrease with distance. The hypothesis that immigration should be higher toward

larger patches because of more frequent encounters with patch boundaries (see Prediction 2;

Lomolino, 1990; Hanksi, 2001) received some limited support. Indeed, model qyjuv (AR+ d)

(Table 3-1.b), which supported the hypothesis of a positive relationship between movement and

the size of the receiving site had a w of 0.17 (Table 3-1.b). Although the model that assumed









higher emigration out of smaller areas for juveniles (qyjuv (AD)) was not parsimonious (Table 3-

1.b), examination of the /f parameter for this model supported this hypothesis.

The fact that we only found evidence of a patch size and distance effect on the monthly

movement probabilities of the juveniles at the regional scale is consistent with Prediction 3.

However, we can only infer that juveniles may respond to distance and size of the destination

site, whereas adults do not (Table 3-1.c.), possibly indicating that adults are responding to other

factors (e.g., habitat quality). Only by including a measure of habitat quality (currently

unavailable) in our models could we test the hypothesis that adult movements are more likely to

be determined by the acquired knowledge of the quality of the available habitats than by the

patch boundary effect. The fact that we found no influence of patch size and distance on monthly

movement among contiguous or moderately isolated wetlands can be explained by the fact that

movements among these wetlands are so frequent that the effect of patch size effect and distance

may be diluted over time (i.e., after a few months birds may not search wetlands blindly

anymore).

If patch size and distance affect movement patterns among patches, one can see how

habitat loss and fragmentation may affect dispersal, particularly for juvenile birds. It is

particularly likely to increase the search cost when animals move to locate new suitable

wetlands.

Inter-annual Pattern of Movement

Despite relatively high average monthly movement probabilities out of regions (e.g.,

average movement probabilities out of E and K were 0.04 and 0.09, respectively), kites exhibited

strong philopatric tendencies to particular regions at an annual scale (e.g., annual estimates of

site tenacity for regions E and K were 0.95 and 0.72, respectively).









This extent of site tenacity is surprising given the high environmental variability that

characterizes the kite's range in Florida (Beissinger, 1986; DeAngelis and White 1994). Indeed

many species that use environments where food resources vary strongly in space and time are

often nomadic (e.g., DeAngelis and White 1994). However movement out of familiar areas may

incur important search costs (starvation, predation). Kites may also benefit from staying in or

returning to familiar regions, as it could contribute to maximizing their breeding output and

chance of survival (e.g. predation avoidance) (Stamps 2001).

In summary, kites movement in this fragmented system varies from site tenacity (between

breeding season and at the regional scale) to nomadism (within region on a monthly scale),

depending on the spatio-temporal scale of observation and hence on the activities of primary

relevance at different times and places. In particular, one may want to distinguish between

breeding (or natal) philopatry and exploratory movements, as the factors governing these

processes may be different. Additionally, our results indicate that Snail Kites move substantially

less between regions that have been isolated by human-induced fragmentation than within these

regions. Thus, many kites may have little familiarity with wetlands located outside their natal

region. A regional disturbance could therefore have significant demographic consequences. Kites

that are familiar with many landscapes within the population's range may survive a regional

drought by moving to other less-affected regions, while survival of birds without knowledge of

alternative wetlands could be dramatically reduced. The drought that occurred in Florida in 2001

provided an opportunity to evaluate the effects of this type of natural disturbance on kites.

Regional Survival and Resistance of the Population to Natural Disturbance

The analysis of annual movement indicates that kite movement was affected by the 2001

drought. As expected, a proportion of birds moved from the most to the least-impacted regions,

which is consistent with Prediction 4.a. (but the drought effect was only significant for kites









moving from O to K). Although models including an age effect on annual movement were not

supported (possibly because of low sample size), no juveniles that had fledged one year prior to

the drought were found to have moved toward refugia (i.e., only adult birds were observed

moving to region K in 2001). This latter observation is not based on any robust estimation

procedure and therefore should be interpreted with caution. However, it is worth pointing it out

as it supports Prediction 4. b, which states that because adults are more familiar with the

surrounding landscapes they are more likely to reach refugia habitats than juveniles.

Despite the fact that a proportion of kites moved from the most to the least-impacted

regions, most birds did not appear to successfully reach refugia habitats and overall, this regional

drought had a substantial demographic effect on the population (Figure 3-2), which is consistent

with Prediction 5.a. The survival analysis conducted over the last 13 years, at the scale of the

whole population, also indicates that apparent survival varied among regions. During non-

drought, adult survival was lower in northern regions (K and J) than in southern regions (E and

O), possibly because of lower apple snail availability in the northern regions (Cattau

unpublished). Juvenile apparent survival was also lower in northern regions than in southern

regions during non-drought years, but differences were not statistically significant. Our results

supported Prediction 5. b, which predicted that survival should be lower in areas most impacted

by the drought than in areas least impacted. Adult apparent survival in regions E, O and J (lowest

DSV), decreased significantly during the drought, while survival in K (highest DSV) did not

decrease (Figure 3-2). Prediction 5.b. could not be tested for juveniles because of low sample

size. When averaging survival over all the regions apparent survival of adults decreased by 16%

during the drought while juvenile apparent survival dropped by more than 86% during the

drought (Figure 3-2). Thus, the drought had a larger effect on juvenile apparent survival than on









adult apparent survival, which is consistent with Prediction 5.c. Interestingly, adult apparent

survival only decreased significantly between 2001 and 2002, while juvenile apparent survival

had already decreased significantly between 2000 and 2001, indicating that juveniles were also

more susceptible to early effects of the drought. A declining trend in juvenile apparent survival is

also evident in Figure 3-2. However, we had no good apriori reason to expect this trend. It could

be due to stochastic variation or unrecognised variations in wetland conditions. Additionally, we

should note that out of 65 juveniles equipped in 2003 with radio transmitters, 36 were observed

alive between March and May 2004 (Martin et al. unpublished data). Therefore, juvenile survival

between 2003 and 2004 rebounded since the drought to at least 0.55 (detection probability was

not accounted for this estimate).

The dry-down effects of the drought began in mid January 2001; most of the birds that

fledged during the previous breeding season (from the 2000 cohort) were approximately 9

months. Because juveniles are somewhat proficient at capturing snails after only 2 months

(Beissinger 1988), by 9 months these birds should be equally efficient at capturing and extracting

snails. Field observations of kite interactions indicate no dominance of adults over juveniles that

are 4 months or older (Martin et al. unpublished data). The only major difference in foraging

abilities between young and older birds, that we are aware of, would be their respective

familiarity with the landscapes. Adults would potentially have explored more wetlands than

juveniles (Bennetts et al. 2002). This may thus explain the weaker effect of the drought on adults

(see Prediction 5.c).

We note that the survival estimates presented in this study are apparent survival estimates,

indicating that the complement of these estimates includes both mortality and permanent

emigration from the study system. Thus, lower survival during drought could be due to both









permanent movement out of the system and lower true survival due to the drought. It is possible

that some kites moved temporarily to peripheral habitats (typically highly disturbed habitats:

agricultural areas, large canal) during drought. Although these habitats will typically retain more

water than major kite habitats during drought, they are unlikely to be suitable for breeding

activity; thus, when conditions improve, most birds should move back to major wetlands. Hence,

because the Snail Kite population in Florida is assumed to be an isolated population (Bennetts et

al. 1999a) and because the geographic scope of our study encompasses the major kite habitats, it

is unlikely that many kites remained outside the sampled areas for three consecutive sampling

seasons after the drought. Even if substantial temporary emigration into unsampled areas

occurred during drought it would not have biased survival if it was followed by movement back

into the study system when conditions improved.

Conclusions and Conservation Implications

Reducing habitat fragmentation has now become almost a rubber-stamp recommendation

for maintaining populations of many species of terrestrial mammals, insects, and even birds with

reduced dispersal abilities. However, the benefits may be less obvious when dealing with species

able to cover several hundred kilometres in one day and whose daily dispersal abilities exceed

the distance separating patches that have been isolated through fragmentation. As suggested by

previous theoretical studies (e.g., Doak, Marino and Kareiva 1992), we found that considering

scale issues was critical to understanding movement of kites in fragmented landscapes. The case

study of the Snail Kite in Florida also provides an example of how fragmentation could

indirectly affect the persistence of species with great dispersal abilities. As suggested by

Bennetts and Kitchens (2000) and Bell (1991), exploratory behaviours may be important for

many animals to resist periodic low food availability events (such as droughts). Thus, if

fragmentation reduces exploratory movements of kites, it could also reduce resistance of the kite









population to disturbance events. Further work to support this hypothesis may be particularly

critical to conserve this endangered species, but may also be relevant to other avian nomads (e.g.,

waterbirds in Australia, see Roshier et al. 2001).










Table 3-1. Multistate models (with survival and detection probabilities equal to 1) of monthly
movement probabilities (q) of adult (ad) and juvenile (juv) Snail Kites among the five
major regions (E, O, K, L, J) and P (peripheral and matrix areas), based on radio-
telemetry data. These models evaluate the effect of patch size, distance and regional
identity alone on movement probabilities.
Model AAICc w K
a-Movement among regions of juvenile and adult modelled simultaneously
q (r) 0 0.69 30
q ad (r) qjuv(AR*d) 1.7 0.30 43
q ad (r) q juv(AD) 8.2 0.01 40
b-Movement among regions modelled using data from juvenile only
Ijuv(AR*d) 0 0.79 13
qfjuv(AR+d) 3.0 0.17 12
qfjuv(AD) 6.5 0.03 10
qfjuv(AR) 8.8 0.01 11
c-Movement among regions modelled using data from adult only
q ad(r) 0 1.00 30
Notes: AICc is the Akaike's Information Criterion. AAICc for the ith model is
computed as AICc,- min (AICc). w refers to AICc weight. K refers to the number of
parameters. Only models with w > 0.01 are presented (see Table B-l in Appendix B,
for models with w <0.01). "r": region (includes 6 states: E, O, K, J L and P
(peripheral and matrix); "AR": Area of the receiving site; "AD": Area of the donor
site; "d": distance.










Table 3-2. Multistate (with survival and detection probabilities equal to 1) models of monthly
movement probabilities (q) of adult (ad) and juvenile (juv) Snail Kites among
wetlands in the E and K region based on radio-telemetry data. These models evaluate
the effect of patch size, distance, season, wetland identity alone on movement
probabilities.
Model AAICc w K
a-Movement within the E region of adult and juvenile Snail Kites
If (seas*cw) 0 1 20
b-Movement within the K region of adult and juvenile Snail Kites
qf (seas) 0 0.67 3
q ad (seas) q juv(seas) 2.6 0.18 6
S(.) 5.4 0.04 1
qi ad (.) Wqjuv(.) 6.5 0.03 2
q (mw) 6.5 0.03 12
f (breed) 7 0.02 2
q (years*seas) 8 0.01 10
Notes: Only models with w > 0.01 are presented (see Table B-2 in
Appendix B, for models with w < 0.01). "cw": contiguous wetland; "mw":
moderately isolated wetland; "seas": season; "breed": breeding season. For
other notations see Table 3-1.









Table 3-3. Multistate models of annual apparent survival (q), sighting (p), and movement
probabilities (q) of adults (ad) and juveniles (juv) Snail Kites based on banding data.
The drought effect on 0. during 2000-2002 was denoted D1-2. The drought effect on q
in 2001 was denoted D1. Constant ( during non-drought years (1992-2000 and 2002-
2004) was denoted ND. Because all models included in Table 3 had region and time
dependent sighting probabilities (p(r*t)), Table 3-3 only includes model structures for
0 and W.
Model AAICc w K
[ [E-O#K-J] KND) bad) ,[E-OJ]
(ad O a=] ))d ) (D1 2) v(r + td) y(r D) 0 0.96 85
(4[EQO#K-J] ,i [EO J]jd AE.0-J
Id ("= D)no d(.) a O (D-2) Ojuv(r + td) )y(r) 7 0.03 78
Notes: Only models with w > 0.01 are presented (see Table B-4 in Appendix B, for models
with w < 0.01). "t": time (years); "r+td": additive effect of region and time on 0, except
during 2001-2002,during which 0 was time dependent only; ".": 0 is constant during 1992-
2004. Superscript indicate region specific 0,; "=": regions have identical 0; ": regions have
different 0. For other notations see Table 3-1.































E 7 J 6
10 8


0 50 km 9







Figure 3-1. Major wetlands used by the Snail Kite in Florida. Regions: Kissimmee Chain of
Lakes (K), Everglades (E), Lake Okeechobee (0), Saint Johns Marsh (J), and
Loxahatchee Slough (L). Moderately isolated wetlands included in K are: East Lake
Tohopekaliga (1), Lake Tohopekaliga (2), Lake Kissimmee (3), as well as the small
lakes coloured in grey within the rectangle. Contiguous wetlands included in E:
Water Conservation Areas 1A (4), 2A (5), 2B (6), 3A (7), 3B (8), Everglades
National Park (9), and Big Cypress National Preserve (10). The grey colouring of the
wetlands indicates the area of the wetlands that were included in this study. The thick
contour lines delimit regions that include several wetlands. The dotted line indicates
the historic Kissimmee-Okeechobee-Everglades watershed which constituted a
network of well connected wetlands (Davis and Ogden 1994; Light and Dineen
1994).



















1.0 V

Qgs -____^___^-___l-- ^- ---^--~-- __--tL ___^__

0.6

0.4

0.2 Adults in E
-- Adults in K
0.0 -o Juveniles

M 7 LoC r_ M00 a\ C
r^ t<- 11 i i>1 1 oo 1 1 ? i
CA rn t n r- W 00 C)

Year


Figure 3-2. Apparent survival ( ) between 1992 and 2003 of adult and juvenile Snail Kites,
obtained using the most parsimonious model in Table 3. Error bars correspond to
95% confidence intervals. During non-drought years (1992-2000 and 2002-2003),
q of adults were similar in E and 0; and in K and J. During drought (2000-2002), of
adults were similar in E, O and J, but different in K. For readability, only q in E and
K are presented for adults. 0 of juveniles were averaged across regions. Arrow
indicates the beginning of the drought that started in January 2001. Estimates between
1992 and 1999 were consistent with Bennetts et al. (2002).









CHAPTER 4
NATAL LOCATION INFLUENCES MOVEMENT AND SURVIVAL OF THE SNAIL KITE

Introduction

In heterogeneous environments, the place of birth (or natal location) is likely to be critical

to the growth, survival, and future reproduction of organisms. Indeed, resource availability and

other factors (e.g., parasitism) that can affect vital rates (e.g., survival and reproduction) often

vary in space and time (Latham and Poulin 2003; Pettorelli et al. 2003). Moreover, it has been

shown that environmental conditions during prenatal and early post-natal stages can have long

term consequences on key life history traits (e.g., lifespan and age of first reproduction)

(Metcalfe and Monaghan 2001). Natal location, by shaping the habitat preferences of many

animals, may also affect their movement and settling decisions during the course of their

lifetimes (Stamps 2001; Davis and Stamps 2004). Thus, natal location may influence the

ecological dynamics of wild populations through its effects on movement, habitat selection and

survival (Stamps 2001; Blums et al. 2003; Davis and Stamps 2004).

Despite the importance of better understanding how natal location affects movement

decisions and survival, rigorous analyses to estimate movement and survival in relation to the

place of birth are lacking (Blums et al. 2003). Most past studies used ad hoc methods to estimate

site fidelity (e.g., return rates, reviewed in Doherty et al. 2002). These ad hoc measures can be

severely biased, because they are also functions of resighting and survival probabilities (Doherty

et al. 2002). Modern analytical techniques, such as multistate models provide more robust

estimates of movement and survival (Doherty et al. 2002). Although a few studies have now

successfully used multistate models to estimate site fidelity (typically to breeding sites)

(Hestbeck et al. 1991; Lindberg et al. 1998), only a handful of studies have used these models to

estimate philopatry to the natal site (Lindberg et al. 1998). In fact, we are not aware of any study









that used robust estimators of movement to compare fidelity to the natal site (at the adult stage)

with fidelity to non-natal sites. Nonetheless, the fact that some animals may exhibit a particular

attraction toward their natal site at the adult stage could be critical to the dynamics of many wild

populations. Indeed, attraction by individuals to the natal site could influence habitat selection,

patterns of patch occupancy, reproduction, and survival (see Schjorring 2002).

We studied patterns and consequences of movement related to the natal site in a

geographically isolated and spatially structured population of Snail Kites (Rostrhamus sociabilis)

in Florida. The Snail Kite is a highly specialized raptor that feeds almost exclusively on

freshwater snails (Beissinger 1988). Because of this food specialization, Snail Kites are confined

to the remaining wetlands in Central and South Florida (Takekawa and Beissinger 1989). The

environment occupied by the Snail Kite in Florida is highly variable spatially and temporally

(Beissinger 1986). Animals such as Snail Kites that occupy habitats whose food abundance is

unpredictable in space and time are generally expected to be nomadic (Bennetts and Kitchens

2000; Wiens 1976). By modeling movement rates among wetlands used by this bird, Bennetts

and Kitchens (2000) measured the extent of nomadism. However, Martin et al. (2006) found that

despite frequent exploratory movement, Snail Kites showed strong site tenacity on an annual

scale during the breeding season. Bennetts and Kitchens (2000) hypothesized that exploratory

movement may familiarize kites with their landscapes. It has been recognized that exploring

unfamiliar habitats is energetically costly (Schjorring 2002). However, if the exploration phase

occurs at times when search costs are minimal (i.e., when food is most abundant), the familiarity

with alternative locations gained by birds during exploration may greatly outweigh exploration

costs by reducing the risk of mortality during a subsequent disturbance, such as a regional









drought, which dramatically decrease snail availability to kites (Beissinger 1995; Bennetts and

Kitchens 2000; Martin et al. 2006).

On the other hand, benefits of staying in or returning to a familiar area range from

increased competitive ability to enhanced predator avoidance (Stamps 2001). Thus staying in or

returning to a familiar habitat may also, in some circumstances, increase an animal's probability

of surviving and/or of breeding successfully. Although Martin et al. (2006) found that kites

exhibit site fidelity (i.e., philopatry) to certain regions during the breeding season, these authors

did not specifically examine the level of site tenacity relative to the natal site. Nonetheless,

distinguishing between site fidelity specific to the natal site as opposed to site fidelity to non-

natal sites, and examining consequences on survival, may be essential to disentangle the

ecological dynamics of many vertebrates.

Hypotheses and Predictions

Prediction 1

If kites prefer their place of birth when compared to any post-dispersal sites they may have

explored in the course of their lifetimes, we expect: (1) movement from post-dispersal sites

toward birds' natal site to be greater than movement from post-dispersal sites toward non-natal

sites; (2) we expect greater movement toward the natal site than away from the natal site; (3)

finally, we predict that philopatry to the natal site should be greater than philopatry to non-natal

sites. Here, we define philopatry as the probability for a kite to be found in a particular region at

year t+ given that it was present in that same region in year t (Martin et al. 2006). Thus,

philopatry to the natal site (or natal philopatry) corresponds to the probability for a kite to be

found in its natal region at year t+ given that it was present in its natal region in year t.

Prediction 2









Wetland conditions affect survival (Beissinger 1995; Bennetts and Kitchens 2000; Martin

et al. 2006) and movement (Takekawa and Beissinger 1989; Bennetts and Kitchens 2000; Martin

et al. 2006) of Snail Kites. During a regional drought a proportion of Snail Kites is likely to

move to the least disturbed areas (Takekawa and Beissinger 1989; Bennetts and Kitchens 2000;

Martin et al. 2006). Consequently, we expect the level of natal philopatry to vary according to

wetland conditions. Kites should be less philopatric to their natal site when it is affected by a

drought. Conversely, during a drought, natal philopatry of birds whose natal areas are located in

refugia habitats (areas least affected by droughts) should be greater than during non-drought

years. Because of their greater familiarity with the paths linking post-dispersal habitats to their

natal habitat, birds hatched in a refugia habitat should also have higher probabilities of moving to

that refugia during a drought.

Prediction 3

If natal location influences adult survival, we expect adult survival to vary substantially

among groups of kites that were hatched in different regions.

Prediction 4

Because we expect kites to be more likely to stay in or return to their natal location (see

Prediction 1), we predict that during a drought, kites whose natal site is located in refugia

habitats should be less impacted by a drought (i.e, their survival should decrease less) than birds

whose natal site is located outside of a refugia habitat.

Study Area

We sampled four major wetland complexes that encompass a very large proportion of all

the landscapes used by the Snail Kite in Florida (Figure 4-1). Most of these wetland complexes

(hereafter referred to as regions) consist of several wetlands that were separated by small

physical barriers (e.g., road, levee or limited extent of non-wetland areas) easily crossed by Snail









Kites (Bennetts 1998). On the other hand, each region was isolated from the others by extensive

areas mostly unsuitable for breeding or foraging (i.e., matrix), and not easily crossed by kites.

We used the same four primary regions as Bennetts et al. (1999a): Everglades "E"; Kissimmee

Chain of Lakes "K"; Lake Okeechobee "L"; Saint Johns "J" (Figure 4-1).

Material And Methods

Field Methods

Capture-mark-recapture

The Snail Kite population in Florida has been monitored since 1992 using capture-mark-

recapture methods (Bennetts et al. 1999a). Because our study focused on movement and survival

related to the natal region of Snail Kites, we only included birds banded as juveniles (before

fledging, at approximately 30 days), whose natal regions we knew (sample size: 1722 birds).

Those birds were captured and marked directly at the nest during the peak of the breeding season

between 1992 and 2004. Birds were then resighted during annual surveys (which also took place

during the peak of breeding season: March through June). Each region was surveyed at least

once using an airboat, and bands were identified using a spotting scope.

Data Analysis

Multistate modeling

Multistate models (Hestbeck et al. 1991; Williams et al. 2002) simultaneously estimate

transition probabilities among geographic states T", sighting probabilities "p", and apparent

survival probabilities (hereafter simply referred to as survival). We defined VQS as the

probability that an animal in region "Q" at time t is in region "S" at time t+ 1, given that it is

alive at t+1. In this example, regions "Q" and "S" are geographic states. We defined "pQ"1 as the

probability that an individual alive in region "Q" in year t is sighted (Williams et al., 2002). We









defined Q'" as the probability of surviving (and not permanently emigrating from the study

system) over the interval [t, t+ 1] for a kite alive in year t in region "Q" (Williams et al., 2002).

Modeled parameters used notation from Martin et al. (2006); time dependency was "t" and no

time effect was ".". We assigned each bird to one of two age classes: juveniles (denoted "J"), 30

days to one year in age; and adults (denoted "AD"), older than one year. Effects embedded in

parameters (i.e., ", T" and "p") were shown using parentheses. A multiplicative effect was

denoted "*". Additional effects included: (1) a drought effect; (2) an age effect denoted "age";

(3) a natal region effect denoted "nr"; and (4) a regional effect denoted "r". The regional effect

"r" allowed parameters of interest (i.e.," ", "T" or "p") to vary among regions. Regions

included both natal and non-natal regions. We used multistate models with four groups (each

group included birds hatched in the same natal region: "E", "L", "K" or "J"); and four

geographic states (each state corresponded to a region: "E", "L", "K" or "J") to estimate

movement and survival of Snail Kites. It is important to recognize the distinction between groups

and geographic states. A particular kite can belong to only one group (which corresponds to its

natal region). In contrast, a kite can move from one geographic state to another. A geographic

state corresponds to a region occupied by a kite at some point in time.

Movement

First, we used a classic parameterization of multistate models which ignored any effects of

natal regions (e.g., Hestbeck et al. 1991, Blums et al. 2003, Martin et al. 2006). For instance,

model "v(r)" allowed transitions to vary among regions (i.e., geographic states) but assumed no

natal region effect. We used an alternative parameterization (hereafter referred as

parameterization "NOA") that allowed us to develop models that estimated three types of

transition probabilities for each group of kites hatched in the same region (each group was









exclusively comprised of birds from the same natal region). The first two types of probabilities

were transitions between the natal region (denoted "N", for "natal") and post-dispersal regions

(denoted "O", for "other"). Post-dispersal region "0" potentially included any of the four major

regions used by kites except for their natal region. The notation NO,, corresponded to transition

probability from region "N" to "O", while ON,, corresponded to transition probability from

region "0" to "N". In addition to ON" and No", we were able to estimate transition OA"

transition from region "0" (occupied at time t) toward any post-dispersal region "A" (at time

t+ ). "A" included any regions except for the natal region and the post-dispersal region occupied

at t while in region "O". Next, we provide an example of the constraints that were used to

estimate ON", No", and OA". For example, for all kites that were hatched in region "E",

we developed models that assumed: (1) EL EK EJ NO (subscripts indicate the natal

region; in this example all transitions probabilities pertain to kites that were hatched in region

"E"); (2) LK LJ= KL KJ JL JK OA,,. and (3) "LE KE JE ON ". We used similar

constraints to model movement of kites hatched in the other three regions. These constraints

were imposed in order to compare models that formalised our apriori hypotheses (see

HYPOTHESES AND PREDICTIONS). We developed four different types of models using the

parameterization "NOA": (1) w[NO'ON'OA](nr) (which assumes No", ON" and T OA" to be

different and to vary among natal regions); (2) W[NO=A'ON](nr) "; (3) w[NO=ONOA](nr)"; and (4)

" [NO=N=OA](nr) ". These four types of models were developed to evaluate whether NO,, ON"

and OA,' differed substantially within each natal region. In addition, we developed a model

that included an interaction between a natal region effect ("nr") and a region effect ("r") on

movement. This model was denoted W(r*nr)". Finally, because a drought occurred between









January and August 2001, we evaluated models that included a drought effect on T" between

2000 and 2001 (see also Martin et al. 2006). This effect was denoted "Dm". We developed a

model which assumed that NO", vON" and OA" differed among groups of birds hatched in

different natal regions but also differed during drought and non-drought years. This model was

denoted i[NOON'OA](nr Dm) ". We also considered a model that included an interaction between

"nr "r" and "Dm". This model was denoted "W(r*nr*Dm".

Based on model averaged estimates (see "MODEL SELECTION") of movement

parameters of all models described above, we derived natal philopatry (denoted NN; the

probability for a kite to be found in region "N" at time t+1, given that it was present in "N" at

time t); and philopatry to non-natal site (denoted oo"; the probability for a kite to be found in

region "O" at time t+1, given that it was present in "O" at time t; in other words Too" is the

probability that a kite located in a particular non-natal region at time t will be found in that same

non-natal region at time t+ 1). These estimates of philopatry were obtained as one minus the

estimated rates of moving away from the area of interest, and the associated variance of these

derived estimates was computed using the delta method (see also Hestbeck et al. 1991; Williams

et al. 2002).

Survival

We developed models that allowed "4" and "p" to vary across time or remain constant for

each age class. Given that environmental conditions are similar among certain groups of regions

as a result of spatial proximity, similar topography, similar latitude or management (see Bennetts

1998), we developed models that assumed similar apparent survival probabilities across groups

of regions. Thus, we developed models based on proximity (see Figure 4-1) that evaluated the

similarity of survival in the northern regions "K" and "J". Even though the boundaries of"E"









and "L" are closer together than "L" and "J", the distance between the centers of"E" and "L" is

greater than the distance between the centers of"L" and "J" (Figure 4-1). Therefore, we

considered models allowing survival in "L" to be different from "K", "J" and "E". An analysis of

hydrological data by Bennetts (1998) showed that wetland conditions in "E" and "L" were the

most positively correlated among all the regions. We therefore developed a model that assumed

apparent survival to be similar in these two regions. To illustrate our notation system we chose a

model with a common survival parameter for two groups of natal regions: nr[E=L#K=J] ". The

superscript "nr" indicates that ")" is "natal region specific". In models with the superscript "nr",

survival was allowed to vary among natal regions (i.e., groups) but it was constrained to be equal

among regions occupied (i.e., geographic states). The superscripts on the right of model

" nry[E=LK=J]", indicate regions to which the survival probabilities pertain ("=" indicates that

" nrE is the same as nrL "; similarly nrK is the same as n J"; indicates that nr E"


and nrL are different from nrK and nr J"). We also constructed models in which adult

survival varied among regions; we referred to these survival rates as "region specific" survival.

In models that estimated "region specific" survival, survival parameters were allowed to vary

among regions occupied (i.e., geographic states), but were constrained to be equal among natal

regions (i.e., groups). "Region specific" survival estimates were obtained using the same

parameterization as Martin et al. (2006) and were denoted by a superscript "r" instead of"nr"

(e.g., 'r[E=LK=J]",). In age structured models, adult survival was denoted 4A and juvenile

survival was denoted (j ". We also created models that included a drought effect on ")". As in

Martin et al. (2006), the drought effect on survival was modeled as a two years effect and was

denoted "D". Conversely, a no drought effect on ")" was denoted "ND". We used this approach









because the drought was likely to affect ")" before and after the 2001 sampling occasion. We

used subscripts on the right of ")" to indicate whether the model structure of a particular

component of ")" pertained to drought years subscriptt d reflected survival during the interval

2000-2002) or non-drought years subscriptt nd reflected survival during the interval 1992-2000

and 2002-2004). For example, the first part of model r[E *L*KJ] d(ND) r[ELJ](D)


indicates that during non-drought years r "rL and r differed

substantially while remaining constant over time. The second part of this model (i.e.,

" D,d(ND) ") indicates that during the period that corresponded to drought years (2000-2002)

there was no drought effect on "r ". The third part of the model (i.e., r[E*L* (D) ") indicates
AD AD, dA

that during drought years there was an effect of the drought on rE ", "r ", and r ". The

symbol between "E", "L" and "J" indicates that r E ",r and r-J were different

during the drought. Because the drought intensity was strongest in "E", "L" and "J", and weakest

in "K" (see "WETLAND CONDITIONS"), some models assumed similar drought effects on

"" in "E", "L", and "J" (e.g., ELJ(D) ") with no drought effect on "r," in K (e.g.,


" (D,d(ND) ") (see also Martin et al. 2006). The absence of subscript "nd' or "cd' indicated that

the survival term reflected the entire period of study (1992-2004). For example model

" AD (r nr)" assumed a multiplicative effect between "r" and "nr" on adult survival for the entire

period of study (1992-2004). This model did not assume any drought effect on adult survival.

Model 4D (r nr D)", on the other hand, assumed a multiplicative effect between "r", "nr" and

"D".









Model Selection, Goodness of Fit and Program Used

We developed a set of models that corresponded to our best apriori hypotheses. Next we

used Akaike information criterion adjusted for small sample size (AICc) (Burnham and

Anderson, 2002) as a criterion to select the model that provided the most parsimonious

description of the variation in the data (i.e., model with the lowest AICc). Models with a value of

AAICc (the difference between the AICc of a particular model and that of the model with the

lowest AICc) less than two were considered to receive a substantial level of empirical support

(Burnham and Anderson 2002). We also used the AICc weight (w) as a measure of relative

support for each model (Burnham and Anderson, 2002). Values ofw range from 0 to 1 (with 0

indicating no support; and 1 indicating maximum support). The sum of weights of all models

including a particular effect was denoted (wt). As recommended by Burnham and Anderson

(2002), we used model averaging to compute estimates of movement and survival. All

computations of the movement and survival rates were carried out using program MARK V 4.1

(White and Burnham 1999). We used the sin link function available from program MARK.

Estimates of standard error (SE) and 95% confidence intervals (95%CI) were obtained directly

from program MARK. We used the delta method to compute estimates of precision for derived

parameters (Burnham and Anderson 2002). All the confidence intervals for the derived estimates

" in our study were approximated as follows: 95%CI [0 ]= 6 + 1.96* SE [0 ]. Goodness of

fit (GOF) tests of multistate models were computed using program U-CARE version V2.22

(Pradel et al. 2003; Choquet et al. 2005). Program U-CARE test the fit of the fully time

dependent "Jolly Move" model (JMV) to the data. To our knowledge there is currently no

rigorous test that directly assesses the fit of the Arnason-Schwarz (AS) model that accounts for

an age effect on survival (" 4(age r t) \(r t) p(r t) "). However, program U-CARE tests the fit









of the "Jolly-move model" (JMV) that accounts for an age effect on ")" (hereafter referred as

model JMVA) (Choquet et al. 2005). This test is also generally valid for the AS model that

accounts for an age effect on survival (Cooch and White 2006). Indeed, in practice the JMV

model is unlikely to show significantly better fit to the data than the AS model (Cooch and White

2006). The JMV model differs from the AS model in that it allows the capture probability for

time t+ to depend on the state at periods t and t+1, whereas the AS model only allows the

encounter probability to depend on the current state and time. However, the dependence of the

capture probability at time t+ on the state at periods t and t+ is unlikely to be often observed

in practice (Cooch and White 2006). The test for the JMVA model requires the summation of the

component WBWA, 3G.Sm, M.ITEC, M.LTEC (see Choquet et al. 2005, for a detailed

description of the procedure). If the test is significant (i.e., P < 0.05), Choquet et al. (2005)

recommend a correction for overdispersion.

Effect Size

We estimated the magnitude of the difference between two estimates of movement (or

survival) by computing the arithmetic difference between these estimates (hereafter referred as

"effect size" denoted ES"). Whenever the 95%CI[ ES] did not overlap 0, we considered the

difference to be statistically significant (Cooch and White 2006). We note that estimates of

95%CI[ ES] take into consideration covariances between estimated movement and survival rates.

Therefore, differences between estimates may be statistically significant based on 95%CI[ ES],

even though 95%CI of the estimates to be compared overlap.

Notes Concerning Regional Specific Survival

We note that the parameterizations used in the present study to estimate "region specific"

survival were similar to the ones described in Martin et al. (2006). Therefore estimates of "region









specific" survival should theoretically not differ substantially between the two studies. Any

differences between these estimates should primarily be due to the fact that the data set used in

the present study included exclusively kites whose natal locations were known. In contrast, all

models assessing the influence of natal location on movement and survival used

parameterizations that differed substantially from any models presented in Martin et al. (2006).

We emphasize that comparing estimates of "region specific" survival (parameterizations used by

Hestbeck et al. 1991; Martin et al. 2006 and other authors) with estimates of "natal region

specific" survival (parameterization specific to this paper), was necessary to fully evaluate the

importance of natal location on regional survival of adult birds. It was also important to make

these comparisons using a common data set that only included individuals whose natal locations

were known (as opposed to using the data set analyzed by Martin et al. 2006). Model averaged

estimates of "region specific" survival were obtained by model averaging estimates from models

that assumed survival to be "region specific" whereas model averaged estimates of "natal region

specific" survival were obtained by model averaging estimates of models that assumed survival

to be "natal region specific".

Wetland Conditions

Martin et al. (2006) determined wetland conditions during drought using hydrological data.

Their results suggest that "E", "L" and "J" were the regions most affected by a drought that

occurred between January and August 2001 and "K" was the least affected. Martin et al. (2007a)

and Bennetts (1998) found high levels of spatio-temporal variation in wetland conditions. Not

surprisingly they found a positive correlation in water levels among wetlands that were located

nearby. The coefficient of correlation decreased as distance increased. Martin et al. (2007a.) also

found that during most multiregional droughts that affected region "E" (e.g., 1981, 1985, 1989,









1990, 1991, 1992 and 2001) at least one of the wetlands within region "K" was wetter (based on

hydrological indicators) than any other wetland within the region "E" (Martin et al. 2007a).

Results

GOF Tests

The GOF test indicated that model JMVA fit the data satisfactorily (X2 137 = 82, P > 0.99).

This test is also generally valid for the most general model in our model set (i.e.,

" (age r* t) \(r t)p(r* t) "). Therefore, there was no need to correct for overdispersion.

Movement

The two most parsimonious models included the component I[NO ONOA](nr) (wt = 0.54,

Table 1). These models assumed movement probabilities between "N" and "0" (" WNO,);

between "0" and "N" (" ON,,); and between "0" and "A" (" OA,,) to be different for each natal

ON,, NO,, OA,,
region. Model averaged movement estimates ON" were greater than either or ^

This was true for all natal regions (Figure 4-2). Differences between ON," and "vA,, were

statistically significant for regions "E" (ES = 0.45, 95%CI = 0.27 to 0.62), "K" (ES = 0.14,

95%CI = 0.04 to 0.25), and "J" (ES= 0.21, 95%CI = 0.05 to 0.38), but not for region "L" (ES=

0.05, 95%CI = -0.09 to 0.18). Differences between ON" and NO" were statistically

significant for regions "E" (ES= 0.46, 95%CI = 0.29 to 0.63), "K" (ES= 0.10, 95%CI = 0.01 to

0.19), and "J" (ES= 0.17, 95%CI = 0.02 to 0.33), but not for region "L" (ES= 0.09, 95%CI = -

0.02 to 0.19).

Models T[NOOA#ON](nr Dm)" received less support from the data than model

" I[NOQOAON]((nr)" (AAICc = 14.2; see also Appendix C). Similarly models [NO=OA'ON](nr Dm)"

received less support from the data than model VI[NO=A'ON](nr)" (difference in AICc between









these two models was 6.8, Table 1). Therefore the hypothesis of a drought effect on movement

received little support from our data. Models [NO'NOA(nr) ", I[NO=ONOA](nr) ",

c" I[NO=ONOA](nr) ", "r)", v(t)" and v(.)" received considerably less support based on AICc


weights (w 0) (see also Appendix C). Models V[NO'ON'OA](nr age) ", v(nr r) ", and

'" V(nr r Dm)" did not reach numerical convergence when optimizing the likelihood.

Comparison of Natal Philopatry and Philopatry to Non-Natal Site

Estimates of natal philopatry were greater than estimates of philopatry to non-natal regions

(Figure 4-3). The differences between these estimates were statistically significant for regions

"E" (ES= 0.49, 95%CI = 0.31 to 0.66) and "L" (ES= 0.16, 95%CI = 0.02 to 0.31), but not for

regions "K" (ES= 0.01, 95%CI = -0.12 to 0.13), and "J" (ES= 0.10, 95%CI = -0.07 to 0.26).

Survival

Adults

Models that measured "region specific" survival received more support than models that

measured "natal region specific" survival (AAICc > 8.4, Table 1). Models that assumed "region

specific" survival to be similar for regions "K" and "J" in the north and "E" and "L" in the south

during non-drought years (e.g., [E J=K](.)") were among the most parsimonious models (wt

= 0.74, Table 1). Models that assumed no drought effect on region "K" were well supported (wt

S1, Table 1), as were models that assumed similar drought effect for regions "E" and "L" (wt ~

0.99, Table 1). Figure 4-4 shows a substantial decrease in survival in regions "E" and "L". On

the other hand, the effect of the drought on region "J" was not clear. Indeed, the AAICc was only

1.7 between the most parsimonious model (i.e.," r Ej=L =K] ( > =J ND) r[ (D) "), which

assumed no drought effect on "J" and the second most parsimonious model which assumed a









drought effect on "J" (i.e.," yELJ=K]( r Dd(ND) rE-LJ](D) ", Table 1). Models D(r *nr)",


, AD(r*nr*D), r1 ,1 1 ) 1 1 r r 1 ND) 1 [E-L=Jr=K] T [ K] (ND)

received almost no support based on AICc weights (w 0) (see also Appendix C). Among

models that estimated "natal region specific" survival, model

" nA[EDLnd JK]) ([K-Jo nLD) was the most parsimonious, and model
nr ,[E =LnJ =K] nr ,t[K=J], nr ,[

nr And [LK] nr ](N) (D) was the second most parsimonious (difference in AICc


between these two models was 2.5, see Table 1).

As pointed out earlier, models that assumed "region specific" survival were better

supported than models that assumed "natal region specific" survival. However, model averaged

survival estimates for these two types of models were similar (ES < 0.06, Figure 4-4), except for

region "J" (ES = 0.14, but the difference was not statistically significant 95%CI = -0.02 to 0.30,

Figure 4-4). Estimates of "natal region specific" survival varied significantly among groups of

kites hatched in different regions (Figure 4-4). During non-drought periods "natal region

specific" survival estimates were not significantly greater in "E" than in "L" (ES = 0.031, 95%CI

= 0 to 0.06, Figure 4-4). "Natal region specific" survival estimates were greater in "L" than in

"K" (ES= 0.08, 95%CI = 0.04 to 0.12, Figure 4-4); and were greater in "L" than in "J" (ES=

0.08, 95%CI = 0.04 to 0.12, Figure 4-4). Overall, during non-drought years, "natal region

specific" survival estimates were significantly greater in the southern regions ("E" and "L") than

in the northern regions ("K" and "J") (ES = 0.11, 95%CI = 0.05 to 0.16, Figure 4-4). During the

time interval when the drought most severely impacted adult survival (2001-2002), "natal region

specific" survival in the southern regions ("E" and "L") was significantly lower than in northern









region "K" (ES= 0.35, 95%CI = 0.23 to 0.48, Figure 4-4); it was also lower than in northern

region "J" (ES= 0.35, 95%CI = 0.19 to 0.52, Figure 4-4).

Juveniles

Models that assumed juvenile survival to be time dependent (i.e., ~(t) ") received the

most support (wt 1, Table 4-1). Models 4j(nr t)", "j(nr D)" and j(.)" received almost no

support (w 0, see also Appendix C).

We do not present juvenile survival estimates because they were not the focus of the

current analysis. Furthermore these estimates were almost identical to the ones published in

Martin et al. (2006).

Detection Probabilities

Models that assumed a multiplicative effect of region and time on detection (i.e., "p(r*t)")

received the most support (wt 1, Table 4-1). Models "p(t)"; "p(age*t)"; "p(age*r*t)"; "p(r)",

"p(.)" received almost no support (wt 0, see also Appendix C).

Discussion

Effect of Natal Region on Movement

This study, which focuses on movement and survival related to the place of birth, shows

that Snail Kites, in addition to exhibiting some high level of site tenacity to most regions during

the breeding season (Martin et al. 2006), also exhibit a particular attraction for their natal region.

Indeed, using multistate models with four geographic states we found that estimates of

movement from post-dispersal regions toward birds' natal region were greater than movement

from post-dispersal regions toward non-natal regions (differences were statistically significant

for regions "E", "K" and "J", Figure 4-2). We also found that estimates of natal philopatry were

greater than estimates of philopatry to non-natal regions (differences were statistically significant









for regions "E" and "L", Figure 4-3). Finally, we found that estimates of movement were greater

toward the natal region than away from the natal region (differences were statistically significant

for regions "E", "J" and "K"). These findings provide evidence supporting Prediction 1. The

extent of affinity to the natal region appeared to depend on the region (Figure 4-2 and Figure 4-

3). Interestingly, our results show that it may be worthwhile to compute simultaneously the three

estimates of natal attraction described in this study. The fact that the effect of natal location on

movement was not statistically significant for all regions when using all estimators may be due

to lack of statistical power (e.g., due to small sample size). Alternatively, one should note that

even if movements toward natal sites are substantially greater than toward non-natal sites, when

movements from post-dispersal sites toward non-natal sites are small, estimates of natal

philopatry may not differ from estimates of philopatry to non-natal sites (Figure 4-2 and Figure

4-3).

Fidelity to the natal region may benefit kites in numerous ways including reducing the

risks associated with movement. Movement often incurs costs (Baker and Rao 2004), it is

energetically costly (Schjorring 2002) and it can increase the risks of predation (Stamps 2001).

Furthermore, familiarity with a particular region may potentially increase the fitness of kites if

performance of activities related to foraging or reproduction can be improved through training

and learning of habitat features specific to a particular region (Stamps 2001).

On the other hand, exploring habitats away from the natal region may be critical to the

survival of most kites, given the unpredictability of the system used by kites (Beissinger 1986;

Bennetts and Kitchens 2000; Martin et al. 2006). A combination of exploratory movement

during the non-breeding period when search costs are relatively low, and a natal-philopatric type









of behavior when birds begin reproductive activities, is an appealing hypothesis to explain the

movement of kites in Florida, but this hypothesis remains to be tested.

Models that included a drought effect on movement were not well supported based on

AICc, and therefore, did not support Prediction 2. This contrasts with a previous study by Martin

et al. (2006) that indicated evidence of a drought effect on movement. The discrepancy between

the two studies may be due to a difference in sample size. In the present study we used only data

on kites for which the natal location was known, which significantly reduced our sample size,

and therefore, our ability to detect drought effect on movement. Given previous findings that

most droughts affect regions differentially, and given the fact that the probability of staying or

returning to a particular region is greatly dependent on the region of birth, we can expect great

influences of the natal region on kite survival. Next we examine effects of the natal region on

kite survival.

Influence of Natal Region on Survival

We found that estimates of adult survival varied substantially among groups of kites that

were hatched in different regions (see "natal region specific" survival, Figure 4-4). This result is

consistent with Prediction 3. When comparing models that assumed "natal region specific"

survival with models that assumed "region specific" survival, "region specific" models received

more support from the data. However, estimates of "region specific" and "natal region specific"

survival were very similar (Figure 4-4). Thus, our results suggest that kites experience survival

rates that are characteristic of the region occupied (i.e., geographic state) when survival is

measured, but because kites have a tendency to stay in or return to their natal region more than to

any other regions, as a consequence adult survival is ultimately influenced by the natal region.

Interestingly, during most years, adult survival rates of birds hatched in the southern

regions (i.e., "E", "L") were higher than for birds hatched in the northern regions (i.e., "K" and









"J" regions). The difference in survival rates between birds hatched in various regions may be

explained by varying quality among habitats. For instance, snail availability to kites appear to be

lower in the northern regions during non-drought conditions (Cattau unpublished data, also

reviewed in Martin et al. 2006).

During the drought that occurred in 2001, adult birds hatched in "E" and "L" suffered high

mortality rates, while birds hatched in "K" were substantially less affected (Figure 4-4). This

finding is consistent with Prediction 4. Martin et al. (2006) found that during a drought kites

moved from regions most affected by the drought toward regions least impacted by the drought.

In the current study, we did not find any evidence of a drought effect on movement (possibly

because of low sample size). Nonetheless, we found that kites exhibit a particular attraction to

their natal region (Figure 4-2 and Figure 4-3). Therefore, at any one time, kites are likely to be

found in their natal region. Thus, many of the birds hatched in "K" may have survived the

drought by staying in region "K", or by moving back to natal region "K" (which as reflected by

the hydrological indicator was the region least impacted by the 2001 drought). Indeed, we expect

birds that were hatched in "K" to be more likely to reach that region than birds hatched in other

regions because of their potentially greater experience with the path linking region "K" to their

post-dispersal regions. The impact of the drought on "J" remains unclear. On one hand, model

averaged estimates of "natal region specific" survival estimates did not indicate any substantial

decrease during the drought for region "J" (Figure 4-4). On the other hand "region specific"

survival indicated a decrease during the drought although not as great as in regions "E" and "L"

and this decrease was not statistically significant (Figure 4-4).

Interestingly, during most multiregional droughts for which kite distribution data were

available (1981-1982, 1985 and 2000-2001), many kites appeared to use region "K" as a refugia









habitat (Martin et al. 2006; Takekawa and Beissinger 1989), suggesting that this region is

generally less impacted than the major southern regions (i.e., "E" and "L"). This is also

confirmed by hydrological data for the last 30 years (Martin et al. 2007a). Indeed, during most

multiregional droughts which affected region "E" (e.g., 1981, 1985, 1989, 1990, 1991, 1992 and

2001), at least one of the wetlands within region "K" was wetter (based on hydrological

indicators) than any other wetland within the region "E" (Martin et al. 2007a).

Thus, birds hatched in region "K" may have lower survival and reproduction rates for

many years, but may be more resistant to multiregional disturbance events, which are believed to

be a major cause of mortality among Snail Kites (Beissinger 1995; Bennetts and Kitchens 2000;

Martin et al. 2006).

Conclusions and conservation implications

Most ecologists would probably recognize that the effect of natal location on movement

and survival may have important conservation implications for species which are notorious for

their high degree of natal philopatry (e.g. albatross species, Bekkum et al. 2006). In contrast,

conservation implications may be less obvious when dealing with vertebrate species which

exhibit more subtle patterns of natal philopatry. The Snail Kite which has been described as a

nomad (e.g. Bennetts and Kitchens 2000), may be one of such species. Indeed, we found that

natal region is critical in influencing movement and survival of Snail Kites in Florida and that

large variations in these vital rates may occur among habitats, in part because of temporal

variation in habitat conditions. Thus, one should be cautious when evaluating the conservation

value of habitats (see also Holt and Gomulkiewicz 2004; Jonze' et al. 2004). In particular, this

study and that of Martin et al. (2006) show that regions in which survival is low for many years

may be critical during disturbance events, such as droughts, by serving as refuges during drought

and possibly by providing the entire population with a pool of individuals (i.e., kites hatched in









refugia habitats) with greater ability to resist such disturbance (see also Holt and Gomulkiewicz

2004).

Aside from its practical implications for conservation, our study highlights the importance

of considering natal location as a potentially important factor affecting the ecological dynamics

of spatially structured populations of animals, which, like Snail Kites, inhabit heterogeneous

environments and use experience to base their settling decisions.










Table 4-1. Multistate models of apparent survival (" 4 ": survival of adults; ")j ": survival of
juveniles) and annual transition probabilities (" w") among the four major wetland
complexes used by Snail Kites in Florida between 1992 to 2004. Factors incorporated
in the models included: age, region, natal region; and a drought effect on movement
and survival.

Model AAICc w K DEV
r,[E=L#J=K],, r,[K=J]NDr,, r,[E=L],r, [NO#ON#OA](nr)
D,[EnLd K]) AD,[Kd dJ (D),) (t) [N ON)Op(r*t) 0.0 0.38 76 2591.2

r,[E=L#J=K], r0 K d(ND) r [E-LDJ]( /t\ \[NO#ON#OA](l\ *,,
AD,nd K] AD(ND) d (D) t) 1.7 0.16 76 2592.9
r,[E=L#J=K] r,[K=J]rND r,[E=L], [NO=OA"ON] (r r )
rAD,nd[E AD,d ) ADd(D) J(t) p[N= A ON] (r t) 2.0 0.14 72 2601.7
r,[E#LJ=K],, r,[K-J]ND,) r,[EL](D) (t) [NO#ON#OA] (nr) 2 *
"AD,nd "AD,d "AD,d J Vt ) p(r) 2.0 0.4 77 2591.1

r,[E-L#J-K], r K ND r[E-L-J](D j, /t\ [NO-OA#ON](r) p(r t)
rAD,n[d ) AD ,d(ND) Dd (D) t1 ) 3.7 0.06 72 2603.3
r [E-=L#JK] rK d(ND) r[E-L-J](D), /(t)\ T[NO#ON#OA](nr) p(r t)
AD,nd () ADd(ND) AD,d 3.8 0.06 77 2592.8
r,[E-=LJK], r d(ND) r[E-LEJ](D) 0(t)0 [NO'ONOA](nr) p(r *t) 5.1 0.03 79 2589.9
AD,nd ADd(ND) D,d 5. 0.03 79 259.
r 1'n r,1 KDd(ND r,[E-Ld J](D) t) ( [NOONOA]0.0 80 2589.8

r 11 r K ,d(ND r,[E#L#J],, [NO#ON#OA](rlr)
rI',na 1[E) D Dr d t) (ND[NOO) nr)] t) 8.1 0.01 82 2586.6
nr [E#L#J=K],, nr,[K-J](N) nr[E L](D), (t) [NOONOA](nr t) 8.4 0.0 77 2597.4
AD,nd AD,d ADd 8.4 01 77 2597.4
ri[E-L#J-K]/^ r [K-J]/TN > r [E-L]/ (D) it) T [NO-OA#ON](nr*Dm) p(r t)
AD,nd G) C Y ( tD,d\U/ 8.8 0.0 80 259
nr [E=L J=K]0. nr [KJ](ND) nr[E L](D) i (t) [NO#ON#OA]() p(r*t)
AD,nd ADd(N D() (t)) p 10.9 0.00 76 2602.1
Notes: AICc: Akaike information criterion. AAICc for the ith model is computed as AICc, min (AICc). w: AICc weight. K:
number of parameters. DEV: deviance as given by program MARK. "nr": natal region; "f": region; "t": time; ".": no time
effect; "age": age effect; "*": multiplicative effect. Superscripts on the right of 4 indicate the natal region survival pertain to
(e.g., E ": survival in region "E"; there were 4 regions: Everglades "E"; Lake Okeechobee "L"; St Johns "J"; Kissimmee
"K"). Superscripts on the left of "4" indicate whether survival is natal-region-specific (denoted ",nr ") or simply region-
specific (denoted "r "). # ": regions have different 4 "; "=": regions have similar 4 (e.g., )[E=L J] ", survival rates are
similar in "E" and "L" but different in "J"). Subscript "nd': survival term reflects non-drought years (1992-2000 and 2002-
2004). Subscript "d': survival term reflects drought years (2000-2002). "ND": no drought effect on )" during the interval
2000-2002. "D": drought effect on 4 "during the interval 2000-2002. Superscripts on the right of indicate the direction of
movement between two regions (" yN ": transition from "N" to "O"; and lOA ,, transition from "O" to "A"). "N": natal
region; "0" is a post-dispersal region ("O" includes all regions except for the natal region); "A" is a post-dispersal region ("A"
includes all regions except for the natal region and the post-dispersal region "O" occupied in the previous year); # ": regions
have different transitions probabilities (e.g., T [NOOON#OA] ,, T NO ,, ON and T OA are different). "Dm": drought effect
on "\Tl" between 2000 and 2001. "p": sighting probability. Only models with AAICc < 11 are presented (see supplementary
materials for models with AAICc > 11).














































Figure 4-1. Major wetland complexes (i.e., regions) used by the Snail Kite in Florida.
Kissimmee Chain of Lakes (K), Everglades (E), Lake Okeechobee (L), and Saint
Johns Marsh (J).












98












0.7

0.6 NO
?? OA
0.5
ION .
^ 0.4 -

0.3 *

0.2 -

0.1

0.0
E L K J
NATAL REGION


Figure 4-2. Movement probabilities (" W ") between natal region "N"; post-dispersal region "0"
(potentially includes all regions except for the natal region); and post-dispersal region
"A" (potentially includes all regions except for the natal region and the post-dispersal
region "0" occupied in the year prior to moving to "A") for Snail Kites hatched in
NO
four regions. Model averaged estimates of movement between "N" and "0" (" x ");
SON ^OA
between "0" and "N" (" \ "); and between "0" and "A" (" W ") are presented.
SON OA
Error bars: 95%CI. Asterisks indicate that differences between and "
were statistically significant (i.e., 95%CI[ ES] do not include 0).










>- 1.0 NPHL PHLNN
I- T
0.8 -





0.4 /


9O.2
0//

0.0
E L K J
REGION


Figure 4-3. Model averaged estimates of natal philopatry (NPHL) and philopatry to non-natal
region (PHLNN). Error bars: 95%CI. Asterisks indicate that differences between
estimates of NPHL and PHLNN were statistically significant (i.e., 95%CI[ ES] do not
include 0).