<%BANNER%>

Biases in Population Estimation for Colonially Nesting Great Egrets (Ardea alba) and White Ibises (Eudocimus albus) in t...

Permanent Link: http://ufdc.ufl.edu/UFE0021317/00001

Material Information

Title: Biases in Population Estimation for Colonially Nesting Great Egrets (Ardea alba) and White Ibises (Eudocimus albus) in the Florida Everglades
Physical Description: 1 online resource (101 p.)
Language: english
Creator: Williams, Kathryn
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: ardea, bias, ciconiiformes, estimation, eudocimus, everglades, indicator, mark, population, recapture, superpopulation, survey
Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Estimating avian breeding population sizes often involves extrapolations derived from estimates of related biological parameters, and may therefore fail to account for bias introduced from several sources. In colonially breeding birds, estimation of breeding population size is complicated by visual biases and by asynchronous nest initiation and turnover. We used artificial landmarks to create quadrats in nesting colonies, and quantified visual biases by counting the number of nests in a quadrat via both walk-in survey and aerial photography. Visual bias due to vegetative occlusion caused aerial surveys of White Ibises (Eudocimus albus) to underestimate numbers of nests by an average of 12.37% (+/-20.65%) when compared with walk-in ground surveys. Misidentification of the species of nesting birds in mixed-species areas proved to be a problem in aerial surveys of Great Egrets (Ardea alba). We examined asynchrony bias by applying a mark-recapture approach to nests. We individually identified nests from the air through the use of natural and artificial landmarks, and followed nest fates through subsequent aerial surveys to obtain presence-absence information over the course of a breeding season. We used the superpopulation approach (a variation on a Jolly-Seber open-population mark-recapture model) to model nest initiation and survival as if nests were animals in a marked population. The superpopulation approach allowed for the estimation of the total number of nest starts throughout the survey period. We found that asynchrony in nest initiation dates caused one-time ?snapshot? counts of nests at the peak of nesting to underestimate true numbers of nest starts by 47?382% for both species. Nest turnover rates were highly variable among years and between colony sites, suggesting that nesting asynchrony must be measured for each season and site of interest. These results indicate that, without bias-correction, peak counts do not allow for useful inter-year comparisons, because each peak count represents a different proportion of that season's total nesting activity. Unless populations are highly synchronous, or this bias is measured on a seasonal basis, one-time 'snapshot' surveys of avian breeding populations probably cannot be used effectively as indices of breeding population size. Researcher disturbance may also bias estimates of population parameters for colonies. Walk-in censuses of White Ibis and Great Egret colonies (initiated after the onset of incubation for the majority of nests) found that while nest success was not significantly affected by researcher disturbance, disturbed areas demonstrated depressed rates of new nest initiation. These species may be sensitive to disturbance early in the nesting cycle, but be resilient to walk-in disturbance after the onset of incubation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kathryn Williams.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Frederick, Peter C.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021317:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021317/00001

Material Information

Title: Biases in Population Estimation for Colonially Nesting Great Egrets (Ardea alba) and White Ibises (Eudocimus albus) in the Florida Everglades
Physical Description: 1 online resource (101 p.)
Language: english
Creator: Williams, Kathryn
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: ardea, bias, ciconiiformes, estimation, eudocimus, everglades, indicator, mark, population, recapture, superpopulation, survey
Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Estimating avian breeding population sizes often involves extrapolations derived from estimates of related biological parameters, and may therefore fail to account for bias introduced from several sources. In colonially breeding birds, estimation of breeding population size is complicated by visual biases and by asynchronous nest initiation and turnover. We used artificial landmarks to create quadrats in nesting colonies, and quantified visual biases by counting the number of nests in a quadrat via both walk-in survey and aerial photography. Visual bias due to vegetative occlusion caused aerial surveys of White Ibises (Eudocimus albus) to underestimate numbers of nests by an average of 12.37% (+/-20.65%) when compared with walk-in ground surveys. Misidentification of the species of nesting birds in mixed-species areas proved to be a problem in aerial surveys of Great Egrets (Ardea alba). We examined asynchrony bias by applying a mark-recapture approach to nests. We individually identified nests from the air through the use of natural and artificial landmarks, and followed nest fates through subsequent aerial surveys to obtain presence-absence information over the course of a breeding season. We used the superpopulation approach (a variation on a Jolly-Seber open-population mark-recapture model) to model nest initiation and survival as if nests were animals in a marked population. The superpopulation approach allowed for the estimation of the total number of nest starts throughout the survey period. We found that asynchrony in nest initiation dates caused one-time ?snapshot? counts of nests at the peak of nesting to underestimate true numbers of nest starts by 47?382% for both species. Nest turnover rates were highly variable among years and between colony sites, suggesting that nesting asynchrony must be measured for each season and site of interest. These results indicate that, without bias-correction, peak counts do not allow for useful inter-year comparisons, because each peak count represents a different proportion of that season's total nesting activity. Unless populations are highly synchronous, or this bias is measured on a seasonal basis, one-time 'snapshot' surveys of avian breeding populations probably cannot be used effectively as indices of breeding population size. Researcher disturbance may also bias estimates of population parameters for colonies. Walk-in censuses of White Ibis and Great Egret colonies (initiated after the onset of incubation for the majority of nests) found that while nest success was not significantly affected by researcher disturbance, disturbed areas demonstrated depressed rates of new nest initiation. These species may be sensitive to disturbance early in the nesting cycle, but be resilient to walk-in disturbance after the onset of incubation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kathryn Williams.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Frederick, Peter C.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021317:00001


This item has the following downloads:


Full Text
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 E20101113_AAAAMQ INGEST_TIME 2010-11-13T16:03:36Z PACKAGE UFE0021317_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 1051969 DFID F20101113_AACLTV ORIGIN DEPOSITOR PATH williams_k_Page_054.jp2 GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
49aa264942f92275a1f05785262219f4
SHA-1
4135ca48238c42fed69402c500ade16c499865be
46066 F20101113_AACMAE williams_k_Page_083.pro
118c2d3af38140a8bfe9418f03d8e75b
75bb608cea9078ffb40748a46c4e5249d53f74df
100647 F20101113_AACLUK williams_k_Page_082.jp2
e0c4c05e3d8301a3b0b1ac09fb81d3c0
d89c7f13f7482e90173961b503c228d47775c06a
876991 F20101113_AACLTW williams_k_Page_055.jp2
3e63a2f83c8435af266599594d4e686c
a44cfba26a36cdfc3c0ddbcdbe4726a13334d1ad
45511 F20101113_AACMAF williams_k_Page_086.pro
1a8c8d2bfb5efa4c7bb51cccf4275ff7
85eb1332abda76142cd0e47426de6e5c455a6969
108339 F20101113_AACLUL williams_k_Page_083.jp2
921f9ef65e8c4e1bf4e2cf67af442172
8127739ce9b87503c345707d75ecc5e3543953bc
1038718 F20101113_AACLTX williams_k_Page_056.jp2
fa02b10c97373dca777a335484377ad8
fb6798441ace10bc0c4b1eb24cf4df87039146f9
12705 F20101113_AACMAG williams_k_Page_088.pro
bccd3de0cfdf8d4b50548f4454a4d24d
249f4d5e94010e17aa1a2796e8b4277a2db9cbed
25271604 F20101113_AACLVA williams_k_Page_005.tif
10ba3ac3e8bdccfd1181c90f712961e1
4bfcf8a8e19eb1af76b03a00d25b530c7a46a4a3
103367 F20101113_AACLUM williams_k_Page_084.jp2
1ed35dceca1e2bb1e640720d2c047f99
0e5e2067d7661a73e1a429d4cae8f97ab90df263
372154 F20101113_AACLTY williams_k_Page_057.jp2
a0973c8088b40650bd2fcaed669e9b31
13689936fa1d6ae2f6c0f41a2930002f250fbfe9
54435 F20101113_AACMAH williams_k_Page_089.pro
3de508ae451a8bf0c2a4fc9a3cfbbcd2
f16ae8c54fe0e0824f07963bcc31146f05430597
F20101113_AACLVB williams_k_Page_007.tif
a917b3c41d235951c4943e0769878f25
b95f55ffb4d24a67716cb328f793906d2ede2fea
114382 F20101113_AACLUN williams_k_Page_085.jp2
34ed6e4795c64a4aabfa779f67183487
f38561eec361e8e9b9ef531afb756cab6be0f23d
4019 F20101113_AACMAI williams_k_Page_090.pro
203887deeb625bba583871f37f857f0a
0dffaab56c3b11127e952a15af98ca8799fefa9a
F20101113_AACLVC williams_k_Page_008.tif
19e7b6930acfbc635ed360919a45d4e9
ae8a273321c3e37295f0271b268632a3d9e0a5b9
98682 F20101113_AACLUO williams_k_Page_086.jp2
743b59c9c0edf6453953e0ef23412f4a
58735a6b2404569077c2f96c1272f1b7fda41511
362602 F20101113_AACLTZ williams_k_Page_058.jp2
4b77f3247403ae9b110dfed8ab4241c4
77e272d7e30c5d8972447cec91b4d72cae834ad7
18178 F20101113_AACMAJ williams_k_Page_091.pro
9cb2652815cb51e89f4a110f0f815c7e
25602a6d79f00b8f3287ccff5eaa8eba9f60e280
1053954 F20101113_AACLVD williams_k_Page_009.tif
f5bd06db764a1cb69af56613d464045e
b3ef3d41df3ba2302adf72c1b1ba2d936ecb3ae1
118414 F20101113_AACLUP williams_k_Page_089.jp2
b25bad7e63d7b5a75016ef41b79b32a6
9ae9546bf670d0fd57a89ba414ebadaa55d2d3b5
63082 F20101113_AACMAK williams_k_Page_092.pro
cadc1c006e3277195ec398579c3003f2
ea1e0f68e9eea7f532ec6beaec28ef39772569c4
F20101113_AACLVE williams_k_Page_010.tif
9faa473b266cfb4b3701cc3e4b845280
e0453cc85b45ea7bf12f0872f3fc7df6f62d62d6
136727 F20101113_AACLUQ williams_k_Page_092.jp2
d00068acbe03c75dd4d69acdbd205cdd
73ed131324b780ff4dacbe60ede7fdb0ccff61e0
56740 F20101113_AACMAL williams_k_Page_094.pro
3fdaf8633c9fe11b9e08bad561086f2a
c795a0e09dd8bd222b42198727bccb35ebd675d6
F20101113_AACLVF williams_k_Page_011.tif
ad5bb968715be36c466d3c46d8d43b3f
2148a43c60f855075ffee12f26a94c178872812b
66431 F20101113_AACMAM williams_k_Page_095.pro
f5fdd0e55a711b9b804232a44f3eada6
103b1aab3c054de838637f4b59dbd7d607021570
F20101113_AACLVG williams_k_Page_012.tif
65f011cfce3dbf49bca85bcd81095c9c
f1937bffefdbfdd9f7695fc7a75a78b3a00c6353
128146 F20101113_AACLUR williams_k_Page_093.jp2
667378915f9114096b54067981a1e71d
bb045f63f8005ab96014a060164a40df2701c8dc
1989 F20101113_AACMBA williams_k_Page_016.txt
8431be6be7a776b7d896a6831d175387
7a604d3075bc1716f4c7a240083b70963e1b9c1b
68580 F20101113_AACMAN williams_k_Page_096.pro
ffcff28845d61c43ece39456ea578e4f
c1c5727c0666476013e79423f4f4e81d0e134377
F20101113_AACLVH williams_k_Page_015.tif
cc988cdfa91bf4cda3763759401e107a
a5f6248fcededcfaed9905a0a6f4e3c4b3d45e0c
140973 F20101113_AACLUS williams_k_Page_096.jp2
829c23d2843118d91c81084d94615954
7c90144a2f2a11fbba61f0eb984c805758ed7bcd
2107 F20101113_AACMBB williams_k_Page_017.txt
47b6ca9dd5d18c30251b1b9454a42bfb
14d8d8b713019aaf3086c26181df7aa1e233f261
64635 F20101113_AACMAO williams_k_Page_097.pro
fed078e616da4fe64ea9210a85e5faeb
d52930d85037b19f74578c0ab75d729fffe08b33
F20101113_AACLVI williams_k_Page_016.tif
e00c9eb6cebb6e9d16db59a3774281b8
8e45a1bc50ab4a33b6256c1f744fdfd826b2f8d4
134388 F20101113_AACLUT williams_k_Page_097.jp2
1c2e72317feb2dbb70f1e2a95cf5d45f
106c938283687b84a66c97a01477ff6f31dd1190
2227 F20101113_AACMBC williams_k_Page_019.txt
96895781a7b7276970dac75a1523e5a6
0ccaffcd5c15b088580b8e685afdbdbd94a3e567
65132 F20101113_AACMAP williams_k_Page_098.pro
f3a75bbf891bd2b5047fb07a23c364fa
19a42e4e01e33e87960c4d764961d75d080dc804
F20101113_AACLVJ williams_k_Page_017.tif
3fc0eee35825d9e0884bd4c625b0aaf8
95169fb850b44f86086a6c9563686e6553585572
132829 F20101113_AACLUU williams_k_Page_098.jp2
10fe545865ff69e5571d5802affad311
d138bf16f24fa19c7c3a1033afb8e4ea48481cbc
2162 F20101113_AACMBD williams_k_Page_020.txt
fdc8980cc6d53e9ef2b533912947337c
155ae9db52fafac0f977f719d4cedaaf535007d9
31622 F20101113_AACMAQ williams_k_Page_100.pro
34afff8ac4a01ddf51b434a14b0995e2
ea14e5ac109df50f0ec2102da3c3d99ddfeefb06
F20101113_AACLVK williams_k_Page_018.tif
b856661c6f45a97a91686e14a5ed406a
2e3c56d59953205e4ea048f0c1f0ea0b8d0e2853
133649 F20101113_AACLUV williams_k_Page_099.jp2
00a7fab81e1dbd080547c1c70ba6e50e
d983f1b285ce65d0861f49a1eaac81fc38ea9cce
1712 F20101113_AACMBE williams_k_Page_021.txt
190f553131858f8882685058adf301ab
9e9b6226d9b3b0ed896b92cc2381eb745142a62e
528 F20101113_AACMAR williams_k_Page_001.txt
f966a79ccc5a3b4599c885406767825c
af50b3cfa023f3ee3f7dce46a2103b294362f36a
F20101113_AACLVL williams_k_Page_019.tif
4dea417c19abcb09bf96aacf0475a819
d68c0f96bb6ac43918a70b3beaf16d30cd3c4df7
25927 F20101113_AACLUW williams_k_Page_101.jp2
e777e1e4a761d8f3a2cf992e05137f70
d584db6ae3cdea40d6ac712f0620870603d1c83b
2042 F20101113_AACMBF williams_k_Page_022.txt
ce76a5fece40578839c0111939e19a45
686421807c18c31065fcff0929c240400afd1a43
91 F20101113_AACMAS williams_k_Page_002.txt
e6927e16562865deaf8b32ed1d869c1c
623c39f8b128e7512c258bb4d8f3a97e5827ca6e
F20101113_AACLVM williams_k_Page_020.tif
e17c86e7e0c5320d3a61bbd99167c6dc
46ed7343693169d8a2689378e7d7622b10aa59bd
F20101113_AACLUX williams_k_Page_001.tif
f9ddebef39c8fa6d81d23c78aa5faf43
186422b95fa9b6a8d72dae4f75ecc807fc735faa
1464 F20101113_AACMBG williams_k_Page_023.txt
fa1020c7c1c67ea5de1af207d2c13982
eb22dc3b8327fbf525e3c8f4d572093e493ae7c4
F20101113_AACLWA williams_k_Page_043.tif
59d2d5a8cfbfc62d78c2c76fd69e32aa
031592022c4d2499e89cbf92b4307bad85e2898f
213 F20101113_AACMAT williams_k_Page_003.txt
3147553aee09e8a0f1d178a582c72e69
7a1eb5ea2ac525f16f3181155382aaf86779f9fd
F20101113_AACLVN williams_k_Page_021.tif
1f16a2e7f3b71c53c8a76fea61f2e7c2
87622be2b3b7a258fc5acd26595edfa41dbff545
F20101113_AACLUY williams_k_Page_002.tif
1074267fcac87ae79a38daa8d7f45069
8f51279a53c0cc91d18859962236bcd00d83fb0a
434 F20101113_AACMBH williams_k_Page_024.txt
515f7f13f964cb38f9868398b769e17a
ba99c52c5835de29e811c72a27c0861e0cbf27df
F20101113_AACLWB williams_k_Page_044.tif
2deaaa641b7f11634af8c792cddc91f5
25e62691a946202c3e70179f232fe23c64e4bd9c
2633 F20101113_AACMAU williams_k_Page_006.txt
136faaf31374764bd664a6252aab504d
28bdcaa6e2605c377946d41e8358f5f2b0333fc6
F20101113_AACLVO williams_k_Page_024.tif
cd30ea354f098182d133df96dde45354
095974eddf7a80ebcc393d0913f9e25aae77163c
F20101113_AACLUZ williams_k_Page_004.tif
cc51c724a9d47835ea3c39041786d50c
5247baad053c38a40b2a0981ad74110828e53628
468 F20101113_AACMBI williams_k_Page_025.txt
fd1af37013564e5e3029a5a775a256d5
59e9dc2269a35bb8007104e7ff0eb9e61cdd9af9
F20101113_AACLWC williams_k_Page_045.tif
d88ee7ce07d7180b67524a569478ac09
2375fa8bdb6abebd7ceb94d6aec22ebe327d4e0b
1912 F20101113_AACMAV williams_k_Page_007.txt
b0e4f293d6189b8d81a6591c876cec1c
c35bfb970df46d117b6f64739ef85d4e2c28699f
F20101113_AACLVP williams_k_Page_025.tif
3991f849055ba511069fc22b0a17e49a
fb32ba9f2de7941cb32fec8ffb619be047f0f5e2
2077 F20101113_AACMBJ williams_k_Page_028.txt
3014c6d50106525b662269e10e6d7168
a059796b5369762a42fcd1df9ea9efe6356a4e12
F20101113_AACLWD williams_k_Page_046.tif
24411c508d3cf85bcf085dacd79e9899
536b2bef4c8825d68f9c76b4386b53c1d4675a51
2120 F20101113_AACMAW williams_k_Page_009.txt
529a35ea43cec0d77e1a2230adb2c83e
8b71c4d13e9a7808b8bfa08b9cac46c58918b6d4
8423998 F20101113_AACLVQ williams_k_Page_026.tif
c67436998400af00acd9646808607809
89a7c1fe574e5da21b8e696018d9f8147578df95
2147 F20101113_AACMBK williams_k_Page_030.txt
1c41e0fda89b4184e7be1feb16d40d5c
b08d227a36d782eb0d7a0c55a1eae1cb39c0fef8
F20101113_AACLWE williams_k_Page_048.tif
57631ba48c578d4da3ef92820ec1c097
f1075f8d7f5a1200b680a48e5c74c86546d52b01
F20101113_AACLVR williams_k_Page_027.tif
2c541bb0400f6d67d42a582263fcf83d
df2bd33f8d7bf3747730a47a95959bd8931cb148
2116 F20101113_AACMBL williams_k_Page_031.txt
9d54aafe71c81d22dfaca259f2450fc8
e99b32d6666fe7ee10103156f57108d9f2aba2b7
F20101113_AACLWF williams_k_Page_049.tif
1ecb5e265bf6324bf8ee230a3eb52312
082b6d9f74c03c226f5d0f3045b0bd57db90d96e
2090 F20101113_AACMAX williams_k_Page_011.txt
de66a39eb9ad2600e14753bd2bf2f11c
7c307eea17c7f96ea2403d7ce03e6bbb25c84c18
634 F20101113_AACMCA williams_k_Page_049.txt
899f76b24411cdb4af02aca28d32594d
2a41f57eefdd53405973fe7625a9ae764e738eef
2143 F20101113_AACMBM williams_k_Page_032.txt
2cd8f58501293ea58de9a7ac5376e3d2
8e17af198fcab47cf6f297c7b6d2c8d967849a04
F20101113_AACLWG williams_k_Page_050.tif
40b804f1c1385358b019de784a2f40ff
dd05f074713b679427d8ea627d57685d18cbf18b
2173 F20101113_AACMAY williams_k_Page_014.txt
a21718a8818ac61b008cf656b0678ec9
e11ac651406d3542fbb1baaec75bd18389c938a4
F20101113_AACLVS williams_k_Page_028.tif
0def4638015c355282f908efe262b692
045b2ea4e53b1588a65bae661e491695e2b6501f
1012 F20101113_AACMCB williams_k_Page_052.txt
4ad9333a76396b9e282a77ad873e6f2b
a828ccd3dae10cbd6d7ef301f4b49b23bd0d773f
2261 F20101113_AACMBN williams_k_Page_033.txt
ebfc05a16ef53ceb4e134363e71a69c4
f02f1b4cad032ca295f23296bceaabe85667de08
F20101113_AACLWH williams_k_Page_051.tif
206050c1db79f833c1123244dd894a37
3e9309eb0b4a2caa28ea00d322b5ae025229ac44
2060 F20101113_AACMAZ williams_k_Page_015.txt
10d09c53e92bce122f43d50dbe3ce4f8
76d49eb8a8127c3f1ef7e2dc737fddb70fed3cfa
F20101113_AACLVT williams_k_Page_029.tif
5ede33ebc3c1d5021a107b8aea848b80
9dcb9e96692db5f8ab7e5d5459b7c8dc707cc23d
236 F20101113_AACMCC williams_k_Page_053.txt
5c73bb9cdd07e7b4a1fb127c4a3bf3a1
13d8ac85626525f42eb7e5d5ea6e8b66d1794abe
2238 F20101113_AACMBO williams_k_Page_034.txt
1f8cffa3d13b1479dc77e43d55a156ca
ef2721d7a49ffee5674a0dd35e3c7080858268c7
F20101113_AACLWI williams_k_Page_052.tif
ce63f44ce8b5f9d35f0ad0dcd985728a
af0c4315984120aaf4f88e9bf4f6e905b66833dd
F20101113_AACLVU williams_k_Page_032.tif
fcdd59a92d8ac0b4df33f56531d14e14
4f490b81722d41204bfcaf5013a515e8ff096ff3
557 F20101113_AACMCD williams_k_Page_054.txt
79fe87daecd9344c8a332177f974b2ef
250ca629a4293801b099826b12686df08d8b06f6
1856 F20101113_AACMBP williams_k_Page_035.txt
ac1ed0c7a044b479fcab911a127f04e7
22689efd3a4a10c1575e5fb3d540dbf315ed18d5
F20101113_AACLWJ williams_k_Page_053.tif
b07400fc64fc2383771f5e6b48bdf931
d38dc920d6d9b6e44a2cfa24337721613d259028
F20101113_AACLVV williams_k_Page_033.tif
48b3dacd9eee3923cf53d3f13be7c660
574205eea4a48796fb51fb62004d6a2cb237fcc9
359 F20101113_AACMCE williams_k_Page_055.txt
19db032a97336f4dfd6256807e4e9b19
d98bf4ae78e52723334c400f867195f4c40d59fd
2151 F20101113_AACMBQ williams_k_Page_037.txt
fb0548ee0df03261fb345048d86c58ba
f5126a8118ecdb802d39a51c8eed511371389b88
F20101113_AACLWK williams_k_Page_054.tif
af939be9e53a37e7b3e53731b33591ae
7dfe45eae50b56c4c2c998baba2e5ba23f1769d1
F20101113_AACLVW williams_k_Page_034.tif
4759055c941de1fe8a4d6a8fd21a8b33
2cc61cba956caa8175d189e9d2a853897d7f0cb5
791 F20101113_AACMCF williams_k_Page_057.txt
362955f1c60d6ac5422fb318a4bde3ea
ef9fd256ae6bb036acb78b0e298872a0b9ca940c
2057 F20101113_AACMBR williams_k_Page_038.txt
2b0118c99133346b793e0328f9257923
1bd4dfef76cdac830d9e31abb05c666465f8b2de
F20101113_AACLWL williams_k_Page_057.tif
6129f0fa629f4f8078b2d38d5976acbb
ef36a25a9b328213a7546275b9e09231c54e3a9d
F20101113_AACLVX williams_k_Page_040.tif
8746edc7ee288e4b93e20ee90d8b8e9b
7e4a57f94300c5d52e47649fa8a4a8fd8cbd05d1
786 F20101113_AACMCG williams_k_Page_058.txt
5c80da8a4552b1632daf07259d27de09
60fa9d12ae06e85edf7d91fc18b33a0be811a2df
F20101113_AACLXA williams_k_Page_078.tif
450868fa8a56229990e0bfca264f4570
a7d9562b2d1bfb684356cc13c7e4c8ae874521e2
2066 F20101113_AACMBS williams_k_Page_039.txt
d8062957e578b39bcd26a6439732945a
9ae1e76b17c331a825015f393aaec2a227612b22
F20101113_AACLWM williams_k_Page_060.tif
e09bf824f9533adfe1e45389e08f6430
5be2731657644de003b7757b383b497830e7bdbf
F20101113_AACLVY williams_k_Page_041.tif
f057a9b20ad7658b183e29be33b3365c
2d2f10fd3ad727618cd518064a648b816a7b6df6
954 F20101113_AACMCH williams_k_Page_059.txt
cba816c42a7ef792dbde82fb2327cb23
489f1a3dd1a839873a028f422a25e585fb0cdbff
F20101113_AACLXB williams_k_Page_080.tif
333a36a7612a55b8d8fcc306f633b080
19e4cd12d5224e73814c394036f1a2d04c545499
2240 F20101113_AACMBT williams_k_Page_040.txt
aa93448dda3c1adabfc06612d1a7ec01
8eb3e8d65ef83eeb69271743fcfc65e5c2f465b2
F20101113_AACLWN williams_k_Page_061.tif
2f2a714357c541ea020dd11534318241
eccec35e623c781102bda491bb426c227821131b
F20101113_AACLVZ williams_k_Page_042.tif
adf784341bb3d62e1cff240c346b4efb
660632bbf5d8710755ecf8695564045df60a8233
2046 F20101113_AACMCI williams_k_Page_062.txt
b469ff21d520dc3285dfe09845c43b72
3a733b053ecc7c1f7f237dcc64065808a017a299
F20101113_AACLXC williams_k_Page_081.tif
bae8c7fd13b84bc56811f381105e85cf
bd8268a789cf3f2637d61eb2a1070a4d991c58d1
2239 F20101113_AACMBU williams_k_Page_041.txt
5fcf4905cfb6fb0e5529086a4f2605d5
0b7bcb902fa784e9a18e42d1fe5843cd3157bef0
F20101113_AACLWO williams_k_Page_062.tif
8becf63da1c932f10119deed6d9ca3f5
54b025f0f71528f12173957cd13b940abeadc3fa
2080 F20101113_AACMCJ williams_k_Page_063.txt
8c9a36a227cb3c46aa7dec9d86c16524
6c6d02e0299f428510b1485ddfc4bc54d1d76c72
F20101113_AACLXD williams_k_Page_082.tif
603d6b113da60d7ce8abea1b6da0533b
cef39d0a6067731d428bf6e68ba877f5049e26a8
2100 F20101113_AACMBV williams_k_Page_042.txt
0fb6ff4d6d6e7f6263ef2d8fd0ada51e
582fa4e765e0026707257373c87950c756757c0e
F20101113_AACLWP williams_k_Page_064.tif
a3d413d9fd2ea32585e2f50b3f7a717e
3e47971c215860e74c1b4174e75e5e6dad90c14f
2231 F20101113_AACMCK williams_k_Page_064.txt
29937c2252077cf885341882b38d4b6b
207315a2c79e1838009c84b5898140df3f3e98ad
F20101113_AACLXE williams_k_Page_086.tif
f4b586dec0735a0b923dfe8ec54e645d
aeb2098501ea12f671e40c74a1b2ee9d518c0285
2074 F20101113_AACMBW williams_k_Page_045.txt
8cf5c180984be5caec92fc3c5b137550
671eaaabcca398d83072a3b304f615a3b81b53a1
F20101113_AACLWQ williams_k_Page_065.tif
6bda239da6abee31f5718c1704d4dc2e
a1969e1dd2396430170cb9be8fdf7eb0b5bd510f
2134 F20101113_AACMCL williams_k_Page_065.txt
57e77202465b768fc1c8360ccae0189b
31f535df9fdb822a027bef065ae4d361c5d4e432
F20101113_AACLXF williams_k_Page_088.tif
22fa5d9ba88c5362bbcb04742d482241
9eb092708baf550d5d728138cf1a9fed28048992
2233 F20101113_AACMBX williams_k_Page_046.txt
2f26bd4124fd3341c7521502facd7b89
ad732d99065f47069b4e01d2e4401ea48cd410e0
F20101113_AACLWR williams_k_Page_068.tif
aa49a614ac207bfc7b4a4ca325cb7c9b
a4eb2d5890a4eab3733d9b79b994f319b17fbab7
2216 F20101113_AACMCM williams_k_Page_067.txt
cc25651ce9704b9eeefec959d233494d
ee3247174903182d12ae1d028f55120383312a3b
F20101113_AACLXG williams_k_Page_090.tif
e7426659f5d3be5fa84e0871fafe81e3
8392de44e5d88bb1a8a5a943604beeee9b3c0bc2
2157 F20101113_AACMBY williams_k_Page_047.txt
44f8fa7f3ae06a3a7cd4ca58beec6b39
b68216e3c28ccd1d26e1ca133a5a4b528f34a233
F20101113_AACLWS williams_k_Page_069.tif
a44519d664ef1e77ec474b78422bf619
ddef791ad6cef74cffd100fcd98ef8be3e5c8434
2102 F20101113_AACMDA williams_k_Page_085.txt
090899b5a5a7e36f16c92ea951c35837
f87def0e828d9dcdf01da1b2e739b9a7799aa46e
2187 F20101113_AACMCN williams_k_Page_068.txt
c991f95299eb2aad3571268a07db9551
433e6102c5d1c484e5c68414cae53035c652b785
F20101113_AACLXH williams_k_Page_092.tif
cc64523f7ee2b7c12277652c93e0f96a
5603d2c4703c70d1f069d447b0c23fafade1fe56
1741 F20101113_AACMBZ williams_k_Page_048.txt
81eef01d9278ad57bfa93d558abc7ddc
4c1564885eb4c598d999a135dd5b7bd3c8274e04
1719 F20101113_AACMDB williams_k_Page_087.txt
18cf67b8371ecbe5d511343ecae09bf5
ccfd4503cc8912ed2ba5905a73b1157a850a2c39
2121 F20101113_AACMCO williams_k_Page_070.txt
85bd496d125c17755d5142f8d32ca77d
04d11b3c3b07dd7f342dc2be058b03ede33324e5
F20101113_AACLXI williams_k_Page_093.tif
d31fb317b8989f20e19a40b817be5619
38f6807efaf39f61071cf2c0183d362d36ff4e5e
F20101113_AACLWT williams_k_Page_070.tif
e92fdc2ffefa02739e0e87231be32c09
5cd59d9419dbc5881be2490f755175b6aedac00f
2218 F20101113_AACMDC williams_k_Page_089.txt
b28bb1cc1ac3a82248e2db9ddc0c26a9
9e0b10c965738c76be1ccfc0fc39488ab0cafb7b
F20101113_AACMCP williams_k_Page_072.txt
6d9b27b5589827525df08994bb5a77f1
3560c971a8cb743c76790cc63d0e9f0f3f20f0d9
F20101113_AACLXJ williams_k_Page_094.tif
652ee1b32b18f03dbd5ab1bcc8f79b18
9a385f655dff69b014a98055738a6fb69041bd65
F20101113_AACLWU williams_k_Page_072.tif
628ea8b449b8e896b6e8dbbdd3dbb7e7
51578d2c79f6ff3532ffc5976f0cdc3a9caac476
778 F20101113_AACMDD williams_k_Page_091.txt
269227d7c7b389750e0e137d3c5eb04c
4f6821d824dfc18309e9039615bb9d5ddb296694
509 F20101113_AACMCQ williams_k_Page_074.txt
6db78cd6f1026fb08dcabb945fdf8fa4
9a220bd0c5df244ee272f5d5bc79ee7ea9242347
F20101113_AACLXK williams_k_Page_095.tif
69e0a778c6822238b4a981faeaa10cb4
fa2fd931b0d8fc4e753de00ea39091ab370644c2
F20101113_AACLWV williams_k_Page_073.tif
5d0191e2682a94cfbb5172eb82d51203
3c098e5eb0504ed4a8ed9f3c4242ff768bb3b7e5
2689 F20101113_AACMDE williams_k_Page_093.txt
c15b3825fe4481764b69aed02f8bb799
09d2baed4cfc6f6efe15176db332dd6929ca4ef6
2790 F20101113_AACMCR williams_k_Page_075.txt
3b568a4acf34d3205b4d7c3783d5cece
de4bd98c0db93c888a360c9a372c1386695b7dfb
F20101113_AACLXL williams_k_Page_096.tif
47adff7650b1bd8cf87fb9a5f6f5d615
bf6ef106f231a59a06000ed28b34939dca46c1b0
F20101113_AACLWW williams_k_Page_074.tif
24205b7d1356187028d2d89c3bc27780
333259d295f95c533a4909948c1576ee66e93a76
2367 F20101113_AACMDF williams_k_Page_094.txt
bd8a37475560854a9e95abc8a6f1c1a1
872302dd63a3cc64a21d9f286b1c668b25fe7759
56738 F20101113_AACLYA williams_k_Page_012.pro
862762c55b51737818dc06b5548ef0ee
d47f0d08fff69152fb2c0fe129b701c2d7c45498
1316 F20101113_AACMCS williams_k_Page_076.txt
4d27fb6c574ba76b88430bdf29578763
a2dfad878cfd7cbd58084a8eeabe0c1ce14ac964
F20101113_AACLXM williams_k_Page_099.tif
f6f7379c29ad78eb7f1d39999d4ddbe9
943a6687675edabbe76bfab0139fb24d8eab87da
F20101113_AACLWX williams_k_Page_075.tif
d0434150939cd839a3262e0d1b077b29
b4f2200a8abc01eeee0df42bf55bfdfb80943d25
2742 F20101113_AACMDG williams_k_Page_095.txt
271e2d37368afcd2570e96d784cb0509
3d5269156aede04b9e8de08686d4f56c6ad97630
54438 F20101113_AACLYB williams_k_Page_014.pro
d10361549b261bccef2bff64b65f518f
d3e2bcbf07bf63a0834a402a1ba89411f3eb1d95
481 F20101113_AACMCT williams_k_Page_077.txt
dba87db3c71f9cc37d93ee0bdb3258ae
ee9ca24a652b07ae98b054931ed3da6388080012
F20101113_AACLXN williams_k_Page_100.tif
aebfa72a8589ed6a6fbbbb53106b352f
aa4546f626a190da99b67a61a5ac09b7792de848
F20101113_AACLWY williams_k_Page_076.tif
07f8bf80c13c0a816b8e60db275c6395
49f89317c12ff97fce5e2d300f40a92775187961
2837 F20101113_AACMDH williams_k_Page_096.txt
df3bc484a32ebdcc150656b424034e3f
d92a65b895fbc3d13222cea80ec5714730b33c94
51779 F20101113_AACLYC williams_k_Page_015.pro
7a71b9f525af871c0c0f323795e9ba81
bae596d37b940e85e340b6a29fff5352a718fd79
470 F20101113_AACMCU williams_k_Page_078.txt
4d75e36681e5fe445a782d0fac398d98
bdc1d67e5243eb60c7cce48dd58bdbedd5075e5a
F20101113_AACLXO williams_k_Page_101.tif
767f95696438d64e6b46e52c1d066ab7
088a6e2f08037df8ad53ce0e0e9b608ae5c2fc89
F20101113_AACLWZ williams_k_Page_077.tif
e6fa924866464996eb7af0cd2178f4c7
77b2c16f5c6c232abfdc13de36e5086d1afed56a
2654 F20101113_AACMDI williams_k_Page_097.txt
4150c42ac904132f437b4e651211c574
05df352ede307251c01978943012003f946330d5
49602 F20101113_AACLYD williams_k_Page_016.pro
2a1d1a4df563f54808db1f0a67759c41
86a8b94114b3449dc244ef29a97d0ec69c921164
550 F20101113_AACMCV williams_k_Page_079.txt
f1e364338d54e8b0b6e437e94a903e29
431fc394d3c61a7dbedc220c723ebbc50fa97943
9377 F20101113_AACLXP williams_k_Page_001.pro
6ddb60da5245711afcec7497886441f5
32b61c8be59a9c0227ff6691dbe765f2817b4059
F20101113_AACMDJ williams_k_Page_098.txt
42cc92ae6bc6c58bfbb851acf64b852b
536dc7ea5f498223221f760085c93e3458585a8a
52265 F20101113_AACLYE williams_k_Page_018.pro
acd76a5cc6fb4a4cefc1d3c20a7e9096
61f02358a9c0bd7bbc9eef24f3b6c11e37326e4f
604 F20101113_AACMCW williams_k_Page_080.txt
5d254c6b411676305843b07fb03749f4
db16fd6484fb4c3958a3e8d161506fffd84d1698
962 F20101113_AACLXQ williams_k_Page_002.pro
f6dcd5b4c9a78299340af9b308299fa8
ce4d737c1d2ca145934bc4b407a6410ed9158e87
1323 F20101113_AACMDK williams_k_Page_100.txt
7e79274a9ab47a036b2321017b6b1e24
e76700494bc41fbcb0996177262e90035efd7101
56947 F20101113_AACLYF williams_k_Page_019.pro
7228d2f09dedb1354996f6416c230254
b7c67adc2100b4592d2530062907cfa44fda4144
3211 F20101113_AACMCX williams_k_Page_081.txt
e609c5f0e4aded15685b9cdea180f1e1
077700dc774328c96a4e4cdd033fe867c4dd0484
2979 F20101113_AACLXR williams_k_Page_003.pro
a0a7057ef72bb199dbc894c1351302af
6e64b8cb429f192c7d526009b6e758a396d80d71
429 F20101113_AACMDL williams_k_Page_101.txt
dd5ff7eefa5901a6a9826c4ca20dece5
8818f24c7420998b9e2430442a9ab53bfa33269c
43074 F20101113_AACLYG williams_k_Page_021.pro
7b4ccc5a28ee39cf5696f5095cd41510
585ab10cf23c061e9c5d0e6adeafe9d8d743e4f1
2246 F20101113_AACMCY williams_k_Page_083.txt
b4d6f1819b0d499e9c830f8b6ff270ab
e58ed4033f4ecf5484e6d65ccd90d21a435f3291
36550 F20101113_AACLXS williams_k_Page_004.pro
a16605f36b6e908fd8d149d9501d720e
a3423b0047fa33752ce697568f5235d83882407d
4442 F20101113_AACMEA williams_k_Page_078thm.jpg
a7bcf5ba05460472b3b7df136196bfac
7d1bdc093de9ce89f7c5221c5ff181eb34ebc30f
2204 F20101113_AACMDM williams_k_Page_001thm.jpg
4d642e6e0fc29e21f6788ad655c8cb85
13e0927d192a9b01045032a07280c0334d3f767a
49467 F20101113_AACLYH williams_k_Page_022.pro
600c2d0dcadfc077cda72dd0d088163e
0a1e4fc9aebe45319fe0ecb3790eeb04ce0306b0
1947 F20101113_AACMCZ williams_k_Page_084.txt
10b120d2811751987363d65d403e90b5
dab2e27cbb0ea36e4f9e1c4dc7ec75f1d6697ca2
75264 F20101113_AACLXT williams_k_Page_005.pro
bab574c8ed83af5f4dfeac241a9ba2a3
3d96bc0b981922555734f347bc4eb6c496ff7252
9296 F20101113_AACMEB williams_k_Page_077thm.jpg
f8573260c6ba90161a61f7bcc23af403
36f3d971980f0dc6c227eb35a2ffd8002ff0abbf
1237506 F20101113_AACMDN williams_k.pdf
ba9e3d182e714e764bf20d5a92c9fb2e
1ccaeef6ee79296baa9ec69fa46b334fe48356dc
8702 F20101113_AACLYI williams_k_Page_024.pro
345693a86dda587b974eb875c04de298
4872052bf47d8c2b063d77e6370b08fa3ce7f5ed
8965 F20101113_AACMEC williams_k_Page_089thm.jpg
9f345fb08556270cdbed9383a583a8b7
499f09aac06c67690598dda76f1bcf48219f01fa
38621 F20101113_AACMDO williams_k_Page_095.QC.jpg
38139a11b05ffef0e93f264986992fb1
6a7fc7562b51109f8b716c09550f139ff3d106dc
10124 F20101113_AACLYJ williams_k_Page_025.pro
20beaeb6e403e850d887d33566f2898f
8b922452e98a0fbd8ed21ad997df10cd732b4b0c
61626 F20101113_AACLXU williams_k_Page_006.pro
df0d340e7f7edf6003392b0c80ceb9e9
ac7c0b1f81002ac18943f0492f52b3176c83df47
37033 F20101113_AACMED williams_k_Page_056.QC.jpg
1dcef48f4faf7ad1bd4d48a998ccd5fa
438eaee25f3e1654d54b78f78743b22db2dc63f3
35886 F20101113_AACMDP williams_k_Page_020.QC.jpg
b4afd8a5bd56cc52d94158add938943e
6e579483ec05ddad91e88b2b59105d1df83626e8
51903 F20101113_AACLYK williams_k_Page_027.pro
d8dca1777aaff4c77bd0d50eeafa2ee9
dc29e2d71ea674ddcb8c24fd68f7a541a3c1f2d8
46375 F20101113_AACLXV williams_k_Page_007.pro
937ff91f1cb48657c025e9206f98b5bc
da72f02510eb2795c496f291c09c5441331b0cc6
4370 F20101113_AACMEE williams_k_Page_057thm.jpg
124258371f7c6d19a75aeabfe3dc7d0c
3e9df989e6ec7d20c74be11a8a6a79d15b960662
37495 F20101113_AACMDQ williams_k_Page_012.QC.jpg
1676fcc2d942a0f1c38780566873abe2
125533bb9af4434ff5e2d7e4c3e509d877367795
50826 F20101113_AACLYL williams_k_Page_029.pro
f4e3ad57e432013987e5968c3bf39943
76517239b8792c960c6af9d342afe533fcd8ac5c
55738 F20101113_AACLXW williams_k_Page_008.pro
37b8656d8fd2465e683246897c59e59d
ebabfa6c61f6d5d43995ec6f4bdaa59c6c59bcde
8731 F20101113_AACMEF williams_k_Page_045thm.jpg
38b10c2ce2a4ab9b31c5ac937cca81e9
394355410794cd2fdc21dd1045d265087eb39e4d
28381 F20101113_AACMDR williams_k_Page_021.QC.jpg
df6281ba4ed1055c1b7e294b46181e27
d627faf163d09cdf5c525b3613c5b79d50975115
54757 F20101113_AACLYM williams_k_Page_030.pro
f098d58c605cff512a1e0c22cc994f21
5b4fc1898244977679039ab38fc3f5ce33f71a58
49470 F20101113_AACLXX williams_k_Page_009.pro
bebe28f27ec2ea676ca7fbf0663d6007
b335d5637f763ee61636eea6374185889e30bb7f
9002 F20101113_AACMEG williams_k_Page_013thm.jpg
b5aa81666eb3c987f009d393c01029cf
3bf2b332b17ad8d05c8cf4744d7a857a3a79a298
56757 F20101113_AACLZA williams_k_Page_044.pro
1e843c428c68f40728723abf10ff5183
641cd493036122b0ff65c81a48780f9101724830
34603 F20101113_AACMDS williams_k_Page_075.QC.jpg
cf72db5ef95a24b1ca41c26990927c14
aee6488d0fd6bf8c3f357fac541f9b78d2a95ee9
52659 F20101113_AACLYN williams_k_Page_031.pro
65090cd2c24118fb471af045e2fed8f3
1e618874a92e1f094fc843066931c7031fe8c8f1
36522 F20101113_AACLXY williams_k_Page_010.pro
c6fdcfba01aed1877ce2caa382962aec
38eeb0fe010dc0cb9c09ec0a39b0d5c3faa2ac7e
38372 F20101113_AACMEH williams_k_Page_013.QC.jpg
1e98cf205a53d028dde72b8c844a4bcc
7facb2f06396df385cf94f819f4df5083dabfd71
57023 F20101113_AACLZB williams_k_Page_046.pro
5348abf977fde74e2f79a0d4e7217004
d301cbd232f9b0a70aafca20b7bc0da0e9585ada
32084 F20101113_AACMDT williams_k_Page_093.QC.jpg
1e29cf41a808ff22451498bc5f0b70e5
897782995157a2a5338ba775f233df246c5fdec9
54457 F20101113_AACLYO williams_k_Page_032.pro
853037f7407912a0b201afebf1561fd3
049138bc5e9127e2446e58a31f59dba00e0f5a65
50386 F20101113_AACLXZ williams_k_Page_011.pro
9486188777f0bdee7106118066c7cbb2
e9a56f69069196d6c3218c4bcc7faf6c5c0e2d13
9640 F20101113_AACMEI williams_k_Page_074.QC.jpg
795fa2c8763cece5630fb3f71b70b0aa
3b5e0e51e93042992fd9e59b09382d05853d023a
43819 F20101113_AACLZC williams_k_Page_048.pro
1bc21888e4ff8aa9f328f7152bc49c76
ed4bfb4bab7b7653436d3a4f97fa8ed98843006e
13497 F20101113_AACMDU williams_k_Page_060.QC.jpg
8ee0c88ee4a9982b1b09089da25b31a9
91ca6da021cd5a5fdc98c0daaa73e498fec2f108
57731 F20101113_AACLYP williams_k_Page_033.pro
0d630020bfe5e68ab7b57e067de10b31
98c2d2167dd1287e6b6bafbc529d539e26f5db9e
37769 F20101113_AACMEJ williams_k_Page_098.QC.jpg
2eb33441f6e22acf4019c1c43fd6ea76
fda1ff23a9e8189ffea9ee01f5cb7882b6be7cbd
15440 F20101113_AACLZD williams_k_Page_049.pro
835900d6c70a77d2328c047f9e5e49bb
c7d980c242eed97300e59552f6abe343e7774fd2
29753 F20101113_AACMDV williams_k_Page_083.QC.jpg
10dea1244845fce3f73abd5fbbff12a1
e43f3b93799ddb7f7e67e9cdbe9e6175724c42c5
57192 F20101113_AACLYQ williams_k_Page_034.pro
0b2acdfd68075df40eeab2418d606156
66ce5b7dea08972e45f6d5a77e463f0e1e12e77c
13539 F20101113_AACMEK williams_k_Page_059.QC.jpg
2fddc8974535b3e5af9513ca641f9889
110f4014d29d36c4920cb0d714433594bc204b7f
50092 F20101113_AACLZE williams_k_Page_050.pro
4db2581b73386485e5b4f27f68cea2d5
3acc44faa3c83f77602301a97eaffbfb98eeffa0
996 F20101113_AACMDW williams_k_Page_003thm.jpg
20c1137c74e9a252eacf6d6937ea570c
737bc34ddceae7f31bfa7c46657c252c507c2163
42875 F20101113_AACLYR williams_k_Page_035.pro
220b321d7a49078aec0db0220b87dec3
5612dfbbbbdfd1bbb356371e71aa5d2902647b53
3319 F20101113_AACMFA williams_k_Page_026thm.jpg
3a2b100b3e7761ddb284630adefc66c8
1fb69540cb9a4ecce64d2e67a63831bd47be8b84
8844 F20101113_AACMEL williams_k_Page_020thm.jpg
498bd56eba6f470bc5ae556a14e7ec13
55faf0ae7c088917e4cfc955d468abb59d8f8dd4
23999 F20101113_AACLZF williams_k_Page_052.pro
af3c476b7b2610860949c7c476c921df
b5572fdac328c24a8031957e5aec955c7b75caab
2571 F20101113_AACMDX williams_k_Page_049thm.jpg
68ce191ac385a5bef2cd4a5d97774856
1b3bc4799279446553fadc0265c9dd147f6a2a59
57471 F20101113_AACLYS williams_k_Page_036.pro
8f4614427a1fc73df1c085433f855332
bd4d26ba3248ecd75f7d735a76096cee632a1293
4175 F20101113_AACMEM williams_k_Page_060thm.jpg
eff7277b78f5d3260b7ad7a0a94275af
031edb4d42f8166ab2318effca508036545b004c
11138 F20101113_AACLZG williams_k_Page_054.pro
3b908c653103604b919e517bdc89573a
57fceb4e60372b9984b985821d6a219e76e5b8f8
36394 F20101113_AACMDY williams_k_Page_089.QC.jpg
6d3cbe7a4ee4b151bc66885ce54ac7b7
4974cdd19237c51137ca7551897e1d59525b85f8
54770 F20101113_AACLYT williams_k_Page_037.pro
1dda9a2c0c8c5a6b0e345bed8ceb6f81
10b219c78c050a42516ba4a3d9af5eaae2815880
35767 F20101113_AACMFB williams_k_Page_037.QC.jpg
99ddb6476988e7f555ffe58af942bd55
0374f5d53d5babea7df861860821191f878c6512
4992 F20101113_AACMEN williams_k_Page_100thm.jpg
05c56d9d602032db82c3636e41cffe38
5587c4b6f515af3531fbdc8ea169d30ff4bb110d
6727 F20101113_AACLZH williams_k_Page_055.pro
64cf14714d7d4d106d666794823cea28
e5349dfa252307f7c138ed7d30eaa7f56e330edd
38562 F20101113_AACMDZ williams_k_Page_096.QC.jpg
21fa15fa3bb74e3a2b6b60c8d7f4e88b
1ddfc1e755d9c863aa380123904dd8cf4b5a792a
51373 F20101113_AACLYU williams_k_Page_038.pro
368298913e6ca6c56c20beede43d35fe
a913ab166a2e0338bd083473cf151b827221a9ae
36170 F20101113_AACMFC williams_k_Page_027.QC.jpg
3ddc742218a7b8ff55364c128a7f6920
6bce22e507d92706e812f85c0706f2ac0810d438
37620 F20101113_AACMEO williams_k_Page_034.QC.jpg
7c4eb79a0d840adcccb586afa160c8dd
dce3de713f16342a8668f967e9e7c5d8c157396f
8756 F20101113_AACLZI williams_k_Page_056.pro
d326b2a19ee33e61d4cedc5a19fed745
9b47a6c3450050eb480f178f07ca6ba195e733bb
37023 F20101113_AACMFD williams_k_Page_077.QC.jpg
52253f1aa9539e077de091ef0635fd65
3080a2fadeec0d06de00c488c13d71836c5c0b77
8810 F20101113_AACMEP williams_k_Page_071thm.jpg
75c47d2de690f90c3e9fcb6c35c4ab61
7ee34d78b72c15178d17360d097aa2c5121c3524
13911 F20101113_AACLZJ williams_k_Page_057.pro
5d0ad07f74b2554db8d6daa673209ef1
e85262c001953234641f0c02a9e539ae6b043628
51208 F20101113_AACLYV williams_k_Page_039.pro
d3f5a6fad3c0e63f546e5d46889a6668
9b802a9d249a045f9456d840ae59507b999c397f
38593 F20101113_AACMFE williams_k_Page_033.QC.jpg
4965a6aa0aa95c286bd24c2f39caa53d
79a55e1f3d717721dbf93134421743611f8b9462
9176 F20101113_AACMEQ williams_k_Page_067thm.jpg
3765edb7cb5188efb6416c3e3b39616a
a271bedc362777d6eec2169597d81777c0fd0726
13394 F20101113_AACLZK williams_k_Page_058.pro
1b10cb739b1e05a2c140b955ff17d3db
e862f9af77acb18b3f079ffc23d81c13ca94dc77
33264 F20101113_AACMFF williams_k_Page_008.QC.jpg
e96d9a8968c4bf96510ff3ed54b7aa18
89f0d83a78123dcd6a7cfc2cfbaaa7e8824c1072
7372 F20101113_AACMER williams_k_Page_035thm.jpg
3cd46d493ec51a9f1790ea4e9d76caa2
4e45ac1b19276e5a7f4c0e51ebff062b690354af
14689 F20101113_AACLZL williams_k_Page_059.pro
917b07b085520099d38bf76d27c211c0
abd1525c14eb6b7b0007ffaa7de192426973820d
57132 F20101113_AACLYW williams_k_Page_040.pro
95e4dab3a0a8c978baffce9a6b6e7fb2
503de983d6e6ae452ac204af3c5fa698c8c9d764
8789 F20101113_AACMFG williams_k_Page_042thm.jpg
ba3f60795e018ff3480ee577fe7d6311
64d48e747c91597ad2ebe2c27975ea087625a578
37083 F20101113_AACMES williams_k_Page_030.QC.jpg
60da396e8dda9d9fceead01b1cae7bc6
56cb7ac771a281d9236cedfbf409aa6280640f16
7734 F20101113_AACLZM williams_k_Page_060.pro
cc8952cf6a658dc0b875390e137ac522
bed03f37efe1e8d8f7e5c5d123f6a4aeec902367
56211 F20101113_AACLYX williams_k_Page_041.pro
df38a1500569049f3683bcdd690a18d8
94f9eb27d1e8bfe470d47ce5578f3443e611888c
8825 F20101113_AACMFH williams_k_Page_046thm.jpg
982e1d0194467a64748f7db2d1b5d53a
e879b59fdd6ebcf384989919e582450e2933f80f
10767 F20101113_AACMET williams_k_Page_080.QC.jpg
ad3327bfabaea406d2ac2d9c2fcef4b4
5f60bca9bc19600080f465d0b0d2be07e7ba120c
55539 F20101113_AACLZN williams_k_Page_061.pro
1e95e146c0e727715ac7a412a937b46b
e39af6da6631210a60dd7c74e9f2952c23f83950
53458 F20101113_AACLYY williams_k_Page_042.pro
aed62ddbcdbb9ba7b7b55b74eeced353
c7e704c797d70f743486f607d097844596993253
27510 F20101113_AACMFI williams_k_Page_050.QC.jpg
717f925b9749a9f8dbaadcd03d87a2f0
e44a2fdbe092f57796dfc3136533d34d320b9ff2
8868 F20101113_AACMEU williams_k_Page_072thm.jpg
bd38699faae1dbead0b6a2729f50676c
3dcbd7ac8b948a24b15e103d1107a43f5c2c8fcd
50965 F20101113_AACLZO williams_k_Page_062.pro
5c27a5ac62af8390375a5dc94bb51348
975dc1a58f8e30d6cc1946e79ff699e94ecacd1e
56701 F20101113_AACLYZ williams_k_Page_043.pro
fb8099ec937157881bc23be8bcb86fc2
396b76cf5054c72e8df5b49864055440070228fa
13556 F20101113_AACMFJ williams_k_Page_058.QC.jpg
c4eacda7118f95450917edc298676091
0cb8e20837850571dc7fbacd91c63f9192b31257
29305 F20101113_AACMEV williams_k_Page_082.QC.jpg
cb1c824ec3a1ad6e24a55d6738b62ec2
b27c24e24da4a400c51ba874c87e7b3739ae6c06
52923 F20101113_AACLZP williams_k_Page_063.pro
5fea4db41a9b6b249e3a166232a343f4
4477a645a3ee07647a8cbd7f73184c6bf59fe0e0
6251 F20101113_AACMFK williams_k_Page_004thm.jpg
07a81feb1dab8c815f1b33a71f5dab30
c42e0ff9fc8fd5d68ff9121f3d2d7ab2c082f5fc
36929 F20101113_AACMEW williams_k_Page_081.QC.jpg
50fae61e245ff480a29dff358ac5dc08
5217f3283f0e4491f8c1937b756c06a950fd2cfc
56736 F20101113_AACLZQ williams_k_Page_064.pro
a1e7c609464e4037f9d3992d8f8a3038
524b113f07de583e9a74965bbfed05296b525329
9074 F20101113_AACMFL williams_k_Page_033thm.jpg
4585fb6a1ea72a21b3584247761386fc
d41a9cdc69df4a6ba6ddb729b530485207104232
9096 F20101113_AACMEX williams_k_Page_012thm.jpg
25d6b284938ff0129ddbd040ee1d98eb
abed81b3c35398ce8f2aeaeffad8ac2c62c65be7
53680 F20101113_AACLZR williams_k_Page_065.pro
e9cfb04f3612e2cb3b8a88c7ae55b770
b305fd97daaeea8d591393d766eee34c1490c43d
35631 F20101113_AACMGA williams_k_Page_032.QC.jpg
5b284cc7f52de52992ecd0997f757f71
94650658992ce07f6bed928cca565f1cd3025640
27938 F20101113_AACMFM williams_k_Page_084.QC.jpg
f454d79fb2f554267a9cf0b5f59dfea8
f9e0f21af617ffbbdbd73135956d0fbca7e70ef2
4385 F20101113_AACMEY williams_k_Page_058thm.jpg
89533909a1d3fcb42b96438e689c012c
5a4f91af4eade7344f53d879dec8e40c266a037a
56580 F20101113_AACLZS williams_k_Page_067.pro
3e14f67ae65fd9e249ed393e600dbaae
22bba6a22f318dcbc14b5b8c69f1f1860c01df8c
7320 F20101113_AACMGB williams_k_Page_054thm.jpg
66e94ec9bbc8627790af32d35b9819be
0d853c44aa7353dfa4849c9c79042516ccabd11c
34260 F20101113_AACMFN williams_k_Page_065.QC.jpg
f8299c0b062311bf745b72dadb616216
2f5cfc6facd9c1f2cdf1e184d6fa96201bd84b29
34903 F20101113_AACMEZ williams_k_Page_062.QC.jpg
e242d1ddb081683e6b723d36d555b6b0
595c7d44b2d75c588156c784f19aab719ff2c9af
55730 F20101113_AACLZT williams_k_Page_068.pro
4f798e3370a5185cffa8b94f0ee9a993
5404c572c482114ff163ca6614b49d62c91b990b
546 F20101113_AACMFO williams_k_Page_002thm.jpg
ef0cc8a40c751fe2e89d6bdba1ac6352
ae41eef160cddc4e626543fa9f92bba48fe666b8
52707 F20101113_AACLZU williams_k_Page_069.pro
3bebbd937eadadc531134795c52d5aa5
f2369223adf48f1c16d6625f6e912c87960beca9
153793 F20101113_AACMGC UFE0021317_00001.xml FULL
e446410beed7971979cbf4cd35e054a9
512f2af3c116cc71e026863064c391d4f86157a0
9036 F20101113_AACMFP williams_k_Page_032thm.jpg
e7506c9c85ebdac60c936a6cacc72b5b
1f03cc4d8238442a1f90b0ba408471928e287ec1
52139 F20101113_AACLZV williams_k_Page_071.pro
6d4401d609e4611223bba26d220ead77
213131c0dbc7b7c71e61a542467f25d2f7c8737e
2379 F20101113_AACMGD williams_k_Page_003.QC.jpg
5737e7f840f174aff3efca551e9a6037
03c66650abcce4878bd44ed252538b523e0ccf06
25982 F20101113_AACMFQ williams_k_Page_004.QC.jpg
8928a2a2a37e7cbaccc07569b658c9b9
86f57604e50582b3bc04c150a2b96e81a2ff988a
7451 F20101113_AACMGE williams_k_Page_005thm.jpg
1bd8c77feb72c4106be14fb51cc2f0a4
5c37de707776dbb49596fb436daccae82b05039c
9189 F20101113_AACMFR williams_k_Page_064thm.jpg
b20355ec4683430251356be329f59c9b
62e3805b7880c971d8f17604b39ab8e325f4adf0
55201 F20101113_AACLZW williams_k_Page_072.pro
42d800ab8aab90d745e2f3ed1e3d3289
17be6fde2ca5c209983ba93e9bcedb39f488906f
29915 F20101113_AACMGF williams_k_Page_005.QC.jpg
fc02f2a66e81e92772247ac4824338d6
6e50f9a529ff6798431b0ee2e4b97d51cafdd03c
1229 F20101113_AACMFS williams_k_Page_002.QC.jpg
5b8a3dbc2c0d69c0dcedd175d9ba863c
bbf9e6a64c32fca655cd1b9599b30d6f7d3fdca0
56170 F20101113_AACLZX williams_k_Page_073.pro
676bccf2fe3c425d82f2ad315c4e71f5
de0fed081f61311b496e34bfdd08e2d0a3bf8203
30411 F20101113_AACMGG williams_k_Page_006.QC.jpg
29f09cfa315ec3f2b59f0d56cbe4cc6f
d64758f7f033d6e68c11b24925e5e54b4c7f0d91
37105 F20101113_AACMFT williams_k_Page_068.QC.jpg
dac282e18e49534b2b076de86f93daf9
3e98953373baa3e23e618df64c20ad1e4b93decb
65972 F20101113_AACLZY williams_k_Page_075.pro
e2ac0343b8025c375ac0bd979cf5feda
23ad1663419274097035474437f81fa5424368a8
32028 F20101113_AACMGH williams_k_Page_009.QC.jpg
f2e44c2388e2ec4d7691d5b002423f9b
37b2bd1dadc5fe66905a6c788425d84dc8d16c93
F20101113_AACLZZ williams_k_Page_077.pro
229aac07677848f486ee3cbfcf9d5903
3688cd09b827c5ca49b5b6d2a1eff98306cf6545
8739 F20101113_AACMGI williams_k_Page_011thm.jpg
61895e2c9bad7cff1c1a61d836a77ce6
a5373ac576e6a6b73655ef88a5a3556e4c18b0bf
9373 F20101113_AACMFU williams_k_Page_053.QC.jpg
48029c68a6164a7f0213dff093e9db8e
de433237cd98da3c79e650a0e43d20c5b8d8e109
35539 F20101113_AACMGJ williams_k_Page_011.QC.jpg
7839c35fa933b63636d7f6915dedbe19
acc4fc805c81fa6cbffcbb4932c77eb519f7aac0
36173 F20101113_AACMFV williams_k_Page_066.QC.jpg
21fc0401193f5f96681d0934a858a5f7
3df068f8ab29785fd1087cb6a445d397eb867259
8617 F20101113_AACMGK williams_k_Page_014thm.jpg
8f40a510ac51c459afdd039741f9168e
cd62eae906698cd105bd89a07a3d90f26fb5c4a7
8431 F20101113_AACMFW williams_k_Page_029thm.jpg
0ff5a7c8e2d03a867118b3513701ed2c
2033e638e1b15756cdcb031456735b985d872082
34801 F20101113_AACMHA williams_k_Page_028.QC.jpg
3658126dcfe740cb4f46481ecf755103
3dc2c62a004b45eba5c30b28c7dfb769ae656296
35260 F20101113_AACMGL williams_k_Page_014.QC.jpg
b4ee4d44d60bc2035ddd8383488fb49a
e6a366a4bb0ab3dd8609d86d0d42f9f5e64620a4
8572 F20101113_AACMFX williams_k_Page_081thm.jpg
640aaf305e1d23c3a7b5e971fb9f3277
d009c290d61c6a3f55b90beeee4e82cc56358b56
34481 F20101113_AACMHB williams_k_Page_029.QC.jpg
f8223d2cadef7017e0932241f287752d
c0cc5c51f9838f72e0d2ea860aad0b06c6527a78
8841 F20101113_AACMGM williams_k_Page_015thm.jpg
16843322932ecb3daa8ae262fb59fc59
6a830369198803a505d76b8aed0da727f2eeff0a
8046 F20101113_AACMFY williams_k_Page_009thm.jpg
9aa438874c0554cc6dfdda4c770860dd
3c28f7a734675e7423d5921fb30289dad90e14f6
33855 F20101113_AACMHC williams_k_Page_031.QC.jpg
8f4db4964c00d4a9c97b5e5367ac4ece
eecc0acfd825c7e3966d5d068048c19eb1eabaab
35838 F20101113_AACMGN williams_k_Page_015.QC.jpg
cd6929469f6f9774b5ca658f98955c0a
6317715b2fe4c6158fe51a70edfc7a2bfa4b45a4
10769 F20101113_AACMFZ williams_k_Page_049.QC.jpg
5f077d20fd7f1fc5e7f4973194453afc
c1b50e8ad5985b8a1c3d1d3d7a297da8fda7339f
8341 F20101113_AACMGO williams_k_Page_016thm.jpg
8ffebcae404766a197b6e21d1d99c6d4
e2d76f690666c767af061efd429dc4950c0bad87
28389 F20101113_AACMHD williams_k_Page_035.QC.jpg
765272d384b0b661da27a671e61eab08
20a180723bf355615fd4c93270e8cc1638746467
34012 F20101113_AACMGP williams_k_Page_016.QC.jpg
7cdd0c0b9360bbf1930337148f2f3402
6dfd7b905d88d96c8ac79b4ee61b6c10e538ca45
9265 F20101113_AACMHE williams_k_Page_036thm.jpg
5996589b4722b19cd83cf7b0fa79fb88
8e3c8c4ee8f79978b7def6a436cfb5f13a9f6057
35245 F20101113_AACMGQ williams_k_Page_017.QC.jpg
6c3ab41fb549fc83e9ffd80b265fb731
cfaab68b04b7e54f72a407ac01e27ae19cbe87c6
38479 F20101113_AACMHF williams_k_Page_036.QC.jpg
dcde6564a6fb375dfc39531fe47468d8
d4c8df5e4ebcaecb89430a5c103869e89c8b3ac6
35687 F20101113_AACMGR williams_k_Page_018.QC.jpg
7095777af952fc16abf5feb6d79e9ba7
3068a55999db9549e9c3152b38b515f8df140957
34947 F20101113_AACMHG williams_k_Page_038.QC.jpg
f0493647a388710c9fd2df67401e8bac
f97700954ccde60d43cb08b65ee8d34d6ce819df
9175 F20101113_AACMGS williams_k_Page_019thm.jpg
a8df8e609d84e7d4c5c00631a7ecff7b
448534e8268f19ef00bc6d2c3710ed2d58ebb244
8545 F20101113_AACMHH williams_k_Page_039thm.jpg
a252c19abcbbc3b221e6c3e22b6f72e2
15e28a43c5e87b8b0cd78ee659ae918fe4ca4e72
37376 F20101113_AACMGT williams_k_Page_019.QC.jpg
46a83e1eb3c3e726ac39c8b2c515035f
fd1d129749552a10d9701683b5ab0a92cf90dfdb
9295 F20101113_AACMHI williams_k_Page_040thm.jpg
6da461546d3f7a25cec79ce6b3a9507f
d9324431f36c97ab6f50d8f95d4c83af3f2cb646
7467 F20101113_AACMGU williams_k_Page_022thm.jpg
b1a9878c367097d7a347d29ada7fc815
676a219ca842ceb6f0718db5150db90e3ae449e5
34681 F20101113_AACMHJ williams_k_Page_042.QC.jpg
9749b8bb1c5c35fdb90fbedc0633f7e7
ab81bfa08637e89172185a311c0cd4baf5e9b85b
29565 F20101113_AACMGV williams_k_Page_022.QC.jpg
e71e6d1370f4c7fcd0581b145be8ae05
91cab28d7d7fd4d2bd23ff02a539820721eefa3b
37964 F20101113_AACMHK williams_k_Page_043.QC.jpg
0db3fc7423b2ee5fc4ea4395bbe9e0f4
4b8e355ecb9e78084e3f019c8aee843353c744e9
5050 F20101113_AACMGW williams_k_Page_023thm.jpg
4fe847c6ba13cdffd8c780e7d36ba94b
314f9dbc6e2e39eafa28336d96a66d5f6821d122
9101 F20101113_AACMHL williams_k_Page_044thm.jpg
b0667c4fb74b9bd012b7258ffb729aa4
413d96c763448c5e90dd4fc6634e5bf35016ebc5
33686 F20101113_AACMGX williams_k_Page_024.QC.jpg
d15f4c02b4b09a1b0305c82e6378e72a
128a81ec7aebbbd27de8056284f4cb7fe37c7f4a
38354 F20101113_AACMIA williams_k_Page_061.QC.jpg
628677ca877e1f9519190d061f51753e
770146f316b76688d53d270cf97d31889fe78ecb
36708 F20101113_AACMHM williams_k_Page_044.QC.jpg
46b415a425b0e07e4da1b09fb00535b2
cd70274fe80773b626aab94003c3942bea87b91f
3470 F20101113_AACMGY williams_k_Page_025thm.jpg
355c5cf73bee6511544d2e683478a3a9
4611b9bf4bd97b27be6922b6eb11dfda49c96512
35493 F20101113_AACMIB williams_k_Page_063.QC.jpg
c9f52f81e9ccecab6eae53d41371ccbe
052a5735fb8e3d262bcd94b6b5f39b2034def313
35721 F20101113_AACMHN williams_k_Page_045.QC.jpg
99ef0ff0549ed4ea5728e522f44f96eb
e60b782be370e02acf7af419db005493c00bd854
10429 F20101113_AACMGZ williams_k_Page_025.QC.jpg
eddc3f4ef424561ab74a2cbba4f6f3c0
a01478c5330e31aadbe796f577ce9eba5f782ff0
8588 F20101113_AACMIC williams_k_Page_065thm.jpg
91864436efe2a6880ff842ec614b9051
f0be09e1bc5c648624d19d915697ebe2019fc7fc
37634 F20101113_AACMHO williams_k_Page_046.QC.jpg
0331c51039ff038e6f6d970e7337f6b0
d040f259e1549999bba6b42ac9d11140588fac1f
9158 F20101113_AACMID williams_k_Page_068thm.jpg
d5dd934e78b8ff3fdc7cbe8dfc3afe12
b04196508c92720225bc01bb6b684ae707bf83a3
8686 F20101113_AACMHP williams_k_Page_047thm.jpg
4b80de8395f748488c3dceea7b4fd860
346a029a1a7be29b18113e25aad9d88bd638607b
7063 F20101113_AACMHQ williams_k_Page_048thm.jpg
440f55458a9fb84c8fcd5060a619ed0d
564de7551433cc807c3d115308cc09287576f1d6
8532 F20101113_AACMIE williams_k_Page_069thm.jpg
3e9c080ca2a41fecd7cc107cca345050
e87abd438e4697480a409b72ca2aa11402e62abe
29220 F20101113_AACMHR williams_k_Page_048.QC.jpg
34d6d0340a89a3f400db15f3d79ac574
f61336e30729f04836114e0c84307348020f3b94
35340 F20101113_AACMIF williams_k_Page_070.QC.jpg
1e5d177947c30d9e0a79abf2c7d8fbff
8093ca1482bd2f1df0b6488e30a7913ef429cb4b
6345 F20101113_AACMHS williams_k_Page_050thm.jpg
e3096d3b4cf5b72af661261c56ed6166
9b366e24dcefe6803ddeff9ecc4863fa1ee8cb9d
34987 F20101113_AACMIG williams_k_Page_071.QC.jpg
8edff1343987282bc50ca464b8d29c60
396b3150f9bb45ce0b0b192de5194ce0b8a4f61e
34612 F20101113_AACMHT williams_k_Page_051.QC.jpg
d112020c67e371543872522eaf645859
4a73e8fc9d2e581129245e52d98b4a8090291307
9102 F20101113_AACMIH williams_k_Page_073thm.jpg
a9e30ebc8325cc5e4f5cb6dcbed18c68
4c845c921dde063a584de5ff318b820a0c07ed11
16429 F20101113_AACMHU williams_k_Page_052.QC.jpg
b646ab3bab267b7fc75fc6c084ccfb6e
9276ae9946e913923377f54306e1b624d6e1d47d
37004 F20101113_AACMII williams_k_Page_073.QC.jpg
4ec7009cd7c0a2d0ba4f1a2ef0ffcc13
734864a7dd5e86d22b127df666def0fd956f35a3
2987 F20101113_AACMHV williams_k_Page_053thm.jpg
a7d2c11f4fd9ea6ead6633dd3ef91a66
fc5b7e12470bc9c9bcf3deaafddfa362ead29784
2444 F20101113_AACMIJ williams_k_Page_074thm.jpg
a2443e249bd6956476383d9770e13170
6d29ea321e710d2f15cb8066e3474165f8b04bd9
27809 F20101113_AACMHW williams_k_Page_054.QC.jpg
4c044dd5e88a9ac9f9aef3c21b36f380
f824b0fcafb8ba6ab87674322fbff29c53fd61a6
7814 F20101113_AACMIK williams_k_Page_075thm.jpg
639675a56ec37c988212dd4b07e95f05
ab71ad7d412c8890c54e469e0d89b53115b9f495
18437 F20101113_AACMHX williams_k_Page_055.QC.jpg
7e597cbe6c20e1f3c815e377a5185132
d1be056fa63f48a4af79437a2a8aefd9f6fdb915
13561 F20101113_AACMJA williams_k_Page_091.QC.jpg
6bcdc999e949b3c4548cc67b36c59466
64e079c28630a27e6bbb38ddc23d582fc27d1e07
3801 F20101113_AACMIL williams_k_Page_076thm.jpg
78919f60dbd14515a28768b3af1b1633
79d9ba37953e752ddb8452e563e21f6bc55c4fd8
9321 F20101113_AACMHY williams_k_Page_056thm.jpg
0181a93be1706f8cf8b2a67bcc1df78f
0ebf41185bbee2591a3e58259bb714af7fca480c
7866 F20101113_AACMJB williams_k_Page_092thm.jpg
3e8babbf975940a819e3f9a4a2136f7e
7df95e15314a7f92c2fb4256f06334ab73aafef4
16054 F20101113_AACMIM williams_k_Page_076.QC.jpg
51ffbe01f59aa70c96428d16384d30b5
85d498621470be9512d50a2b5563a3fb5779b9e7
13940 F20101113_AACMHZ williams_k_Page_057.QC.jpg
3b61ed34afaf1ba48f4433f41174afb4
5ab5b06c1839ccad12c7d3bc5f3e15bcf20a6520
34151 F20101113_AACMJC williams_k_Page_092.QC.jpg
e042b3d4b9878ae22edf9e3ef5257959
020640d1bb2b78e7bfab4dd60199f8d048342835
16165 F20101113_AACMIN williams_k_Page_078.QC.jpg
04ff73ef7c3d66337508227f3d97373e
ba391d483dba9b6f7b027432d8b04fc63f360792
7519 F20101113_AACMJD williams_k_Page_093thm.jpg
4ccd6e6e60202a4f71ce817122e41af0
c33fa5589c68d8e674f92b2fc6a3dd5e29f230df
10043 F20101113_AACMIO williams_k_Page_079.QC.jpg
bf068f6bb946bb258e72d7240bb20681
23e705c3b2f59c2da83cf74db70242544a890ade
8448 F20101113_AACMJE williams_k_Page_094thm.jpg
5e96d0bfaa83e7da935b46d3ff32372f
41868abdf37f6b29a7c7db99c6457778f603be99
3190 F20101113_AACMIP williams_k_Page_080thm.jpg
4b59eb8c279e213fa7876f05ac0a5ff8
d65778f3bd2d743ea125df16bd51558a915609ee
7526 F20101113_AACMIQ williams_k_Page_082thm.jpg
b4b95223bd0483e982ae584dee0b5328
86fadac3865a081ae57aa5852488b866fa72ea59
32678 F20101113_AACMJF williams_k_Page_094.QC.jpg
d381216a3aaecfb36719798605f9f232
5be1956af70522a3d9dbc80251a04417c38eca0f
7265 F20101113_AACMIR williams_k_Page_084thm.jpg
5ed32ba2cec4ee09b590e31566805f33
847737f0055680ea10608d89ee25048b9afd83ec
9829 F20101113_AACMJG williams_k_Page_095thm.jpg
d1336da5cd860eb563d6ca8961674724
2767fe136e7a9b16b161eba2e31e08eb480fae3c
8635 F20101113_AACMIS williams_k_Page_085thm.jpg
0fb87aa93dd476fd4355a706ba025379
2c2475521cb9a26cb2b69a5829ae67f3f0679999
9605 F20101113_AACMJH williams_k_Page_096thm.jpg
ae0a9a21a5879770fec3253f982c5aff
5c9d9a1ce7d2edf6d2532a1c94c37cd8e36d0988
34469 F20101113_AACMIT williams_k_Page_085.QC.jpg
e5f6582bb0be691f9ca157ec8a54ac0e
d67e0edc4bcb3fd8f0246c3a9fb51c78a2d3752f
37473 F20101113_AACMJI williams_k_Page_097.QC.jpg
00f144d5f8c53700561b05e5a590680a
d5ef77f5694e01183b3baa21272e83aa2f753537
30651 F20101113_AACMIU williams_k_Page_086.QC.jpg
66e9627297613d2cd528c2098cbc1b28
53c4893e11203bee55f34b7903b0a966eeee1039
9371 F20101113_AACMJJ williams_k_Page_098thm.jpg
73fa324d47e4f05f88e7b8648012aae7
736440e91a25597a723bc1eeaea3eca06fce8afe
7574 F20101113_AACMIV williams_k_Page_087thm.jpg
295788cb8babc2f9c46ed95b4cb9f2c5
a1de71e611fa7118eccc6ed07a5d757c52d181a8
9484 F20101113_AACMJK williams_k_Page_099thm.jpg
99bbb93a6f6aa4c696e0bba9c7bcda86
a983f7814fcf4edc7800677dbd857a8c2fe71e61
30715 F20101113_AACMIW williams_k_Page_087.QC.jpg
24e073ad1837f15f0adcd68c1205422f
1a044c9fb59d248f4512f354cb3221f63e577eed
36985 F20101113_AACMJL williams_k_Page_099.QC.jpg
ce30292eff4c3424fa4deeee82487034
f465b04b371a748787d546b5b5af838bcf51c2bd
1104 F20101113_AACMIX williams_k_Page_090thm.jpg
3bae08e90f9ecc1d6d2c567c87f99e91
80fd845b888f32dd97acdca19a4d89def5dc0737
4174 F20101113_AACMIY williams_k_Page_090.QC.jpg
3f86b542e413b6473330efa4a544d376
02a13ed61fb58803530ec5f33948911714b91c5a
3543 F20101113_AACMIZ williams_k_Page_091thm.jpg
95bad061eda66d20bbb16047efab28c1
9ba3a7f96a39212725525a5785f4a16f895b75a2
F20101113_AACLIC williams_k_Page_030.tif
93e42495b4a000c3cc08f3998cc32c7a
f44d42077c444ca01457c91ec2ab208e44232e74
106899 F20101113_AACLID williams_k_Page_022.jp2
f7533a095a03ee6ba51c8fce8fcef68c
a7eb32b265ca5deb53173e6ad9d87eaa797ee2ce
F20101113_AACLIE williams_k_Page_022.tif
f2f3ae7a0bcf860eba9ac1c1e6335d19
15538ce2b120281ca36a727a5781bf72c590a8b3
F20101113_AACLIF williams_k_Page_006.tif
ca43aa441b1b8fbf93cb11ae472327ce
d2be7c9deb87b74edad9a5d61cd807f6638c084e
2226 F20101113_AACLIG williams_k_Page_012.txt
79a8aa25d109d0717d57f4c952928e7a
f728724d6e805a2f550c7af762eb9b1cf18bcdb1
F20101113_AACLIH williams_k_Page_079.tif
42fcf69be7eabb5273d1e8e856f536b4
99db520712c90cf0205937039bab57e2d7cbdf4b
35307 F20101113_AACLII williams_k_Page_069.QC.jpg
ae5eb8679bb2a60bc446f42bf6ee657b
b4049d5f35d194061f2d3056a21701e2c357f33f
93739 F20101113_AACLIJ williams_k_Page_087.jp2
95c78f259ce49e3f70fc0b03c61924c1
442945d305207e46e5a2274d069cbf245db25a52
30837 F20101113_AACLIK williams_k_Page_076.pro
1c0aeeb516a6b823bea403122cb29449
a08d1f36f0861152e7d949162fe67654c8beaa57
F20101113_AACLJA williams_k_Page_097.tif
b03810b70bc6dc061860d671147f2b15
17f4d244b7e1c295927053a58f4173789681c4bc
55306 F20101113_AACLIL williams_k_Page_013.pro
de6a139c09f2c6ccdebedbc2c4b743b6
b8241b36a87d9e9a1a6f54649adc5dfc3806a8b4
5272 F20101113_AACLJB williams_k_Page_053.pro
0a47079fbf24ce910b5621218ff7466c
7b939b332469832c76467cc24d072d37dbb6e844
36626 F20101113_AACLIM williams_k_Page_072.QC.jpg
a4f4a228205c74c05e047005731215d3
5ad29dc281496e92d9e8961ffc694ae33443f484
F20101113_AACLJC williams_k_Page_066.tif
91861eafa3a4e9f62a5072fc734ac657
d719626892edc5b5117c76172399d3b33ad50302
F20101113_AACLIN williams_k_Page_037.tif
9b979ad902da26c68d6cdf353dca0a7d
f60694b855dcc8fe1a47bb84ed49c588fc35478b
1825 F20101113_AACLJD williams_k_Page_086.txt
9e87376111b30fd76041919f813c8d94
7788d522b4d3225b732fd820fc3111d0b58d623a
F20101113_AACLIO williams_k_Page_063.tif
2af239c6cae3ccb5c5d8b272efcbd6a2
dce4929ac29ffa92f263e538c529e4904d631ca5
44548 F20101113_AACLJE williams_k_Page_084.pro
3b9b14e989374edf6e9b4e9c0fd7529a
3594aa8b71c4755d99c40783ba25332a1b742e56
117873 F20101113_AACLIP williams_k_Page_094.jp2
17189de076d70a650fa50c1028d2783d
f6b59db1fd2c4e15bf74093a9a20b708aca41baf
95203 F20101113_AACLJF williams_k_Page_021.jp2
3f507191b7dd417b58073da8e672fd48
09a3e03f6d734b4c1b5b924885c52736d6931678
112534 F20101113_AACLIQ williams_k_Page_011.jp2
9a1b1be370a3c2cbf2b135910d813378
3c5f694bac380384e1a82660201f32f48bbe1f3a
F20101113_AACLIR williams_k_Page_058.tif
f427e9885c4d840692bc6fee99d4c9c3
6e057f5d46995990560bd31e89c7e690a81c76b7
4906 F20101113_AACLJG williams_k_Page_055thm.jpg
d1c8d6b89e508008394ee027a1002ff9
ebc207d0ff3ddb388d1836a875cd362eacc861ac
2219 F20101113_AACLIS williams_k_Page_044.txt
47c23b513bfad0eb28be0c845daa2722
c8c52c13f5cc381e1d7d40fd4e71e54c181bd15f
8284 F20101113_AACLJH williams_k_Page_008thm.jpg
d535571e04c6294e9742c4b13481606a
b1fe922490558e1a1120ce36b538e94ef37a86c9
8264 F20101113_AACLIT williams_k_Page_024thm.jpg
a5ebd2779ff295fd54b756b49567a424
418d3f876ffc4b1ed6dae163967637b2df5eed5c
10099 F20101113_AACLJI williams_k_Page_088.QC.jpg
f84db53337a12c3b3c7cbdf7dedd47ef
8a38c27f2fc9429bd39584946739e2cb4edbaaa8
55262 F20101113_AACLIU williams_k_Page_066.pro
91428ca55863ebed88894c2c783f3443
8774986e70f0c8cf80228411a1c52b2338090f71
7060 F20101113_AACLJJ williams_k_Page_021thm.jpg
442d753e32f23cfcd32e85a15c613afc
040a4bb587a8b619a73a312e5b72d0663a2dd230
F20101113_AACLIV williams_k_Page_056.tif
65571a7595fcf052accd0df1f87c2782
547178683b31b36067221d0c2d3e9cbbff38eb22
8086 F20101113_AACLJK williams_k_Page_083thm.jpg
8c8b79514403dc07340ec0bee903b7c6
56b886422461ac4ac030d2f48584be25f9c69765
37521 F20101113_AACLIW williams_k_Page_067.QC.jpg
13f00f995058727f8e3f2034f56a1bec
0192906ed294410af1e4413375f4830075a6bde3
2249 F20101113_AACLJL williams_k_Page_036.txt
2e79ed2f95853f1079b48a8880ba214e
af70271a3c0b2de7fda1f3b9573abd976f920675
F20101113_AACLIX williams_k_Page_039.tif
a75a301037c29a14d6935b1638ab485c
99856e987b0cb542eb0bc184d08aac6294341d1e
8300 F20101113_AACLKA williams_k_Page_031thm.jpg
52bc46afd7ffd990627f1171a64d69f9
683c4e817861977b03e36d6b7b590e286fadd9e3
10606 F20101113_AACLJM williams_k_Page_026.QC.jpg
54c9c944fb00d44c034bc2bb211e0d7e
ecae194803af999a48495b54220d2bd6141b7dad
8609 F20101113_AACLIY williams_k_Page_017thm.jpg
e6e2536a4fd319f43fc3772bf5286ff7
73b3822ad54e22b0b6ae4c151285255da6f1b69d
108678 F20101113_AACLKB williams_k_Page_047.jpg
c68d44477435dc3a8e6905af04effdd2
db66d978058ae42eb55964b8af780310f21be4c4
417 F20101113_AACLJN williams_k_Page_060.txt
0b5639d70ccd18253a9a0fb13b67cd8e
cba71668a9dd585c639d4a3219e75a81162989c8
12051 F20101113_AACLIZ williams_k_Page_090.jp2
b033bc651d2cfb28bb916f4df2e3cc7b
918d3eb4cacc05dc81730e48bcae8f32ee8b8b37
F20101113_AACLKC williams_k_Page_059.tif
bafbe76add9cdfebf9abee8e944cb28d
bc21f7b32cdc0c7b7fbce077e5fce2cc40b99046
7905 F20101113_AACLJO williams_k_Page_051thm.jpg
fb985ba2b4eaedd91d394f687a891db2
87b3e59e5de6c464af31d9b72c2bd9bc7fcd8801
F20101113_AACLKD williams_k_Page_047.tif
8e2472e08ec477106c8ac020bda3a1fc
710cdcb91ab64e771efbd22ad4e3d2d8ebd818b2
112593 F20101113_AACLJP williams_k_Page_044.jpg
1596ea00818e25bc9321e84ac0ad51dd
e937242f17c60ee070bd85f4f1f2a55b524e5ac2
77482 F20101113_AACLKE williams_k_Page_023.jpg
3d8ca595947aec76b14c753c1583db69
065da44a6dafde4740cd400a456d4b3d77b92f09
111897 F20101113_AACLJQ williams_k_Page_038.jp2
2f5084d378391526917a1c950cef7480
904240afe36883c14e3f702f5454341097c82d3c
121381 F20101113_AACLKF williams_k_Page_073.jp2
0b3f196bebad5b9a6563a888906a5d99
f63b1c122f14af1ec10030151417c8b686e143cd
F20101113_AACLJR williams_k_Page_071.tif
85386c8cdc2a825c3d8563f26f983702
fbc3a91727badffb84d509643c3560ce7e1ef6ef
37268 F20101113_AACLKG williams_k_Page_041.QC.jpg
2b2dc43caf70fe6d81f75b511beea43b
db5f239fc9bfb243aec145a7248db3ee62e5585e
96353 F20101113_AACLJS williams_k_Page_048.jp2
c5325072cc5059afe3a844c14549a6b7
d58690638bc8917290c444ef07ea2a66e0a71b0c
6302 F20101113_AACLJT williams_k_Page_010thm.jpg
28091fa7a5e855bc24b843011a4684f1
4e4ff6f06657ebcdd05dac3a424a240f3503170c
113936 F20101113_AACLKH williams_k_Page_071.jp2
677dd18762f0f6a711c9d3332e270861
49ba558bb632db420e722bfab43814ffaf6ffd22
36558 F20101113_AACLJU williams_k_Page_047.QC.jpg
9b35367b7e53ec88776369a55174218a
ca824e98983a7d5b6e58d859b0823bc1ad2482b2
106664 F20101113_AACLKI williams_k_Page_018.jpg
8107aec0988df887aa9e46f4b7fb0c3a
2ef7dcc777baa62fc7bd285a815b2e6a4a5a1e73
F20101113_AACLJV williams_k_Page_038thm.jpg
67ed5fc7534b23a3d6206a52e4c6be50
18821a2450a5dd2fab4ad34a4cc6c2eb3d2e6e52
F20101113_AACLKJ williams_k_Page_013.tif
0a4febf5abe8cc402f4ad92bd91ef14c
9a0bcc3d7ac98c34c9a6ff7108f671611b619051
52803 F20101113_AACLJW williams_k_Page_028.pro
dc516dffa8e6dc36e601ecb23ba74491
48aa632530e825b981333f5fe1e4afebe3a9b782
123379 F20101113_AACLKK williams_k_Page_033.jp2
0bb36708c51c03f0234154eb445938e1
4c8b6f6f17d6bd817306b727043c7aebf1858368
1941 F20101113_AACLJX williams_k_Page_082.txt
d902077968d5079994cba5dc93ae21eb
e6dd6ddf1cfd1c6ae51e000d9abb62e87a0aaba1
53977 F20101113_AACLLA williams_k_Page_070.pro
b9dfbbbaa8a47c722e8635b19dade597
9eb6c6925072b888025972d8a2edbc252823c2df
F20101113_AACLKL williams_k_Page_031.tif
3cf7348ee0ae049a707c26d3248a5f5b
0f11655915cb1c8e5062eeead4b9a9f4fb8ec33e
F20101113_AACLJY williams_k_Page_038.tif
7d6621b019b40b8200f60e12e48960be
5c51903dd8d72a8ed626383f13f1e3386614f361
F20101113_AACLLB williams_k_Page_023.tif
d0830aaaa074b23ce43f44dfd736e97c
7354fd594460c92c6cf5ccd9fa24ddd7b958ad34
F20101113_AACLKM williams_k_Page_036.tif
4eb57e4d590abbe893be2fb599ebba23
16b21e2c79d67dd512141a96b85fb275bf1efd8d
108230 F20101113_AACLJZ williams_k_Page_032.jpg
5fbe4ae7af8b89c387afd919a4801042
4edd1679804394e8e5d52136ed57c9790f8aae01
1453 F20101113_AACLLC williams_k_Page_010.txt
abe6b0f19e286f468b0950751a637060
066ab690b2368cfdbce1e5d562f55cf05c22255e
42032 F20101113_AACLKN williams_k_Page_082.pro
020d008e2eb6a2cc023899d231f1618b
4ec05f08410476fded354dae987e3eb9be798cd9
115942 F20101113_AACLLD williams_k_Page_032.jp2
8784e9e660f332059ef96d1f8582ba91
50a6d0d3284214b21824bbba6924b6a2f0512ece
37354 F20101113_AACLKO williams_k_Page_023.pro
972a05a162b606ee571c462ae16afdb7
b3d36bbc8b84bb5f3aefe34f5a9aef3fc5c068d9
4143 F20101113_AACLLE williams_k_Page_052thm.jpg
3dc6706d5fa7763a28baf30b9d80b087
eab0dacc25e1b181d7ef10df54ae807df72fb929
117343 F20101113_AACLKP williams_k_Page_047.jp2
cda1194e47804ff0eb1c6b07baa23396
a06baa68f982c6b878d20eaea793022a09baaeb2
51698 F20101113_AACLLF williams_k_Page_085.pro
6172a2c3f0410e82cf16a7ec2f5cec34
3f7622a2acfaae3a60a7cdd7f7f7d6432681d8f8
92900 F20101113_AACLKQ williams_k_Page_086.jpg
748b940304faa932ff32e1dd1fe468e6
a301ff2d1e01edbc4c0a5c979e58867d90e54679
9277 F20101113_AACLLG williams_k_Page_034thm.jpg
f210616ee67f87270bf73aa4ab9a0de4
5a275d0bf7bfa01431bb28e06b0bbd31ec1a386a
F20101113_AACLKR williams_k_Page_067.tif
2fad73ecb5436acd863b11abac23147c
efd24f17b0598484fbb8b0a7f36cbf750427f77a
163 F20101113_AACLLH williams_k_Page_090.txt
97f6bc3c87fafd401c0e1e5e7e0af64e
6ba12f0c43eecabf7d742e7d642addbdcd54c3ce
2092 F20101113_AACLKS williams_k_Page_071.txt
a0b5db112b994ce805e1864cdba43135
2a7cb3aef2424c1ee3fbd74663e5bf8a959d32b2
7543 F20101113_AACLKT williams_k_Page_086thm.jpg
8fc9c32f9bc3fbc5682c15a7c35aa552
bc8fb9b9c33ae7a782b09a554630e6b502398fcb
8723 F20101113_AACLLI williams_k_Page_027thm.jpg
90a7cc9de8cdd8f889c2aea69bfe8aed
e9b93074520591e5394830c6b30dbb686a8817b9
51996 F20101113_AACLKU williams_k_Page_017.pro
800c24a97bae238a31c6f98dfa4ff6e6
6d317990c1d5502f58f008c2f462a4cc264da2ee
80887 F20101113_AACLLJ williams_k_Page_010.jp2
bb03c61fb85caa9aad063bb1184b7c7a
31c4b6131cbd470c06cb15573d3f4bbbe018bc8e
32242 F20101113_AACLKV williams_k_Page_088.jp2
2838e1a34bfebd6de2e38832c2128f43
e228244c2fb1923c434347d808e2b01617f585b4
27918 F20101113_AACLLK williams_k_Page_007.QC.jpg
f661bac4274a4cb0e6c83b931d689b5b
3afe32ffb6a257c610194c502f6d840eaf9a920a
2198 F20101113_AACLKW williams_k_Page_073.txt
5811fe052afaa824bfca0289d7991f71
844c2c8c12fbac78d49186568bda8a092edd7271
24537 F20101113_AACLMA williams_k_Page_010.QC.jpg
e5926453853fc6f39184311ed9fd1578
686079608366130bcaf24ad3dc69799cd5a76acf
121725 F20101113_AACLLL williams_k_Page_064.jp2
019e22e3b6dc331cb6a769200714ed53
662038703af7c897cf5dc8e80360ff0d9cf874b2
F20101113_AACLKX williams_k_Page_035.tif
bc9dd74a77fb24b74e48ebf4435df682
7a05c3b426ebf1c77ad0ce0c6129484d73279f47
F20101113_AACLMB williams_k_Page_083.tif
ac1d30644fd36f27ccdda0f6d61d6ee7
696be8f7866bd15da4c335ed9348453c78adb56a
9172 F20101113_AACLLM williams_k_Page_061thm.jpg
b35308bd7e3304ce2df242ea1d7987e8
61211b81ee757ccc3eb55a86d2fd43d69be4cac8
30501 F20101113_AACLKY williams_k_Page_074.jp2
fccc58abefcf2f04fc45b4d440ad24a4
38fed9c90108abe338af65b62fe9c2f8f1cc94da
3323 F20101113_AACLMC williams_k_Page_079thm.jpg
5d2514bc3a235b3a59a232d882961476
1237f8759bc98a8f541183d8f837d804e13c30b5
2732 F20101113_AACLLN williams_k_Page_092.txt
4b581905f42dc0ee3cb7383ee703a138
be37c3929b69f1e05e5e951467ced39b149a68ec
8422 F20101113_AACLKZ williams_k_Page_062thm.jpg
ce01230554d473c58b97157531b077f1
84a036ef1eb43076b5fed2b828edd0c5f9b2326e
52779 F20101113_AACLMD williams_k_Page_045.pro
b119c6d65fda88aef8722a1c82c7d43f
1b35e4ecf498cdc8190c841582aa4c4ae8da16fb
55071 F20101113_AACLLO williams_k_Page_020.pro
dd01338ca94c14ebab0982e29b41bbab
18780683626eeffc84dd8886ca2c13fa3bb17537
137997 F20101113_AACLLP williams_k_Page_075.jp2
ccb47d5a1c6f1536069deccfee9e8d33
bd374a278a5f3710b5a8c7fa8a21542ef262f537
89005 F20101113_AACLME williams_k_Page_048.jpg
16621f3c7de382ae0156f90fdbc836de
4e645ff80a856e546394bc405dc8c3823829297f
9410 F20101113_AACLLQ williams_k_Page_097thm.jpg
85ff3e43aaa8bbf8d79ffbaca5f47275
fce01979058d876ec83c0709309ea336755d676d
8769 F20101113_AACLMF williams_k_Page_037thm.jpg
a66176f06c31e290056913ab0fbc723a
5947d37aa814feb802a17259cb2d0cf0edbecb17
2008 F20101113_AACLLR williams_k_Page_029.txt
9dce121e7a91687df7030c981c5f51bc
30320f49dc54b181f4f53abd743cdf3b0be8cc6d
22419 F20101113_AACLMG williams_k_Page_023.QC.jpg
9bb149ee1ba378e370142730435da3dd
6afe36b5971ef87a5c9c6a6942740304e7b8fca0
2167 F20101113_AACLLS williams_k_Page_066.txt
7ae5b2fc7b25ef80999c35e51505e4d8
53ff41a5910ec60107b937c32da6f79519782c1a
120495 F20101113_AACLMH williams_k_Page_068.jp2
a5d13077aa632e3d225d2552c879fd65
f16ada8ef12404507b28164414604041a4a62513
93846 F20101113_AACLLT williams_k_Page_054.jpg
22eafa6506426e88ebe86b34ed035c25
1a9176c09ce66ff2add32d6f075121880de60ed9
2153 F20101113_AACLMI williams_k_Page_027.txt
c06b97be3aa3a22a87f4f63ecef3de79
b2f2c8fdee48abed6af5a785492f159f0a1c81c1
9124 F20101113_AACLLU williams_k_Page_001.QC.jpg
b95a596180a44412162dfbe0e51640a7
180f865f0df9a490128de64475871dc23825e35d
2653 F20101113_AACLLV williams_k_Page_088thm.jpg
ad8984945635493477e1e7801e818f34
a4e42c80a742ccb22d3227b27bfb9afd0aff1950
116096 F20101113_AACLMJ williams_k_Page_063.jp2
164842dc47a7ec601437096cec7d0a94
e54280827421deb1e8e5cfca911c1d4f3914efa1
62497 F20101113_AACLLW williams_k_Page_076.jpg
6b54a700ada31f42cf0ec886aa93625a
f01ad4bd8681d8b0d1095871178e405e57f7a20e
114061 F20101113_AACLMK williams_k_Page_013.jpg
5abb779b6be37cd0bbd0050f64ec2aa0
d031e98edbdeb35ed4bd741384e8fc684a0cee64
2114 F20101113_AACLLX williams_k_Page_069.txt
dc1e82885fbff8eff66a1ba24f373589
ffe1c64cb29351a8c742308f4aefce485a887a8c
6712 F20101113_AACLNA williams_k_Page_007thm.jpg
44f7ff41950e1c838d77badc63dfc2fa
a8818645fda93ef4c0e323a809488144724a63d0
11978 F20101113_AACLML williams_k_Page_026.pro
092e7cf22b218e0a19cf8762a700a205
c77f180829b828cbbe4882a0bad4372fd2b2d21d
2210 F20101113_AACLLY williams_k_Page_013.txt
26d648eb16b9d4aeca4f03713d624484
03901d06757e1f3d2adf0465b9861322b3a710d6
F20101113_AACLNB williams_k_Page_084.tif
5e30adcf1485a48f9d05bba60401abee
9de1dc4e53b55405367ffa18f755760b3500efdb
43529 F20101113_AACLMM williams_k_Page_091.jp2
5773d9658dd25acd94f7c8dea9e6550d
a0508f2fa8904026b3464de501e3f585471c7e69
F20101113_AACLLZ williams_k_Page_089.tif
02b9975e2edc9dbe4a7b836e19db9886
da28162a3caa9f45ec405734eb7dad597cd7e57c
117585 F20101113_AACLNC williams_k_Page_066.jp2
10363c787af2776a29229e2ffab85b2c
26094857cfc17301a53afe2bb6eb8ea296b03687
F20101113_AACLMN williams_k_Page_056.txt
fbc128660dc3dc25b43bc33a4fff3c60
fd4a0ae9b942ad5eb3832bdae76cc56ee5bbf4ad
53796 F20101113_AACLND williams_k_Page_051.pro
514459c74e6d743fc4b9457769305d8c
799fd38e2d50e937cf2f2bce69d59b481e8017b7
37778 F20101113_AACLMO williams_k_Page_064.QC.jpg
8df0932c63c3559ac30dc6286eead21f
9218930d874b6fc790e06e3704a9ebc4c6cb9781
113998 F20101113_AACLNE williams_k_Page_018.jp2
61ed83a595ad511dd151eb67a61a53bb
7795c6d3291dcb152f21d9c201b4a7fac416eb9b
1980 F20101113_AACLMP williams_k_Page_101thm.jpg
92d0ec6befdc46b60abeb9b2e444cf7f
973951e3e41987f011175d3ab30fbfa6c475b005
27156 F20101113_AACLNF williams_k_Page_053.jpg
c84df30d9108c12b9567e320815ebc1b
b52c12022d264dc68899be666d8a5b77aa687f1d
114289 F20101113_AACLMQ williams_k_Page_065.jp2
9065a2ac1d20f3c60f720695069f084a
23d0726f6b43871c5d2d6c3aede9ef7d4eb03455
38943 F20101113_AACLNG williams_k_Page_049.jpg
d253941d0e3d2b50906efeb13c6e84b8
8284902b4c632d0ffd81f78476fd54c80411b5d7
123946 F20101113_AACLMR williams_k_Page_099.jpg
64f2a49144eb32e71978ba1c165ecbc1
5d557ea1f0f7c82bd0c5fb1b1de3fbe67324255b
54694 F20101113_AACLNH williams_k_Page_047.pro
575d90e58bd212db2769e3eaacbf3e6e
e842f99d71e8cadc278d04b6dc30d5686298f426
2104 F20101113_AACLMS williams_k_Page_018.txt
595559f7c49123ec81a5d142bfe77e34
49a707a0b2fed90385c0aef2bda138e91ee37174
9836 F20101113_AACLNI williams_k_Page_101.pro
388c9224af12f3355bab078fff46eea1
c83fbbd35fb18a23fbaac193f39911f31e9b707e
2149 F20101113_AACLMT williams_k_Page_050.txt
d182fdefce431c304ed543ed0f9491df
0a1a8bc3d340ed74e417100cf23a83f04fea13a6
42318 F20101113_AACLNJ williams_k_Page_087.pro
1601fd821819d9bc501615caf8666dfc
a934f824a91773677585514587092a8db0d3397f
F20101113_AACLMU williams_k_Page_061.txt
3b9321accc1f69b255d60f0f6f6ae9d0
c21864e5ba113a6db6b15defce3c2185e687256f
3272 F20101113_AACLMV williams_k_Page_005.txt
6b7f4f581eeb2b5ed3417dd54a785897
928ca5c4e3db8fa70e3043643f1a2ff3cef9b3c0
F20101113_AACLNK williams_k_Page_098.tif
d30646dea982cc5de36ab8bb828a0d64
29cf263bae6efcd56ccf2b444eb8c249e39ebcca
1051938 F20101113_AACLMW williams_k_Page_095.jp2
b98061a100a23427397f81cb250a453b
cd9f704f6851e6560edf8a5d658478d3a18adea3
121043 F20101113_AACLOA williams_k_Page_046.jp2
796ba82c198b5596419d7ea8d6b75714
61b755991343fb7343081c8934e95a724ae1f1d9
F20101113_AACLNL williams_k_Page_051.txt
b85219fde3bf5ca7a9d19ec3e9bc26dd
48386dfeb8873641a7d1bc82c713a9624ae4fd9f
1038053 F20101113_AACLMX williams_k_Page_077.jp2
c3a2b62ee0707ae4d526e725a8ad9d0a
d1eba390a6c4120b6015e7917300cc08f9620075
F20101113_AACLOB williams_k_Page_014.tif
a79827aa8fbdbaed7da3463707ca243d
55c739c03f4c7565f5097214826a3c52f58b1495
8976 F20101113_AACLNM williams_k_Page_066thm.jpg
226075105e77b9440eccf6e0b1ede55a
4f1e026bfb1ad4ba8ad2f2adac6bb6e90b3f8313
105198 F20101113_AACLMY williams_k_Page_050.jp2
5c84692649a9b945ff9e140d86c94353
09a8a6cad6f473d0e9eefe989b69bdc67e5ea654
53382 F20101113_AACLOC williams_k_Page_078.jpg
b035d68b395185e4c9135dec5accf672
f1e11c528454349eb65644dd4f24d0d3c6324c69
34653 F20101113_AACLNN williams_k_Page_039.QC.jpg
3f52c4cbbc172caa0c578be8a0bd1d09
67ff8ee625dd53a1d9776ccb47ec4dde53e2efd6
8799 F20101113_AACLMZ williams_k_Page_070thm.jpg
96e8a7362bad4ccb409ed1eeb2c82930
fd255e2912f275ab210457c8186a3a262c7af3d5
30212 F20101113_AACLOD williams_k_Page_088.jpg
a7e9d0ce0b526fc357b445bf8a55cdef
fd7437337af18e8555e46bd59f763aea4aa2cc46
8582 F20101113_AACLNO williams_k_Page_063thm.jpg
992da029b4ade5d30f761c8e41f8029b
acdf17fadc9ea2325656e9a45e0edcc4a64d5f90
43349 F20101113_AACLOE williams_k_Page_059.jpg
1e54799619045491ac68a5a8194fc720
36b9a90efa54766a5872abb924cdb2a830593732
640 F20101113_AACLNP williams_k_Page_026.txt
41dc7ba570ac148635dd32af3c72ff80
bf58b86339adabeaea77846d28c8e6cf4684978c
116165 F20101113_AACLOF williams_k_Page_070.jp2
92720f5f6b9ba9b72497cff400c2f22a
63096f5857741dfa2aacc33acb283c26c20851d4
8995 F20101113_AACLNQ williams_k_Page_041thm.jpg
5f60810e24b230be46ca551c31ee487b
cc52e04fae42e5b93ca32802e7c82f923578e4a7
2298 F20101113_AACLOG williams_k_Page_008.txt
91cc1df33c68bd0905f4a5d035a0295c
fac8df34d072263f9a1202c0aa88dd28990dffa0
119930 F20101113_AACLNR williams_k_Page_044.jp2
708a69a68da85f21a46715f809a913d0
62b8ff5f6b46d66d4181c531527faa5c394cdefd
7944 F20101113_AACLOH williams_k_Page_101.QC.jpg
a486eac044286a8dc032e4926c20e459
1627f8c0ceee695acccb0ea6562c47f4d5c36feb
2671 F20101113_AACLNS williams_k_Page_099.txt
7d098aa566e61b5b2434a21ac0327d66
f7a137951468b7c486664476404cd6cff86b07ad
19833 F20101113_AACLOI williams_k_Page_100.QC.jpg
1f66937029b4a754ebb9fe8079bb2aeb
181316dd276766ae26fbee43bc5e3a91c88d49c7
69284 F20101113_AACLNT williams_k_Page_100.jp2
08b4d997df37f85ef6bddbf8dc97bc18
174de2943a820a82eb0e518da5557052b7af15b4
4181 F20101113_AACLOJ williams_k_Page_002.jpg
ec1778e4056573b8a8ce0a3e0393021e
490eb9651406d10cf914db397fea9a5b3a4481b0
8567 F20101113_AACLNU williams_k_Page_018thm.jpg
88c4f91a7b42f877053ebce7577a10ab
6aa11e9ab7d6ac7acf2ea77bc89feddda2f39754
64489 F20101113_AACLOK williams_k_Page_099.pro
87cc409d813c1e606785ac049c6883b0
8a3b8a2582479e52acdc8a3edcbca68115b7ac0a
F20101113_AACLNV williams_k_Page_043.txt
904bc5d72841ca5681d7a80fa4516958
de5c6b2d50b6da447ca8b6ce9658c8fcdad426fe
8979 F20101113_AACLNW williams_k_Page_030thm.jpg
ca2ab701d2235c2f7d9d97b5a28820ae
1b86f17b49d9b5143a791e84facf9c8807498391
114768 F20101113_AACLOL williams_k_Page_019.jpg
1d9bd7451ac57012fb428e1df17bcdf8
1bf3eafdf1890539abb818267ded0d5045f13df0
F20101113_AACLNX williams_k_Page_091.tif
aa2e448590912ade496d00470941f49c
a09e92f93e478f7b82b4cf3f53a131deff222235
F20101113_AACLPA williams_k_Page_055.tif
669852685bfac394b844ecf14c69c58d
50c5b283791be61174a7b741882f009f75c755e6
38808 F20101113_AACLOM williams_k_Page_040.QC.jpg
108ba8238abb9f4b675ba375c8ecc460
657132178d1d2ccc197ecff357bd46c758c1a7ea
105918 F20101113_AACLNY williams_k_Page_094.jpg
502b07da812b38e2414097cc7598a991
5104c41cc1fc6658bde805e30b291027e10dff98
144444 F20101113_AACLPB williams_k_Page_081.jp2
edcdb049399629229f939f3ec21de912
8e83b0e02c40ed198de619c76aef2e7646021dfa
31868 F20101113_AACLON williams_k_Page_025.jpg
5700183186ba654b59253413f52f6996
1bea8aac0e45a965d1a493274c9e30a93bfbfce1
8501 F20101113_AACLNZ williams_k_Page_028thm.jpg
69280487f8024962a835d9a86a8f08c2
fc91f364f2cd48b75ca5b7c482613b2a354d0717
121448 F20101113_AACLPC williams_k_Page_019.jp2
cf1f8484e655a98e02e07e1bce60cfa0
f690d2c3a742dfd69c33f57c47ecc2538701048f
F20101113_AACLOO williams_k_Page_087.tif
4b4a2a0c8beee66f76514b30f6aee6df
d325ca3dea046ca0b04486865765fa9565edd7b6
112702 F20101113_AACLPD williams_k_Page_069.jp2
349ffe12079ff6f4de5be9f41909d370
89d8b3ae914b43e269b34e0b29c47971ac084992
4189 F20101113_AACLOP williams_k_Page_059thm.jpg
7e5f381e2622dbe9636badccafe6c1b9
f8826d7aafa1ee02f4f7f49a2d8d0423b32b00a9
118830 F20101113_AACLPE UFE0021317_00001.mets
fdb648787baf6b8a24e8e3eb480efb2c
7658aab6fcde393d6a71e8f2e3acb1a9664ffc75
7398 F20101113_AACLOQ williams_k_Page_006thm.jpg
0d831168e9eac58a5b25ffbfb4b58174
ea1bc120c8e55b0ae5e639e25122482875d70e1a
F20101113_AACLOR williams_k_Page_003.tif
fe79cbe772b7a630d14b3df9ad7b6b9c
35c1c43eaed9e3e5011ebf79be9b4c3f04579d7d
12607 F20101113_AACLOS williams_k_Page_074.pro
f52ba4179e3aa96709772398d227d66d
9a08646d06641891d4da165fe4a40a5d43978ced
29868 F20101113_AACLPH williams_k_Page_001.jpg
783c762866d3cfcc9fe7a002fe2d2b8c
fc2ec5636c97651398a1a2d4439882876eb2aaad
9067 F20101113_AACLOT williams_k_Page_043thm.jpg
3ecdea3e2df19b85f394db3c2601690c
1eb91163e717ad5b5dedb116ddf5eddbbd8990fc
9071 F20101113_AACLPI williams_k_Page_003.jpg
261a38a83871333b6b7db2555d3507eb
220a483f60253598493bed5df8f055ca336d06b4
537 F20101113_AACLOU williams_k_Page_088.txt
aea754443f850c0683e6b7ad484ad509
1d8bd2e2263edbcc28d247ce44b2420b4322d2bc
79784 F20101113_AACLPJ williams_k_Page_004.jpg
4a8790ad09d0a99fc57c5d0baf81c2f4
734b9486541ff700f3f860cdd6e50e2dd1a0c815
61771 F20101113_AACLOV williams_k_Page_093.pro
f0ebe89fb418d4fe6811e6c2ba92d1d6
803d50a1aa2cb378dcb59a7b81e9c493d24a7f03
130531 F20101113_AACLPK williams_k_Page_005.jpg
d4f69d49b957871f5a6a5f291592e218
eacc70c1e881976fb38f528a2b3f183b3cd13f02
F20101113_AACLOW williams_k_Page_085.tif
8a099f7c2cee602b09b14b231378bc8e
3ccb73c5492b44e7fd2cddd7beff274742da9cdf
127224 F20101113_AACLPL williams_k_Page_006.jpg
aee46d0ec91e2f67878ae5cb2c384406
957ada265e8800fc6e0c49807b9b6fde73ab8b6b
303893 F20101113_AACLOX williams_k_Page_026.jp2
a964896b8a6ab7a1e7d74c344e33fda6
338b8f2dc9b45d8d7682c65d91410208a1db71ba
109671 F20101113_AACLQA williams_k_Page_027.jpg
fe59a1cd590fdb8df2ff3b14fb6e726d
75cd639164d2a41377197e27f4601a6e4bfe584b
1504 F20101113_AACLOY williams_k_Page_004.txt
c42db1953be730f6b5cc78a7ee3651ba
71a7426863f52978b315b96885fcd4867e10ec1e
106952 F20101113_AACLQB williams_k_Page_028.jpg
bb21aafb9158a87a433fa390cbaa88ca
4a86fa82018e9f09972bd22440cf2a453817e263
100506 F20101113_AACLPM williams_k_Page_007.jpg
181e06f82080f7fb02eaf0313e5d9ed2
ea40d3276d2098fd6fd802b2ada2e88b69375c60
114601 F20101113_AACLOZ williams_k_Page_012.jpg
db88e3e3235eccb7a19e86f1010ee9ab
5665eb924f44f1ef033955f1c2ea7d0cff227938
105168 F20101113_AACLQC williams_k_Page_029.jpg
18389557299711e8fc191c26b8a46b72
0ed6d9502a8d4695433167fbaf929fcb649d5f77
119050 F20101113_AACLPN williams_k_Page_008.jpg
1b09e29f80d623bdec050449ce107c1c
d4883bb1f8ac34b1fba6fe6ff75c10ad1d7075cd
112813 F20101113_AACLQD williams_k_Page_030.jpg
d9807b41a5413728fc4ce0aa2602cfef
3ebdddc7c047e33fd8ba0f8c5277bc849a433d66
104028 F20101113_AACLPO williams_k_Page_009.jpg
baa0776a410fc6b7b9c4bf720741201f
fa0cb3d53c9c854e923e6b67f03afd1fbf108334
106501 F20101113_AACLQE williams_k_Page_031.jpg
0b4a047674487e9672d9376ecc8a8a5d
1213552d3850cfcb679dbbf3732db366bc3ec5dc
74453 F20101113_AACLPP williams_k_Page_010.jpg
8406d7da17a078810d83cc23759b9890
1841e3c3e1bd87e2f2286fa6dccdebf9c885e02b
117349 F20101113_AACLQF williams_k_Page_033.jpg
2e969ba3aba9feb0ba2f4697cd23f67c
7779f65fc7279b3c6ddfd1b234d56dc10941ac11
106670 F20101113_AACLPQ williams_k_Page_011.jpg
70be1b428ac1fb260da2f2a5b1494b08
67c2289dd9dfc5531150dd90ea02351aef42ea39
116403 F20101113_AACLQG williams_k_Page_034.jpg
05fc5e6ac552ff4d57b4b5d81429e1f3
261488f788cc159bfee7c18706af9f1e552e9d37
109564 F20101113_AACLPR williams_k_Page_014.jpg
824789d886e55d138207c6112303a279
8b96e097983df99c50222932b164de6b5279249f
86371 F20101113_AACLQH williams_k_Page_035.jpg
5763e24bc41aae634a1bb86af7f3d386
93ee04fc98f24ce09f0e480ba40f4f541bd10c90
106006 F20101113_AACLPS williams_k_Page_015.jpg
37c12c38f523b1e459325125a302b252
0557bc3596b59f264fc267cea374a3c9f32e6125
116917 F20101113_AACLQI williams_k_Page_036.jpg
309326c8313ab82ec09669676441a6f8
949098247f2f8e72448a662558931b2604007c9e
101737 F20101113_AACLPT williams_k_Page_016.jpg
b459c80c71779bd83669382796fb54bf
3dc3466981283e1b1d6e6aedd5fcbd06e1fa2971
109799 F20101113_AACLQJ williams_k_Page_037.jpg
52050965af2c2efd668ffb8aae92c649
4d7bac2338cba573cbbae7b8aeb1e9cf43b1db9a
106309 F20101113_AACLPU williams_k_Page_017.jpg
1d69df43d71174248bd87c3b3955a765
12640b845aac019c7e059bc3e90ebcc2d467eaaa
104005 F20101113_AACLQK williams_k_Page_038.jpg
553840446c2b0e976a30434ea0e7b1d0
48269ec244effa8d8e02a32a63cc281319f85c82
110338 F20101113_AACLPV williams_k_Page_020.jpg
940171ae426febe385ee7c80dbc90fcf
eaf585dc2e058369c93e0b8108487ec53a6b7d3e
106253 F20101113_AACLQL williams_k_Page_039.jpg
6a62ef177de07719a5045c7d8541c954
43452afdbb2ad6f70a7221ec814b79f46c72e8a4
88514 F20101113_AACLPW williams_k_Page_021.jpg
af831ce7a542f56ad8ad418c61d1653e
526bba8bc4d19645b7967714af248801ca259a4a
117577 F20101113_AACLRA williams_k_Page_061.jpg
9b61e9d2285ca9d449517898ec80115d
581546e16809086e63ede812b30f38fb9ddd5198
116005 F20101113_AACLQM williams_k_Page_040.jpg
e5bd341a34760811b9a46265bb6d4300
712a159dae7fdf0a4bc0a6e2c1ccbb675d780252
103510 F20101113_AACLPX williams_k_Page_022.jpg
9460dcba9a24d7dcc4e88db3558983d6
f8d5d081826d1d18322bc3420307d1486b27df58
120834 F20101113_AACLPY williams_k_Page_024.jpg
d49be73b12c6364d1582da4cfc12b8ee
f50f8cf7e6fdae8a1a5ab0697dd7325441e0eb38
104174 F20101113_AACLRB williams_k_Page_062.jpg
74f56339fe27ac0953a80afcbd3e6d6b
088f49239e3cc4e30927ac8e0af92d09c649932a
113117 F20101113_AACLQN williams_k_Page_041.jpg
c0890829b2f34299f9c37548725c015c
861639d12120792524a571de1a9d9fa038aa81b5
33410 F20101113_AACLPZ williams_k_Page_026.jpg
802e0629d9ca518e3aac7c0dfc19085e
431a498fb06ff4e3f8ebdff53431a98c9bd377c4
109452 F20101113_AACLRC williams_k_Page_063.jpg
0aba44ca6694e8d9d526a7be49a19afb
756488deadb08d094e8380e75779af5faaebdb38
107855 F20101113_AACLQO williams_k_Page_042.jpg
1b33bf2958f8fd2ba3f2ae5d896480cb
92a08621f78f02ee550d0b5e18d345337fbcb2a8
115099 F20101113_AACLRD williams_k_Page_064.jpg
bbe885663363ea352ffb917743119eac
9116fd942df7d089358ab70c98df9bfd30516ab4
113627 F20101113_AACLQP williams_k_Page_043.jpg
a855b545058ec7c0986b68b0ef315a3e
3bb26df309815a241cae9765e381bdca052deb2a
106882 F20101113_AACLRE williams_k_Page_065.jpg
fe011d506db921190359c88ce7e57570
6d8f08c94a71e5e67f532b72d0a6cb0bb7789ee4
107310 F20101113_AACLQQ williams_k_Page_045.jpg
449e1c59cdaca0924338d7982195a0ef
c7348b314580a97ef7bec911c2e2d00cb004b23d
112357 F20101113_AACLRF williams_k_Page_066.jpg
80b695a506df6c76e00d4f8acb2e87b9
f6e4b852dbc0afc451d22c09aea0e5b85d4220c8
114139 F20101113_AACLQR williams_k_Page_046.jpg
d8271695381b765c4ea0876b620a7b3f
7c69dca20a5f02aed7f0c6db9ad7730d341d6c14
114973 F20101113_AACLRG williams_k_Page_067.jpg
4edf75bf808139ac09fc881ddb4b3596
13e26d97cb69c39c0c39a48049ee6b8524337e25
101207 F20101113_AACLQS williams_k_Page_050.jpg
271f2c1f6893751a9a8cb9be2baafec7
d6766b408ed2e363895563d7d48e7a819a203a73
114185 F20101113_AACLRH williams_k_Page_068.jpg
aa4fac25b4525b78537d914bd581e22c
8a04df4698c7916998ef7930ff3fa913ceffe2e9
121673 F20101113_AACLQT williams_k_Page_051.jpg
6823a21cd4d00896ee57223475d1e09f
737089fcdff319974cc8b5bc6bdff1e4abf2cd7b
106614 F20101113_AACLRI williams_k_Page_069.jpg
ab07f4b72b74958f647402c72238f255
e376d35e45c811d17b0c722188d54550e74068b2
54023 F20101113_AACLQU williams_k_Page_052.jpg
451e094c5f832f706ccfcad588f05d40
62c3003d513b25bba96eb30199ea51ac4145784e
110130 F20101113_AACLRJ williams_k_Page_070.jpg
27e2154c0f13e0e891d0ca6763c6a5b7
43d5124400a0ac1433ba440e0714faf6e0119cea
73116 F20101113_AACLQV williams_k_Page_055.jpg
17e8e3ce013a3402eee4291107120ba2
e3a1ee64fa1ac95bc870b798dbafde785635211d
105739 F20101113_AACLRK williams_k_Page_071.jpg
50c32e04eef3d349e307459bf9df579e
fc15601dc6ea1f2b8df68c35bc190c60c8110813
118643 F20101113_AACLQW williams_k_Page_056.jpg
d04c83ea8aab99bc40f434c382ba3180
5a29be41e0564a1a602a10d750dbda697a23a647
110791 F20101113_AACLRL williams_k_Page_072.jpg
7c1a2bdf39583b7a097ded5c63302a30
9bd68dfb18e499c5ccdd7241c09d5c39494bfd9b
43463 F20101113_AACLQX williams_k_Page_057.jpg
65f16106ca87b6e663bc62b27de6acbe
b26ef46accdf7f265edcc778288919bc3b6e263c
42550 F20101113_AACLSA williams_k_Page_091.jpg
8e4538eb118e8e08267dbe76cd59ca3b
6c6853d031511d4d7ab1a3467b3514ffc34fe864
113769 F20101113_AACLRM williams_k_Page_073.jpg
060683d4497fd0d3c931377995a8e7e1
05b74965b686c6830eae9a7276c8158d7663ba1e
43194 F20101113_AACLQY williams_k_Page_058.jpg
ac94f9b72f9c93059c781476bd961227
3af2efae47b90001cd8c1e3a8a10666ae91f3e22
129615 F20101113_AACLSB williams_k_Page_092.jpg
6f61153bbafdf86c80903736cd5367d6
206ccd03a1ec8a0a44dff9034591cb6b188fb568
28297 F20101113_AACLRN williams_k_Page_074.jpg
c29d2baa219ea5b3bc917f3844b85280
70c61e371e09bef3bd3a57dd50a90b0ce9b5b005
37631 F20101113_AACLQZ williams_k_Page_060.jpg
4ac6dda6962d869932dc4fa554295f11
6d4b6c523d3738bf0f58d0d4b08a4a3c152b929f
124746 F20101113_AACLSC williams_k_Page_093.jpg
35a02ca84e5c38593c056924dbdd2ce2
89c3951b435904e4d3e4d7fd6040ce268dbd45aa
133047 F20101113_AACLSD williams_k_Page_095.jpg
5a0f743b0b35f71a6f4f38f5a8a0e387
8a8a5028870b044c98d0bc0f6e783a0a35524be6
133618 F20101113_AACLRO williams_k_Page_075.jpg
6737a79a884289c6ed452eb30e509e20
03d0a7d9dcd52c77ccd77b5eb8e2dbb02e168717
132927 F20101113_AACLSE williams_k_Page_096.jpg
93352f0997c3c6244d543a7cf0506191
7921b416685dd198c11a896b72b327f795b902ab
118674 F20101113_AACLRP williams_k_Page_077.jpg
6b99373499512c8b81690f26889359c7
2d01f91d570ab8ef2d94e4977c0227437de02117
125439 F20101113_AACLSF williams_k_Page_097.jpg
093a781f2b3567a29d3c5768c1d8dcda
c8022d8451561a812b89af67c57e1ec5751887a7
31429 F20101113_AACLRQ williams_k_Page_079.jpg
7191bf2906fb8f814c6e1024f4fb2df0
e9622c96cedab928bfa64879485a4f3802160301
129663 F20101113_AACLSG williams_k_Page_098.jpg
d7949094e3dfd44d67670bc517bafc0c
5982ba602735fb8d46256aa62d6416aa89770a72
35107 F20101113_AACLRR williams_k_Page_080.jpg
dc1f597d40f388bce7b877eb703c2568
d962b642aa61b66358f404ad79187b0fa349929b
65309 F20101113_AACLSH williams_k_Page_100.jpg
a74635d9607f58bc5c5b5aecd58e570f
d1f52eb4bb87d5b078404347ee053e13ac0537d2
139549 F20101113_AACLRS williams_k_Page_081.jpg
71032554a2b27320872be7721252e75e
1560c5aae9295add3088904fcb2ff6d34279a9f5
25005 F20101113_AACLSI williams_k_Page_101.jpg
f0572dbe9dbb21e15f27396796c2865b
f90a3cdce515d6104e5c73e61f7962c8e6383c92
93515 F20101113_AACLRT williams_k_Page_082.jpg
1c32e8af25296c5eba2e5b94979289ac
2015bb6fa84ae043803e701e00b52203e379a0e0
28647 F20101113_AACLSJ williams_k_Page_001.jp2
ae47436ec5bf7cc64a75f37097d44a12
fdcc34cfa87b7a42d48cc830edb1849b5440259c
100911 F20101113_AACLRU williams_k_Page_083.jpg
9b94c1beba8c0a2430a33c964473d455
4d39c12d87fdce4ca2358d698d0b0ffc4c6e73cd
5628 F20101113_AACLSK williams_k_Page_002.jp2
cb00c263e1d84bb32fad6d9bb0729273
d430415a02ab1086b0f623e3fda299132e789fdd
96507 F20101113_AACLRV williams_k_Page_084.jpg
01bbad0f5b5c439f5e27bec2292c750b
c03b7c7d6dc4f6f9ce2754360403ac89be889c00
10034 F20101113_AACLSL williams_k_Page_003.jp2
9158df634b13ceeb711a412fc331820b
a5b9ca054d43de93a3c365809ca9d0cdf6a3ef4f
109173 F20101113_AACLRW williams_k_Page_085.jpg
dc1ddc350676c9c4b6f092133d52ce83
912306287642f7c4cfb899b5d99ebbb9003babcc
1051859 F20101113_AACLTA williams_k_Page_024.jp2
6930ab0a35284e732f6ba366e2be46b3
5d6e4b36c4ab1257dceaea2417610c297175f1e6
81937 F20101113_AACLSM williams_k_Page_004.jp2
a33e23a3c8469b73512f9dcadc19ba7d
d4957ca3f8cd3d3a3630ccab8c465daa70ba8394
89497 F20101113_AACLRX williams_k_Page_087.jpg
a7f3391a5b2e974f3fb5aeedbe68cfd4
b906c8c0d8369f5c6eac97e2b51b2c4373cff1ee
30727 F20101113_AACLTB williams_k_Page_025.jp2
3cc502b3134522d089da9a10680e902d
971f991a2282a9e59f27f7f26a18c407991af8c7
1051936 F20101113_AACLSN williams_k_Page_005.jp2
62f8a88f693d8ecff50ab14ef218c3d7
e33d77f5a7322df149abdb4da82ee24bd55d9e06
112964 F20101113_AACLRY williams_k_Page_089.jpg
32c9fb08fde93b4c31ecce4b6d07f9f7
39d8aea2a369736a9fbd20cb1cc4a7db222d6bc6
114083 F20101113_AACLTC williams_k_Page_027.jp2
c9c9eef98e8732be19f70328df3dfc78
d33473d24b74ff8e681e45d6933dc935e700b33c
1051973 F20101113_AACLSO williams_k_Page_006.jp2
6707edb986d64aa6d870db22fc25b4e2
4d8b089c59d3025b019a6a31758f19f650fae767
10981 F20101113_AACLRZ williams_k_Page_090.jpg
931fa7bede740176bfd990cf7ba52439
a0c9a84828a22abb45bc2348f41b00a10e2f7cf2
113818 F20101113_AACLTD williams_k_Page_028.jp2
9936f7f7cc460f580dc5ad68c33cecbf
9383fdc9908afc5b1c6516fb39c64024e1963809
112131 F20101113_AACLTE williams_k_Page_029.jp2
70fcc6d28c21feb8442b5c846b6b2819
cb41518d6504474648e8c6ec5bddb91c2f8647de
F20101113_AACLSP williams_k_Page_007.jp2
2264e426e4a95a8101ac31d19d608603
32d29132384879a1d83e14e44bc3db26990e0e11
119916 F20101113_AACLTF williams_k_Page_030.jp2
ad159042719afa08c9db424ae23bf6d0
a85119683a4d5b2fbabc48d383a31cccf8de2fc2
1051984 F20101113_AACLSQ williams_k_Page_008.jp2
820383f38b2dc101b9898b819ab06a85
aa3f14d8cc2a0788457feafeaeeecd16660bfc5a
112989 F20101113_AACLTG williams_k_Page_031.jp2
a932aef45c357bdae7fd4c7877bc9260
9f8ad057b4bbb344dcfd9f2c8f793a38507acfe7
109148 F20101113_AACLSR williams_k_Page_009.jp2
c11b3555d8e9288ae125730b62ab750b
8bbd9580e6eb65867d46c6228d88861f81e91065
121605 F20101113_AACLTH williams_k_Page_034.jp2
3a2f2cc5551750419507b9a9b9570586
27d8cf268d3e1faff91bbbc1272639af9ff4823c
123248 F20101113_AACLSS williams_k_Page_012.jp2
aaed1cea3f01da1dc66ea597341925b0
51fa678c9666d3aef3ec3711aa60d7980cbc5922
92400 F20101113_AACLTI williams_k_Page_035.jp2
7b42c8a521660f0a06c6003af3fc8b9b
ce4b1be8b771f758be2af4d8936683f789b6e00d
121271 F20101113_AACLST williams_k_Page_013.jp2
9a4abd650fbc2282fe645207935338d5
745b92603a264a19dacd4349dd7c73a3f6a3b888
123885 F20101113_AACLTJ williams_k_Page_036.jp2
b5bc7e9473406bda605d89b4d1497db6
60b05c2892f43d7fd6d4bf875ce3c5bbf6db3d89
117401 F20101113_AACLSU williams_k_Page_014.jp2
c6415063286072e9ac9b11a8f7818198
392df16a2d011c0c9222b37a5d2706a9e014639d
116309 F20101113_AACLTK williams_k_Page_037.jp2
1a24673ec6f6b112e47806fc7396a59c
2434680481c3eaefd7fc13d9ef0ab94890226b77
112522 F20101113_AACLSV williams_k_Page_015.jp2
1db01d3ad10958253df8c8bc061a34b1
3d6f830df298bdc367b1a636ac0487bfafb6befc
111893 F20101113_AACLTL williams_k_Page_039.jp2
0de080a9ae2a758010911aec6811cf39
3039120e4090c20fd82843508093949ba4353bf6
108257 F20101113_AACLSW williams_k_Page_016.jp2
b8f3d3a24e0609ef5660f02ebe387c01
2a99d0af3dadeb4ea70bfe5a0c4750629ec5f972
123811 F20101113_AACLTM williams_k_Page_040.jp2
c4f149a1b2029c3ce58f55f9dffd0185
4d0644c22a5a64691ff02c06cd5d3d2e718a7a99
112675 F20101113_AACLSX williams_k_Page_017.jp2
223693e41282b49261395a36158ee041
f738a0640212900f0ce242cdcb048d50a13eb2ca
395459 F20101113_AACLUA williams_k_Page_059.jp2
9362d8d73ad67a5f92120fa3debfd0f3
4d8a910eacd23f25321fc7216144b057a5f45b8e
120647 F20101113_AACLTN williams_k_Page_041.jp2
06fa8929b7fb3bf4ec6559b0edd36a42
89c08ef3edd1c282c715308ab7a55f4ac9f1693f
117365 F20101113_AACLSY williams_k_Page_020.jp2
dca7acc1facc69de8aed46fd33761900
426c07ff8e14bc10a8321b6bc69bcac314541a1d
308035 F20101113_AACLUB williams_k_Page_060.jp2
d9aa958027b99cadeba75247ba493e56
0f0e2b3446b26ab02a7e0d49126ece63311c875f
114249 F20101113_AACLTO williams_k_Page_042.jp2
a559ed699e30df40252f7486889f8eae
7813cc9f41403fa2fe039c492789b9f0660bd5f6
81378 F20101113_AACLSZ williams_k_Page_023.jp2
d81ef90e503ed744d9a473a26ff0b95e
a8f1e6be5f87849558efd8a10bd17a943aca03e7
123719 F20101113_AACLUC williams_k_Page_061.jp2
620c455152638f3b06beae09afc1c46d
3860d9124b3d382c3e597a4db4f10a12c32f6f46
120064 F20101113_AACLTP williams_k_Page_043.jp2
b70de76887d30b5fae6a7d60ef4a0e5a
a4cf81af2a5c7d1a69f84abe3b954b141c89d5fc
112780 F20101113_AACLUD williams_k_Page_062.jp2
38427173ee292689de4dabf62e09c04d
8482cd6a26180c96735a25a09126ad4ddde45ae9
119723 F20101113_AACLUE williams_k_Page_067.jp2
0ef02f542068d7542200ebe7c747b6ba
56265a5695b3ffead78ad6197b1e7edb4cf4ad07
114218 F20101113_AACLTQ williams_k_Page_045.jp2
8d9269434d4059c213d59e153db7427c
bebabf91e1c39a8bb981e36d3b652bede0a46b5d
117328 F20101113_AACLUF williams_k_Page_072.jp2
ca533fb551daa9ccd1d3372aa6528fbd
5020b5a8a2c8b13dc76f971aaca0d10f215c9917
37768 F20101113_AACLTR williams_k_Page_049.jp2
c913836582ac082265a47e9b4604a734
2ad6a2b1f7273b59df70deed7b79ef4b7a1023c8
9232 F20101113_AACMAA williams_k_Page_078.pro
4d9a08e83532daad5686972e455d5221
4c11e8af7e6afb5d66626e8514c0beb3a6c48dea
62577 F20101113_AACLUG williams_k_Page_076.jp2
cdcbba1c4bc46f12b59f0b4cda7e9796
a459674e362a016087ee575e5edc336e3528b689
119649 F20101113_AACLTS williams_k_Page_051.jp2
cb46fbd914f6942620c6371cf9f2fa7e
adedb590544b7a3fc8051add549a0b7ef637fc67
9469 F20101113_AACMAB williams_k_Page_079.pro
c25502e2e428c943be868206bdc64245
a26a45848e21e07552d82da802eb6fd24626b5de
53147 F20101113_AACLUH williams_k_Page_078.jp2
2f67ee2b83e0bc72d8c34aadf98991f1
e68eaac45e86079e34ebe022bb29cee00e58242f
56818 F20101113_AACLTT williams_k_Page_052.jp2
99e3c4138e6d0c6a526eace3e11f4b31
c4b3a1749d05fe59d3f39bd752d4b02a57b42698
12309 F20101113_AACMAC williams_k_Page_080.pro
cf31b0e0dee3e4ec65126b13bad69026
f78098370866d7eb72525efb4c6eb1d2a6eafcfc
29101 F20101113_AACLUI williams_k_Page_079.jp2
acc44b83b53d6975943c972426a403b7
f50edb0ccd8ee3af0b5f61996d03d9c839ecd351
237147 F20101113_AACLTU williams_k_Page_053.jp2
5ffa0131bb274efcb5aa1c46d8d5b803
8491c370648b99ebd9683e79ead35a12426b212a
69627 F20101113_AACMAD williams_k_Page_081.pro
006837fca70de0768a086c3bcac9f422
e6cad0dbc5e989717cdbd134624798d700c784c1
32888 F20101113_AACLUJ williams_k_Page_080.jp2
5401fa630c61633d540345732bdfd7d0
f15922a9d92ed27f76607c102ac469a13f2b71a8







BIASES IN POPULATION ESTIMATION FOR COLONIALLY NESTING GREAT EGRETS
(Ardea alba) AND WHITE IBISES (Eudocimus albus) IN THE FLORIDA EVERGLADES


















By

KATHRYN A.WILLIAMS


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2007

































O 2007 Kathryn A. Williams





































To my parents, Susan and Douglas Williams, for their unflagging love and support. You guys
are the best









ACKNOWLEDGMENTS

I would like to thank my committee members, Drs. Scott Robinson and Kathryn Sieving,

and most particularly my advisor and committee chair, Dr. Peter Frederick. Peter has been a

great mentor these past several years, and I am grateful that I have had the opportunity to work

with him. I am a better scientist and, I believe, a better person as a result. Dr. Jim Nichols has

also provided me with invaluable guidance, assistance, and instruction, all with great grace and

good humor.

I owe great thanks to the people who helped me in the field, including Christy Hand, Sam

Edmonds, Carolyn Enloe, Brad Shoger, Becki Smith, Andrew Spees, Eric Trum, and particularly

my coworker John Simon, without whom this thesis might never have been written. I also owe

additional thanks to my pilots, Kenny Lung, Matt Alexander, and Steve Diehl, along with

everyone else at Unusual Attitudes, Inc.

Julien Martin kindly came to my assistance during superpopulation analyses. My lab

mates, Evan Adams, Rena Borkhataria, and Nilmini Jayasena, provided much-welcomed

suggestions and commiseration, as did my roommates, Dina Liebowitz and Cyndi Langin.

Last of all, I would like to express my appreciation for my family. Mom, Dad and Jen--I

am incredibly lucky to have you guys. Thanks for being so supportive when I said I wanted to

move to the Everglades and drive airboats for a living.












TABLE OF CONTENTS


Page

ACKNOWLEDGMENT S .............. ...............4.....


LI ST OF T ABLE S ................. ...............7................


LIST OF FIGURES .............. ...............8.....


AB S TRAC T ......_ ................. ............_........9


CHAPTER


1 VISUAL BIAS IN AERIAL ESTIMATES OF COLONIALLY NESTING WHITE
IBISES AND GREAT EGRETS ............ ..... .__ ...............11..


Introducti on ............ ..... ._ ...............11....
M ethods .............. ...............13....

Sam ple Areas ............... ... ......_ ...............14...
Counts and Photographic Analyses ............ .....__ .......__ ........ 1
Error Calculations and Tests for Covariates ................. ...............15...............
Results ................. ...............17.................
Great E grets ................. ...............17.......... .....
W hite Ibises ................... .... ......... .......... .. ............1
Correction for Visibility Bias in Colony Counts ................ ..............................18
Discussion............... ...............1


2 ESTIMATING BREEDING POPULATION SIZE FOR ASYNCHRONOUSLY
NESTING COLONIAL WADING BIRDS .............. ...............27....


Introducti on ................. ...............27.................
Estim ation Error .............. ...... ... .. .. ... .. .. .. ...................2
A New Strategy For Estimating Breeding Population Size For Unmarked
Popul ati ons ................. ...............29........... ...
Methods .................... ...............30.
Survey Methodology .............. ...............30....
Artificial Geographic Markers .............. ...............31....
Photographic Analysis.................. .. ... .................3
The Superpopulation Modeling Approach to Population Estimation .............................33
Subsampling Alley North Colony .............. .. ...............36...
Sub sampling to Determine Necessary Survey Frequency............... ...............3
R results ............._.. .. ... ...... .. ...... .... ..... .......... .... .......3
Interobserver Error in Tallying Individual Nest Histories From Photographs................39
Superpopulation Estimates .............. ...... .. .............3
Sub sampling to Determine Necessary Survey Frequency............... ...............4
Discussion................. .................4
Limitations and Assumptions ................. ...............43....._.._......











M odel fit ........._. ._....... ......_ ......... ...............45
Refinements of the superpopulation technique ................. ......_.. .............. .46
Incorporation of renesting ................ ....._._ ......_ ....._._ ...............47
Implications of Results ................. ...............47................

3 REMOTELY MEASURED EFFECTS OF RESEARCHER DISTURBANCE ON
GREAT EGRETS (Ardea alba) AND WHITE IBISES (Eudocimus albus) .........................61

Introducti on ............ ...... .. ...............61...
M ethods .............. ...............62...
Ground Transects............... ...............6
Aerial Survey Methodology .............. ...............64....
Photographic Analysis...................... ......... ............6
Analysis of Presence-Absence Databases: A Mark-Recapture Scenario.............._._......66
Results ........._..... .. ... .. .._._ ....... ...............69
Vacation Island Great Egret Colony ........._._. ...._. ...............69
Alley North W hite Ibis Colony .............. ...............70....
D iscussion...................... ... .........7
Violations of Model Assumptions ................. ...............71................
Effects of Disturbance on Nest Survival ................ .......................... ..........71
Effects of Disturbance on Nest Initiation ................. .................. ................72


APPENDIX


A SUPERPOPULATION MODEL OUTPUT (POPAN DATA TYPE) FOR ALL
C OLONIE S ................. ...............8.. 1......... ....


B CALCULATION OF STANDARD ERRORS FOR EXTRAPOLATED
SUPERPOPULATION ESTIMATES FROM ALLEY NORTH SAMPLES........................85


C SAMPLE BUDGET FOR DEVELOPING PEAK COUNT AND
SUPERPOPULATION ESTIMATES ................. ...............89.................


D MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON
GREAT EGRET S ............ ..... ._ ...............92...


E MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON
W HITE IBISES .............. ...............93....


LIST OF REFERENCE S ............ ...... ._ ...............94..

BIOGRAPHICAL SKETCH ............ ...... ._ ...............101..










LIST OF TABLES


Table Page

1-1 Quadrats from 2005 and 2006 seasons from the Everglades of southern Florida.. ...........22

1-2 Visibility bias-corrected peak counts for five Great Egret and White Ibis colonies
from the 2005 and 2006 breeding seasons. .............. ...............23....

2-1 Names and locations of study sites.. ............ ...............49.....

2-2 All colonies for which we followed nest fate through time in aerial survey
photographs and calculated a superpopulation estimate.. ................. .................5

2-3 All samples of Great Egret and White Ibis populations from large colony (Alley
North). ................. ...............51._._._......

3-1 Characteristics of the two best fit, lowest AIC models describing population
parameters for Great Egret nests in the central Everglades, Florida. .........._... ..............75

3-2 Characteristics of the best fit, lowest AIC models describing population parameters
for White Ibis nests in the central Everglades, Florida. .................. ................7

3-3 Comparison of model-averaged survival and entry probabilities between disturbance
groups for the Alley North White Ibis colony. ............. ...............76.....

A-1 For each colony or sample, we tested a set of four candidate models .............. .... ........._..81

C-1 Sample budgets for superpopulation estimation and peak count estimation of
breeding population size for a hypothetical Great Egret colony. ............_.._ .................. 91

D-1 All models for Vacation Island Great Egret colony' s analysis from Chapter 3. ......._......92

E-1 All models for Alley North White Ibis colony' s analysis from Chapter 3 ................... .....93










LIST OF FIGURES


Fiare Page

1-1 A 10x10 meter White Ibis quadrat at Alley North colony in WCA-3A (04/07/06).. ........24

1-2 Proportional error in aerial counts of White Ibis quadrats ................. .......................25

1-3 Proportion error in aerial estimates of White Ibis quadrats by vegetative cover. ..............26

2-1 Types of data that should be incorporated into estimates of breeding population size. ....53

2-2 Satellite photograph of the modern Everglades, with maj or watershed regions
outlined in yellow.. ............ ...............54.....

2-3 Satellite photograph of Alley North colony's tree island (WCA-3A). .............. ...............55

2-4 Two types of artificial landmarks used in this study .............. ...............56....

2-5 Multi-observer comparison of superpopulation estimates for Vacation Island 2006
colony. ........... ..... .. ...............57....

2-6 Peak counts and associated superpopulation estimates for three Great Egret colonies
in W CA-3 A (2005-2006) ................. ...............58........... ....

2-7 Peak counts and extrapolated superpopulation estimates for White Ibises in 2005 and
2006 and Great Egrets in 2006 at Alley North colony (WCA-3A). ............. ..................59

2-8 The proportional difference between superpopulation estimates and peak counts for
2005 and 2006 ................. ...............60........_. ....

3-1 Two types of artificial landmarks used in this study .............. ...............77....

3-2 The difference in entry probabilities between disturbance groups at Vacation Island
Great Egret colony as a function of time. ......_.._._ ......._.. ...._. ..........7

3-3 Model-averaged weekly survival probabilities for both disturbance groups at Alley
North White Ibis colony as a function of survey interval (time).. ............ ....................79

3-4 Model-averaged weekly entry probabilities for both disturbance groups at Alley
North White Ibis colony as a function of survey interval (time).. ............ ....................80









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

BIASES IN POPULATION ESTIMATION FOR COLONIALLY NESTING GREAT EGRETS
(Ardea alba) AND WHITE IBISES (Eudocimus albus) IN THE FLORIDA EVERGLADES

By

Kathryn A. Williams

August 2007

Chair: Peter C. Frederick
Major: Wildlife Ecology and Conservation

Estimating avian breeding population sizes often involves extrapolations derived from

estimates of related biological parameters, and may therefore fail to account for bias introduced

from several sources. In colonially breeding birds, estimation of breeding population size is

complicated by visual biases and by asynchronous nest initiation and turnover. We used

artificial landmarks to create quadrats in nesting colonies, and quantified visual biases by

counting the number of nests in a quadrat via both walk-in survey and aerial photography.

Visual bias due to vegetative occlusion caused aerial surveys of White Ibises (Eudocimus albus)

to underestimate numbers of nests by an average of 12.37% (+20.65%) when compared with

walk-in ground surveys. Misidentifieation of the species of nesting birds in mixed-species areas

proved to be a problem in aerial surveys of Great Egrets (Ardea alba).

We examined asynchrony bias by applying a mark-recapture approach to nests. We

individually identified nests from the air through the use of natural and artificial landmarks, and

followed nest fates through subsequent aerial surveys to obtain presence-absence information

over the course of a breeding season. We used the superpopulation approach (a variation on a

Jolly-Seber open-population mark-recapture model) to model nest initiation and survival as if

nests were animals in a marked population. The superpopulation approach allowed for the









estimation of the total number of nest starts throughout the survey period. We found that

asynchrony in nest initiation dates caused one-time "snapshot" counts of nests at the peak of

nesting to underestimate true numbers of nest starts by 47-3 82% for both species. Nest turnover

rates were highly variable among years and between colony sites, suggesting that nesting

asynchrony must be measured for each season and site of interest. These results indicate that,

without bias-correction, peak counts do not allow for useful inter-year comparisons, because

each peak count represents a different proportion of that season's total nesting activity. Unless

populations are highly synchronous, or this bias is measured on a seasonal basis, one-time

"snapshot" surveys of avian breeding populations probably cannot be used effectively as indices

of breeding population size.

Researcher disturbance may also bias estimates of population parameters for colonies.

Walk-in censuses of White Ibis and Great Egret colonies (initiated after the onset of incubation

for the maj ority of nests) found that while nest success was not significantly affected by

researcher disturbance, disturbed areas demonstrated depressed rates of new nest initiation.

These species may be sensitive to disturbance early in the nesting cycle, but be resilient to walk-

in disturbance after the onset of incubation.









CHAPTER 1
VISUAL BIAS IN AERIAL ESTIMATES OF COLONIALLY NESTING WHITE IBISES
AND GREAT EGRETS

Introduction

Birds are one of the most frequently used bioindicators (Temple and Wiens 1989, Stolen et

al. 2004). Close monitoring of bird populations may improve our knowledge of the relationship

between environmental conditions and demography and allow the populations to be used as

indicators of ecological change. Estimates of population size also have value in their own right,

by allowing for more informed management of birds. However, the estimation of avian breeding

population size must often be an approximation derived from estimates of related biological

parameters (Frederick et al. 2006, Bibby 2000, Bock and Jones 2004) and may therefore fail to

account for bias introduced from several sources (Erwiin and Custer 1982; e.g., Ogden 1994,

Gratto-Trevor et al. 1998). In colonially breeding birds, there may be several sources of bias

involved in estimating breeding population size, especially for large breeding aggregations.

These include species or groups of birds that nest asynchronously over the course of a breeding

season, nest in multispecies colonies, and whose nests may be concealed by vegetation during

aerial surveys.

A great amount of effort and scientific inquiry has been directed toward understanding bias

in counts of animals (e.g., Prenzlow and Loworn 1996, Erwiin 1982, Bart and Earnst 2002,

Rosenstock et al. 2002, Tomialojc and Verner 1990, Dodd and Murphy 1995, Frederick et al.

2003, Frederick et al. 1996, Gibbs et al. 1988, Bayliss and Yeomans 1990, Rodgers et al. 1995).

Detectability has been recognized as a problem in line transects (Bibby 2000, Dodd and Murphy

1996), perimeter counts (Dodd and Murphy 1996), point counts (Bibby 2000, Nichols et al.

2000), and other types of population surveys using vocal or visual counts. Detectability can be a









source of bias for several reasons, including individual heterogeneity in behavior (such as

frequency of song) and visual bias caused by the varying distances of birds from the observer.

Although estimates of the size of bird aggregations taken from aircraft or other aerial

platforms have many advantages, they introduce detectability concerns related to variation in

visibility of nests (Gibbs et al. 1988, Frederick et al. 2003, Bayliss and Yeomans 1990) and

species misidentification (Rodgers et al. 2005, Barbraud and Gelinaud 2005, King 1976). Visual

occlusion of nests as seen from the air may be due to several factors, including vegetative

occlusion and miscounting because of high nest densities (Dodd and Murphy 1995, Prater 1979,

Bayliss and Yeomans 1990). For example, while Great Egrets (Ardea alba) and American Wood

Storks (M~ycteria amnericana) often nest at or near the top of the tree canopy, where the nests are

generally highly visible from the air, White Ibises (Eudocimus albus) and small herons (Egretta

spp.) usually nest in both canopy and midstory, where their nests may be occluded from above.

Additionally, high nest density in species such as White Ibises may make it difficult to

differentiate between individual nests. Misidentification of species in aerial surveys of mixed-

species colonies has also been found to be a problem in at least one study, although this problem

has not been well explored in the literature. Rodgers et al. (1995) found that observers' difficulty

in distinguishing Great Egret and Wood Stork nests from the air led to large and highly variable

estimates of aerial bias in wading bird counts (Order Ciconiiformes).

When estimating regional breeding populations it is vital that accurate estimates be

obtained for large colonies, as they contain a large percentage of the total breeding population

and errors in population estimation for large colonies will disproportionately affect total

estimates. We studied visual bias in estimates of Great Egret (Ardea alba) and White Ibis

(Eudocimus albus) population sizes because they are conspicuous colonial nesters that nest in










large colonies, and because the two species have different nesting habits and potentially different

nest visibilities. In this chapter we compare counts taken from both the ground and the air in the

same large marked quadrats. Resulting estimates of visibility bias were used to adjust colony

counts for encounter probability. We predicted that, for both species, visual bias would result in

an underestimation of numbers of nests. We also expected that degree of error would increase

with higher vegetative density and nest density.

Methods

We studied Great Egrets and White Ibises in the Water Conservation Areas (WCAs) of the

Florida Everglades (Broward and Palm Beach Counties, FL) in March through May of 2005 and

2006. WCA-3A and WCA-1 are large areas of primarily sawgrass and wet prairie in the central

Everglades that are controlled by, respectively, the South Florida Water Management District

and the U.S. Fish and Wildlife Service. Colony counts during this time ranged up to 1,193 Great

Egret nests and 13,566 White Ibis nests in a single colony; these counts were derived from

monthly aerial surveys conducted from January-June of 2005 and 2006. We conducted surveys

in a Cessna Skyhawk (172) high wing aircraft flying at approximately 177 kph and 244 meters

above ground level. We took photographs from the copilot' s seat through an open window,

using a Canon EOS 20D with a 28-135mm image stabilizing lens. We accepted the largest

single-month count of the breeding season from these aerial photographs as the maximum or

"peak" seasonal count for each colony and species. This type of count has been commonly used

in the past as a minimum breeding population estimate for wading birds in the Everglades. We

present data here from five colonies for one or both species and years.

For visual bias studies, we worked in the mixed-species colony at Alley North (WCA-3A;

N 260 11.179', W -800 31.431'), which included both Great Egrets and White Ibises in different

areas of the single large tree island. We also worked in the White Ibis colony New Colony 3 in









WCA-1 (N 260 32.013', W -800 17.879'), which consisted of over twenty smaller tree islands in

close proximity. The emergent vegetation in both colonies was composed primarily of willow

(Salix caroliniana) with small numbers of pond apples (Anona glabra). White Ibises at Alley

North also nested in lower vegetation on the outskirts of the tree island, primarily in cattail

(Typha latifolia).

Sample Areas

We created rectangular quadrats in Great Egret and White Ibis colonies to compare aerial

and ground counts of nests within these areas. White Ibis quadrats varied in size from 100 m2 to

roughly 200 m2, and Great Egret quadrats were either 400 m2 or 900 m2. Each quadrat consisted

of four artificial physical landmarks that were large and conspicuous enough to be seen from

aircraft at 100-200 meters above ground level. Markers were of three types:

* A roughly 1-2 meter wide area of vegetation painted with a dilute white latex paint (1:3
water:paint) dispensed from a backpack sprayer of the type common in landscaping. These
markers were only utilized during the 2005 season.

* A 3-meter tall, 10-cm diameter white vertically mounted PVC pipe with a 1.5-meter
horizontal X on the top. These markers were primarily utilized during the 2006 season.

* A lxl-meter piece of white or pale-colored cotton cloth, tied at the corners to vegetation in
a horizontal position (Figure 1-1). This type of marker was used exclusively during the
2006 breeding season.

We delineated the edges of quadrats with plastic flagging tied to vegetation at regular

intervals between corner markers, except in cases at New Colony 3 where the sample area

consisted of an entire, well-defined tree island. The creation of a quadrat and the counting of all

nests within it required approximately 30 minutes to 1.5 hours, depending upon size of quadrat,

ease of movement, and nest density.

We marked a total of 21 rectangular quadrats in Great Egret and White Ibis colonies in

order to compare aerial and ground counts of nests within these areas. In 2005 we erected









markers for four quadrats in the White Ibis colony at Alley North. In 2006, we erected markers

for ten quadrats in Alley North (four Great Egret quadrats and six White Ibis quadrats). We also

erected markers for seven White Ibis quadrats in New Colony 3.

Counts and Photographic Analyses

On the date that markers for a quadrat were erected, we counted the number of nests in the

quadrat on the ground, and marked each nest with flagging when it was counted in order to avoid

double counting. Within 24-36 hours of the ground count, we photographed the quadrat from

the air in a Cessna Skyhawk (172) at an altitude of approximately 152 meters. Photographs were

taken from the copilot' s seat with the door removed, so as to take photos as close to vertically as

possible, using a Canon EOS 20D with a 28-135mm image stabilizing lens. Photographs were

analyzed using either Adobe Photoshop Elements version 2.0 or the shareware program

Paint.Net version 2.0, using the same general procedure as for colony-wide counts. On the

computer screen we delineated quadrat edges using colored lines (Figure 1-1), and marked nests

with colored dots as they were counted.

Error Calculations and Tests for Covariates

For most of the figures in this paper, we present aerial bias in nest counts as

Percent bias in aerial count = ((aerial count/ground count)-1)x100

A -12% error, for instance, indicates that the aerial count of a quadrat underestimated the

true number of nests found during the ground count by 12%. Note that in Table 2, estimates are

presented in the slightly less intuitive format of (ground count/aerial count) in order to form a

ratio with maximum colony counts.

During the ground count for each quadrat, we subj ectively ranked the level of tall

occlusive vegetative cover present as low (more than 50% cattails or tall grass cover and less

than 50% willows and other tree cover), medium (between 10% and 50% cattails or grass cover,









50-90% small willows and other tree cover) or high (more than 90% willow and other tall tree

cover). We compared White Ibis quadrats for statistically significant differences between

vegetation groups, using a nonparametric Kruskal-Wallis test for k independent samples, since

the ratio data were sufficiently non-normal that transformation did not produce a normal or

symmetric distribution for the data. We also used nonparametric tests to determine if date of

quadrat creation or colony location affected aerial estimation bias in 2006. We examined effect

of colony location (Alley North or New Colony 3) using a 2-tailed Mann-Whitney U-test, and

examined the effect of Julian date of quadrat creation using a 2-tailed Spearman rank correlation.

The four Great Egret quadrats were all located in an area where other species also nested,

including Snowy Egrets (Egretta thula), Little Blue Herons (Egretta caerulea), Tricolored

Herons (Egretta tricolor), and Black-crowned Night Herons (Nycticorax nycticorax). Adult

Snowy Egrets and Great Egrets can be difficult to differentiate from the air, as they are both

white and the difference in body size may not be obvious from a distance. Small heron nests can

be reliably differentiated from Great Egret nests by sight on the ground, but small heron nests in

the genus Egretta cannot be readily differentiated from each other until the chicks are 10-14

days old. For this reason, the true number of Snowy Egret nests present in the quadrats was

unknown.

We averaged the information from all quadrats for a given species to derive visual bias

estimates for that species in any colony, by taking the ratio between the ground and aerial count

for each quadrat, summing the ratios for all quadrats for that species, and dividing by the number

of quadrats, as in the equation below:

Visual bias=E(ground counti/aerial counti)/n









where i is a given quadrat and n is the sample size of quadrats for the species of interest.

For each species, we used the resulting bias estimate for the left side of the ratio equation

(quadrat ground count)/(quadrat aerial count) = (bias-corrected count)/(colony count)

to correct colony peak counts for visibility bias. We compared the new bias-corrected

counts and associated confidence intervals to original colony-wide aerial counts to determine the

significance of the results. We calculated the associated confidence intervals (CIs) for each

species' visibility bias as follows: CIs=Cl+(z*(sd/#\n)). For 95% confidence intervals, z=1.96; CI

was the sample mean error rate for the species; sd was the sample standard error for each

species; and n was the number of quadrats observed for each species. The mean error and

confidence intervals from the samples were then multiplied by the maximum total count for each

colony for the season, in order to obtain a new bias-adjusted colony count and 95% CIs.

Results

Great Egrets

Numbers of nests identified from the air were greater than the ground counts for two out of

four Great Egret quadrats, and we strongly suspect that some percentage of Snowy Egrets were

mistaken for Great Egrets in aerial photos (Table 1-1). This source of bias was apparently larger

on average than any bias due to vegetative occlusion, which would have caused aerial

underestimates rather than overestimates. Aerial counts ranged from -14% to 260% of ground

counts, with a mean of 70.28% (sd 131.20%). Due to the small sample size, statistical analyses

were not conducted on the Great Egret quadrats. The common nest location preferences of this

species meant that all four quadrats were placed in areas of high vegetative cover.

White Ibises

White Ibis nest numbers were generally underestimated in aerial surveys (Figure 1-2).

Although the severity of this underestimation varied, the mean bias in aerial surveys was










approximately -12.37% (sd=20.65%/; n=17). Visual bias did not increase with density of nests

(Figure 1-2), and in 2006 also did not vary by colony (Z=-0.714, p=0.475, n=13) or by date of

quadrat initiation (p=-0.275, p=0.363, n=13). Visual bias for White Ibises did not significantly

increase with increased vegetative cover (Figure 1-3; X2=0.163, p=0.922, n=17, df=2) although

the direction of the difference in median error values between the low and medium cover groups

suggests that with a larger sample size for the high vegetative cover, we might see a progression

in median error rate from low to high canopy cover (Figure 1-3). Median visual error for

quadrats in low vegetative cover was -11.07% (IQR -14.97% to -5.40%, n=6); for quadrats in

medium cover, -16.50% (IQR -18.71% to 8.96%, n=9); and for quadrats in high vegetative

cover, -6.98% (IQR -15.47% to 1.51%).

Correction for Visibility Bias in Colony Counts

Uncorrected maximum colony counts for White Ibis colonies lay outside the 95%

confidence intervals for the new bias-corrected estimates in both 2005 and 2006 (Table 1-2).

These uncorrected peak counts may thus be judged to be statistically significant underestimates

of the true numbers of nests present. The uncorrected counts for Great Egret colonies were not

statistically significantly different from the bias-corrected estimates, as the 95% confidence

interval for the bias-corrected counts overlapped with the original counts (Table 1-2). However,

the sample size of quadrats and colonies was small, and there was large variation in degree of

bias among quadrats. Visibility bias seemed to be quite substantial in some locations, if not on

average.

Discussion

There is evidence that visual bias in animals varies by species (Short and Bayliss 1985),

ground cover (Short and Bayliss 1985), observer quality (Erwin 1982), and other factors.

However, several studies have found a consistent underestimation of true numbers of birds using









aerial surveys (Pollock and Kendall 1987; e.g., Gibbs et al. 1988, Dodd and Murphy 1995). Our

results reinforce the conclusion that visual bias can be an important source of error in estimates

of avian breeding population size, and that the evaluation of detection probability is essential for

sampling efforts that use counts as population indices (Pollock and Kendall 1987). While the

degree of visual bias seems likely to vary with colony site and species, the amount of variation

we found in aerial bias among quadrats was similar to that found in other studies (Rodgers et al.

1995, Gibbs et al. 1988, Dodd and Murphy 1995). For ibises, the degree of bias (-12%) was

similar to that for other estimates for wading birds (-15% for Squacco Herons Ardeola ralloides,

Barbraud et al. 2004; -16%, -10% and -11%, respectively, for Great Egrets, Snowy Egrets, and

White Ibises, Kushlan 1979). The bias for ibises did not appear to increase with greater nest

density, in contrast to Great Blue Herons (Ardea herodia~s; Gibbs et al. 1988). This may be due

in part to the fact that many of the White Ibis quadrats were located in areas of relatively low

vegetative cover, which limited the number of potential nest strata and allowed individual nests

to be relatively easily distinguished.

There is little evidence in support of the vegetative cover prediction. The locations in

which quadrats could be placed were unfortunately limited by the necessity for ground access,

and thus a disproportionate percentage of the quadrats at Alley North colony were located in

areas of low and medium vegetative cover. An increased sample size for White Ibis quadrats in

high vegetative cover is needed before level of vegetative cover can be ruled out as an important

covariate for survey bias in White Ibises, but in our study the variation in aerial bias was much

larger within vegetative cover groups than between groups. Initial evidence for this species does

not support the hypothesis that visual bias due to vegetative occlusion increases with vegetative

complexity. For Great Egrets, a larger sample size is also needed, as well as a finer-scale









stratification of vegetative cover level. Great Egrets nest almost exclusively in areas that,

according to the criteria used in this study, would be classified as high cover (K. Williams pers.

obs.). However, Great Egrets nest largely above the canopy, and it may be (as Rodgers et al.

found for Wood Storks; 1995), that Great Egret visibility bias is likewise not significantly

affected by level of vegetative cover.

The uncorrected maximum colony counts for Great Egret colonies were not statistically

significantly different from the bias-corrected estimates, as the 95% confidence intervals for the

bias-corrected counts overlapped with the original colony counts. However, the sample size of

quadrats and colonies was small, and there was clearly a large amount of variation in bias for this

species in mixed-species groups, due to both visual occlusion and to difficulty in distinguishing

Great Egrets from other white species during aerial counts. The resulting bias-corrected peak

count of Great Egrets for Alley North in 2006 (with the associated confidence intervals) reflects

these dual sources of bias (Table 1-2).

The quadrat method of quantifying bias due to visual occlusion is necessary for situations

in which nests are located on multilayered substrates. Vegetation is often the cause of this sort of

visual bias (Frederick et al. 2003), but similar problems could be evident in colonies in which

nests are occluded by rock crevices or artificial structures. However, our results indicate that

ground truthing may not always be a gold standard for correcting visibility bias. It has been

suggested that ground counts may be inaccurate in highly vegetated areas where movement

within the colony is difficult, and canopy nests are difficult to see or count (Prater 1979,

Frederick et al. 2003). Additionally, if ground counts and aerial surveys are not conducted

within a very short time span, some nest turnover may occur and cause increased bias between

the two estimates (Frederick et al. 2003). In our study, we believe that ground counts were not










hampered by nest location or difficulty of movement, and aerial counts were conducted within as

short a time span as possible after ground counts. Nevertheless, ground/aerial comparisons could

not adequately address bias due to species misidentification (see also Rodgers 1995). Even the

ground surveys could not distinguish the non-target species (Snowy Egrets) from other small

herons, which meant that we were unable to correct for aerial bias due to species

misidentification. Since the two types of error (visual occlusion and misidentification) may be

inseparable in multispecies colonies (King 1976), it is important that such estimates be applied to

single-species colonies with extreme caution. In the future, separation of vegetative occlusion

from species misidentification bias could be achieved by comparing bias estimates from single-

and mixed-species areas. Alternatively, surveys could be conducted using helicopters, which

allow species to be more easily differentiated.

Our results indicate that White Ibis aerial surveys are likely to be significantly biased due

to vegetative occlusion. However, this bias did not increase significantly with vegetative

complexity or nest density. Our small sample size for Great Egret aerial surveys did not indicate

systematic bias due to vegetative occlusion, but species misidentification appears to be a

significant problem for this species. We suggest that visual bias due to both vegetative occlusion

and species misidentification should be explicitly measured in aerial surveys of colonies, as both

of these sources of error can affect estimates of breeding population size.










Table 1-1. Quadrats from 2005 and 2006 seasons from the Everglades of southern Florida.
White Ibis quadrats were 100 m2 (OT in One case 200 m2) and Great Egret quadrats
were either 400 or 900 m2. White Ibis nest densities varied from 0.08 to 4.45 nests
per m .


% error in
aerial
COUnic
-13
-61
9
-50
-16
-14


Vegetative
COVeTd
Low
Medium
High
Low
Low
High


Ground county
23
54
10

205
GREG 14
SH 35
GREG 24
SH 7, SNEG 1
GREG 5
SH 19
GREG 16
SH 18
72
33
30
18
67
414
445
139
303
96
390
114


Aerial county
20
21
11
4
173
GREG 12
SNEG 19
GREG 19
SNEG 6
GREG 18
SNEG 6
GREG 25
SNEG 10
69
30
24
20
73
467
491
113
253
73
319
99


Year
2005
2005
2005
2005
2006
2006


Species
WHIB
WHIB
WHIB
WHIB
WHIB
GREG/mixed


Location
Alley North
Alley North
Alley North
Alley North
Alley North
Alley North

Alley North

Alley North

Alley North

Alley North
Alley North
Alley North
Alley North
Alley North
New Colony 3
New Colony 3
New Colony 3
New Colony 3
New Colony 3
New Colony 3
New Colony 3


2006 GREG/mixed

2006 GREG/mixed

2006 GREG/mixed


-21 High


High

High

Low
Low
Medium
Medium
Medium
Medium
Low
Medium
Medium
High
Medium
Medium


2006
2006
2006
2006
2006
2006
2006
2006
2006
2006
2006
2006


WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB
WHIB


a. WHIB= White Ibis; GREG= Great Egret; SNEG= Snowy Egret; SH= unidentified small
heron.
b. Ground counts and aerial counts are estimated numbers of nests present in the quadrat.
c. Values in "% error" column are ((aerial count/ground count)-1)x100.
d. Vegetative cover values are subj ective measurements (low = more than 50% cattails or tall
grass cover and less than 50% tree cover; medium = between 10% and 50% cattails or grass
cover and 50-90% small willows and other tree cover; high = more than 90% willow and other
tall tree cover).









Table 1-2. Visibility bias-corrected peak counts for five Great Egret and White Ibis colonies
from the 2005 and 2006 breeding seasons.

Colony (species)a Year Peak count Visibility errorb (Sd) New estimate (LCI-UCI)"
Vulture (GREG) 2005 121 0.84 (0.46) 101 (46-156)
Vulture (GREG) 2006 458 0.84 (0.46) 383 (176-591)
Vacation Island (GREG) 2005 79 0.84 (0.46) 66 (30-102)
Vacation Island (GREG) 2006 155 0.84 (0.46) 130 (59-200)
Cypress City (GREG) 2005 107 0.84 (0.46) 90 (41-138)
Cypress City (GREG) 2006 173 0.84 (0.46) 145 (66-223)
Alley North (GREG) 2005 850 0.84 (0.46) 711 (326-1,097)
Alley North (GREG) 2006 1193 0.84 (0.46) 998 (458-1,539)
Alley North (WIHIB) 2005 12750 1.23 (0.43) 15,702 (13,078-18,327)
Alley North (WIHIB) 2006 13566 1.23 (0.43) 16,707 (13,915-19,499)
New Colony 3 (WHIB) 2006 4800 1.23 (0.43) 5,911 (4,924-6,899)
a. GREG=Great Egret; WHIB=White Ibis.
b. Error calculated for each species as (E(ground counti/aerial counti))/n, where i is a given
quadrat and n=number of quadrats.
c. Confidence intervals for visibility error were calculated as CIs== p+(z*(sd/9\n)), where CI is the
sample mean error, z=1.96 for 95% confidence limits, sd=sample standard deviation, and
n=sample size (number of quadrats for the species). New bias-corrected estimates and
confidence intervals were derived from multiplying mean proportion visibility error and error
CIs by the peak count for each colony.








































Figure 1-1. A 10x10 meter White Ibis quadrat at Alley North colony in WCA-3A (04/07/06).
The ground count for this quadrat was 72 nests; the aerial count from this photograph
was 69 nests. Written labels and lines between the four artificial markers were
inserted into the photograph using Adobe Photoshop Elements v. 2.0.














#g' *


*
*
* *


Ground Count of Quadrat


Figure 1-2. Proportional error in aerial counts of White Ibis quadrats. Proportion error is
calculated as (aerial count/ground count)-1, where the ground count of number of
nests was conducted during a walk-in survey of the quadrat and the aerial count was
obtained via digital photography of the quadrat site from small aircraft.













0.2




c0.0-




O

a,-0.4




-0.6






-0.8
Low Medium High

Vegetative Cover


Figure 1-3. Proportion error in aerial estimates of White Ibis quadrats by vegetative cover.
Vegetative cover values are subj ective measurements (low = more than 50% cattails
or tall grass cover and less than 50% tree cover; medium = between 10% and 50%
cattails or grass cover and 50-90% small willows and other tree cover; high = more
than 90% willow and other tall tree cover).









CHAPTER 2
ESTIMATING BREEDING POPULATION SIZE FOR ASYNCHRONOUSLY NESTING
COLONIAL WADING BIRDS

Introduction

The need for reliable avian demographic information has increased as scientists attempt to

use attributes of birds to measure both ecological degradation and restoration (Thompson 2002,

Rosenstock et al. 2002, Stolen et al. 2004). Birds are one of the most frequently suggested

bioindicators (Temple and Wiens 1989, Stolen et al. 2004), but accurate estimates of population

parameters can be hampered by a number of different kinds of bias. It is rare that entire breeding

populations can be counted, for instance, and most attempts at estimating population size are

approximations derived from estimates of related biological parameters, such as numbers of

nests (Bibby 2000, Williams et al. 2002, Bock and Jones 2004, Frederick et al. 2006). However,

there is a large amount of uncertainty in survey counts of nests and in the relation of those nest

counts to population size (Thompson 2002, Bock and Jones 2004, Frederick et al. 2006). Inter-

observer error, visibility bias, and species that nest asynchronously over the course of a breeding

season or nest in multispecies groups can present significant problems when trying to estimate

breeding population size. In the case of colonially breeding birds, large aggregation size may

also cause difficulties when attempting to estimate population size. However, it is vital that

accurate estimates be obtained for large colonies, as they may contain a large percentage of a

species' breeding population, and errors in population estimation for large colonies will

disproportionately affect total estimates.

Estimation Error

A huge amount of effort and scientific inquiry has been directed toward understanding bias

in nest counts (e.g., Erwiin 1982, Gibbs et al. 1988, Tomialojc and Verner 1990, Dodd and

Murphy 1995, Frederick et al. 1996, Prenzlow and Loworn 1996, Bart and Earnst 2002,









Rosenstock et al. 2002, Frederick et al. 2003). Aerial estimates of the population sizes of

colonial nesters have been found to be biased due to visibility problems, interobserver variation

in estimation error, and misidentification of species (Bayliss and Yeomans 1990, Frederick et al.

2003, Rodgers et al. 1995). In addition, most studies of estimation error to date have focused on

estimating nests that are present at the time of survey, and have rarely recognized the problem of

estimating nests that occur during the breeding period of interest but are not present at the time

of survey (Frederick et al. 2006, Cherenkov 1998).

Asynchronous breeding, even within a defined breeding season, is common in colonially

nesting birds, including terns (Hernandez-Matias et al. 2003), prions (Liddle 1994), martins

(Magrath 1999), and wading birds (Ciconiiformes; Piazza and Wright 2004). In many species,

nesting may therefore occur before surveys begin during a season, after they end, or in between

consecutive surveys. Estimates of unmarked nests on any given date are inherently

underestimating the true numbers of nests, partly because nests may be occurring outside the

dates of survey, and partly because novel nests on any given survey may be confused with nests

that were present on a previous survey date. This is true even if the counts on each date are

highly accurate. In such a situation, where nests cannot be individually distinguished, the best

guess for the population size is often the largest or "peak" count of concurrent nests on a single

survey date. These circumstances occur in several taxa, including wading birds, seabirds, loons,

and shorebirds (Frederick et al. 2006; e.g., Sagar and Stahl 2005, Earnst et al. 2005, Morrison et

al. 1994). While the resulting bias may be negligible when nesting is highly synchronous and

little renesting occurs, the degree of error in less synchronous birds can be substantial (e.g., 47%

in wading birds; Frederick et al. 2006).









In addition to problems with nesting asynchrony, using numbers of nests as a proxy for

breeding population size can be complicated by the degree of occurrence of renesting or multiple

nesting (Thompson et al. 2001, Nagy and Holmes 2004). Renesting and multiple brooding are

rare in many populations of high-latitude nesting birds, but they can be common among

temperate and tropical species. We wish to make clear in this chapter that we are addressing the

problem of estimating numbers of nests, and make no assumptions about degree of renesting or

double brooding.

A New Strategy For Estimating Breeding Population Size For Unmarked Populations

The problems described above illustrate a need for the development of new methods for

estimating the size of breeding bird populations (Figure 2-1). Visibility bias has been addressed

elsewhere (Chapter 1) and has been well explored in the literature; therefore we will primarily

focus here on methods to account for nesting asynchrony. This problem may be approached by

estimating the turnover of individually marked nests and incorporating this information into a

population estimate using a mark-resight method. We have used the superpopulation approach

(Schwarz and Arnason 1996), a variation on a Jolly-Seber open-population capture-recapture

model. In this case nests are treated as individuals in an animal population, where the population

is equivalent to the estimated total numbers of nest starts during a period of interest such as a

breeding season. We chose to study bias derived from nesting asynchrony in Great Egrets

(Ardea alba) and White Ibises (Eudocimus albus) because these two birds are conspicuous

colonial nesters with very different life histories and temporal nesting patterns. We believe that

the methods outlined here provide a new and much improved technique for the monitoring and

estimation of large unmarked populations of nesting birds.









Methods


Survey Methodology

We conducted monthly aerial surveys between January and June of 2005 and 2006 in

which we searched for and counted colonies of wading birds in the central and northern

Everglades (Broward, Dade and Palm Beach counties, FL). We conducted surveys in a Cessna

Skyhawk (172) high wing aircraft flying at approximately 177 kph and 304 meters altitude, with

one observer on each side of the aircraft. When we located colonies we circled them for aerial

visual estimates and aerial photography at 244 meters above ground level. We took photographs

from the copilot' s seat through an open window, using a Canon EOS 20D high-resolution digital

camera with a 28-135mm image stabilizing lens. We edited photographs and counted nests

using either Adobe Photoshop Elements version 2.0 or the shareware program Paint.Net version

2.0, using colored dots to mark nests as they were counted and delineating overlapping

photograph edges using colored lines. We accepted the largest single-month count of the

breeding season from these aerial photographs as the peak seasonal count for each species.

In addition to these "peak count" aerial surveys, we also took photographs on a roughly

semiweekly basis of groups of individually identifiable nests. This allowed us to estimate

turnover (losses of nests and new entries to the colony). Groups of birds were identified through

proximity to natural or artificial landmarks (see below). We took photographs at an altitude of

approximately 152 meters from the copilot' s seat with the door removed, so as to take photos as

close to vertically as possible. We generally conducted survey flights between 08:00 and 10:30

am or 16:00 to 18:00 pm, but flight times varied with weather, plane availability, and other

factors. Midday light was better for survey photographs, but turbulence, wind and poor weather

also tended to be more severe in the heat of the day. We conducted these surveys in three small

Great Egret colonies and one large mixed-species colony (Table 2-1) in Water Conservation









Area 3A (Figure 2-2) in one or both of the 2005 and 2006 breeding seasons. The mixed-species

colony, Alley North (Figure 2-3) was too large to conduct a complete survey. We therefore

concentrated on photographing certain landmarks that could be reliably located from the air.

Artificial Geographic Markers

The repeated identification of individual nests in aerial surveys required some kind of

identifying landmarks. We used natural landmarks such as clearings within the tree island, but

we also erected artificial markers in areas of colonies that were both accessible on the ground

and where birds nested in substantial numbers. These markers were of three types, all of which

were easily identifiable from 100-300 meters above ground level:

* A roughly 1-2 meter wide area of vegetation painted with a dilute white latex paint (1:3
water:paint) dispensed from a backpack sprayer of the type common in landscaping. These
markers were only utilized during the 2005 season.

* A 3-meter tall, 10-cm diameter white vertically mounted PVC pipe with a 1.5-meter
horizontal X on the top (Figure 2-4A). These markers were primarily utilized during the
2006 season.

* A lxl-meter piece of white or pale-colored cotton cloth, tied at the corners to vegetation in
a horizontal position (Figure 2-4B). This type of marker was used exclusively during the
2006 breeding season.

Photographic Analysis

We used printed photographs of the same areas from different dates to identify individual

nests and obtain presence-absence data for each nest on each subsequent survey date. We

uniquely numbered nests consecutively from the earliest date, and entered presence-absence

information into the database using the following format:

* Nest inactive (0)- this included cases in which nest and parent/chicks were not visible; nest
was visible but empty (no sign of parents/chicks); or there was no photo of the nest area
available on that date

* Nest active (1)- parent and/or chicks were visible










Since some nests were temporarily inactive or not visible in photos from a particular date,

we only assumed that a nest had failed if it was found to be inactive on four consecutive survey

dates. After that time, we treated any nest in that location as a new nest start. Although we used

the presence of a large white bird as an initial indication of nesting, these could have been

roosting birds or birds temporarily away from their nests. Before we analyzed each presence-

absence database, we therefore eliminated first observations of all nests from the database, and

so deleted from the database all "nests" that were seen only once. In actuality, we believe that

some small proportion of these single-appearance nests were indeed nests. In addition, the use of

the four consecutive zero rule (as above) may have missed some nests that failed, restarted, and

failed again within the period of four visits. However, both these types of possible errors seem

unlikely given the relatively long courtship and egglaying periods of these birds (4-10 days for

Great Egrets, according to McCrimmon et al. 2001; 9-10 days for White Ibises, as cited in

Kushlan and Bildstein 1992) and the semiweekly frequency of aerial surveys. In any case, if

either of these assumptions is incorrect, the effect is in the same direction, to underestimate the

true numbers of nests. We eliminated the first sighting of nests that were seen on multiple dates,

in addition to those nests seen only once, in order to avoid artificially inflating survival

estimates.

Independent observers. Three independent observers analyzed the same set of

photographs for one colony (Vacation Island) from the 2006 breeding season. This allowed for a

comparison of photo analysis methods between observers, and served as a test of the

repeatability of the method. We used Observer 3's results (the median estimate from the three

observers) as the final superpopulation estimate for the 2006 Vacation Island colony, as that

observer also conducted the analyses for all other colonies.









The Superpopulation Modeling Approach to Population Estimation

The superpopulation approach (Schwarz and Arnason 1996) is a variation on a Jolly-Seber

open-population capture-recapture model that includes as a derived parameter the gross births

within the population. This parameter includes all animals that enter the population during the

sample period and either survive until the next survey, or emigrate or die before they are

available to be sampled (Schwarz and Arnason 1996, Schwarz et al. 1993). In our study, nests

were equivalent to individuals in a population; the number of new nests present at each survey

represented immigration into the population, and the number of "marked," or individually

identifiable, nests that failed between consecutive surveys indicated the level of emigration from

the population. The model's calculated gross superpopulation size represented the total number

of nest starts over the entire sampling period. Detectability of nests is included in the model as

an encounter probability term for each survey.

We fit regression models to the capture-recapture data from colonies using Program

MARK version 4.3 (White and Burnham 1999). We used the POPAN data type (Arnason and

Schwarz 1995), which utilizes a robust parameterization of the Jolly-Seber model (Schwarz and

Arnason 1996). We set time intervals to fractions of weeks between each set of consecutive

surveys, and allowed the four parameters (survival cp, encounter probability p, entry probability

pent, and superpopulation size N) to vary with time, depending upon the model. For example,

for a database with six survey dates, there were a total of seventeen possible parameters (five

survival interval probabilities, five entry interval probabilities, six encounter probabilities, and

one superpopulation size). For each database, we tested a set of four candidate models: a general

(fully time-dependent) model, a so-called "dot model," in which survival and entry variables

were held constant through time, and two models in which either encounter probability or

survival probability was allowed to vary with time while the other was held constant (Cooch and









White 2007). Probability of entry into the population would not be expected to hold constant

throughout the season, since for both species there tends to be a surge of initial nesting in an area

followed by lower levels of nest initiation thereafter (McCrimmon et al. 2001, Kushlan and

Bildstein 1992). Thus, entry probabilities were allowed to be time-dependent in all models. For

models in which survival probability cp and encounter probability p varied with time, not all

parameters in the model were estimable (Schwarz and Arnason 1996). We set p1=p2 and pk Pk-1

so that all survival parameters were estimable in the model (J. Nichols pers. comm.). In the

model in which survival was held constant and encounter probability varied, the initial p value

was still inestimable, so for this model we set p1=p2 and allowed all other encounter probabilities

to vary. We used a sin link function to estimate survival and encounter parameters, a mlogit(1)

function to estimate entry parameters, and a log link function to estimate superpopulation size N.

The gross superpopulation size N* is a derived parameter of the POPAN model. It

includes the net superpopulation size (all animals that enter the population between two

consecutive surveys and are available to be captured during the second survey) as well as

animals that enter and leave the population between consecutive surveys and thus are never

available to be sampled. We used MARK's model averaging capability to find the weighted

average values for all parameters across all models. Using the counts from each survey date and

the estimated encounter and survival probabilities from each survey date or interval (adjusted for

time elapsed between surveys), the number of new entries ('births") into the population between

each consecutive set of surveys could be calculated (Schwarz et al. 1993, Schwarz and Arnason

1996). The gross superpopulation size in MARK is derived from summing these gross entries

between each consecutive set of survey dates, and adding the sum to the estimated number of

nests present during the first survey (after Schwarz and Arnason 1996, Schwarz et al. 1993):









N. n./p.

B. = N.+ N. (c.)^'t.


Bi = I n(gipt.)


N* = N1 + Ik-1 Bi

where Ni is the estimated total number of individuals in population at occasion i; ni is the

number of individuals seen at occasion i; pi is the encounter prob. at occasion i; Bi is the

estimated number of individuals entering the population between sampling occasions i and i+1;

cpi is the survival probability at occasion i; ti is the time between surveys i and i+1(as a proportion

of a week); B~i is the estimated gross number of individuals entering population between i and

i+1; N* is the estimated gross superpopulation size; and k is the total number of surveys.

We fit the four models to the data from each colony's presence-absence database

(Appendix A). We used an information-theoretic approach to model selection, and considered

that a model with a AAICc value of less than 2 was relatively well supported by the data, while a

model with a AAICc value of greater than 10 was not supported by the data (Williams et al.

2002). We quantified the goodness of fit (GOF) of the most general model using chi-square tests

for each survey interval to see if observed values varied from the expected number of surviving

and encountered nests. These tests evaluated the following assumptions inherent in Jolly-Seber

models (Cooch and White 2007, Pollock et al. 1990):

* There is no heterogeneity in capture probability, either among individuals or among
cohorts (cohorts in this case meaning all nests that were seen for the first time on the same
date) .

* There is no heterogeneity in survival probability among individuals or cohorts.









We calculated chi-square values using Program RELEASE (Burnham et al. 1987),

available within MARK. The chi-square values generated for each survey interval were summed

for the entire sampling period, and were divided by the degrees of freedom to obtain a variance

inflation factor, c-hat (a measure of overdispersion in the data). Following Cooch and White

(2007), we accepted that a c-hat value of 1 indicated good model fit, values of 1-3 indicated

moderately good fit, and >3 indicated probable violation of model assumptions--that is, that

none of the models in the tested model suite may be a good fit for the data. A c-hat value of less

than 1 essentially means that the data are underdispersed, and there is little agreement in the

literature about what this means or what to do about it (Cooch and White 2007). Following the

recommendation of Cooch and White (2007), we used a c-hat value of 1 when the c-hat

calculated in RELEASE was less than that. Otherwise, we multiplied the calculated c-hat values

presented in the tables in Appendix A (c-hat values of less than 1 are not presented) by the

model-based variance and covariance estimates for each model suite. This adjusted model

weights to compensate for the overestimation of precision caused by overdispersion.

Subsampling Alley North Colony

For the smaller Great Egret colonies (tree island lengths of approximately 120-200

meters), all or most of the colony could be covered in one or two photographic passes. For the

very large colony (length of tree island approximately 1900 meters; Figure 2-3), we subsampled

the colony during aerial photography, resulting in a collection of geographically distinct year-

and species-specific estimates of superpopulations with standard errors. Samples were surveyed

for different lengths of time during the breeding season, depending upon nesting patterns and the

availability of good-quality photographs for each area. We calculated a superpopulation estimate

for each sample, and compared it to the raw count for each sample, which was the number of

nests seen in the sample on the survey date closest to the colony's peak count date. This raw










count was assumed to be the number of nests in the sample unit that would have been seen and

included in the peak count on the peak count date. We assumed that the ratio between the

samples' superpopulation estimates and raw counts also held for the entire colony, and used that

proportional difference in a ratio with the entire colony's peak count to find an extrapolated

superpopulation size for the entire colony. For example, if on average 70% of nests present

during the season in the sample areas were not present at the survey date closest to the peak

count (when the raw counts were taken), then the colony peak count likewise would be assumed

to be about a 70% underestimate of the total number of nest starts.

We found the proportional difference between the superpopulation estimate and the raw

count for each sample, and then took the average of these proportions across all samples. We

took the ratio between each sample's values and averaged the individual ratios, rather than

obtaining an overall ratio between the two types of estimates for all samples (e.g., calculating the

difference between the summed raw counts and superpopulation estimates across all samples), as

it is unlikely that there is one underlying level of nest turnover within the colony; different

nesting cohorts within the colony are likely to have different levels of nest turnover at different

times within the breeding season. The ratio of averages (summed value), while in some cases

shown to be less biased and have a smaller variance estimator than the average of ratios (Rao

2005), weights all nests in all samples equally and assumes that there is one underlying ratio

between raw counts and superpopulation estimates for the colony. The average of ratios

recognizes that each sample may represent an area with a different ratio between the raw count

and superpopulation estimate (due to date of cohort initiation, nest density, or other factors), and

thus takes natural variation in turnover rate into account when estimating an overall proportion

for the colony.









The average ratio of this proportional error across all samples was our estimated colony-

wide proportional difference between the peak count and extrapolated superpopulation count, as

follows:

(E(N~i/RCi))/n = (N~colony/PCcolony)

N~i is the superpopulation estimate for sample i; RCi is the raw count for sample i; n is the

number of samples in the colony for that year and species. Cumulatively, (E(N~i/RCi))/n is the

average proportion of the total number of nest starts that were seen in the raw counts. PCcolony

is the peak count for the colony. We solved for "N~colony," the entire colony's extrapolated

superpopulation estimate. We also calculated the variance of the ratio estimate, in order to

calculate 95% confidence intervals for the colony-wide superpopulation estimate (Appendix B).

These superpopulations were compared to colony peak counts to determine the level of bias

introduced into peak counts by 1) asynchronous nesting activity, expressed through changes in

the identities and numbers of nests present over the course of the season due to changes in

survival and entry probabilities; and 2) imperfect detectability of nests in aerial surveys,

expressed for each survey term as an encounter probability.

Subsampling to Determine Necessary Survey Frequency

We subsampled the 2006 Cypress City dataset, a very complete survey dataset with 17

surveys over the course of two months, to determine if flying approximately twice a week as we

did for this research is necessary for the success of the technique. We subsampled the dataset to

manipulate both the number of surveys conducted within a specified time period and the

regularity of the survey intervals. We calculated a superpopulation estimate for each subsampled

dataset and compared these estimates to the inclusive superpopulation estimate for the colony, in

order to examine the loss of estimation accuracy associated with reduced flight frequency.









Results

Interobserver Error in Tallying Individual Nest Histories From Photographs

Three independent observers analyzed the same photographic library from surveys of the

Vacation Island Great Egret colony during the 2006 breeding season (Figure 2-5). The peak

count for this colony in 2006 was 155 nests. Observer l's presence-absence database resulted in

a superpopulation estimate of 509 nests (4.45 nests); Observer 2's estimate was 409 nests

(2.65); and Observer 3's estimate was 480 nests (3.28). Average interobserver error rate was

roughly 13%. Colony information presented in the superpopulation and combined results

(below) is from Observer 3's analysis, as this observer also conducted analyses for all other

colonies.

Superpopulation Estimates

Superpopulation estimates for the smaller Great Egret colonies were 147% to 482% of the

associated peak counts for the same colonies (Figure 2-6), suggesting that asynchrony in nest

initiation has a large effect upon the likely number of nest starts. The extrapolated

superpopulation estimates for Alley North White Ibis colonies in 2005 and 2006 and the Great

Egret colony in 2006 were 213% to 300% of peak counts for the same breeding populations

(Figure 2-7). In all eight cases, the peak counts for colonies were well outside the 95%

confidence intervals for the superpopulation estimates (Table 2-2).

The superpopulation estimates and raw counts for Great Egret and White Ibis samples in

the very large Alley North colony are shown in Table 2-3. The peak count for 2005 was 12,750

White Ibis nests on March 20th; the closest superpopulation survey was also on the 20th, so the

raw counts for each sample were taken from this date. Using the superpopulation estimates and

raw counts for each sample as described above and in Appendix B, we estimated the

superpopulation size for the entire White Ibis colony to be 38,275 nests (1,941 nests). In 2006,









the peak count for White Ibises was 13,566 nests on April 19th. The two closest superpopulation

survey dates were on April 17th and 21st, so we used the average of the two raw counts from

these dates. We estimated the superpopulation size for the entire colony to be 29,287 nests

(1306 nests). The peak count for Great Egrets in 2006 was 1,193 nests on March 16th. The

closest superpopulation survey date for four samples was the 17th. Two samples were only

partially surveyed during the season (surveys of these areas were initiated later in the season, as a

new cohort of nests entered the area), and for these samples the closest survey date was the 21st

We estimated the superpopulation size for the entire colony to be 2,538 nests (32 nests).

Although peak counts consistently underestimated numbers of nest starts, the degree of

underestimation varied widely among colonies and years (Figure 2-8); the Cypress City Great

Egret colony, for instance, was more asynchronous in 2005 than 2006, and thus the peak count in

2005 represented a much smaller proportion of the superpopulation-derived estimate than did the

peak count for 2006. However, at Vacation Island colony, only 25 kilometers away, the level of

nesting asynchrony actually increased slightly from 2005 to 2006. Interestingly, although levels

of nest turnover varied by season and location, turnover did not appear to vary consistently by

species (Figure 2-8).

Subsampling to Determine Necessary Survey Frequency

There was a clear tradeoff between frequency of survey and accuracy of superpopulation

estimate (Table 2-4). Although conducting survey flights 17 times over the course of a 3-month

period provided the highest superpopulation estimates for the Cypress City colony in 2006, each

survey flight cost roughly $180, and this frequency was expensive to sustain. Of the frequencies

tested in Table 2-4, flying every five days provided the best combination of accuracy and number

of nests added to the superpopulation estimate per additional survey flight. Superpopulation

estimates (including the costs of weekly surveys and the labor costs involved in photographic










analysis) appear to be around five times as expensive as peak count estimates to produce

(Appendix C).

Discussion

The estimates of numbers of nest starts that incorporated asynchrony and detectability bias

were significantly different from the peak counts in all eight colonies and years. Asynchrony in

nest initiation and failure, rather than problems with nest detectability, is the maj or source of bias

in our peak count estimates. In Chapter 1, we compared ground and aerial counts of nests and

indicated that visual bias alone may cause errors on the order of 12% underestimates and 70%

overestimates for White Ibises and Great Egrets, respectively. Kushlan (1979), using apparently

similar methods, found underestimates of 11% for White Ibises and 16% for Great Egrets. In

general, such independent visual bias estimates are at least an order of magnitude smaller than

our estimates of combined asynchrony and visual bias in this chapter, which ranged from 47% to

382%. Moreover, the detectability term in the superpopulation model, although functionally

equivalent to an independent visual bias estimate, is actually smaller than such an independent

estimate would be. The detectability term is based solely upon those types of visual bias that

may be detected from repeated aerial surveys (e.g., adult birds temporarily off of nests,

vegetative occlusion due to the angle of a particular photograph), and does not include sources of

bias that may only be detected through a comparison of aerial and ground counts. Nests that are

so heavily occluded by vegetation that they are never seen during any aerial survey, for instance,

are not included in our superpopulation modeling, and species misidentification in aerial surveys

may only be recognized through the comparison of aerial to ground counts. Thus, if the

estimates were truly equivalent, the difference between simple visual bias estimates and our

combined estimates would probably be larger than is represented here. Regardless, visibility

bias appears to affect estimates of nest numbers to a relatively small degree. The vast maj ority









of error in our combined estimates appears to be from the effects of asynchronous nesting and

nest failure (together taken as turnoverr"), as the numbers of individually identifiable nests were

several times the peak count values for all colonies. We therefore address the importance of

superpopulation modeling below solely in regard to the problem of turnover and asynchrony in

nest initiation.

Our results indicate that there was not a large difference between White Ibis and Great

Egret nest turnover rates. This was surprising, since the two typically have different

reproductive phenologies, timing of nesting, and nest failure rates (Frederick and Collopy 1989a,

Kushlan and Bildstein 1992, McCrimmon et al. 2001). However, the inter-year comparisons

(Figure 2-8) illustrated that the degree of turnover varied widely between years and, moreover,

between colonies during the same year (see also Frederick et al. 2006). This suggests that nest

turnover rates are not constant and will have to be estimated on an individual colony basis for

each season of interest. This is unfortunate, since it makes measuring nest turnover much more

difficult and costly.

However, it is apparent from these results that the amount of bias introduced through the

failure to incorporate nesting asynchrony into estimates is also unacceptably large. Estimates of

total number of nest starts at a colony over the course of a season are a vast improvement over

peak counts for two reasons. First, we believe that our estimates of total number of nest starts

are much closer to the actual breeding population size for these colonies than are peak count

estimates. Peak counts have been accepted as minimum breeding population size estimates for

the purposes of developing an index of breeding activity between seasons, but (with good

reason) they have never been presented as estimates of actual breeding population size. Second,

some authors have claimed that peak counts and related methods may be of use as indices of










population growth (James et al. 1996, Link and Sauer 1998), even if they are not reliable

estimates of true population size (Link and Sauer 1998, Link and Sauer 2002). Indices are useful

indicators only if they reliably reflect true relative differences between periods of interest. In the

case of wading birds, we have shown that peak counts are not reliable indicators of relative

differences between annual numbers of nests, and are reflecting different proportions of the

actual breeding population size in different seasons. Given the variability in turnover between

seasons, it appears likely that monitoring schemes using peak counts may be failing to serve their

intended purpose of providing even a reliable index of population size.

Superpopulation estimates of numbers of nests, in contrast, are comparable between years

and may be used as an index of breeding activity. Since renesting may occur, the total number of

nest starts is still not an accurate estimate of actual breeding population size; however, by

examining the total number of nest starts each season (along with an estimate of nest success),

one can obtain a fairly good idea of reproductive conditions for wading birds in a given season.

At Vulture colony in 2005, for instance, the estimate of total number of nest starts was almost

five times the peak count. This colony was completely destroyed by a mammalian predator early

in the season, and was repopulated with new nests five weeks later. The peak count does not

provide an accurate depiction of breeding conditions within Vulture colony for this season, but

our estimate of total number of nest starts, along with a rough estimate of nest failure rate, was

more informative. We therefore feel that despite any shortcomings of the methods used (see

below), the superpopulation technique of estimating nest effort in colonies is likely to be more

accurate than the peak count method.

Limitations and Assumptions

There are several apparent biases and limitations that we have noted in the use of these

estimation techniques. First, detectability estimates in superpopulation modeling do not include









nests that were so heavily occluded by vegetation that they are impossible to see from the air

(and thus are never included in any survey counts). The relative frequency of these completely

occluded nests may only be estimated by comparing ground and aerial counts of the same areas

in colonies (Chapter 1). This inherent bias in the superpopulation-aerial photographic method

probably leads to a slight underestimation of true number of nest starts. This is particularly true

for birds that typically nest subcanopy (e.g., White Ibises), as opposed to those nesting

predominantly in the canopy (Great Egrets). In addition, the total number of nest starts in each

colony was also probably underestimated because the first sightings of all nests were removed

from a database before it was analyzed. Although most of the one-sighting "nests" probably

were not true nest starts, the small proportion that were could cause a bias of unknown

magnitude and lead to an underestimate of true nest starts. We believe, however, that this error

is likely to have been small in comparison to other sources.

Not all of the colonies or samples within colonies were surveyed for the same period of

time, due to photo quality and survey restraints. For example, many of the sample

superpopulation estimates for the very large Alley North colony are likely to be underestimates

of true numbers of nest starts, since only about 20% of samples were surveyed for the entirety of

the breeding season. Surveys of a particular area were usually initiated when markers were

placed in the area, except in cases where the survey area was identifiable in one or multiple

survey photographs before the markers were erected, and surveys were halted when there were

very few or no visible birds remaining in the area. Since there are often different nesting cohorts

in a colony of this size, and nest initiation rates varied during the season, we did not feel that it

was biologically justified to attempt to extrapolate superpopulation sizes for the entire survey

period for samples that were partially surveyed. However, it must be noted that a more thorough









colony-wide, pre-season marking methodology would enable more accurate estimates of

superpopulation size (see below), and likely further increase superpopulation estimates for this

large colony.

Model fit

The calculated c-hat values for several of the model sets were larger than 3, indicating

problems with model fit and possible violations of model assumptions. Survival probability is

probably temporally heterogeneous in colonies, because nest survival varies among stages of

nesting (Mayfield 1975, Frederick and Collopy 1989b, Torres and Mangeaud 2006) and as a

result of environmental variability within the season (Frederick and Collopy 1989a). Likewise,

encounter probability is unlikely to be the same for all nests, because nests that are located in

taller vegetation are more highly visible from the air and more likely to be seen on every survey.

Many of the violations of model assumptions appeared to be related to the test for homogeneity

in capture probability. If this heterogeneity was the result of visibility, as above, it might help in

the future to categorize nests according to visibility and conduct analysis in MARK with

visibility as a grouping variable. Fortunately, however, both survival and encounter probability

estimates in Jolly-Seber models tend to be robust to this type of heterogeneity, so long as average

encounter probability is high (>0.5; Pollock et al. 1990), which it was for all colonies and

samples we examined.

It is also possible that our models fit better in reality than was indicated by the calculated

c-hat values in RELEASE. Due to the large sample sizes in our study, we had very good power

for our goodness-of-fit tests to resolve even small differences between observed and expected

values. We examined the chi-square tables in RELEASE and suggest that the differences found,

while in some cases statistically significant, were probably not biologically significant in terms

of numbers of surviving or encountered nests.









Refinements of the superpopulation technique

We recommend that the first two flights of the season be close together (within 2-4 days),

as the first survey flight will be dropped from the database before analysis. This maximizes the

usable period of observation. Aerial photographic surveys should also be initiated as soon as

incubation begins for the maj ority of the colony. For wading birds, the end of breeding and

onset of incubation corresponds to a decrease in noise and movement within the colony. From

an aerial view, birds in incubation seem settled and sitting still; few are flapping around or

displaying with any frequency (Frederick 2006). After the second survey flight, our analysis

suggests that survey interval can be reduced to once every Hyve to seven days without large loss

of information. While some information can be obtained from less frequent surveys, it becomes

increasingly difficult to accurately identify nests with greater time elapsed between sequential

photographs.

Very large colonies will require random or stratified random sampling designs in order to

capture the full heterogeneity in nest turnover among cohorts within the colony. Since landmarks

must be used to identify nests, our sampling areas were not located randomly. Even for areas

where we used artificial markers, the difficulty of moving through high-density vegetation

restricted our choice of sites. By comparing the ratio of superpopulation estimates to raw counts

for all samples, we examined the relative proportion of nests missed in the peak count, rather

than absolute numbers missed. However, nest turnover rate may vary with nest density, a

possibility that has not been examined in wading birds. If true, in order for the average sample

ratio of superpopulation estimate to raw count to hold true for the entire colony, our samples

would have to be representative of the range and relative proportions of different nest densities

within the colony. The ideal solution would be to distribute markers throughout the colony site

prior to the beginning of the nesting season, in order to allow for random placement of samples.









The design of such a marker must be 1) small enough to fit into small aircraft, 2) small enough to

be deployed from the aircraft safely, 3) large and conspicuous enough to be easily visible from

the air, 4) biodegradable, and 5) of a nature such that it will rest on the canopy of nesting

substrate. The colonies we studied were densely vegetated, remote, and very difficult to place

markers in, and may represent a worse case scenario. For colonies that are more readily

accessible, artificial landmarks might be more readily placed by hand.

We consider the large sample sizes of nests obtainable through the use of the aerial

photographic technique to be central to encompassing the inherently large variation in nest

success and synchrony that exists in large colonies. It is clear that the repeated photo technique

represents a significant increase in manpower and cost by comparison with the peak count

method (Appendix C), but we would suggest that there is no bargain to be had in using a cheaper

method that yields uninterpretable results. If funds are limited, a potential strategy may be to

survey intensively every two to three years, rather than on an annual basis.

Incorporation of renesting

We have focused in this paper on providing estimates of total numbers of nest starts

occurring throughout a nesting season. Since nest failure is common (Frederick and Collopy

1989c), and the nesting season is long (3-4 months), renesting may be frequent. For this reason,

the nesting population (numbers of pairs or breeding females) would be smaller than the numbers

of nest starts (Piazza and Wright 2004). Since this difference could be substantial, we strongly

recommend that this technique be used in conjunction with studies of renesting frequency.

Implications of Results

The results of this study indicate that breeding population sizes for colonially breeding

birds may be considerably larger than previously supposed. However, actual numbers of

breeding pairs are probably somewhat smaller than suggested by the estimated numbers of nest









starts in this study, because of the probability of renesting (see above). Likewise, it should be

clear that the larger estimates resulting from the methodology we used do not represent an actual

increase in population, but rather an increase in our ability to measure the population. The

combined estimates of breeding population size allow us to more accurately measure inter-

annual and inter-site variation, and thus to see variation in breeding effort that was previously

masked by bias associated with peak counts or similar surveys. These new estimates are

therefore both more accurate and offer associated confidence estimates, which peak counts and

related one-time surveys lack.

We suggest that peak or snapshot counts can be useful as indices in situations where the

study species has high nest synchrony, or alternatively where sources of bias are measured on a

seasonal basis and incorporated into estimates. However, our research indicates that unless

demonstrated otherwise, snapshot surveys of avian breeding populations probably cannot be used

effectively as indices of breeding population size. Moreover, there is no way to tell from

snapshot counts alone just how biased these estimates of population size may be. In this study

we have examined an extreme case (long nesting period, highly variable nest failure, potential

for poor nest visibility), but many avian species probably show similar characteristics, if to a

smaller degree. Given the strong biases that have now been demonstrated (this study, Frederick

et al. 2006) even lesser degrees of asynchrony are likely to alter estimates of breeding population

size by a considerable amount.









Table 2-1. Names and locations of study sites. WCA-3A is Water Conservation Area 3A, a
large area of primarily sawgrass and wet prairie in the Everglades controlled by the
South Florida Water Management District. The city of Homestead lies south and
slightly east of the maj ority of WCA-3A.
Colony Location Latitude Longitude
Alley North WCA-3A N 260 11.179 W -800 31.431
Cypress City WCA-3A N 260 07.468 W -800 30.283
Vacation Island WCA-3A N 250 54.939 W -800 37.813
Vulture WCA-3A N 260 01.470 W -800 32.240
Homestead General Airport (KX-51) Homestead N 250 30.0 W -800 33.3









Table 2-2. All colonies for which we followed nest fate through time in aerial survey
photographs and calculated a superpopulation estimate. For smaller Great Egret
colonies, number of aerial surveys conducted during the season and the final number
of individually identified nests (database size) are listed. For Alley North colonies,
surveys were conducted of samples within the colony rather than of the colony in its
entirety; data from all samples is in Table 2-3. Peak counts are the maximum one-
time counts of number of nests in colonies, taken from a series of monthly aerial
surveys during the breeding season. Superpopulation estimates for the same colonies
incorporate detectability bias and bias due to nesting asynchrony, and thus are much
higher estimates of nest numbers than peak counts. Superpopulation estimates also
have measures of uncertainty associated with them, which peak counts lack.
Year Colony Species" Peak count Number of Database size Superpopulation estimate
surveys in (# nests)" N* (LCI-UCI)d
databaseb

2005 Vulture GREG 121 7 499 583 (521-645)
2005 Vacation Island GREG 79 7 215 233 (226-240)
2006 Vacation Island GREG 155 21 474 480 (477-483)
2005 Cypress City GREG 107 6 244 268 (259-277)
2006 Cypress City GREG 173 17 249 254 (251-256)
2006 Alley North GREG 1,193 --2,538 (2,474-2,601)
2005 Alley North WHIB 12,750 --38,275 (34,392-42,157)
2006 Alley North WHIB 13,566 --29,287 (28,674-29,899)
a. GREG=Great Egrets; WHIB=White Ibises.
b. Number of surveys in database does not include first survey, as all first sightings of nests were
eliminated from the database before analysis (see text).
c. The database size is the number of nests in the sample's presence-absence database after the
first sightings were eliminated.
d. N*=weighted model-averaged superpopulation estimate, calculated in Program MARK. LCI-
UCI are 95% confidence intervals.









Table 2-3. All samples of Great Egret and White Ibis populations from large colony (Alley
North) for which we followed nest fate through time in aerial survey photographs and
calculated a superpopulation estimate.
Year Samplea SpecieSb # N* (LCI-UCI)d Raw count Proportional difference
surveys for between N* and raw county
in sample
databases
2006 Q1 GREG 10 56 (54-59) 24 2.33 (2.25-2.46)
2006 Q2 GREG 9 59 (58-60) 28 2.11 (2.07-2.14)
2006 Q3 GREG 9 58 (56-60) 20 2.90 (2.80-3 .00)
2006 GE SE 1 GREG 17 43 (43-44) 22 1.95 (1.95-2.00)
2006 GE SE 2 GREG 16 73 (73-74) 57 1.28 (1.28-1.30)
2006 GE SE 3 GREG 10 54 (51-58) 19 2.84 (2.68-3.05)
2005 IT 1 WHIB 5 51 (47-54) 28 1.82 (1.68-1.93)
2005 IT 3 WHIB 6 67 (63-70) 36 1.86 (1.75-1.94)
2005 GST 1 WHIB 8 19 (16-22) 3 6.33 (5.33-7.33)
2005 GST 2 WHIB 5 30 (25-34) 14 2.14 (1.79-2.43)
2005 GST 3 WHIB 8 12 (8-15) 4 3.00 (2.00-3.75)
2006 Q5-N WHIB 5 94 (93-95) 54.5 1.72 (1.71-1.74)
2006 Q5-S WHIB 5 69 (69-70) 38 1.82 (1.82-1.84)
2006 Q6 WHIB 9 118 (115-121) 41.5 2.84 (2.77-2.92)
2006 Q6-S WHIB 8 124 (121-127) 33.5 3.70 (3.61-3.79)
2006 Q7 WHIB 3 34 (33-35) 25.5 1.33 (1.29-1.37)
2006 Q8 WHIB 5 35 (32-38) 22 1.59 (1.45-1.73)
2006 Q9 WHIB 12 31 (28-34) 13.5 2.30 (2.07-2.52)
2006 Q10 WHIB 12 98 (96-101) 50.5 1.94 (1.90-2.00)
a. Alley North samples are labeled for convenience by location and nearby landmarks.
b. GREG=Great Egrets; WHIB=White Ibises.
c. Number of surveys in database does not include first survey, as all first sightings of nests were
eliminated from the database before analysis (see text).
d. N*=weighted model average superpopulation estimate, calculated in Program MARK. LCI-
UCI are 95% confidence intervals.
e. Raw counts are the numbers of nests seen in sample areas during the survey closest to the
colony peak count date. If two survey dates were equally close to the colony peak count date,
the average of the two raw counts from these surveys is presented here.
f. Difference calculated as N*/Raw count. Proportion CIs calculated using CIs for N*.









Table 2-4. Subsampling of Cypress City 2006 database to determine minimum survey frequency
for superpopulation method. Peak count for same colony during this period is 173 nests.
Database Number of Database size N* (LCI-UCI)a
surveys in (# nests)


database (not
including
first; see text)
17


Inclusive (flights ~twice/week,
irregular spacing)
Every third survey deleted after first
two (irregular spacing)
Flights every 5 days
Every other survey deleted after
first two (flights ~once/week,
irregular spacing)
Weekly flights (after first two
surveys)
Flights every ten days (after first
two surveys)
Bimonthly flights (after first two
surveys)
a. N*=superpopulation estimate.
weighted model averages.


249

229

240
229


253 (250-255)

235 (232-239)


247 (243-
237 (233-


-251)
-241)


232 (228-236)

224 (219-229)

207 (201-213)


Superpopulation estimates and confidence intervals are

















VisibiliyBias ~


SAsynchrony


Renesting )


Figure 2-1. Types of data that should be incorporated into estimates of breeding population size.


Estima~tion of
Breeding
Population Size








































Figure 2-2. Satellite photograph of the modern Everglades, with maj or watershed regions
outlined in yellow. All research presented in this paper was conducted in Water
Conservation Area 3A (WCA-3A) in the central Everglades. Photo from
http://sofia.usgs.gov/; available at
http ://sofia.usgs.gov/publications/circular/1275/images/cover-landcvrdataspot.j pg.


I~S~H~i~Y~F~P~~i~~JPB
~.~~ rdk'k'~
KY~p~iS~:l)


































0 500 1000 1500 Meters



Figure 2-3. Satellite photograph of Alley North colony' s tree island (WCA-3A). Alley North
colony was subsampled in order to develop a superpopulation estimate, as it is too
large to conduct photo analyses in its entirety.





































~~ ra IBP~ ~m ~7 a ~ ~B.











Figure 2-4. Two types of artificial landmarks used in this study. A) Markers made of 10-cm
diameter PVC pipe pounded into the ground and anchored with a steel fence post,
topped with a 1.5-meter wide X. B) Markers made of lxl meter white cotton cloth,
tied to vegetation.










600
328%
310%
S 500-
e e* 264%

o 400-
I 409

E 300-


;s 200-


we1 00 -1 1 5 5



Observer 1 Observer 2 Observer 3 Peak Count

Estimate Source

Figure 2-5. Multi-observer comparison of superpopulation estimates for Vacation Island 2006
colony, derived from photographic analyses by three independent observers.
Percentage values are the percent by which each observer' s superpopulation estimate
is greater than the colony's peak count. Numbers inside the bars are labels of
estimates.










700
m 2005 Peak Count
600 0 2005 Superpopulation Estimate
cnM 2006 Peak Count
I I II I 2006 Superpopulation Estimate
Z 500
482%

400 310%


S300
a I 147%
I I II 250%
~o 295%
E 200

LI

100

Vulture Vacation Island Cypress City

Colony

Figure 2-6. Peak counts and associated superpopulation estimates for three Great Egret colonies
in WCA-3A (2005-2006). The percentage difference between the two estimates for
each colony and year are located above the bars.












50000

38275 1 Superpopn Estimate
40000


c~ 300% 29287
S30000
na, 216%


20000

E 12750 13566
LI.
10000-
21 3%
2538
1193

White Ibis (2005) White Ibis (2006) Great Egret (2006)

Species (Year)

Figure 2-7. Peak counts and extrapolated superpopulation estimates for White Ibises in 2005 and
2006 and Great Egrets in 2006 at Alley North colony (WCA-3A). Each bar is labeled
with the value (in estimated number of nests) it represents; the percentage difference
between the two estimates for each colony and year are located above the bars.












cct 3.5 -
--o

o m 3.0 -

a. 2.5 -





com
-

EO .5 -

05o


r\o~\5`
\laca


\~~-"~

C~`""SC~~


Colony (Species)


Figure 2-8. The proportional difference between superpopulation estimates and peak counts for
2005 and 2006. GREG=Great Egret colony; WHIB=White Ibis colony.


I 2006









CHAPTER 3
REMOTELY MEASURED EFFECTS OF RESEARCHER DISTURBANCE ON GREAT
EGRETS (Ardea alba) AND WHITE IBISES (Eudocimus albus)

Introduction

For colonially breeding birds, the disturbance created by researchers walking very close to

or within the colony may affect reproduction, especially early in the nesting cycle (Bouton et al.

2005, Giese 1996, Ellison and Cleary 1978, Tremblay and Ellison 1979). Indeed, the effects of

disturbance on nest success may be more severe for colonially breeding bird species than for

solitary nesters, due to the number of nests potentially affected by a single disturbance (Carney

and Sydeman 1999, Burger 1981, Kadlec and Drury 1968, Vos et al. 1985). Potential effects of

disturbance may include abandonment (Cairns 1980, Safina and Burger 1983, Carlson and

Mclean 1996), reduced clutch or brood size (Ellison and Cleary 1978, Lock and Ross 1973),

increased rate of predation (Ellison and Cleary 1978, Kury and Gochfeld 1975, Hunt 1972),

reduced rate of new nest initiation (Tremblay and Ellison 1979, Safina and Burger 1983), and

reduced hatching and fledging success due to exposure of eggs and chicks to extreme

temperatures (Hunt 1972, Ellison and Cleary 1978, Cairns 1980). In Great Blue Herons (Vos et

al. 1985, Carlson and Mclean 1996) and various species of Pelecaniformes, Ciconiiformes and

Charadriiformes (Rodgers and Smith 1995), ground disturbances in which people walk through

colonies have been found to have a more severe effect than boating or mechanical disturbances.

However, walk-in surveys are often used to measure nest success (e.g., Cairns 1980, Frederick

and Collopy 1989b, Hunt 1972) or to collect biological samples or other census information

(Frederick et al. 1999, Rodgers et al. 2005, Lock and Ross 1973). If ground surveys themselves

significantly decrease nest success or other reproductive parameters, it may defeat the purpose of

the research (Rodgers and Burger 1981). However, at least in wading birds (Ciconiiformes),

there is some debate as to what level of disturbance is biologically or demographically










significant, and what constitutes "acceptable" disturbance levels may vary widely by species

(Carney and Sydeman 1999, Erwin 1989).

The effects of human disturbance are inherently difficult to measure, since measurement

itself usually requires some level of human intrusion (Shields and Parnell 1986, Frederick and

Collopy 1989b). While this has been addressed through the comparison of colonies with

different rates of human disturbance (e.g., Frederick and Collopy 1989b, Piatt et al. 1990, Davis

and Parsons 1991, Tremblay and Ellison 1979, Burger 1981), it is not clear that location effects

can be entirely controlled in such studies. Comparison has also been made of areas within a

single colony, but this has been limited to rare situations in which birds can be observed from

distant vantage points (Pratt 1970, Hulka and Stirling 2000, Bouton 1999, Galbraith 1987). In

this paper we describe a study of human disturbance effects as inferred by observations made

from aircraft, comparing disturbed and undisturbed areas of the same colonies. As has been

indicated by other studies on ciconiiform species, we predicted that areas with walk-in

disturbance would experience increased nest failure rates and a decreased rate of new nest

initiation.

Methods

Ground Transects

We monitored nesting in two breeding colonies located in Water Conservation Area 3A of

the Everglades (Broward County, FL) in March-May of 2006. Both colonies were located in tree

islands dominated by willow (Salix spp.) set within an emergent marsh of primarily sawgrass

(Cladium jamaicense) and cattail (Typha spp). Vacation Island colony (approximately 1 ha,

located at N25054.939 by W-80037.813) was composed of roughly 90% Great Egrets (Ardea

alba), though a small number of Great Blue Herons (Ardea herodia~s) and Anhingas (Anhinga

anhinga) nested in the taller trees. This colony was distant from other sources of human









disturbance (12.5 km to nearest road and 600 meters to nearest airboat trail) and we believe our

visits to be the only form of human disturbance to which this colony was exposed. The second

colony, Alley North (N25011.179, W-80031.431), was a much larger (45 ha) mixed-species

colony that included Great Egrets, Snowy Egrets (Egretta thula), Little Blue Herons (Egretta

caerulea), Black-crowned Night Herons (Nycticorax nycticorax), Tricolored Herons (Egretta

tricolor), White Ibises (Eudocimus albus), and other species. The northern end of the colony

(approximately 30 ha) was exclusively composed of White Ibis nests, and we studied two small

areas within this single-species area of the colony (hereafter referred to as "Alley North

colony"). The colony was distant from sources of non-researcher human disturbance (3.2 km to

nearest road and 400 meters to nearest airboat trail). One other research team was also working

in the colony during the time period in which we conducted surveys. The maj ority of this

research was based on remotely monitoring radiotelemetry signals from outside the tree island,

but researchers did occasionally enter the colony as well.

We used both aerial photography and repeated walk-in censuses of marked nests to

monitor nesting. At Vacation Island, the "disturbed" area around the walk-in census trail

encompassed the northern 40% of the nesting area. The southern 60% of the tree island (all nests

>15m from the census trail) comprised the "undisturbed" sample area, and was never surveyed

on foot during the breeding season. Walk-in censuses of Alley North were located in the

northeastern portion of the colony, and encompassed less than 1% of the area of nesting. One

100-m2 area was censused using walk-in methods only once, and then aerial methods thereafter;

another 200 m2 area was censused four times on the ground in addition to aerial surveys. The

undisturbed (single-visit) area used in this study was located approximately 200 meters south of









the disturbed site; apart from the initial visit to erect markers to delineate the sample area, we did

not come closer to this control site than 200 meters during the breeding season.

We began walk-in censuses after incubation was already underway for most nests. We

approached the edge of the colony via airboat to within 15 meters (for Great Egrets) or 75 meters

(for White Ibises) of the closest nest. Nest checks were conducted every 5-7 days, and each visit

to a colony took roughly one hour with 2-5 people present on each visit. Each nest was briefly

examined for nest contents, using a mirror pole if necessary, but the contents were not removed.

If abandonment was suspected, the eggs were sometimes touched to determine temperature.

Nests were monitored until abandonment (all eggs in nests were cold) or failure (eggs were

broken or gone from nest, or chicks were dead or gone from nest before fledge date).

Aerial Survey Methodology

We conducted aerial photographic surveys of nests that were individually identifiable from

the air in order to examine nest turnover rates. Surveys were conducted twice a week at irregular

intervals during the breeding season in a Cessna Skyhawk (172) aircraft, and photographs were

taken using a Canon EOS 20D high-resolution digital camera with a 28-135mm image-

stabilizing zoom lens. Camera settings were primarily on auto. We took photographs from an

altitude of approximately 152 meters out the copilot' s door (right side of plane), and the door

was removed for all survey flights so that photographs could be taken as close to vertically above

nests as possible. We conducted survey flights between 08:00 to 10:30 am or 16:00 to 18:00 pm,

but flight times varied with weather, plane availability, and other factors.

To allow for the identification of walking routes from the air, we erected artificial

landmarks around the outer perimeter of each walk-in survey route. For Vacation Island,

markers were placed within 1-3 meters of the edge of the transect path at roughly 15-meter

intervals, and we considered all nests within 15 meters of a marker to be in the "disturbance










zone," while nests outside this area were identified as "undisturbed." We used the length of the

X at the top of the PVC markers (below) in each photograph as a metric for scaling ground

distances in the colony. For the Alley North colony, we placed a marker at the four corners of

both the disturbed and undisturbed study sites. The "disturbed" site had regular walk-in surveys

and the "undisturbed" site had no further ground disturbances after the markers were erected.

The markers were of two types:

* A 3-meter tall, four-inch PVC pipe with a 1.5-meter X made of PVC attached to the top of
the pole (Figure 3-1A).

* A lxl-meter piece of white or pale-colored cotton cloth tied at the corners to vegetation
(Figure 3-1B). These were placed in sheltered areas where they would be less likely to
flap in the wind.

Photographic Analysis

We edited photos in Adobe Photoshop Elements version 2.0. Printed copies of photos

were used to identify individual nests and obtain presence-absence data for each nest on each

subsequent survey date. Nests in the photos were individually numbered and presence-absence

data for nests was entered into the database using the following format:

* Nest inactive (0)- this included cases in which nest and parent/chicks were not visible; nest
was visible but empty (no sign of parents/chicks); or there was no photo of the nest area
available on that date

* Nest active (1)- parent and/or chicks were visible


We compared photos of the date we were currently working on with photos of all previous

dates to ensure that all nests were assigned unique and non-overlapping identities. We used a

conservative definition for nest failure in the photographic censuses, due to visibility constraints

and because some nests were temporarily unoccupied at the moment the picture was taken.

Nests that were apparently inactive for four consecutive survey dates were categorized as failed,

and nests in the same location were thereafter treated as new nest starts. Although we used the










presence of a large white bird as an initial indication of nesting, these could have been roosting

birds or birds temporarily away from their nests. Before we analyzed each presence-absence

database, we therefore eliminated first observations of all nests from the database, and so deleted

from the database all "nests" that were seen only once. Although it is possible that some

proportion of these single-tally birds were actually short-lived nest starts, this was considered to

be a rare occurrence given the amount of time required for breeding and nest-building in these

species (4-10 days for Great Egrets, according to McCrimmon et al. 2001; 9-10 days for White

Ibises, as cited in Kushlan and Bildstein 1992). We eliminated the first sighting of nests that

were seen on multiple dates (in addition to those nests seen only once) in order to avoid

artificially inflating survival estimates.

Analysis of Presence-Absence Databases: A Mark-Recapture Scenario

We used aerial photographs of individually identifiable nests and a variation on a Jolly-

Seber (JS) open-population capture-recapture model called the superpopulation method

(Schwarz and Arnason 1996) in order to estimate rates of nest initiation and abandonment in

large colonies of nesting birds (see also Chapter 2). JS models estimate survival and encounter

probabilities for marked animals, and the superpopulation approach also estimates the net

immigration into a population. The application of the superpopulation method to nests allows

for the monitoring of nests initiated in a colony during a period of interest, such as a breeding

season. In our study, nests were equivalent to individuals in a population; the number of new

nests present at each survey represented immigration into the population, and the number of

"marked," or individually identifiable, nests that failed between consecutive surveys indicated

the level of emigration from the population. Encounter probabilities indicate the chance of

seeing a nest on a particular survey date, given that it has survived and is actually present.









The first survey of the colony was deleted from the database, since due to the first-

observation nest adjustment (described above), it contained only zeroes. The presence-absence

database input into MARK thus started with the second survey date. We fit regression models to

the capture-recapture data from colonies using the POPAN data type (Schwarz et al. 1993,

Schwarz and Arnason 1996) in Program MARK version 4.3 (White and Bumnham 1999). We set

the time intervals to fractions of weeks between each set of consecutive surveys and assigned all

nests to either the disturbed or undisturbed attribute group in the model. The four variables

(survival, cp; encounter probability, p; entry probability, pent; and superpopulation size, N) were

allowed to vary with time, vary by attribute group, or some combination thereof, depending upon

the model. The "null hypothesis" model was a so-called "dot" model, in which survival,

encounter probability, and entry probability were not allowed to vary by group or to change

through time. N, population size, was allowed to vary with group for all models, as the number

of nests seen over the course of the season was different in each location (n=279 for disturbed

area and n=194 for undisturbed area of Vacation Island colony; n=163 for disturbed area and

n=201 for undisturbed area in Alley North colony). For alternative models in which survival

probability cp and encounter probability p varied with time, not all parameters in the model were

estimable (Schwarz and Amnason 1996). We set p1=p2 and pk pk-1 for each group (if group was

included in the model) so that all survival parameters were estimable (J. Nichols pers. comm.).

In the models in which survival was held constant or varied by group and encounter probability

was time-dependent, the initial p value is still inestimable, so for this model we set p1=p2 foT

each group and allowed all other encounter probabilities to vary. We used a sin link function to

estimate survival and encounter parameters, mlogit functions to estimate entry parameters, and a

log link function to estimate population size N of each group.









We used AIC (Akaike's Information Criterion) to distinguish among competing models

when testing the validity of our predictions (Williams et al. 2002). Each model had an associated

AAIC value (a measurement of the difference between it and the best model), and Akaike

weights that expressed the relative probability of each model being the best fit for the data. We

considered AAIC values of less than 2 to indicate relatively well-supported models, while a

AAIC value of greater than 10 indicated that the data did not support the model (Williams et al.

2002). We also used MARK's model averaging capability to Eind the weighted average values

for all parameters across all models.

Additionally, we quantified the model fit of the most general model in the tested suite,

using chi-square tests for each survey interval, to see if observed values varied from the expected

number of surviving or encountered nests. These tests evaluated the following two assumptions

inherent in Jolly-Seber models (Pollock et al. 1990, Cooch and White 2007): that there is no

heterogeneity in capture probability within groups, either among individuals or among cohorts

(cohort in this case meaning all nests that were seen for the first time on the same date); and that

there is no heterogeneity in survival probability within groups, either among individuals or

cohorts. We calculated chi-square values using Program RELEASE (Burnham et al. 1987),

available within Program MARK. The chi-square values generated for each survey interval were

summed for the entire sampling period, and were divided by the degrees of freedom to obtain a

variance inflation factor (c-hat). Following Cooch and White (2007), we interpreted a c-hat

value of 1 to indicate good model fit, values of roughly 1-3 to indicate moderately good fit, and

>3 to indicate probable violation of model assumptions. We adjusted model weights for the

Great Egret colony using the calculated c-hat for the dataset, so the information criterion values

presented for these models are thus quasilikelihood (QAICc) values. This modification adjusts









for the overestimated precision of parameter estimates caused by overdispersion in the data

(Burnham et al. 1987; Burnham and Anderson 1998; Williams et al. 2002). The calculated c-hat

value for the White Ibis colony was <1, which essentially means the data were underdispersed.

As there is little agreement in the scientific literature as to what this means or what to do about it,

we followed the recommendation of Cooch and White (2007) in using a c-hat value of 1 for

model weighting in this case.

Results

Vacation Island Great Egret Colony

After adjusting the model selection used in MARK for the calculated c-hat value (c-

hat=3.21), two models were equally supported by the data (Table 3-1). All other models in the

model-testing suite had less than 1% support (Appendix D for full list). Due to this relatively

high c-hat value, models with fewer parameters were much more highly supported; in the two

best models, both encounter probability and survival probability were held constant over time.

Time had an important effect on probability of entry into the population (new nests being

initiated) in both models. Disturbance category was also retained as an important covariate in

one model, though its effect was additive rather than multiplicative. The undisturbed group had

slightly higher probability of entry into the population across all time periods than did the

disturbed nest group (Figure 3-2). According to the most parsimonious model, this difference

did not vary appreciably with the survey interval, regardless of how the entry probabilities

themselves changed. Although it appears that there may have been an interaction with time on a

linear scale (Figure 3-2), the interaction was not included in the best models, either because the

interaction was examined on a logit scale or because including an interaction term would have

added an additional 19 parameters to the model, and consequently was not supported by the

QAICc model selection criteria.









Alley North White Ibis Colony

Two models were well supported by the data, and several others had some moderate level

of support (Table 3-2; Appendix E for list of all tested models). Time had an important effect on

all three population parameters in all supported models, but the interaction of time with group

effect varied between models. All supported models indicated some effect of disturbance group

on survival and encounter probabilities, and the maj ority of the models also indicated a

disturbance group effect on entry probability (Table 3-2). The two well-supported models both

had group effect as an additive effect on survival probability. According to the best-fit models,

this difference did not vary appreciably with survey interval, regardless of how the survival

probabilities themselves changed. Encounter probability varied with both group and time in an

interactive fashion in the two best models. For the best fit model, entry probability was time-

dependent but did not vary with group; for the next-best model (with a AAIC value of <2,

indicating some support for the model), entry probability varied with time and group in an

additive manner, as for survival above.

Comparison of model-averaged encounter probabilities between disturbance groups for the

Alley North White Ibis colony showed that encounter probability did not differ consistently

according to disturbance regime. In some survey intervals encounter probability was higher for

the disturbed group, and for others it was higher for the undisturbed group. Encounter

probability seemed to vary widely among survey intervals and between geographic areas,

regardless of level of researcher disturbance. Comparison of model-averaged survival estimates

indicated that survival probability was higher in the disturbed area than the undisturbed area

(Figure 3-3, Table 3-3). Entry probability in the disturbed area was lower than or equal to that in

the undisturbed area (Figure 3-4, Table 3-3). Although the absolute differences in probabilities









between the two groups were considerably larger for survival estimates than for entry probability

estimates, entry probabilities showed a proportionally larger effect of disturbance (Table 3-3).

Discussion

Nest success did not appear to be negatively impacted by researcher disturbance during

incubation and nestling periods for either Great Egrets or White Ibises. However, model-

averaged parameter estimates indicated that nest initiation was lower in the disturbed areas for

both species when compared to control sites. This suggests that breeding birds may be sensitive

to disturbance in the colony early in the nesting cycle.

Violations of Model Assumptions.

The c-hat value for the Vacation Island (Great Egret) model suite was larger than 3,

indicating potential problems with model Sit. There are biologically sound reasons why model

assumptions regarding encounter and survival homogeneity may be false for nesting wading

birds, and the lack of model fit we saw for this dataset may be partially due to one of these

factors. Fortunately, however, both survival and encounter probability estimates in Jolly-Seber

models tend to be robust to this type of heterogeneity, so long as average encounter probability is

high (e.g., >0.5; Pollock et al. 1990); for Vacation Island, model-averaged encounter probability

was 0.945, and for Alley North, it was 0.823.

Effects of Disturbance on Nest Survival

According to the two best-supported models for the Great Egret colony, researcher

disturbance had no significant effect on nest survival or encounter probability. Researcher

disturbance after the early egg-laying stage appears to cause no appreciable decrease in nest

success for Great Egrets. This agrees with similar Eindings of lack of nest failure due to human

disturbance later in the nesting cycle in Tricolored Herons, Egretta tricolor (Frederick and

Collopy 1989b), Black-crowned Night Herons, Nycticorax nycticorax (Parsons and Burger










1982), and Snowy Egrets, Egretta thula (Davis and Parsons 1991). Although walk-in surveys do

affect nest success for many taxa, perhaps resistance to disturbance in later nest stages is a

common characteristic for wading birds.

Likewise, our single or multiple site visits to the White Ibis colony during incubation and

early nestling stages did not appreciably decrease nest success. Moreover, nest survival was

significantly higher in the disturbed area of the White Ibis colony than in the undisturbed area.

This result is counter to generally accepted theory regarding the effects of walk-in ground

disturbance on colonially nesting birds, but may be attributable to one of two several possible

causes. First, the nests in the two sites may have belonged to different cohorts; based on the

single walk-in survey of the undisturbed area, the average nesting stage appeared to be several

days to a week behind the disturbed area. Nest initiation date has been found to be very

important for nest success in White Ibises, presumably due to environmental variables associated

with initiation date (Frederick and Collopy 1989a). A second potential cause of higher nest

failure in the undisturbed area is that it was not as "undisturbed" as we thought; there was

another team of researchers studying White Ibises at the same colony, and their frequency of

visitation to different areas may have affected nest survival rates. Thirdly, it is possible that

researcher presence may have discouraged the presence of predators in that area of the colony,

although we consider this possibility to be unlikely, as researcher disturbance has been

postulated to actually facilitate predation at a White Ibis colony in North Carolina (Shields and

Parnell 1986).

Effects of Disturbance on Nest Initiation

It makes biological sense for entry probability to be time-dependent in all supported

models, as for both species there tends to be a surge of initial nesting in a given area followed by

lower levels of nest initiation thereafter (McCrimmon et al. 2001, Kushlan and Bildstein 1992).









However, for both Great Egrets and White Ibises, one of the two best models also included

disturbance group as a factor in entry probability. For Great Egrets, the two models were so

equally weighted that for all practical purposes they were indistinguishable using QAICc. This

lends some support to the hypothesis that regular researcher disturbance causes depressed levels

of nest initiation as compared to areas where nests remained undisturbed. Nest initiation rates

are not often measured by researchers studying effects of disturbance (e.g., Carlson and McLean

1996, Burger et al. 1995), but decreased initiation has also been seen in Black Skimmers,

Rynchops niger (Safina and Burger 1983) and Black-crowned Night Herons, Nycticorax

nycticorax (Tremblay and Ellison 1979). Depressed nest initiation may not be universal;

Frederick and Collopy (1989b) presented evidence that Tricolored Heron nest initiation may not

be affected by researcher disturbance, although this question was not a primary focus of the

research. However, the effect of disturbance on nest initiation likely varies by species and

frequency of visitation (Goitmark 1992), and the evidence for both Great Egrets and White Ibises

indicates that further research is warranted on this topic.

We have found some effect of researcher disturbance on nest initiation in wading birds, but

no discernible effect on nest success. This suggests that wading birds are vulnerable to

disturbance early in the nesting period when they are considering nesting locations, but that they

may be less sensitive once nesting has begun. It is important to note, however, that disturbance

may affect choice of nesting area in the future, a possibility that has not been examined in

wading bird species. Even when birds do not overtly react to disturbance during the current

breeding season, they may avoid areas in ensuing seasons where they have been disturbed in the

past, as has been found in several raptor species (Platt 1977, as cited in Goitmark 1992; White

and Thurow 1985) and for Adelie penguins (Pygoscelis adeliae) in Antarctica (Wilson et al.









1990, Young 1990; but see Fraser and Patterson 1997). Renesting rate may also be reduced in

disturbed areas (Goitmark 1992), but renesting rates in wading birds are unknown.

Perhaps more important than its evidence for disturbance effects, this study serves to

reinforce earlier work in using remote means to infer effects of disturbance. We hope this new

and widely applicable method of remotely measuring disturbance will prove useful for future

research on this topic.









Table 3-1. Characteristics of the two best fit, lowest AIC models describing population
parameters for Great Egret nests in the central Everglades, Florida. Model parameters
are probability of survival (phi), encounter probability (p), probability of entry into
the population (pent), and size of population (N). Each parameter was held constant
over the entire survey period (.), held constant for each group separately (group),
allowed to vary with time (t), allowed to vary with both time and group (t~g) or
allowed to vary with both time and disturbance group in an additive fashion
(t+group). Both models are about equally supported. Model 1 does not include
disturbance group as an important factor in the probability of new nests entering the
population over time; the second model does include the group parameter. Only
models with greater than 1% support are included.
Model Delta QAICc QAICc weights # parameters
{p(.)phi(.)pent(t)N(group)} 0 0.50147 22
{p(.)phi(.)pent(t+group)N(group)} 0.0119 0.4985 23




Table 3-2. Characteristics of the best fit, lowest AIC models describing population parameters
for White Ibis nests in the central Everglades, Florida. Model parameters are
probability of survival (phi), encounter probability (p), probability of entry into the
population (pent), and size of population (N). Each parameter was held constant over
the entire survey period (.), held constant for each group separately (group), allowed
to vary with time (t), allowed to vary with both time and group (t~g) or allowed to
vary with both time and disturbance group in an additive fashion (t+group). Both
highly supported models (AAICc<2) include disturbance group as an additive effect
on nest survival and a multiplicative effect on encounter probability. There is some
support for the importance of disturbance group on entry probability, but the most
highly supported model (55% support) does not include disturbance group as an
important factor in entry probability. Only models with greater than 1% support are
included.
Model AAICc AICc weights # parameters
{ phi(t+group)p(t~group)pent(t)N(group) } 0.000 0.545 19
(phi(t+group)p(t~group)pent(t+group)N(grou) 1.697 0.233 20
( phi (t+group)p (t+group)p ent(t)N(group) } 3.939 0.076 16
(phi(t~group)p(t~group)pent(t)N(group)} 4.111 0.070 23
{ phi /t*group~pt*group)pent(t+group)N(group) 5.783 0.030 24
{ phi(t+group)p(t~group)pent(t~group)N(grop 6.108 0.026 21
Sphi /t*group~p/t*group~pent(t*g roup)NNgroup) 7.420 0.013 26









Table 3-3. Comparison of model-averaged survival and entry probabilities between disturbance
groups for the Alley North White Ibis colony. Phi=survival probability; pent-entry
probability; interval number indicates survey interval (e.g., interval 1 denotes the time
period between surveys 1 and 2). The percent different in daily probabilities between
groups indicates the difference in the percentage of nests that survive or are initiated
in the disturbed group in comparison to the undisturbed group; a negative difference
indicates that the parameter estimate for the disturbed group was lower than that for
the undisturbed group. Daily probabilities are calculated for phi as ((disturbed
weekly phi)m//(undisturbed weekly phi)"n)-1 x 100. For pent, this value was
calculated as ((disturbed weekly pent) /(undisturbed weekly pent) )-1 x 100.
Parameter % Difference Between Disturbed and
Undisturbed Groups (Daily Probabilities)
phi (interval 1) 2.72
phi (interval 2) 1.57
phi (interval 3) 2.73
phi (interval 4) 3.96
phi (interval 5) 11.00
pent (interval 1) -29.75
pent (interval 2) -48.99
pent (interval 3) -48.01
pent (interval 4) 0.00
pent (interval 5) 0.00





































~~ ra IBP~ ~m ~7 a ~ ~B.











Figure 3-1. Two types of artificial landmarks used in this study. A) Markers made of 10-cm
diameter PVC pipe pounded into the ground and anchored with a steel fence post,
topped with a 1.5-meter wide X. B) Markers made of lxl meter white cotton cloth,
tied to vegetation.













b 70 5


o E 6e-5-


S 5e-5-


-c 4e-5-




2 e-5-





0 0 1 2 1 4 1 6 1 8 1
Surve

Figure3-2.Tedfeec netypoaiiisbtwe itrac rusa aainIln
Gra ge oon safntono ie simtsaefo scn upre
moel ph()()ettgopNgru) -xsi h rbaiiyo e et en






Figre3-2 TD ifferenc i nr ob iltes between itrac groups aresih u eanpstv vralsre Vaaintervals.














SDisturbed Group
0-Q O Undisturbed Group
a, 0.8 -



S0.6 -











0.0
0 1 2 3 4 5

Survey Interval


Figure 3-3. Model-averaged weekly survival probabilities for both disturbance groups at Alley
North White Ibis colony as a function of survey interval (time). The disturbed group
had higher nest survival across all survey intervals than the undisturbed group.
















*Disturbed Group
V Undisturbed Group


I _


0.07


0.06


0.05


0.04


0.03


0.02


0.01


0.00


Survey Interval


Figure 3-4. Model-averaged weekly entry probabilities for both disturbance groups at Alley
North White Ibis colony as a function of survey interval (time). The undisturbed
group had higher probability of nest initiation for the first three survey intervals.
Both groups had near zero entry probability for the last two survey intervals.









APPENDIX A
SUPERPOPULATION MODEL OUTPUT (POPAN DATA TYPE) FOR ALL COLONIES

Table A-1. For each colony or sample, we tested a set of four candidate models: a general (fully
time-dependent) model, a so-called "dot model," in which survival and entry
variables were held constant through time, and two models in which either encounter
probability or survival probability was allowed to vary with time while the other was
held constant (Cooch and White 2007). Probability of entry into the population (pent)
would not be expected to hold constant throughout the season, since for both species
there tends to be a surge of initial nesting in an area followed by lower levels of nest
initiation thereafter. Thus, entry probabilities were allowed to be time-dependent in
all models. For models in which survival probability cp (phi) and encounter
probability p varied with time, not all parameters in the model were estimable. We
set p1=p2 and pk pk-1 so that all survival parameters were estimable in the model. In
the model in which survival was held constant and encounter probability varied, the
initial p value was still inestimable, so for this model we set p1=p2 and allowed all
other encounter probabilities to vary. We used a sin link function to estimate survival
and encounter parameters, a mlogit(1) function to estimate entry parameters, and a
log link function to estimate superpopulation size N. A "t" notation indicates that the
parameter is general (not restricted to be equal over time); a dot notation means that
the parameter is restricted to be the same over all survey intervals. Models that were
completely unsupported by the data (AICc weight of zero) are not included in the
table. C-hat is a measure of the goodness of fit of the most general model; when c-hat
was greater than 1 (indicating a lack of model fit), it was used to adjust the model-
based variance and covariance estimates, and thus affected the relative model weights
within the tested suite. C-hat values of less than 1 are not presented in the table.
Year Colony/ Species c-hat
sample
2005 Vulture GREG Model A AICc AICc Estimated SE
weights N*
{phi;t/p\t/pentn~t)N} 0 0.82215 581.6979 34.8814
{phiat)p(.)pent(t)N} 3.062 0.17785 589.5230 17.2920
2005 Vacation GREG 3.05 Model A QAICc QAICc Estimated SE
Island weights N*
{phi(.)p(.)npentt)N} 0 0.87187 233.5 631 6.6651
{phi(t)p(.)pent(t)N} 4.3784 0.09765 228. 1620 5.9515
{phi(.)p(t)pent(t)N} 7.5307 0.02019 229.5506 4.9093
{phi\LtrpLtr~pent\t)N} 8.8803 0.01028 227.1754 5.0314
2006 Vacation GREG 5.90 Model A QAICc QAICc Estimated SE
Island weights N*
{phi(.)p(.)pent(t)N} 0 1 479.9585 3.2767









Table A-1. Continued.
Year Colony/ Species
sample
2005 Cypress GREG
City


c-hat


2.13 Model


A QAICc QAICc
weights
0 0.54436
A QAICc QAICc
weights
0 0.9909
9.3855 0.00908
21.5337 0.00002
A AICc AICc
weights
0 0.83334
4.4872 0.0884
4.7363 0.07805
16.5396 0.00021
A AICc AICc
weights
0 0.96697
7.6176 0.02144
8.8484 0.01159
A AICc AICc
weights


Estimated
N*
267.7905
Estimated
N*
253.6486
253.7785
252.8228
Estimated
N*
50.6863
50.6636
50.4216
52.3654
Estimated
N*
66.6686
66.6758
66.6722
Estimated
N*

19.0817
19.5700
19.3125
Estimated
N*

29.8978
29.3304
29.2213
28.2547
Estimated
N*

11.5185


SE

6.5382
SE

2.1806
2.1385
2.1116
SE

1.9522
1.8088
1.5666
2.1018
SE

1.9037
1.9551
1.9652
SE


1.6889
2.1732
1.9709
SE


2.6669
2.2952
1.8967
1.8534
SE


1.9025


{phi(t)p(t)pent(t)N}
GREG 3.03 Model


2006 Cypress
City




2005 Alley
North/IT 1






2005 Alley
North/IT 3




2005 Alley
North/
GST 1


{p(.)phi(.)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
Model


{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(.)p(.)pent(t)N}
Model


{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(t)p(t)pent(t)N}
Model


{phi(.)p(.)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}
Model


{phi(t)p(.)pent(t)N}
{phi(.)p(.)pent(t)N}
{phi(t)p(t)pent(t)N}
{phi(.)p(t)pent(t)N}
Model


{phi(.)p(.)pent(t)N}


WHIB







WHIB





WHIB


0
15.8962
18.1392
A AICc


0
0.0292
2.0753
5.0119
A AICc


0.99953
0.00035
0.00012
AICc
weights

0.41299
0.407
0.14632
0.0337
AICc
weights


2005 Alley
North/
GST 2





2005 Alley
North/
GST 3


WHIB








WHIB









Table A-1. Continued.
Year Colony/ Species
sample
2006 Alley GREG
North/Q 1


c-hat


Model

{phi(.)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
Model


Estimated
N*
55.9440
56.1124
56.0022
Estimated
N*
59.0446
58.6230
58.7090
59.4892
Estimated
N*
58.1478
58.0003
58.1257
Estimated
N*

43.4679
43.4806
Estimated
N*

73.4119
73.5128
Estimated
N*

54.2707
54.5926
54.2082
54.8939
Estimated
N*

94.3730
94.4772
94.2479


SE

1.1204
1.2738
1.0206
SE

0.3013
0.2728
0.9285
1.1264
SE

1.4997
0.6814
1.5748
SE


0.1566
0.1608
SE


0.1206
0.1368
SE


1.7800
2.0213
1.7665
2.1892
SE


0.2798
1.2067
0.1886


A QAICc QAICc
weights
0 0.99628
11.1937 0.0037
21.4261 0.00002
A AICc AICc
weights
0 0.64806
2.1356 0.22278
3.256 0.12723
11.6317 0.00193
A QAICc QAICc
weights
0 0.9937
10.2855 0.0058
15.2309 0.00049
A QAICc QAICc
weights


2006 Alley
North/Q2





2006 Alley
North/Q3




2006 Alley
North/
GE SE 1



2006 Alley
North/
GE SE 2


GREG


{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(.)p(.)pent(t)N}
GREG 1.83 Model

{phi(.)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
GREG 1.18 Model


{phi(.)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}
GREG 1.53 Model


0
10.678
A AICc


0
19.8337
A QAICc


0
7.8966
7.9083
15.0858
A QAICc


0
0.9321
6.6281


0.99522
0.00478
AICc
weights

0.99995
0.00005
QAICc
weights

0.96247
0.01856
0.01846
0.00051
QAICc
weights

0.60102
0.37712
0.02186


{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
Model


2006 Alley
North/
GE SE 3





2006 Alley
North/
Q5-N


GREG


{phi(.)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(t)p(t)pent(t)N}
WHIB 1.46 Model


{phi(t)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(.)p(t)pent(t)N}









Table A-1. Continued.
Year Colony/ sample Species c-hat


2006 Alley North/
Q5-S


Model

{phi(.)p(t)pent(t)N}
{phi(t)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
Model

{phi(t)p(.)pent(t)N}
{phi(t)p(t)pent(t)N}
{phi(.)p(t)pent(t)N}
Model

{phi(t)p(t)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
Model

{p(.)phi(.)pent(t)}
{ p(.)phi (t)p ent(t)}
{phi(t)p(t)pent(t)N}
{phi(.)p(t)pent(t)N}
Model

{phi(t)p(t)pent(t)N}
{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(.)p(.)pent(t)N}
Model

{phi(.)p(t)pent(t)N}
{phi(t)p(.)pent(t)N}
{phi(t)p(t)pent(t)N}
Model

{phi(t)p(t)pent(t)N}


Estimated SE
N*


WHIB


A AICc AICc
weights
0 0.58865
0.8875 0.37769
5.7235 0.03365
A QAICc QAICc
weights
0 0.95935
6.9465 0.02976
8.9562 0.01089
A QAICc QAICc
weights
0 0.91789
5.5923 0.05603
7.1216 0.02608
A AICc AICc
weights
0 0.44667
0.2508 0.39403
2.694 0.11614
4.674 0.04316
A AICc AICc
weights
0 0.75027
3.0504 0.16324
4.3348 0.08589
14.2487 0.0006
A AICc AICc
weights
0 0.99985
17.7047 0.00014
24.3684 0.00001
A AICc AICc
weights
0 1


69.2716
69.2778
69.1884
Estimated
N*
117.7637
117.8758
136.0672
Estimated
N*
123.7821
123.6286
123.3685
Estimated
N*
34.1004
34.3604
34.3604
34.1004
Estimated
N*
34.9722
35.9532
35.1331
36.2411
Estimated
N*
30.9664
31.1015
31.0495
Estimated
N*


0.1492
0.1658
0.1357
SE

1.9492
1.9548
47.3142
SE

2.0607
2.2899
1.9028
SE

0.5192
0.7027
0.7027
0.5192
SE

1.1473
2.7331
1.9966
2.0097
SE

1.5501
1.2478
1.1575
SE


2006 Alley North/Q6 WHIB 2.05





2006 Alley North/Q6-S WHIB 2. 12





2006 Alley North/Q7 WHIB







2006 Alley North/Q 8 WHIB 1.00







2006 Alley North/Q9 WHIB





2006 Alley North/Q10 WHIB 1.15


98.3859 1.2228









APPENDIX B
CALCULATION OF STANDARD ERRORS FOR EXTRAPOLATED SUPERPOPULATION
ESTIMATES FROM ALLEY NORTH SAMPLES

We conducted aerial surveys and photo analysis of sample areas from the large Alley

North wading bird colony (WCA-3A) in 2005 (White Ibises) and 2006 (White Ibises and Great

Egrets). Samples were surveyed for different amounts of time, based upon where nesting was

occurring and the availability of good-quality photographs for each area. We calculated a gross

superpopulation estimate (N*) for each sample for however long it was surveyed, using MARK's

POPAN data function. We took the number of nests present in each sample on the survey date

closest to the colony peak count date as the raw count (the equivalent of the peak count for the

individual sample). The superpopulation estimates with standard errors were then compared to

the raw counts for all samples in order to obtain a ratio expression of the difference between the

two estimates. We divided the superpopulation estimate by the raw count for each sample, to

obtain each sample ratio, then averaged these ratios across all samples. We averaged individual

ratios rather than finding a summed ratio across all samples, as it is unlikely that there is one

underlying level of nest turnover within the colony; different nesting cohorts within the colony

are likely to have different levels of nest turnover at different times within the breeding season.

The average ratio of this proportional error across all samples was our estimated colony-

wide proportional difference between the peak count and extrapolated superpopulation count, as

follows:

(E(N~i/RCi))/n = (N~colony/PCcolony)

N~i is the superpopulation estimate for sample i; RCi is the raw count for sample i; n is the

number of samples in the colony for that year and species. Cumulatively, (E(N~i/RCi))/n is the

average ratio of the total number of nest starts to the number of nests seen in the raw count.









PCcolony is the peak count for the colony. We solved for "N~colony," the entire colony's

extrapolated superpopulation estimate.

We also calculated the variance of the ratio estimate, in order to calculate 95% confidence

intervals for the colony-wide superpopulation estimate. Both the sample size and the raw counts

of the samples are fixed values, and thus the only variance that remains to be calculated in the

ratio estimate is the superpopulation variance for each estimate. The variance of the average

ratio, var(CI), is equal to 1 divided by the square of the sample size n, multiplied by the sum

(across all samples) of 1 divided by the square of the sample raw count, multiplied by the

variance of the sample superpopulation estimate. This variance of the superpopulation estimate

for a sample, var(N*), is equal to the square of the standard error of the estimate (which is

calculated by Program MARK). Therefore it is a simple task to calculate var(CI), the variance of

the ratio of superpopulation estimates to raw counts for all samples:

(E(N~i/RCi))/n = CI

var(CI)=var(E(N" i/RCi)/n)=(1/n2) Var(E(N" i/RCi)) =(1/n2)-E((1 /RCi2)-var(N" i))

var(N~i)= SEi2

where RC-peak count, N*=superpopulation estimate, n=sample size, CI=ratio of raw

counts to superpopulation estimates for all samples, and SE-standard error of the

superpopulation estimate. This calculation ignores any covariance between the raw count and

superpopulation estimate. However, judging by the comparison of peak counts to

superpopulation estimates for the small Great Egret colonies examined in this study (Figure 2-6),

we would suggest that this covariance is probably fairly small, and may be safely excluded from

our calculations.









Thus the average ratio of raw counts to superpopulation estimates for all samples (CI), with

its associated variance, may be used with the peak count for the entire colony in order to estimate

the superpopulation estimate for the entire colony. We assumed that the average ratio between

the superpopulation estimates of samples and the raw counts of samples is equivalent to the ratio

between the superpopulation estimate for the entire colony and the peak count for the colony.

The 95% confidence intervals for CI were calculated as Ci2-\(var(CI)). As the peak count

for the colony is considered to be a fixed variable (it does not have an associated variance

estimate), these 95% CIs may be used to directly estimate the 95% CIs for the colony

superpopulation estimate.

Example: The 2005 White Ibis samples. The sample raw counts, superpopulation

estimates, and standard errors for the superpopulation estimates are shown below.

Sample Raw count N* SE of N*
IT 1 28 51 1.91
IT 3 36 67 1.91
GST 1 3 19 1.69
GST 2 14 30 2.38
GST 3 4 12 1.90


Substituting these values into the equations listed above, we find an average proportional

difference between superpopulation estimate and raw count of 3.00198, and a variance for this

value of 0.02318:

CI = (E(N~i/RCi))/n = ((5 1/28)+(67/3 6)+(19/3 )+(3 0/14)+(12/4))/5 = 3.00198

var(CI)= (1/n2),(( /i2)-var(N" i))= (1/n2)-E((1 /RCi2)- SEi2)

= (1/52)-(((1/(282))- 1.912)+((1/(362~))-1.912)+((1/(32~))-1.692)+((1/(142~))2.3 82)

((1/(42))-1.902)) = 0.02318

Confidence intervals for the average sample ratio CI may then be calculated as follows:









95% CIs for CI = C1i2-\(var(CI)) 3.00198129(10.02318) = 2.69743, 3.30643

The peak count for this colony in 2005 was 12,750 nests. As this colony peak count is a

fixed value, we simply multiply this peak count by CI and its associated CIs to find the

extrapolated superpopulation estimate (38,275 nests) and confidence intervals for the entire

colony (34,392; 42,157):

Colony N* = 12750-3.00198 = 38,275

Colony LCI= 12750-2.69743 = 34,392

Colony UCI = 12750-3.3 0643 = 42,157









APPENDIX C
SAMPLE BUDGET FOR DEVELOPING PEAK COUNT AND SUPERPOPULATION
ESTIMATES

The superpopulation method for estimating asynchrony bias is much more labor- and

flight-intensive than many current monitoring techniques, as frequent surveys are required to

follow individual nests through time from the air. For the purposes of this budgetary analysis we

assumed that weekly surveys provided sufficient accuracy in superpopulation estimation (Table

2-4), but the necessary frequency of surveys will depend upon the synchrony in nesting of the

species of interest and upon the desired precision of the result. For this exercise, we calculated

labor and flight costs for a 3-month breeding season, surveying a Great Egret colony with a peak

count of 458 nests and a superpopulation estimate of 666 nests. The hypothetical colony was 45

km from the airport, and in this exercise it was the only colony being surveyed. Estimated labor

time, flight time, and associated costs are listed in Table C-1. According to these sample

calculations, superpopulation estimation can be roughly four times as costly as peak count

estimation in terms of flying and 20 times as costly in terms of labor for photographic analysis;

overall, the superpopulation estimate is roughly five times more costly to produce. Costs of

transportation to and from the airport, overhead costs, etc. are not included in this analysis.

Clearly these relative values will vary depending upon length of breeding season, number

of flights conducted per month for the superpopulation estimation, labor costs per hour, cost of

plane rental per hour, and other factors. However, we hope that this example provides some

guidance for managers trying to determine the proper tradeoff between monetary cost and

estimation accuracy for monitoring breeding colonies. As a concluding comment, we would like

to note that a cheaper survey method that gives unreliable (or uninterpretable) estimates is no

bargain; cash-strapped managers may want to consider surveying intensively every 2-3 years









(and obtaining useful estimates), rather than surveying annually using a peak count or related

method and failing to obtain useful information.









Table C-1. Sample budgets for superpopulation estimation and peak count estimation of
breeding population size for a hypothetical Great Egret colony.

Superpopulation Peak count
estimate estimate
Hours of editing and
printing photos: 10 0
Hours of photo analysis: 50 3
Cost of labor/hr: $10 $10
Total analysis costs: $600 $30

km to colony: 45 45
km/min flight time: 1.75 1.75
Flight time per survey
(min): 51.48 51.48
Flights per season: 12 3
Season length (weeks): 12 12
Hours flight time for
season: 10.30 2.57
Cost per hour flying: $160 $160
Cost of flights for season: $1,647 $412
Cost of labor flying time: $103 $26
Total flying costs: $1,750 $43 8

Total cost: $2,350 $468









APPENDIX D
MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON GREAT
EGRETS

Table D-1. All models for Vacation Island Great Egret colony' s analysis from Chapter 3. A "t"
notation indicates that the parameter is general (not restricted to be equal over time);
"group" means that the parameters is allowed to vary by group but not by time;
"group~t" indicates that the variable is allowed to vary with time and to change over
time in a different way in each group; "group+t" means that the parameter is allowed
to vary with time, but that the variations change in a similar way between groups over
each survey interval (essentially, that the difference is additive rather than
multiplicative), and a dot notation means that the parameter is restricted to be the
same over all survey intervals. N, population size, is allowed to vary by group in all
models, as the number of nests in each group is different. The entire model suite for
these parameters included 125 potential models. Most of these models were not run
and are not included in the table, either because it became clear during modeling that
they were not going to be a good fit for the data or because the models failed to
converge.


QAICc
weights
0.50147
0.4985
0.00001
0.00001
0.00001
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0


Model
{p(.)phi(.)pent(t)N(group)}
{p(.)phi(.)pent(t+group)N(group)}
{p(t~phi(.)pent(t)N(group)}
{p(t)phi(group)pent(group+t)N(group)}
{p(t~phi(group~pent(t)N(group)}
{ p(.)phi(group)pent(group~t)N(group) }
{p(t)phi(group+t)pent(group+t)N(group)}
{p(t~phi(t~pent(t)N(group)}
{p(group)phi(group)pent(group~t)N(group)}
{p(t)phi(group)pent(group~t)N(group)}
{p(group*t~phi(.)pent(.)N(group)}
{p(t)phi(.)pent(group~t)N(group)}
{p(t)phi(group~t)pent(group)N(group)}
{p(group~t)phi(.)pent(group~t)N(group)}
{p(.)phi(group~pent(.)N(group)}
{p(.)phi(group)pent(group)N(group)}
{p(group~phi(.)pent(.)N(group)}
{p(.)phi(group*t~pent(.)N(group)}
{p(t)phi(group)pent(group)N(group)}
{p(.)phi(.)pent(group)N(group)}
{p(group~t)phi(.)pent(group)N(group)}
{p(.)phi(.)pent(.)N(group)}


AQAICc
0
0.0119
21.4128
22.4091
22.4092
3 1.94
49.3695
53.0026
53.5716
54.2869
54.29
55.3114
74.404
88.5037
14636.2824
14640.2926
14641.7839
14688.8833
14801.6243
14820.9384
14837.2927
26347.0759


# parameters
22
22
40
42
42
42
59
58
42
60
44
59
60
79
5
7
6
41
25
4
44
4





























Model
( phi (t+g)p(t" g)p ent(t)N(g)}
(phi (t+g)p(t" g)p ent(t+g)N(g)}
( phi (t+g)p(t+g)p ent(t)N(g)}
( phi (t "g)p(t "g)p ent(t)N(g)}
(phi(t*g~p(t*g~pent(t+g)N(g)}
( phi (t+g)p(t" g)p ent(t" g)N(g)}
(phi (t "g)p(t "g)p ent(t" g)N(g)}
(phi (t+g)p(t+g)p ent(t "g)N(g) }
( phi (t "g)p(t)p ent(t)N(g)}
{phi (t+g)p(t+g)p ent(t+g)N(g) }
( phi (t)p(t" g)p ent(t)N(g)}
(phi (g)p(t "g)p ent(t)N(g)}
(phi (t "g)p(t)p ent(t" g)N(g)}
( phi (g)p(t "g)p ent(t" g)N(g)}
(phi (t+g)p(t" g)p ent(. )N(g)}
{ phi (t)p(t)p ent(t)N(g) }
(phi (g)p(t "g)p ent(g)N(g)}
{phi(.)p(t~pent(t)N(g ) }
{ phi(t~p(.)pent(t)N(g) }
(phi (t)p(t)p ent(. )N(g) }
{ phi (t)p(t)p ent(. )N(g) }
{phi(.)p(.)pent(t)N(g>)
(phi(.)p(.)pent(.)N(g>)}


# parameters
19
20
16
23
24
21
26
18
19
17
18
17
22
19
17
14
14
1 1
10
14
11
6
4


AAICc
0
1.69680
3.93850
4.11060
5.78330
6.10840
7.41990
9.4766
12.8334
13.0595
14.3942
16.5615
16.6314
17.5562
21.8437
32.8912
43.9314
45.5 52
46.3739
52.6684
109.867
325.708
361.315


APPENDIX E
MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON WHITE
IBISES

Table E-1. All models for Alley North White Ibis colony' s analysis from Chapter 3. A "t"
notation indicates that the parameter is general (not restricted to be equal over time);
"group" means that the parameters is allowed to vary by group but not by time;
"group~t" indicates that the variable is allowed to vary with time and to change over
time in a different way in each group; "group+t" means that the parameter is allowed
to vary with time, but that the variations change in a similar way between groups over
each survey interval (essentially, that the difference is additive rather than
multiplicative); and a dot notation means that the parameter is restricted to be the
same over all survey intervals. N, population size, is allowed to vary by group in all
models, as the number of nests in each group is different. The entire model suite for
these parameters included 125 potential models. Most of these models were not run
and are not included in the table, because it became clear during modeling that they
were not going to be a good fit for the data.


AICc
weights
0.54463
0.23316
0.07601
0.06974
0.03022
0.02569
0.01333
0.00477
0.00089
0.00079
0.00041
0.00014
0.00013
0.00008
0.00001
0
0
0
0
0
0
0
0










LIST OF REFERENCES


Arnason, A. N., and C. J. Schwarz. 1995. POPAN-4: enhancements to a system for the analysis
of mark-recapture data from open populations. Journal of Applied Statistics 22:785-
800.

Barbraud, C., and G. Gelinaud. 2005. Estimating the sizes of large gull colonies taking into
account nest detection probability. Waterbirds 28:53-60.

Barbraud, C., Y. Kayser, D. Cohez, M. Gauthier-Clerc, and H. Hafner. 2004. Detection
probability of nests of Squacco Herons in southern France. Journal of Field Ornithology
75:172-175.

Bart, J., and S. Earnst. 2002. Double sampling to estimate density and population trends in birds.
Auk 119:36-45.

Bayliss, P., and K. M. Yeomans. 1990. Use of low-level aerial photography to correct bias in
aerial survey estimates of Magpie Goose and Whistling Duck density in the Northern
Territory. Australian Wildlife Research 17:1-10.

Bibby, C. J. 2000. Bird census techniques. Academic Press, London, UK.

Bock, C. E., and Z. F. Jones. 2004. Avian habitat evaluation: should counting birds count?
Frontiers in Ecology 2:403-410.

Bouton, S. N. 1999. Ecotourism in wading bird colonies in the Brazilian Pantanal: biological and
socioeconomic implications. Unpublished M. S. thesis, University of Florida, Florida,
USA.

Bouton, S. N., P. C. Frederick, C. Dosualdo Rocha, A. T. Barbosa Dos Santos, and T. C. Bouton.
2005. Effects of tourist disturbance on Wood Stork nesting success and breeding
behavior in the Brazilian Pantanal. Waterbirds 28:487-497.

Burger, J. 1981. Effects of human disturbance on colonial species, particularly gulls. Colonial
Waterbirds 4:28-36.

Burnham, K. P. and D. R. Anderson. 1998. Model selection and inference: a practical
information-theoretic approach. Springer-Verlag, New York, New York, USA.

Burnham, K. P., D. R. Anderson, G. C. White, C. Brownie, and K. P. Pollock. 1987. Design and
analysis of methods for fish survival experiments based on release-recapture. American
Fisheries Society Monograph 5:1-437.

Cairns, D. 1980. Nesting density, habitat structure, and human disturbance as factors in Black
Guillemot reproduction. Wilson Bulletin 92:352-361.










Carlson, B. A., and E. B. Mclean. 1996. Buffer zones and disturbance types as predictors of
fledging success in Great Blue Herons (Ardea herodia~s). Colonial Waterbirds 19:124-
127.

Camney, K. M., and W. J. Sydeman. 1999. A review of human disturbance effects on nesting
colonial waterbirds. Waterbirds 22:68-79.

Cherenkov, S.E. 1998. Accuracy of one-visit censuses of forest passerine birds during a breeding
season. Zoologichesky Zhurnal 77:474-485.

Cooch, E., and G. C. White. 2007. Program MARK: A gentle introduction. 5th edition. Available
online at: http://www.phidot. org/software/mark/docs/book/.

Davis, W. E. Jr., and K. C. Parsons 1991. Effects of investigator disturbance on the survival of
Snowy Egret nestlings. Journal of Field Ornithology 62:432-43 5.

Dodd, M. G., and T. Murphy. 1995. Accuracy and precision of techniques for counting Great
Blue Heron nests. Journal of Wildlife Management 59:667-673.

Dodd, M. G., and T. M. Murphy. 1996. The status and distribution of wading birds in South
Carolina, 1988-1996. Report SG9610-A. South Carolina Marine Resources, Columbia,
South Carolina, USA.

Eamnst, S. L., R. A. Stehn, R. M. Platte, W. W. Larned, and E. J. Mallek. 2005. Population size
and trend of Yellow-billed Loons in Northern Alaska. Condor 107:289-304.

Ellison, L. N., and L. Cleary. 1978. Effects of human disturbance on breeding of Double-Crested
Cormorants. Auk 95:510-517.

Erwin, R. M. 1982. Observer variability in estimating numbers: an experiment. Journal of Field
Omithology 53:159-167.

Erwin, R. M. 1989. Responses to human intruders by birds nesting in colonies: Experimental
results and management guidelines. Colonial Waterbirds 12: 104-108.

Erwin, R. M., and T. W. Custer. 1982. Estimating reproductive success in colonial waterbirds: an
evaluation. Colonial Waterbirds 5:49-56.

Fraser, W. R., and D. L. Patterson. 1997. Human disturbance and long-term changes in Adelie
penguin populations: a natural experiment at Palmer Station, Antarctic Peninsula. Pages
445-452 in B. Battaglia, J. Valencia, and D. W. H. Walton, editors. Antarctic
Communities: Species, Structure and Survival. Cambridge University Press,
Cambridge, UK.

Frederick, P. C. 2006. Recommendations for monitoring breeding populations of long legged
waders (Herons, egrets, ibises, storks, spoonbills, order Ciconiiformes). Appendix 1,
Il-I41 in Southeast United States Waterbird Conservation Plan. U. S. Fish and Wildlife
Service (USFWS), Division of Migratory Bird Management.










Frederick, P. C., and M. W. Collopy. 1989a. Nesting success of five ciconiiform species in
relation to water conditions in the Florida Everglades. Auk 106:625-634.

Frederick, P. C., and M. W. Collopy. 1989b. Researcher disturbance in colonies of wading birds:
effects of frequency of visit and egg-marking on reproductive parameters. Colonial
Waterbirds 12:152-157.

Frederick, P. C., and M. W. Collopy. 1989c. The role of predation in determining reproductive
success of colonially nesting wading birds in the Florida Everglades. Condor 91:860-
867.

Frederick, P. C., J.A. Heath, R. E. Bennetts, and H. Hafner. 2006. Estimating nests not present at
the time of breeding surveys: an important consideration in assessing nesting
populations. Journal of Field Ornithology 77:212-219.

Frederick, P. C., B. A. Hylton, J. A. Heath, and M. Ruane. 2003. Accuracy and variation in
estimates of large numbers of birds by individual observers using an aerial survey
simulator. Journal of Field Omithology 74:281-287.

Frederick, P. C., M. G. Spalding, M. S. Sepulveda, G.E. Williams Jr., L. Nico, and R. Robbins.
1999. Exposure of Great Egret nestlings to mercury through diet in the Everglades of
Florida. Environmental Toxicology and Chemistry 18:1940-1947.

Frederick, P. C., T. Towles, R. J. Sawicki, and G. T. Bancroft. 1996. Comparison of aerial and
ground techniques for discovery and census of wading bird (Ciconiiformes) nesting
colonies. Condor 98:837-840.

Galbraith, H. 1987. Marking and visiting Lapwing Vanelhts vanelhts nests does not affect clutch
survival. Bird Study 34:137-138.

Gibbs, J. P., S. Woodward, M. L. Hunter, and A. E. Hutchinson. 1988. Comparison of techniques
for censusing Great Blue Heron nests. Journal of Field Ornithology 59: 130-134.

Giese, M. 1996. Effects of human activity on Adelie Penguin Pygoscelis adeliae breeding
success. Biological Conservation 75:157-164.

Goitmark, F. 1992. The effects of investigator disturbance on nesting birds. Pages 63-104 in D.
M. Power, editor. Current Ornithology, volume 9. Plenum Press, New York, New
York, USA.

Gratto-Trevor, C. L., V. H. Johnston, and S. T. Pepper. 1998. Changes in shorebird and eider
abundance in the Rasmussen Lowlands, NWT. Wilson Bulletin 110:316-326.

Hernandez-Matias, A., L. Jover, and X. Ruiz. 2003. Predation on Common Temn eggs in relation
to sub-colony size, nest aggregation and breeding synchrony. Waterbirds 26:280-289.

Hulka, S., and J. Stirling. 2000. A study of breeding Black-throated Divers Gavia arctica based
on observation from vantage points. Bird Study 47:117-121.










Hunt, G. L. 1972. Influence of food distribution and human disturbance on the reproductive
success of Herring Gulls. Ecology 53: 105 1-1061.

James, F. C., C. E. McCulloch, and D. A. Wiedenfield. 1996. New approaches to the analysis of
population trends in land birds: a comment on statistical methods. Ecology 77:13-27.

Kadlec, J. A., and W. H. Drury. 1968. Structure of the New England Herring Gull population.
Ecology 49:644-676.

King, K. A. 1976. Colonial wading bird survey and census techniques. Pages 155-159 in A.
Sprunt IV, J. C. Ogden, and S. Winckler, editors. Wading Birds: Research Report No.7
of the National Audubon Society. Plenum Press, New York, New York, USA.

Kury, C. R., and M. Gochfeld. 1975. Human interference and gull predation in cormorant
colonies. Biological Conservation 8:23-34.

Kushlan, J. A. 1979. Effects of helicopter censuses on wading bird colonies. Journal of Wildlife
Management 43:756-760.

Kushlan, J. A., and K. L. Bildstein. 1992. White Ibis (Eudocimus albus). Volume 9 in A. Poole,
P. Stettenheim, and F. Gill, editors. The Birds of North America. Academy of Natural
Sciences, Philadelphia, Pennsylvania, USA.

Liddle, G. M. 1994. Interannual variation in the breeding biology of the Antarctic Prion
Pachyptila desolata at Bird Island, South Georgia. Journal of Zoology 234: 125-129.

Link, W. A., and J. R. Sauer. 1998. Estimating population change from count data: application to
the North American Breeding Bird Survey. Ecological Applications 8:258-268.

Link, W. A., and J. R. Sauer. 2002. A hierarchical analysis of population change with application
to Cerulean Warblers. Ecology 83:2832-2840.

Lock, A. R., and R. K. Ross. 1973. The nesting of the Great Cormorant (Phalacrocorax carbo)
and the Double-crested Cormorant (Phalacrocorax auritus) in Nova Scotia in 1971.
Canadian Field-Naturalist 87:43-49.

Magrath, M. J. L. 1999. Breeding ecology of the Fairy Martin. Australian Journal of Zoology
47:463-477.

Mayfield, H. F. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456-466.

McCrimmon, D. A., J. C. Ogden, and T. Bancroft. 2001. Great Egret (Ardea alba). Volume 570
in A. Poole and F. Gill, editors. The Birds of North America. The Birds of North
America, Inc., Philadelphia, Pennsylvania, USA.

Morrison, R. I. G., C. Downes, and B. Collins. 1994. Population trends of shorebirds on fall
migration in eastern Canada 1974-1991. Wilson Bulletin 106:431-447.










Nagy, L. R., and R. T. Holmes. 2004. Factors influencing fecundity in migratory songbirds: is
nest predation the most important? Journal of Avian Biology 35:487-491.

Nichols, J. D., J. E. Hines, J. R. Sauer, F. W. Fallon, J. E. Fallon, and P. J. Heglund. 2000. A
double-observer approach for estimating detection probability and abundance from
point counts. The Auk 117:393-408.

Ogden, J. C. 1994. A comparison of wading bird nesting dynamics, 193 1-1946 and 1974-1989,
as an indication of changes in ecosystem conditions in the southern Everglades. Pages
533-570 in S. Davis and J. C. Ogden, editors. Everglades: the ecosystem and its
restoration. St. Lucie Press, Del Ray Beach, Florida, USA.

Parsons, K. C., and J. Burger. 1982. Human disturbance and nestling behavior in Black-crowned
Night-Herons. Condor 84: 184-187.

Piatt, J. F., B. D. Roberts, W. W. Lidster, J. L. Wells, and S. A. Hatch. 1990 Effects of human
disturbance in breeding Least and Crested Auklets at St. Lawrence Island, Alaska. Auk
107:342-350.

Piazza, B. P., and V. L. Wright. 2004. Within-season nest persistence in large wading bird
rookeries. Waterbirds 27:362-367.

Platt, J. B. 1977. The breeding behavior of wild and captive Gyrfalcons in relation to their
environment and human disturbance. Unpublished M.S. thesis, Comell University,
New York, USA.

Pollock, K. H., and W. L. Kendall. 1987. Visibility bias in aerial surveys: a review of estimation
procedures. Journal of Wildlife Management 51:5 02-510.

Pollock, K. H., J. D. Nichols, C. Brownie, and J. E. Hines. 1990. Statistical inference for capture-
recapture experiments. Wildlife Monographs 107:1-97.

Prater, A. J. 1979. Trends in accuracy of counting birds. Bird Study 26: 198-200.

Pratt, H. M. 1970. Breeding biology of Great Blue Herons and Common Egrets in Central
Califomnia. Condor 72:407-416.

Prenzlow, D. M., and J. R. Lowomn. 1996. Evaluation of visibility correction factors for
waterfowl surveys in Wyoming. Journal of Wildlife Management 60:286-297.

Rao, P. S. R. S. 2005. Ratio Estimators--II. Pages 6978-6983 in S. Kotz, C. B. Read, N.
Balakrishnan, and B. Vidakovic, editors. Encyclopedia of Statistical Sciences, 10.

Rodgers, J. A., and J. Burger. 1981. Concluding remarks: symposium on human disturbance and
colonial waterbirds. Colonial Waterbirds 4:69-71.

Rodgers, J. A., P. S. Kubilis, and S. A. Nesbitt. 2005. Accuracy of aerial surveys of waterbird
colonies. Waterbirds 28:230-237.










Rodgers, J. A., S. B. Linda, and S. A. Nesbitt. 1995. Comparing aerial estimates with ground
counts of nests in Wood Stork colonies. Journal of Wildlife Management 59:656-666.

Rodgers, J. A., and H. T Smith. 1995. Set-back distances to protect nesting bird colonies from
human disturbance. Conservation Biology 9:89-99.

Rosenstock, S. S., D. R. Anderson, K. M. Giesen, T. Leukering, and M. F. Carter. 2002.
Landbird counting techniques: current practices and an alternative. Auk 119:46-53.

Safina, C., and J. Burger. 1983. Effects of human disturbance on reproductive success in the
Black Skimmer. Condor 85:164-171.

Sagar, P. M., and J.-C. Stahl. 2005. Increases in the numbers of breeding pairs in two populations
of Buller's Albatross (Thala~ssarche bulleri bulleri). Emu 105:49-55.

Schwarz, C. J., and A. N. Arnason. 1996. A general methodology for the analysis of capture-
recapture experiments in open populations. Biometrics 52:860-873.

Schwarz, C. J., R. E. Bailey, J. R. Irvine, and F. C. Dalziel. 1993. Estimating salmon spawning
escapement using capture-recapture methods. Canadian Journal of Fisheries and
Aquatic Sciences 50: 1181-1197.

Shields, M. A., and J. F. Parnell. 1986. Fish Crow predation on eggs of the White Ibis at Battery
Island, North Carolina. Auk 103:531-539.

Short, J., and P. Bayliss. 1985. Bias in aerial survey estimates of kangaroo density. Journal of
Applied Ecology 22:415-422.

Stolen, E. D., D. R. Breininger, and P. C. Frederick. 2004. Using waterbirds as indicators in
estuarine systems: Successes and perils. Pages 409-423 in S. A. Bortone, editor.
Estuarine indicators. CRC Press, Boca Raton, Florida, USA.

Temple, S. A., and J. A. Wiens. 1989. Bird populations and environmental changes: can birds be
bio-indicators? American Birds 43:260-270.

Thompson, B. C., G. E. Knadle, D. L. Brubaker, and K. S. Brubaker. 2001. Nest success is not
an adequate comparative estimate of avian reproduction. Journal of Field Ornithology
72:527-536.

Thompson, W. L. 2002. Towards reliable bird surveys: accounting for individuals present but not
detected. Auk 119: 18-25.

Tomialoj c, L., and J. Verner. 1990. Do point counting and spot mapping produce equivalent
estimates of bird densities? Auk 107:447-450.

Torres, R., and A. Mangeaud. 2006. Factors affecting the nesting success of the Cattle Egret
(Bubulcus ibis) in Laguna Mar Chiquita, Central Argentina. Ornitologia Neotropical
17:63-71.










Tremblay, J., and L. N. Ellison. 1979. Effects of human disturbance on breeding of Black-
crowned Night Herons. Auk 96:364-369.

Vos, D. K., R. A. Ryder, and W. D. Graul. 1985. Response of breeding Great Blue Herons to
human disturbance in Northcentral Colorado. Colonial Waterbirds 8:13-22.

White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from populations
of marked animals. Bird Study 46S: 120-138.

White, C. M. and T. L. Thurow. 1985. Reproduction of Ferruginous Hawks exposed to
controlled disturbance. Condor 87:14-22.

Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal
populations. Academic Press, New York, New York, USA.

Wilson, K.-J., R. H. Taylor, and K. J. Barton. 1990. The impact of man on Adelie penguins at
Cape Hallett, Antarctica. Pages 183-190 in K. R. Kerry and G. Hempel, editors.
Antarctic ecosystems: ecological change and conservation. Springer-Verlag, Berlin,
Germany.

Young, E. C. 1990. Long-term stability and human impact in Antarctic Skuas and Adelie
Penguins. Pages 232-236 in K. R. Kerry and G. Hempel, editors. Antarctic ecosystems:
ecological change and conservation. Springer-Verlag, Berlin, Germany.









BIOGRAPHICAL SKETCH

Kathryn A. Williams was born in Maine and grew up on the Scarborough Marsh. She

attended Tufts University in Medford, Massachusetts from 2000 to 2004 and graduated with a

double B.S degree in biology and environmental science. After brief forays into retail and trawl

fishing, she began working on the South Florida Wading Bird Proj ect in 2005.





PAGE 1

1 BIASES IN POPULATION ESTIMATION FOR COLONIALLY NESTING GREAT EGRETS ( Ardea alba ) AND WHITE IBISES ( Eudocimus albus ) IN THE FLORIDA EVERGLADES By KATHRYN A.WILLIAMS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Kathryn A. Williams

PAGE 3

3 To my parents, Susan and Douglas Williams, for their unflagging love and support. You guys are the best

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my committee members, Drs. Scott Robinson and Kathryn Sieving, and most particularly my advisor and committee chair, Dr. Peter Frederick. Peter has been a great mentor these past several years, and I am grateful that I have had the opportunity to work with him. I am a better scientist and, I believe, a better person as a result. Dr. Jim Nichols has also provided me with invaluable guidance, assistance, and instruction, all with great grace and good humor. I owe great thanks to the people who helped me in the field, including Christy Hand, Sam Edmonds, Carolyn Enloe, Brad Shoger, Becki Smith, Andrew Spees, Eric Trum, and particularly my coworker John Simon, without whom this thesis might never have been written. I als o owe additional thanks to my pilots, Kenny Lung, Matt Alexander, and Steve Diehl, along with everyone else at Unusual Attitudes, Inc. Julien Martin kindly came to my assistance during superpopulation analyses. My lab mates, Evan Adams, Rena Borkhataria and Nilmini Jayasena, provided much welcomed suggestions and commiseration, as did my roommates, Dina Liebowitz and Cyndi Langin. Last of all, I would like to express my appreciation for my family. Mom, Dad and Jen I am incredibly lucky to have you gu ys. Thanks for being so supportive when I said I wanted to move to the Everglades and drive airboats for a living.

PAGE 5

5 TABLE OF CONTENTS Page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 7 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 ABSTRACT ................................ ................................ ................................ ................................ ..... 9 CHAPTER 1 VISUAL BIAS IN AERIAL ESTIMATES OF COLONIALLY NESTING WHITE IBISES AND GREAT EGRETS ................................ ................................ ............................ 11 Introduction ................................ ................................ ................................ ............................. 11 Methods ................................ ................................ ................................ ................................ .. 13 Sample Areas ................................ ................................ ................................ ................... 14 Counts and Photographic Analyses ................................ ................................ ................. 15 Error Calculations and Tests for Covariates ................................ ................................ .... 15 Results ................................ ................................ ................................ ................................ ..... 17 Great Egrets ................................ ................................ ................................ ..................... 17 White Ibises ................................ ................................ ................................ ..................... 17 Correction for Visibility Bias in Colony Counts ................................ ............................. 18 Discussion ................................ ................................ ................................ ............................... 18 2 ESTIMATING BREEDING POPULATION SIZE FOR ASYNCHRONOUSLY NESTING COLONIAL WADING BIRDS ................................ ................................ ........... 27 Introduction ................................ ................................ ................................ ............................. 27 Estimation Error ................................ ................................ ................................ .............. 27 A New Strategy For Estimating Breeding Population Size For Un marked Populations ................................ ................................ ................................ ................... 29 Methods ................................ ................................ ................................ ................................ .. 30 Survey Methodology ................................ ................................ ................................ ....... 30 Artificia l Geographic Markers ................................ ................................ ........................ 31 Photographic Analysis ................................ ................................ ................................ ..... 31 The Superpopulation Modeling Approach to Population Estimation ............................. 33 Subsampling Alley North Colony ................................ ................................ ................... 36 Subsampling to Determine Necessary Survey Frequency ................................ ............... 38 Results ................................ ................................ ................................ ................................ ..... 39 Interobserver Error in Tallying Individual Nest Histories From Photographs ................ 39 Superpopulation Estimates ................................ ................................ .............................. 39 Subsampling to Determine Necessary Survey Frequency ................................ ............... 40 Discussion ................................ ................................ ................................ ............................... 41 Limitations and Assumptions ................................ ................................ .......................... 43

PAGE 6

6 Model fit ................................ ................................ ................................ ................... 45 Refinements of the superpopulation technique ................................ ........................ 46 Incorporation of renesting ................................ ................................ ........................ 47 Implications of Results ................................ ................................ ................................ .... 47 3 REMOTELY MEASURED EFFECTS OF RESEA RCHER DISTURBANCE ON GREAT EGRETS ( Ardea alba ) AND WHITE IBISES ( Eudocimus albus ) ......................... 61 Introduction ................................ ................................ ................................ ............................. 61 Methods ................................ ................................ ................................ ................................ .. 62 Ground Transects ................................ ................................ ................................ ............. 62 Aerial Survey Methodology ................................ ................................ ............................ 64 Photographic Analysis ................................ ................................ ................................ ..... 65 Analysis of Presence Absence Databases: A Mark Recapture Scenario ........................ 66 Results ................................ ................................ ................................ ................................ ..... 69 Vacation Island Great Egret Colony ................................ ................................ ................ 69 Alley North White Ibis Colony ................................ ................................ ....................... 70 Discussion ................................ ................................ ................................ ............................... 71 Violations of Model Assumptions. ................................ ................................ .................. 71 Effects of Disturbance on Nest Survival ................................ ................................ ......... 71 Effects of Disturb ance on Nest Initiation ................................ ................................ ........ 72 APPENDIX A SUPERPOPULATION MODEL OUTPUT (POPAN DATA TYPE) FOR ALL COLONIES ................................ ................................ ................................ ............................. 81 B CALCULATION OF STAND ARD ERRORS FOR EXTRAPOLATED SUPERPOPULATION ESTIMATES FROM ALLEY NORTH SAMPLES ........................ 85 C SAMPLE BUDGET FOR DEVELOPING PEAK COUNT AND SUPERPOPULATION ESTIMATES ................................ ................................ .................... 89 D MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON GREAT EGRETS ................................ ................................ ................................ ................... 92 E MODEL SUITE FOR STUDY OF EFFECTS OF GROUND DISTURBANCE ON WHITE IBISES ................................ ................................ ................................ ...................... 93 LIST OF REFERENCES ................................ ................................ ................................ ............... 94 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 101

PAGE 7

7 LIST OF TABLES Table Page 1 1 Quadrats from 2005 and 2006 seasons from the Everglades of southern Florida.. ........... 22 1 2 Visi bility bias corrected peak counts for five Great Egret and White Ibis colonies from the 2005 and 2006 breeding seasons. ................................ ................................ ........ 23 2 1 Names and locations of study sites.. ................................ ................................ .................. 49 2 2 All colonies for which we followed nest fate through time in aerial survey photographs and calculated a superpopulation estimate.. ................................ .................. 50 2 3 All samples of Great Egret and White Ibis populations from large colony (Alley North). ................................ ................................ ................................ ................................ 51 3 1 Characteristics of the two best fit, lowest AIC models describing population parameters for Great Egret nests in the central Everglades, Florida.. ................................ 75 3 2 Characteristics of the best fit, lowest AIC models describing population parameters for White Ibis nests in the central Everglades, Florida. ................................ ..................... 75 3 3 Comparison of model averaged survival and entry probabilities between disturbance groups for the Alley North White Ibis colony. ................................ ................................ .. 76 A 1 For each colony or sample, we tested a set of four candidate models ............................... 81 C 1 Sample budgets for superpopulation estimation and peak count estimation of breeding population size for a hyp othetical Great Egret colony. ................................ ....... 91 D 1 ............... 92 E 1 All mod ....................... 93

PAGE 8

8 LIST OF FIGURES Figure Page 1 1 A 10x10 meter White Ibis quadrat at Alley North colony in WCA 3A (04/07/06).. ........ 24 1 2 P roportional error in aerial counts of White Ibis quadrats.. ................................ ............... 25 1 3 Proportion error in aerial estimates of White Ibis quadrats by vegetative cover. .............. 26 2 1 Types of data that should be incorporated into estimates of breeding population size. .... 53 2 2 Satellite photograph of the modern Everglades, with major watershed regi ons outlined in yellow.. ................................ ................................ ................................ ............ 54 2 3 3A). .............................. 55 2 4 Two types of artificia l landmarks used in this study ................................ ......................... 56 2 5 Multi observer comparison of superpopulation estimates for Vacation Island 2006 colony. ................................ ................................ ................................ ................................ 57 2 6 Peak counts and associated superpopulation estimates for three Great Egret colonies in WCA 3A (2005 2006). ................................ ................................ ................................ .. 58 2 7 Peak counts and extrapolated superpopulation estimates for White Ibises i n 2005 and 2006 and Great Egrets in 2006 at Alley North colony (WCA 3A). ................................ .. 59 2 8 The proportional difference between superpopulation estimates and peak counts for 2005 and 2006. ................................ ................................ ................................ ................... 60 3 1 Two types of artificial landmarks used in this study ................................ ......................... 77 3 2 The difference in entry probabilities between disturbance groups at Vacation Island Great Egret colony as a function of time.. ................................ ................................ ......... 78 3 3 Model averaged weekly survival probabilities for both disturbance groups at Alley North White Ibis colony as a function of survey inte rval (time).. ................................ ..... 79 3 4 Model averaged weekly entry probabilities for both disturbance groups at Alley North White Ibis colony as a function of survey interval (time).. ................................ ..... 80

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science BIASES IN POPULATION ESTIMATION FOR COLON IALLY NESTING GREAT EGRETS ( Ardea alba ) AND WHITE IBISE S ( Eudocimus albus ) IN THE FLORIDA EVE RGLADES By Kathryn A. Williams August 2007 Chair: Peter C. Frederick Major: Wildlife Ecology and Conservation Estimating avian breeding population sizes often involves extrapolations derived from estimates of related biological parameters, and may therefore fail to account for bias introduced from several sources. In colonially breeding birds, estimation of breeding population size is complicated by visual biases and by asynchronous nest initiation and turnover. We used artificial landmarks to create quadrats in nesting colonies, and quantified visual biases by counting the number of nests in a quadrat via both walk in survey and aerial photography. Visual bias due to vegetative occlusion caused aerial surveys of Wh ite Ibises ( Eudocimus albus ) to underestimate numbers of nests by an average of 12.37% (20.65%) when compared with walk in ground surveys. Misidentification of the species of nesting birds in mixed species areas proved to be a problem in aerial surveys o f Great Egrets ( Ardea alba ). We examined asynchrony bias by applying a mark recapture approach to nests. We individually identified nests from the air through the use of natural and artificial landmarks, and followed nest fates through subsequent aerial surveys to obtain presence absence information over the course of a breeding season. We used the superpopulation approach (a variation on a Jolly Seber open population mark recapture model) to model nest initiation and survival as if nests were animals i n a marked population. The superpopulation approach allowed for the

PAGE 10

10 estimation of the total number of nest starts throughout the survey period. We found that asynchrony in nest initiation dates caused one ne sting to underestimate true numbers of nest starts by 47 382% for both species. Nest turnover rates were highly variable among years and between colony sites, suggesting that nesting asynchrony must be measured for each season and site of interest. These results indicate that, without bias correction, peak counts do not allow for useful inter year comparisons, because populations are highly synchronous, or t his bias is measured on a seasonal basis, one time of breeding population size. Researcher disturbance may also bias estimates of population parameters for co lonies. Walk in censuses of White Ibis and Great Egret colonies (initiated after the onset of incubation for the majority of nests) found that while nest success was not significantly affected by researcher disturbance, disturbed areas demonstrated depres sed rates of new nest initiation. These species may be sensitive to disturbance early in the nesting cycle, but be resilient to walk in disturbance after the onset of incubation.

PAGE 11

11 CHAPTER 1 VISUAL BIAS IN AERIA L ESTIMATES OF COLON IALLY NESTING WHITE IBISES AND GREAT EGRETS Introduction Birds are one of the most frequently used bioindicators (Temple and Wiens 1989 Stolen et al. 2004) Close monitoring of bird populations may improve our knowledge of the relationship between environmental conditions and demography and allow the populations to be used as indicators of ecological change. Estimates of population size also have value in their own right, by allowing for more informed management of birds. However, the estimation of avian breeding population size must of ten be an approximation derived from estimates of related biological parameters (Frederick et al. 2006 Bibby 2000 Bock and Jones 2004) and may therefore fail to account for bias introduced from several sources (Erwin and Custer 1982 ; e.g., Ogden 1994 Gratto Trevor et al. 1998) In colonially breeding birds, there may be several sources of bias involved in estimating b reeding population size, especially for large breeding aggregations. These include species or groups of birds that nest asynchronously over the course of a breeding season, nest in multispecies colonies, and whose nests may be concealed by vegetation duri ng aerial surveys. A great amount of effort and scientific inquiry has been directed toward understanding bias in counts of animals (e.g., Prenzlow and Loworn 1996, Erwin 1982 Bart and Earnst 2002 Rosenstock et al. 2002 Tomialojc and Verner 1990 Dodd and Murphy 1995 Frederick et al. 2003 Frederick et al. 1996 Gibbs et al. 1988 Bayliss and Yeomans 1990 Rodgers et al. 1995) Detectability has been recognized as a problem in line transects (Bibby 2000, Dodd and Murphy 1996) perimeter counts (Dodd and Murphy 1996) point counts (Bibby 2000, Nichols et al. 2000) and other types of population surveys using vocal or visual counts. Detectability can be a

PAGE 12

12 source of bias for several reasons, including individual heterogeneity in behavior (such as frequency of song) and visual bias caused by the varying distances of birds from the observer. Although estimates of the size of bird aggregations taken from ai rcraft or other aerial platforms have many advantages, they introduce detectability concerns related to variation in visibility of nests (Gibbs et al. 1988 Frederick et al. 2003 Bayliss and Yeomans 1990) and species misidentification (Rodgers et al. 2005, Barbraud and Glinaud 2005, King 1976) Visual occlusion of nests as seen from the air may be due to several factors, including vegetative occlusion and miscounting because of high nest densities (Dodd and Murphy 1995 Prater 1979 Bayliss and Yeomans 1990) For example, while Great Egrets ( Ardea alba ) and American Wood Storks ( Mycteria americana ) often nest at or near the top of the tree canopy, where the nests are generally highly visible from the air, White Ibises ( Eudocimus albus ) and small herons ( Egretta spp.) usually nest in both canopy and midstory, where their nests may be occluded from above. Additionally, high nest density in species such as White Ibises may make it difficult to differentiate b etween individual nests. Misidentification of species in aerial surveys of mixed species colonies has also been found to be a problem in at least one study, although this problem has not been well explored in the literature. Rodgers et al. (1995) found t in distinguishing Great Egret and Wood Stork nests from the air led to large and highly variable estimates of aerial bias in wading bird counts (Order Ciconiiformes). When estimating regional breeding populations it is vital tha t accurate estimates be obtained for large colonies, as they contain a large percentage of the total breeding population and errors in population estimation for large colonies will disproportionately affect total estimates. We studied visual bias in estim ates of Great Egret ( Ardea alba ) and White Ibis ( Eudocimus albus ) population sizes because they are conspicuous colonial nesters that nest in

PAGE 13

13 large colonies, and because the two species have different nesting habits and potentially different nest visibilit ies. In this chapter we compare counts taken from both the ground and the air in the same large marked quadrats. Resulting estimates of visibility bias were used to adjust colony counts for encounter probability. We predicted that, for both species, vis ual bias would result in an underestimation of numbers of nests. We also expected that degree of error would increase with higher vegetative density and nest density. Methods We studied Great Egrets and White Ibises in the Water Conservation Areas (WCAs) of the Florida Everglades (Broward and Palm Beach Counties, FL) in March through May of 2005 and 2006. WCA 3A and WCA 1 are large areas of primarily sawgrass and wet prairie in the central Everglades that are controlled by, respectively, the South Florida Water Management District and the U.S. Fish and Wildlife Service. Colony counts during this time ranged up to 1,193 Great Egret nests and 13,566 White Ibis nests in a single colony; these counts were derived from monthly aerial surveys conducted from Jan uary June of 2005 and 2006. We conducted surveys in a Cessna Skyhawk (172) high wing aircraft flying at approximately 177 kph and 244 meters using a Canon EOS 20D wit h a 28 135mm image stabilizing lens. We accepted the largest single month count of the breeding season from these aerial photographs as the maximum or in the pas t as a minimum breeding population estimate for wading birds in the Everglades. We present data here from five colonies for one or both species and years. For visual bias studies, we worked in the mixed species colony at Alley North (WCA 3A; N 26 11.179 W 80 31.431), which included both Great Egrets and White Ibises in different areas of the single large tree island. We also worked in the White Ibis colony New Colony 3 in

PAGE 14

14 WCA 1 (N 26 32.013, W 80 17.879), which consisted of over twenty smaller tree islands in close proximity. The emergent vegetation in both colonies was composed primarily of willow ( Salix caroliniana ) with small numbers of pond apples ( Anona glabra ). White Ibises at Alley North also nested in lower vegetation on the outskirts of the tree island, primarily in cattail ( Typha latifolia ). Sample Areas We created rectangular quadrats in Great Egret and White Ibis colonies to compare aerial and ground counts of nests within these areas. White Ibis quadrats varied in size from 100 m 2 to roughly 200 m 2 and Great Egret quadrats were either 400 m 2 or 900 m 2 Each quadrat consisted of four artificial physical landmarks that were large and conspicuous enough to be seen from aircraft at 100 200 meters above ground level. Markers were of three types: A roughly 1 2 meter wide area of vegetation painted with a dilute white latex paint (1:3 water:paint) dispensed from a backpack sprayer of the type common in landscaping. These markers were only utilized during the 2005 season. A 3 meter tal l, 10 cm diameter white vertically mounted PVC pipe with a 1.5 meter horizontal X on the top. These markers were primarily utilized during the 2006 season. A 1x1 meter piece of white or pale colored cotton cloth, tied at the corners to vegetation in a hor izontal position (Figure 1 1). This type of marker was used exclusively during the 2006 breeding season. We delineated the edges of quadrats with plastic flagging tied to vegetation at regular intervals between corner markers, except in cases at New Colon y 3 where the sample area consisted of an entire, well defined tree island. The creation of a quadrat and the counting of all nests within it required approximately 30 minutes to 1.5 hours, depending upon size of quadrat, ease of movement, and nest densit y. We marked a total of 21 rectangular quadrats in Great Egret and White Ibis colonies in order to compare aerial and ground counts of nests within these areas. In 2005 we erected

PAGE 15

15 markers for four quadrats in the White Ibis colony at Alley North. In 20 06, we erected markers for ten quadrats in Alley North (four Great Egret quadrats and six White Ibis quadrats). We also erected markers for seven White Ibis quadrats in New Colony 3. Counts and Photographic Analyses On the date that markers for a quadra t were erected, we counted the number of nests in the quadrat on the ground, and marked each nest with flagging when it was counted in order to avoid double counting. Within 24 36 hours of the ground count, we photographed the quadrat from the air in a Ce ssna Skyhawk (172) at an altitude of approximately 152 meters. Photographs were possible, using a Canon EOS 20D with a 28 135mm image stabilizing lens. Ph otographs were analyzed using either Adobe Photoshop Elements version 2.0 or the shareware program Paint.Net version 2.0, using the same general procedure as for colony wide counts. On the computer screen we delineated quadrat edges using colored lines (F igure 1 1), and marked nests with colored dots as they were counted. Error Calculations and Tests for Covariates For most of the figures in this paper, we present aerial bias in nest counts as Percent bias in aerial count = ((aerial count/ground count) 1) x100 A 12% error, for instance, indicates that the aerial count of a quadrat underestimated the true number of nests found during the ground count by 12%. Note that in Table 2, estimates are presented in the slightly less intuitive format of (ground coun t/aerial count) in order to form a ratio with maximum colony counts. During the ground count for each quadrat, we subjectively ranked the level of tall occlusive vegetative cover present as low (more than 50% cattails or tall grass cover and less than 50 % willows and other tree cover), medium (between 10% and 50% cattails or grass cover,

PAGE 16

16 50 90% small willows and other tree cover) or high (more than 90% willow and other tall tree cover). We compared White Ibis quadrats for statistically significant differ ences between vegetation groups, using a nonparametric Kruskal Wallis test for k independent samples, since the ratio data were sufficiently non normal that transformation did not produce a normal or symmetric distribution for the data. We also used nonpa rametric tests to determine if date of quadrat creation or colony location affected aerial estimation bias in 2006. We examined effect of colony location (Alley North or New Colony 3) using a 2 tailed Mann Whitney U test, and examined the effect of Julian date of quadrat creation using a 2 tailed Spearman rank correlation. The four Great Egret quadrats were all located in an area where other species also nested, including Snowy Egrets ( Egretta thula ), Little Blue Herons ( Egretta caerulea ), Tricolored Heron s ( Egretta tricolor ), and Black crowned Night Herons ( Nycticorax nycticorax ). Adult Snowy Egrets and Great Egrets can be difficult to differentiate from the air, as they are both white and the difference in body size may not be obvious from a distance. S mall heron nests can be reliably differentiated from Great Egret nests by sight on the ground, but small heron nests in the genus Egretta cannot be readily differentiated from each other until the chicks are 10 14 days old. For this reason, the true numbe r of Snowy Egret nests present in the quadrats was unknown. We averaged the information from all quadrats for a given species to derive visual bias estimates for that species in any colony, by taking the ratio between the ground and aerial count for each quadrat, summing the ratios for all quadrats for that species, and dividing by the number of quadrats, as in the equation below: Visual bias= i /aerial count i )/n

PAGE 17

17 where i is a given quadrat and n is the sample size of quadrats for the species of interest. For each species, we used the resulting bias estimate for the left side of the ratio equation (quadrat ground count)/(quadrat ae rial count) = (bias corrected count)/(colony count) to correct colony peak counts for visibility bias. We compared the new bias corrected counts and associated confidence intervals to original colony wide aerial counts to determine the significance of the results. We calculated the associated confidence intervals (CIs) for each was the sample mean error rate for the species; sd was the sample standard error for each species; and n was the number of quadrats observed for each species. The mean error and confidence intervals from the samples were then multiplied by the maximum total count for each colony for the season, in order to obtain a new bias adjusted colony count and 95% CIs. Results Great Egrets Numbers of nests identified from the air were greater than the ground counts for two out of four Great Egret quadrats, and we strongly suspect that some percentage of Snowy Egrets were mistaken for Great Egrets in aerial photos (Table 1 1). This source of bias was apparently larger on average th an any bias due to vegetative occlusion, which would have caused aerial underestimates rather than overestimates. Aerial counts ranged from 14% to 260% of ground counts, with a mean of 70.28% (sd 131.20%). Due to the small sample size, statistical analy ses were not conducted on the Great Egret quadrats. The common nest location preferences of this species meant that all four quadrats were placed in areas of high vegetative cover. White Ibises White Ibis nest numbers were generally underestimated in aeri al surveys (Figure 1 2). Although the severity of this underestimation varied, the mean bias in aerial surveys was

PAGE 18

18 approximately 12.37% (sd=20.65%; n=17). Visual bias did not increase with density of nests (Figure 1 2), and in 2006 also did not vary by colony (Z= 0.714, p=0.475, n=13) or by date of quadrat initia 0.275, p=0.363, n=13). Visual bias for White Ibises did not significantly increase with increased vegetative cover (Figure 1 2 =0.163, p=0.922, n=17, df=2) although the direction of the difference in median error values between the low and med ium cover groups suggests that with a larger sample size for the high vegetative cover, we might see a progression in median error rate from low to high canopy cover (Figure 1 3). Median visual error for quadrats in low vegetative cover was 11.07% (IQR 14.97% to 5.40%, n=6); for quadrats in medium cover, 16.50% (IQR 18.71% to 8.96%, n=9); and for quadrats in high vegetative cover, 6.98% (IQR 15.47% to 1.51%). Correction for Visibility Bias in Colony Counts Uncorrected maximum colony counts for White Ibis colonies lay outside the 95% confidence intervals for the new bias corrected estimates in both 2005 and 2006 (Table 1 2). These uncorrected peak counts may thus be judged to be statistically significant underestimates of the true numbers of nests pr esent. The uncorrected counts for Great Egret colonies were not statistically significantly different from the bias corrected estimates, as the 95% confidence interval for the bias corrected counts overlapped with the original counts (Table 1 2). However the sample size of quadrats and colonies was small, and there was large variation in degree of bias among quadrats. Visibility bias seemed to be quite substantial in some locations, if not on average. Discussion There is evidence that visual bias in ani mals varies by species (Short and B ayliss 1985) ground cover (Short and Bayliss 1985) observer quality (Erwin 1982) and other factors. However, several studies have found a consistent underestimation of true numbers of birds using

PAGE 19

19 aerial surveys (Pollock and Kendall 1987 ; e.g., Gibbs et al. 1988 Dodd and Murphy 1995) Our results reinforce the conclusion that visual bias can be an important source of error in estimates of avian breeding population size, and that the evaluation of detection probability is essential for sampling efforts that use counts as population indices (Pollock and Kendall 1987) While the degree of visual bias seems likely to vary with colony site and species, the amount of variation we found in aerial bias among quadrats was similar to that found in other studies (Rodgers et al. 1995 Gibbs et al. 1988 Dodd and Murphy 1995) For ibises, the degree of bias ( 12%) was si milar to that for other estimates for wading birds ( 15% for Squacco Herons Ardeola ralloides Barbraud et al. 2004 ; 16%, 10% and 11%, respectively, for Great Egrets, Snowy Egrets, and White Ibises, Kushlan 1979) The bias for ibises did not appear to increase with greater nest density, in contrast to Great Blue Herons ( Ardea herodias ; Gibbs et al. 1988). This may be due in part to the fact that many of the White Ibis qua drats were located in areas of relatively low vegetative cover, which limited the number of potential nest strata and allowed individual nests to be relatively easily distinguished. There is little evidence in support of the vegetative cover prediction. T he locations in which quadrats could be placed were unfortunately limited by the necessity for ground access, and thus a disproportionate percentage of the quadrats at Alley North colony were located in areas of low and medium vegetative cover. An increas ed sample size for White Ibis quadrats in high vegetative cover is needed before level of vegetative cover can be ruled out as an important covariate for survey bias in White Ibises, but in our study the variation in aerial bias was much larger within vege tative cover groups than between groups. Initial evidence for this species does not support the hypothesis that visual bias due to vegetative occlusion increases with vegetative complexity. For Great Egrets, a larger sample size is also needed, as well a s a finer scale

PAGE 20

20 stratification of vegetative cover level. Great Egrets nest almost exclusively in areas that, according to the criteria used in this study, would be classified as high cover (K. Williams pers. obs.). However, Great Egrets nest largely abo ve the canopy, and it may be (as Rodgers et al. found for Wood Storks; 1995), that Great Egret visibility bias is likewise not significantly affected by level of vegetative cover. The uncorrected maximum colony counts for Great Egret colonies were not st atistically significantly different from the bias corrected estimates, as the 95% confidence intervals for the bias corrected counts overlapped with the original colony counts. However, the sample size of quadrats and colonies was small, and there was cle arly a large amount of variation in bias for this species in mixed species groups, due to both visual occlusion and to difficulty in distinguishing Great Egrets from other white species during aerial counts. The resulting bias corrected peak count of Grea t Egrets for Alley North in 2006 (with the associated confidence intervals) reflects these dual sources of bias (Table 1 2). The quadrat method of quantifying bias due to visual occlusion is necessary for situations in which nests are located on multilayer ed substrates. Vegetation is often the cause of this sort of visual bias (Frederick et al. 2003), but similar problems could be evident in colonies in which nests are occluded by rock crevices or artificial structures. However, our results indicate that ground truthing may not always be a gold standard for correcting visibility bias. It has been suggested that ground counts may be inaccurate in highly vegetated areas where movement within the colony is difficult, and canopy nests are difficult to see or count (Prater 1979, Frederick et al. 2003). Additionally, if ground counts and aerial surveys are not conducted within a very short time span, some nest turnover may occur and cause increased bias between the two estimates (Frederick et al. 2003) In our study, we believe tha t ground counts were not

PAGE 21

21 hampered by nest location or difficulty of movement, and aerial counts were conducted within as short a time span as possible after ground counts. Nevertheless, ground/aerial comparisons could not adequately address bias due to sp ecies misidentification (see also Rodgers 1995). Even the ground surveys could not distinguish the non target species (Snowy Egrets) from other small herons, which meant that we were unable to correct for aerial bias due to species misidentification. Sin ce the two types of error (visual occlusion and misidentification) may be inseparable in multispecies colonies (King 1976), it is important that such estimates be applied to single species colonies with extreme caution. In the future, separation of vegeta tive occlusion from species misidentification bias could be achieved by comparing bias estimates from single and mixed species areas. Alternatively, surveys could be conducted using helicopters, which allow species to be more easily differentiated. Our r esults indicate that White Ibis aerial surveys are likely to be significantly biased due to vegetative occlusion. However, this bias did not increase significantly with vegetative complexity or nest density. Our small sample size for Great Egret aerial s urveys did not indicate systematic bias due to vegetative occlusion, but species misidentification appears to be a significant problem for this species. We suggest that visual bias due to both vegetative occlusion and species misidentification should be e xplicitly measured in aerial surveys of colonies, as both of these sources of error can affect estimates of breeding population size.

PAGE 22

22 Table 1 1. Quadrats from 2005 and 2006 seasons from the Everglades of southern Florida. White Ibis quadrats were 100 m 2 (or in one case 200 m 2 ) and Great Egret quadrats were either 400 or 900 m 2 White Ibis nest densities varied from 0.08 to 4.45 nests per m 2 Year Species a Ground count b Aerial count b % error in aerial count c Vegetative cover d Location 2005 WHIB 23 20 13 Low Alley North 2005 WHIB 54 21 61 Medium Alley North 2005 WHIB 10 11 9 High Alley North 2005 WHIB 8 4 50 Low Alley North 2006 WHIB 205 173 16 Low Alley North 2006 GREG/mixed GREG 14 GREG 12 14 High Alley North SH 35 SNEG 19 2006 GREG/mi xed GREG 24 GREG 19 21 High Alley North SH 7, SNEG 1 SNEG 6 2006 GREG/mixed GREG 5 GREG 18 260 High Alley North SH 19 SNEG 6 2006 GREG/mixed GREG 16 GREG 25 56 High Alley North SH 18 SNEG 10 2006 WHIB 72 69 4 Low Alley North 2006 WH IB 33 30 9 Low Alley North 2006 WHIB 30 24 20 Medium Alley North 2006 WHIB 18 20 10 Medium Alley North 2006 WHIB 67 73 8 Medium Alley North 2006 WHIB 414 467 11 Medium New Colony 3 2006 WHIB 445 491 9 Low New Colony 3 2006 WHIB 139 113 19 Medium New Colony 3 2006 WHIB 303 253 17 Medium New Colony 3 2006 WHIB 96 73 24 High New Colony 3 2006 WHIB 390 319 19 Medium New Colony 3 2006 WHIB 114 99 13 Medium New Colony 3 a. WHIB= White Ibis; GREG= Great Egret; SNEG= Snowy Egret; SH= unidentified small heron. b. Ground counts and aerial counts are estimated numbers of nests present in the quadrat. 1)x100. d. Vegetative cover values are subjective measurements (low = more than 50% cattails or tall grass cover and less than 50% tree cover; medium = between 10% and 50% cattails or grass cover and 50 90% small willows and other tree cover; high = more than 90% willow and other tall tree cover).

PAGE 23

23 Table 1 2. Visibility bias corrected peak counts for five Great Egret and White Ibis colonies from the 2005 and 2006 breeding seasons. Colony (species) a Year Peak count Visibility error b (sd) New estimate (LCI UCI) c Vulture (GREG) 2005 121 0.84 (0.46) 101 (46 156) Vulture (GREG) 200 6 458 0.84 (0.46) 383 (176 591) Vacation Island (GREG) 2005 79 0.84 (0.46) 66 (30 102) Vacation Island (GREG) 2006 155 0.84 (0.46) 130 (59 200) Cypress City (GREG) 2005 107 0.84 (0.46) 90 (41 138) Cypress City (GREG) 2006 173 0. 84 (0.46) 145 (66 223) Alley North (GREG) 2005 850 0.84 (0.46) 711 (326 1,097) Alley North (GREG) 2006 1193 0.84 (0.46) 998 (458 1,539) Alley North (WHIB) 2005 12750 1.23 (0.43) 15,702 (13,078 18,327) Alley North (WHIB) 2006 13566 1.23 ( 0.43) 16,707 (13,915 19,499) New Colony 3 (WHIB) 2006 4800 1.23 (0.43) 5,911 (4,924 6,899) a. GREG=Great Egret; WHIB=White Ibis. b. Error calculated for each species as ( i /aerial count i ))/n, where i is a given quadrat and n=number of quadrats. sample mean error, z=1.96 for 95% confidence limits, sd=sample standa rd deviation, and n=sample size (number of quadrats for the species). New bias corrected estimates and confidence intervals were derived from multiplying mean proportion visibility error and error CIs by the peak count for each colony.

PAGE 24

24 Figure 1 1. A 10x10 meter White Ibis quadrat at Alley North colony in WCA 3A (04/07/06). The ground count for this quadrat was 72 nests; the aerial count from this photograph was 69 nests. Written labels and lines between the four artificial markers were inserted int o the photograph using Adobe Photoshop Elements v. 2.0.

PAGE 25

25 Figure 1 2. Proportional error in aerial counts of White Ibis quadrats. Proportion error is calculated as (aerial count/ground count) 1, where the ground co unt of number of nests was conducted during a walk in survey of the quadrat and the aerial count was obtained via digital photography of the quadrat site from small aircraft.

PAGE 26

26 Figure 1 3. Proportion error in aerial estimates of White Ibis quadrats by vegetative cover. Vegetative cover values are subjective measurements (low = more than 50% cattails or tall grass cover and less than 50% tree cover; medium = between 10% and 50% cattails or grass cover and 50 90% small willows and other tree cover; high = more than 90% willow and other tall tree cover).

PAGE 27

27 CHAPTER 2 ESTIMATING BREEDING POPULATION SIZE FOR ASYNCHRONOUSLY NESTI NG COLONIAL WADING BIRD S Introduction The need for reliable avian demographic information has increased as scientists attempt to use attributes of birds to measure both ecological degrad ation and restoration (Thompson 2002 Rosenstock et al. 2002, Stolen et al. 2004) Birds are one of the most frequently suggested bioindicators (Temple and Wiens 1989 Stolen et al. 2004) but accurate estimates of population parameters can be hampered by a number of different kinds of bias. It is rare that entire breeding populations can be counted, for instance, and most attempts at estimating population size are approximations derived from estimates of related biological parameters, such as numbers of nests (Bibby 2000, Williams et al. 2002, Bock and Jones 2004, Frederick et al. 2006) However, there is a large amount of uncertainty in survey counts of nests and in the relation of those nest counts to population size (Thompson 2002, Bock and Jones 2004, Frederick et al. 2006) Inter observer error, visibility bias, and species that nest asynchronously over the course of a breeding season or nest in multispecies groups can present significant problems when trying to estimate breeding population size. In the case of colonially breeding birds, large aggregation size may also cause difficulties when attempting to e stimate population size. However, it is vital that accurate estimates be obtained for large colonies, as they may contain a large percentage of a disproportionately affect total estimates. Estimation Error A huge amount of effort and scientific inquiry has been directed toward understanding bias in nest counts (e.g., Erwin 1982 Gibbs et al. 1988 Tomialojc and Verner 1990 Dodd and Murphy 1995 Frederick et al. 1996 Prenzlow and Loworn 1996 Bart and Earnst 2002

PAGE 28

28 Rosenstock et al. 2002 Frederick et al. 2003) Aerial estimates of the population sizes of colonial nesters have been found to be biased due to visibility problems, interobserver variation in estimation error, and misidentification of species (Bayliss and Yeomans 1990, Frederick et al. 2003, Rodgers et al. 1995) In addition, most studies of estimation error to date have focused on estimating nests that are present at the time of survey, and have rarely recognized the problem of estimating nests that occur during the breeding period of interest but are not present at the time of survey (Frederick et al. 2006, Cherenkov 1998) Asynchronous breeding, even within a defined breeding season, is common in colonially nesting birds, including terns (Hernandez Matias et al. 2003) prions (Liddle 1994) martins (Magrath 1999) and wading birds (Ciconiiformes; Piazza and Wright 2004). In many species, nesting may therefore occur before surveys begin during a season, after they end, or in between consecutive surveys. Estimates of unmarked nests on any given date are inherently underestimating the true numbers of nests, partly because nests may be occurring outside the dates of survey, and partly because novel nests on any given survey may be confused with nests that were present on a previous surve y date. This is true even if the counts on each date are highly accurate. In such a situation, where nests cannot be individually distinguished, the best s urvey date. These circumstances occur in several taxa, including wading birds, seabirds, loons, and shorebirds (Frederick et al. 2006 ; e.g., Sagar and Stahl 2005 Earnst et al. 2005 Morrison et al. 1994) While the resulting bias may be negligible when nesting is highly synchronous and little renesting occurs, the degree of error in less synchronous birds can be substantial (e.g., 47% in wadi ng birds; Frederick et al. 2006).

PAGE 29

29 In addition to problems with nesting asynchrony, using numbers of nests as a proxy for breeding population size can be complicated by the degree of occurrence of renesting or multiple nesting (Thompson et al. 2001, Nagy and Holmes 2004) Renesting and multiple brooding are rare in many populations of high latitude nesting birds, but they can be common among temperate and tropical species. We wish to make clear in this chapter that we are addressing the problem of estimating numbers of nests, and make no assumptions about degree of renesting or double brood ing. A New Strategy For Estimating Breeding Population Size For Unmarked Populations The problems described above illustrate a need for the development of new methods for estimating the size of breeding bird populations (Figure 2 1). Visibility bias has been addressed elsewhere (Chapter 1) and has been well explored in the literature; therefore we will primarily focus here on methods to account for nesting asynchrony. This problem may be approached by estimating the turnover of individually marked nests and incorporating this information into a population estimate using a mark resight method. We have used the superpopulation approach (Schwarz and Arnason 1996), a variation on a Jolly Seber open population capture recapture model. In this case nests are treated as individuals in an animal population, where the population is equivalent to the estimated total numbers of nest starts during a period of interest such as a breeding season. We chose to study bias derived from nesting asynchrony in Great Egrets ( Ardea alba ) and White Ibises ( Eudocimus albus ) because these two birds are conspicuous colonial nesters with very different life histories and temporal nesting patterns. We believe that the methods outlined here provide a new and much improved technique for the monitoring and estimation of large unmarked populations of nesting birds.

PAGE 30

30 Methods Survey Methodology We conducted monthly aerial surveys between January and June of 2005 and 2006 in which we searched for and counted colonies of wading birds in t he central and northern Everglades (Broward, Dade and Palm Beach counties, FL). We conducted surveys in a Cessna Skyhawk (172) high wing aircraft flying at approximately 177 kph and 304 meters altitude, with one observer on each side of the aircraft. Whe n we located colonies we circled them for aerial visual estimates and aerial photography at 244 meters above ground level. We took photographs resolution digital camera with a 28 1 35mm image stabilizing lens. We edited photographs and counted nests using either Adobe Photoshop Elements version 2.0 or the shareware program Paint.Net version 2.0, using colored dots to mark nests as they were counted and delineating overlapping photog raph edges using colored lines. We accepted the largest single month count of the breeding season from these aerial photographs as the peak seasonal count for each species. roughly semiweekly basis of groups of individually identifiable nests. This allowed us to estimate turnover (losses of nests and new entries to the colony). Groups of birds were identified through proximity to natural or artificial landmarks (see below). We took photographs at an altitude of close to vertically as possible. We generally conducted survey flights between 08:00 and 10:30 am or 16:00 to 18:00 pm, but flight times varied with weather, plane availability, and other factors. Midday light was better for survey photographs, but turbulence, wind and poor weather also tended to be more severe in the heat of the day. We conducted these surveys in three small Great Egret colonies and one large mixed species colony (Table 2 1) in Water Conservation

PAGE 31

31 Area 3A (Figure 2 2) in one or both of the 2005 and 2006 breeding seasons. The mixed species colony, Alley North (Figure 2 3) was too large to conduct a comple te survey. We therefore concentrated on photographing certain landmarks that could be reliably located from the air. Artificial Geographic Markers The repeated identification of individual nests in aerial surveys required some kind of identifying landmark s. We used natural landmarks such as clearings within the tree island, but we also erected artificial markers in areas of colonies that were both accessible on the ground and where birds nested in substantial numbers. These markers were of three types, a ll of which were easily identifiable from 100 300 meters above ground level: A roughly 1 2 meter wide area of vegetation painted with a dilute white latex paint (1:3 water:paint) dispensed from a backpack sprayer of the type common in landscaping. These markers were only utilized during the 2005 season. A 3 meter tall, 10 cm diameter white vertically mounted PVC pipe with a 1.5 meter horizontal X on the top (Figure 2 4A). These markers were primarily utilized during the 2006 season. A 1x1 meter piece of white or pale colored cotton cloth, tied at the corners to vegetation in a horizontal position (Figure 2 4B). This type of marker was used exclusively during the 2006 breeding season. Photographic Analysis We used printed photographs of the same areas fro m different dates to identify individual nests and obtain presence absence data for each nest on each subsequent survey date. We uniquely numbered nests consecutively from the earliest date, and entered presence absence information into the database using the following format: Nest inactive (0) this included cases in which nest and parent/chicks were not visible; nest was visible but empty (no sign of parents/chicks); or there was no photo of the nest area available on that date Nest active (1) parent an d/or chicks were visible

PAGE 32

32 Since some nests were temporarily inactive or not visible in photos from a particular date, we only assumed that a nest had failed if it was found to be inactive on four consecutive survey dates. After that time, we treated any ne st in that location as a new nest start. Although we used the presence of a large white bird as an initial indication of nesting, these could have been roosting birds or birds temporarily away from their nests. Before we analyzed each presence absence da tabase, we therefore eliminated first observations of all nests from the database, and some small proportion of these single appearance nests were indeed nest s. In addition, the use of the four consecutive zero rule (as above) may have missed some nests that failed, restarted, and failed again within the period of four visits. However, both these types of possible errors seem unlikely given the relatively lon g courtship and egglaying periods of these birds (4 10 days for Great Egrets, according to McCrimmon et al. 2001 ; 9 10 days for White Ibises, as cited in Kushlan and Bildstein 1992) and the semiweekly frequency of aerial surveys. In any case, if either of these assumptions is incorrect, the effect is in the same direction, to underestimate the true numbers of nests. We eliminated the first sighting of nests that were seen on multiple dates, in addition to those nests seen only once, in order to avoid artificially inflating survival estimates. Indep endent observers. Three independent observers analyzed the same set of photographs for one colony (Vacation Island) from the 2006 breeding season. This allowed for a comparison of photo analysis methods between observers, and served as a test of the repe observers) as the final superpopulation estimate for the 2006 Vacation Island colony, as that observer also conducted the analyses for all other colonies.

PAGE 33

33 The Superp opulation Modeling Approach to Population Estimation The superpopulation approach (Schwarz and Arnason 1996) is a variation on a Jolly Seber open population capture recapture model that includes as a derived parameter the gross births within the population This parameter includes all animals that enter the population during the sample period and either survive until the next survey, or emigrate or die before they are available to be sampled (Schwarz and Arnason 1996, Schwarz et al. 1993) In our study, nests were equivalent to individuals in a population; the number of new nest s present at each survey identifiable, nests that failed between consecutive surveys indicated the level of emigration from ss superpopulation size represented the total number of nest starts over the entire sampling period. Detectability of nests is included in the model as an encounter probability term for each survey. We fit regression models to the capture recapture data f rom colonies using Program MARK version 4.3 (White and Burnham 1999) We used the POPAN data type (Arnason and Schwarz 1995) which utilizes a robust parameterization of the Jolly Seber model (Schwarz and Arnason 1996) We set time intervals to fractions of weeks between each set of consecutive pent, and s uperpopulation size N) to vary with time, depending upon the model. For example, for a database with six survey dates, there were a total of seventeen possible parameters (five survival interval probabilities, five entry interval probabilities, six encoun ter probabilities, and one superpopulation size). For each database, we tested a set of four candidate models: a general (fully time dependent) model, a so were held constant through time, and two models in which either encounter probability or survival probability was allowed to vary with time while the other was held constant (Cooch and

PAGE 34

34 White 2007) Probability of entry int o the population would not be expected to hold constant throughout the season, since for both species there tends to be a surge of initial nesting in an area followed by lower levels of nest initiation thereafter (McCrimmon et al. 2001 Kushlan and Bildstein 1992) Thus, entry probabilities wer e allowed to be time dependent in all models. For parameters in the model were estimable (Schwarz and Arnason 1996) We set p 1 =p 2 and p k =p k 1 so that all survival parameters were estimable in the model (J. Nichols pers. comm.). In the model in which survival was held constant and encounter probability varied, the i nitial p value was still inestimable, so for this model we set p 1 =p 2 and allowed all other encounter probabilities to vary. We used a sin link function to estimate survival and encounter parameters, a mlogit(1) function to estimate entry parameters, and a log link function to estimate superpopulation size N. The gross superpopulation size N* is a derived parameter of the POPAN model. It includes the net superpopulation size (all animals that enter the population between two consecutive surveys and are ava ilable to be captured during the second survey) as well as animals that enter and leave the population between consecutive surveys and thus are never available to be sampled. average values fo r all parameters across all models. Using the counts from each survey date and the estimated encounter and survival probabilities from each survey date or interval (adjusted for population between each consecutive set of surveys could be calculated (Schwarz et al. 1993, Schwarz and Arnason 1996) The gross superpopulation size in MARK is derived from summing these gross entries between each consecutive set of survey dates, and adding the sum to the estimated number of nests present during the first survey (after Schwarz and Arnason 1996, Schwarz et al. 1993):

PAGE 35

35 N i = n i /p i B i = N i+1 N i i )^t i B i = B i i ^t i ) i ^t i ) 1 N = N 1 k 1 B i where N i is the estimated total number of individuals in population at occasion i; n i is the number of individuals seen at occasion i; p i is the encounter prob. at occasion i; B i is the estimated number of individuals entering the population between sampling occasions i and i+1; i is the survival probability at occasion i; t i is the time between surveys i and i+1(as a proportion of a week); B i is the estimated gross number of individuals enterin g population between i and i+1; N* is the estimated gross superpopulation size; and k is the total number of surveys. absence database (Appendix A). We used an information theoretic approach t o model selection, and considered that a model with a (Williams et al. 2002) We quantified the goodness of fit (GOF) of the most general model using chi square tests for each survey interval to see if observed values varied from the expected number of surviving and encountered nests. These tests evaluated the following assumptions inherent in Jolly Seber models (Cooch and White 2007, Pollock et al. 1990) : There is no heterogeneity in capture probability, either among individuals or among cohorts (cohorts in this case meaning all nests that were seen for the f irst time on the same date). There is no heterogeneity in survival probability among individuals or cohorts.

PAGE 36

36 We calculated chi square values using Program RELEASE (Burnham et al. 1987) available within MARK. The chi square values generated for each survey interval were summed for the entire sampling period, and were divided by the degrees of freedom to obtain a variance inflation factor, c hat (a measure of overdispersion in the data). Following Cooch and White (2007), we accepted that a c hat value of 1 indicated good model fit, val ues of 1 3 indicated moderately good fit, and >3 indicated probable violation of model assumptions that is, that none of the models in the tested model suite may be a good fit for the data. A c hat value of less than 1 essentially means that the data are underdispersed, and there is little agreement in the literature about what this means or what to do about it (Cooch and White 2007) Following the recommendation of Cooch and White (2007), we used a c hat value of 1 when the c hat calculated in RELEASE was less than that. Otherwise, we multipl ied the calculated c hat values presented in the tables in Appendix A (c hat values of less than 1 are not presented) by the model based variance and covariance estimates for each model suite. This adjusted model weights to compensate for the overestimati on of precision caused by overdispersion. Subsampling Alley North Colony For the smaller Great Egret colonies (tree island lengths of approximately 120 200 meters), all or most of the colony could be covered in one or two photographic passes. For the ve ry large colony (length of tree island approximately 1900 meters; Figure 2 3), we subsampled the colony during aerial photography, resulting in a collection of geographically distinct year and species specific estimates of superpopulations with standard e rrors. Samples were surveyed for different lengths of time during the breeding season, depending upon nesting patterns and the availability of good quality photographs for each area. We calculated a superpopulation estimate for each sample, and compared it to the raw count for each sample, which was the number of

PAGE 37

37 count was assumed to be the number of nests in the sample unit that would have been seen and include d in the peak count on the peak count date. We assumed that the ratio between the nd an extrapolated superpopulation size for the entire colony. For example, if on average 70% of nests present during the season in the sample areas were not present at the survey date closest to the peak count (when the raw counts were taken), then the c olony peak count likewise would be assumed to be about a 70% underestimate of the total number of nest starts. We found the proportional difference between the superpopulation estimate and the raw count for each sample, and then took the average of these proportions across all samples. We obtaining an overall ratio between the two types of estimates for all samples (e.g., calculating the difference between the sum med raw counts and superpopulation estimates across all samples), as it is unlikely that there is one underlying level of nest turnover within the colony; different nesting cohorts within the colony are likely to have different levels of nest turnover at d ifferent times within the breeding season. The ratio of averages (summed value), while in some cases shown to be less biased and have a smaller variance estimator than the average of ratios (Rao 2005) weights all nests in all samples equally and assumes that there is one underlying ratio between raw counts and superpopulation estimates for the colony. The average of ratios recognizes that each sample may represent an area with a different ratio between the raw count and superpopulation estimate (due to date of cohort initiation, nest density, or other factors), and thus takes natural variation in turnover rate into account when estimating an overall proportion for the colony.

PAGE 38

38 The average ratio of this proportional error across all samples was our estimated colony wide proportional difference between the peak count and extrapo lated superpopulation count, as follows: i /RC i ))/n = (N*colony/PCcolony) N* i is the superpopulation estimate for sample i; RC i is the raw count for sample i; n is the i /RC i ))/n is the average proportion of the total number of nest starts that were seen in the raw counts. PCcolony superpopulation estimate. We also calculated the va riance of the ratio estimate, in order to calculate 95% confidence intervals for the colony wide superpopulation estimate (Appendix B). These superpopulations were compared to colony peak counts to determine the level of bias introduced into peak counts b y 1) asynchronous nesting activity, expressed through changes in the identities and numbers of nests present over the course of the season due to changes in survival and entry probabilities; and 2) imperfect detectability of nests in aerial surveys, expres sed for each survey term as an encounter probability. Subsampling to Determine Necessary Survey Frequency We subsampled the 2006 Cypress City dataset, a very complete survey dataset with 17 surveys over the course of two months, to determine if flying appr oximately twice a week as we did for this research is necessary for the success of the technique. We subsampled the dataset to manipulate both the number of surveys conducted within a specified time period and the regularity of the survey intervals. We ca lculated a superpopulation estimate for each subsampled dataset and compared these estimates to the inclusive superpopulation estimate for the colony, in order to examine the loss of estimation accuracy associated with reduced flight frequency.

PAGE 39

39 Results In terobserver Error in Tallying Individual Nest Histories From Photographs Three independent observers analyzed the same photographic library from surveys of the Vacation Island Great Egret colony during the 2006 breeding season (Figure 2 5). The peak count absence database resulted in bserver error rate was roughly 13%. Colony information presented in the superpopulation and combined results colonies. Superpopulation Estimates Superpopulation estimates for the smaller Great Egret colonies were 147% to 482% of the associated peak counts for the same colonies (Figure 2 6), suggesting that asynchrony in nest initiation has a large effect upon the likely number of nest starts. The extrapolated sup erpopulation estimates for Alley North White Ibis colonies in 2005 and 2006 and the Great Egret colony in 2006 were 213% to 300% of peak counts for the same breeding populations (Figure 2 7). In all eight cases, the peak counts for colonies were well outs ide the 95% confidence intervals for the superpopulation estimates (Table 2 2). The superpopulation estimates and raw counts for Great Egret and White Ibis samples in the very large Alley North colony are shown in Table 2 3. The peak count for 2005 was 12,750 White Ibis nests on March 20 th ; the closest superpopulation survey was also on the 20 th so the raw counts for each sample were taken from this date. Using the superpopulation estimates and raw counts for each sample as described above and in Appen dix B, we estimated the superpopulation size for the entire White Ibis colony to be 38,275 nests (1,941 nests). In 2006,

PAGE 40

40 the peak count for White Ibises was 13,566 nests on April 19 th The two closest superpopulation survey dates were on April 17 th and 21 st so we used the average of the two raw counts from these dates. We estimated the superpopulation size for the entire colony to be 29,287 nests (306 nests). The peak count for Great Egrets in 2006 was 1,193 nests on March 16 th The closest superpop ulation survey date for four samples was the 17 th Two samples were only partially surveyed during the season (surveys of these areas were initiated later in the season, as a new cohort of nests entered the area), and for these samples the closest survey date was the 21 st We estimated the superpopulation size for the entire colony to be 2,538 nests (32 nests). Although peak counts consistently underestimated numbers of nest starts, the degree of underestimation varied widely among colonies and years (Fi gure 2 8); the Cypress City Great Egret colony, for instance, was more asynchronous in 2005 than 2006, and thus the peak count in 2005 represented a much smaller proportion of the superpopulation derived estimate than did the peak count for 2006. However, at Vacation Island colony, only 25 kilometers away, the level of nesting asynchrony actually increased slightly from 2005 to 2006. Interestingly, although levels of nest turnover varied by season and location, turnover did not appear to vary consistently by species (Figure 2 8). Subsampling to Determine Necessary Survey Frequency There was a clear tradeoff between frequency of survey and accuracy of superpopulation estimate (Table 2 4). Although conducting survey flights 17 times over the course of a 3 m onth period provided the highest superpopulation estimates for the Cypress City colony in 2006, each survey flight cost roughly $180, and this frequency was expensive to sustain. Of the frequencies tested in Table 2 4, flying every five days provided the best combination of accuracy and number of nests added to the superpopulation estimate per additional survey flight. Superpopulation estimates (including the costs of weekly surveys and the labor costs involved in photographic

PAGE 41

41 analysis) appear to be aroun d five times as expensive as peak count estimates to produce (Appendix C). Discussion The estimates of numbers of nest starts that incorporated asynchrony and detectability bias were significantly different from the peak counts in all eight colonies and ye ars. Asynchrony in nest initiation and failure, rather than problems with nest detectability, is the major source of bias in our peak count estimates. In Chapter 1, we compared ground and aerial counts of nests and indicated that visual bias alone may ca use errors on the order of 12% underestimates and 70% overestimates for White Ibises and Great Egrets, respectively. Kushlan (1979), using apparently similar methods, found underestimates of 11% for White Ibises and 16% for Great Egrets. In general, such independent visual bias estimates are at least an order of magnitude smaller than our estimates of combined asynchrony and visual bias in this chapter, which ranged from 47% to 382%. Moreover, the detectability term in the superpopulation model, although functionally equivalent to an independent visual bias estimate, is actually smaller than such an independent estimate would be. The detectability term is based solely upon those types of visual bias that may be detected from repeated aerial surveys (e.g. adult birds temporarily off of nests, vegetative occlusion due to the angle of a particular photograph), and does not include sources of bias that may only be detected through a comparison of aerial and ground counts. Nests that are so heavily occluded by vegetation that they are never seen during any aerial survey, for instance, are not included in our superpopulation modeling, and species misidentification in aerial surveys may only be recognized through the comparison of aerial to ground counts. Thus if the estimates were truly equivalent, the difference between simple visual bias estimates and our combined estimates would probably be larger than is represented here. Regardless, visibility bias appears to affect estimates of nest numbers to a relati vely small degree. The vast majority

PAGE 42

42 of error in our combined estimates appears to be from the effects of asynchronous nesting and several times the peak c ount values for all colonies. We therefore address the importance of superpopulation modeling below solely in regard to the problem of turnover and asynchrony in nest initiation. Our results indicate that there was not a large difference between White Ibi s and Great Egret nest turnover rates. This was surprising, since the two typically have different reproductive phenologies, timing of nesting, and nest failure rates (Frederick and Collopy 1989a, Kushlan and Bildstein 1992, McCrimmon et al. 2001) However, the inter year comparisons (Figure 2 8) illustrated that the degree of turnover varied widely between years and, moreover, between colonies dur ing the same year (see also Frederick et al. 2006). This suggests that nest turnover rates are not constant and will have to be estimated on an individual colony basis for each season of interest. This is unfortunate, since it makes measuring nest turnov er much more difficult and costly. However, it is apparent from these results that the amount of bias introduced through the failure to incorporate nesting asynchrony into estimates is also unacceptably large. Estimates of total number of nest starts at a colony over the course of a season are a vast improvement over peak counts for two reasons. First, we believe that our estimates of total number of nest starts are much closer to the actual breeding population size for these colonies than are peak coun t estimates. Peak counts have been accepted as minimum breeding population size estimates for the purposes of developing an index of breeding activity between seasons, but (with good reason) they have never been presented as estimates of actual breeding p opulation size. Second, some authors have claimed that peak counts and related methods may be of use as indices of

PAGE 43

43 population growth (James et al. 1996 Link and Sauer 1998) even if they are not reliable estimates of true population size (Link and Sauer 1998 Link and Sauer 2002) Indices are useful indicators only if they reliably reflect true relative differences between periods of interest. I n the case of wading birds, we have shown that peak counts are not reliable indicators of relative differences between annual numbers of nests, and are reflecting different proportions of the actual breeding population size in different seasons. Given the variability in turnover between seasons, it appears likely that monitoring schemes using peak counts may be failing to serve their intended purpose of providing even a reliable index of population size. Superpopulation estimates of numbers of nests, in contrast, are comparable between years and may be used as an index of breeding activity. Since renesting may occur, the total number of nest starts is still not an accurate estimate of actual breeding population size; however, by examining the total numbe r of nest starts each season (along with an estimate of nest success), one can obtain a fairly good idea of reproductive conditions for wading birds in a given season. At Vulture colony in 2005, for instance, the estimate of total number of nest starts wa s almost five times the peak count. This colony was completely destroyed by a mammalian predator early in the season, and was repopulated with new nests five weeks later. The peak count does not provide an accurate depiction of breeding conditions within Vulture colony for this season, but our estimate of total number of nest starts, along with a rough estimate of nest failure rate, was more informative. We therefore feel that despite any shortcomings of the methods used (see below), the superpopulation technique of estimating nest effort in colonies is likely to be more accurate than the peak count method. Limitations and Assumptions There are several apparent biases and limitations that we have noted in the use of these estimation techniques. First, detectability estimates in superpopulation modeling do not include

PAGE 44

44 nests that were so heavily occluded by vegetation that they are impossible to see from the air (and thus are never included in any survey counts). The relative frequency of these completel y occluded nests may only be estimated by comparing ground and aerial counts of the same areas in colonies (Chapter 1). This inherent bias in the superpopulation aerial photographic method probably leads to a slight underestimation of true number of nest starts. This is particularly true for birds that typically nest subcanopy (e.g., White Ibises), as opposed to those nesting predominantly in the canopy (Great Egrets). In addition, the total number of nest starts in each colony was also probably underest imated because the first sightings of all nests were removed from a database before it was analyzed. Although most of the one were not true nest starts, the small proportion that were could cause a bias of unknown magnitude and l ead to an underestimate of true nest starts. We believe, however, that this error is likely to have been small in comparison to other sources. Not all of the colonies or samples within colonies were surveyed for the same period of time, due to photo quali ty and survey restraints. For example, many of the sample superpopulation estimates for the very large Alley North colony are likely to be underestimates of true numbers of nest starts, since only about 20% of samples were surveyed for the entirety of the breeding season. Surveys of a particular area were usually initiated when markers were placed in the area, except in cases where the survey area was identifiable in one or multiple survey photographs before the markers were erected, and surveys were halt ed when there were very few or no visible birds remaining in the area. Since there are often different nesting cohorts in a colony of this size, and nest initiation rates varied during the season, we did not feel that it was biologically justified to atte mpt to extrapolate superpopulation sizes for the entire survey period for samples that were partially surveyed. However, it must be noted that a more thorough

PAGE 45

45 colony wide, pre season marking methodology would enable more accurate estimates of superpopulat ion size (see below), and likely further increase superpopulation estimates for this large colony. Model fit The calculated c hat values for several of the model sets were larger than 3, indicating problems with model fit and possible violations of model a ssumptions. Survival probability is probably temporally heterogeneous in colonies, because nest survival varies among stages of nesting (Mayfield 1975, Frederick and Collopy 1989b, Torres and Mangeaud 2006) and as a result of environmental variability within the season (Frederick and Collopy 1989a) Likewise, encounter probability is unlikely to be the same for all nests, be cause nests that are located in taller vegetation are more highly visible from the air and more likely to be seen on every survey. Many of the violations of model assumptions appeared to be related to the test for homogeneity in capture probability. If t his heterogeneity was the result of visibility, as above, it might help in the future to categorize nests according to visibility and conduct analysis in MARK with visibility as a grouping variable. Fortunately, however, both survival and encounter probab ility estimates in Jolly Seber models tend to be robust to this type of heterogeneity, so long as average encounter probability is high (>0.5; Pollock et al. 1990) which it was for all colonies and samples we examined. It is also possible that our models fit better in reality than was indicated by the calculated c hat values in RELEASE. Due to the large sample sizes in our study, we had very good power for our goodness of fit tests to resolve even small differences between observed and expected values. We examined the chi square t ables in RELEASE and suggest that the differences found, while in some cases statistically significant, were probably not biologically significant in terms of numbers of surviving or encountered nests.

PAGE 46

46 Refinements of the superpopulation technique We reco mmend that the first two flights of the season be close together (within 2 4 days), as the first survey flight will be dropped from the database before analysis. This maximizes the usable period of observation. Aerial photographic surveys should also be initiated as soon as incubation begins for the majority of the colony. For wading birds, the end of breeding and onset of incubation corresponds to a decrease in noise and movement within the colony. From an aerial view, birds in incubation seem settled and sitting still; few are flapping around or displaying with any frequency (Frederick 2006) After the second survey flight, our analysis suggests that survey interval can be reduced to once every five to seven days without large loss of information. While some information can be obtained from less frequent surveys, it becomes increasingly difficult to accurately identify nests with greater time elapsed between sequential photographs. Very large colonies will require random or stratified random sampling designs in order to capture the full heterogeneity in ne st turnover among cohorts within the colony. Since landmarks must be used to identify nests, our sampling areas were not located randomly. Even for areas where we used artificial markers, the difficulty of moving through high density vegetation restricted our choice of sites. By comparing the ratio of superpopulation estimates to raw counts for all samples, we examined the relative proportion of nests missed in the peak count, rather than absolute numbers missed. However, nest turnover rate may vary with nest density, a possibility that has not been examined in wading birds. If true, in order for the average sample ratio of superpopulation estimate to raw count to hold true for the entire colony, our samples would have to be representative of the range a nd relative proportions of different nest densities within the colony. The ideal solution would be to distribute markers throughout the colony site prior to the beginning of the nesting season, in order to allow for random placement of samples.

PAGE 47

47 The desig n of such a marker must be 1) small enough to fit into small aircraft, 2) small enough to be deployed from the aircraft safely, 3) large and conspicuous enough to be easily visible from the air, 4) biodegradable, and 5) of a nature such that it will rest o n the canopy of nesting substrate. The colonies we studied were densely vegetated, remote, and very difficult to place markers in, and may represent a worse case scenario. For colonies that are more readily accessible, artificial landmarks might be more readily placed by hand. We consider the large sample sizes of nests obtainable through the use of the aerial photographic technique to be central to encompassing the inherently large variation in nest success and synchrony that exists in large colonies. I t is clear that the repeated photo technique represents a significant increase in manpower and cost by comparison with the peak count method (Appendix C), but we would suggest that there is no bargain to be had in using a cheaper method that yields uninter pretable results. If funds are limited, a potential strategy may be to survey intensively every two to three years, rather than on an annual basis. Incorporation of renesting We have focused in this paper on providing estimates of total numbers of nest starts occurring throughout a nesting season. Since nest failure is common (Frederick and Collopy 1989c) and the nesting season is long (3 4 months), renesting may be frequent. For this reason, the nesting population (numbers of pairs or breeding females) woul d be smaller than the numbers of nest starts (Piazza and Wright 2004) Since this difference could be substantial, we strongly recommend that this technique be used in conjunction with studies of renesting frequency. Implications of Results The results of this study indicate that breedi ng population sizes for colonially breeding birds may be considerably larger than previously supposed. However, actual numbers of breeding pairs are probably somewhat smaller than suggested by the estimated numbers of nest

PAGE 48

48 starts in this study, because of the probability of renesting (see above). Likewise, it should be clear that the larger estimates resulting from the methodology we used do not represent an actual increase in population, but rather an increase in our ability to measure the population. T he combined estimates of breeding population size allow us to more accurately measure inter annual and inter site variation, and thus to see variation in breeding effort that was previously masked by bias associated with peak counts or similar surveys. Th ese new estimates are therefore both more accurate and offer associated confidence estimates, which peak counts and related one time surveys lack. We suggest that peak or snapshot counts can be useful as indices in situations where the study species has high nest synchrony, or alternatively where sources of bias are measured on a seasonal basis and incorporated into estimates. However, our research indicates that unless demonstrated otherwise, snapshot surveys of avian breeding populations probably canno t be used effectively as indices of breeding population size. Moreover, there is no way to tell from snapshot counts alone just how biased these estimates of population size may be. In this study we have examined an extreme case (long nesting period, hig hly variable nest failure, potential for poor nest visibility), but many avian species probably show similar characteristics, if to a smaller degree. Given the strong biases that have now been demonstrated (this study, Frederick et al. 2006) even lesser d egrees of asynchrony are likely to alter estimates of breeding population size by a considerable amount.

PAGE 49

49 Table 2 1. Names and locations of study sites. WCA 3A is Water Conservation Area 3A, a large area of primarily sawgrass and wet prairie in the Ever glades controlled by the South Florida Water Management District. The city of Homestead lies south and slightly east of the majority of WCA 3A. Colony Location Latitude Longitude Alley North WCA 3A N 26 11.179 W 80 31.431 Cypress City WCA 3A N 26 07 .468 W 80 30.283 Vacation Island WCA 3A N 25 54.939 W 80 37.813 Vulture WCA 3A N 26 01.470 W 80 32.240 Homestead General Airport (KX 51) Homestead N 25 30.0 W 80 33.3

PAGE 50

50 Table 2 2. All colonies for which we followed nest fate through time in aerial survey photographs and calculated a superpopulation estimate. For smaller Great Egret colonies, number of aerial surveys conducted during the season and the final number of individually identified nests (database size) are listed. For Alley North colonies, surveys were conducted of samples within the colony rather than of the colony in its entirety; data from all samples is in Table 2 3. Peak counts are the maximum one time counts of number of nests in colonies, taken from a series of monthly aer ial surveys during the breeding season. Superpopulation estimates for the same colonies incorporate detectability bias and bias due to nesting asynchrony, and thus are much higher estimates of nest numbers than peak counts. Superpopulation estimates also have measures of uncertainty associated with them, which peak counts lack. Year Colony Species a Peak count Number of surveys in database b Database size (# nests) c Superpopulation estimate N* (LCI UCI) d 2005 Vulture GREG 121 7 499 5 83 (521 645) 2005 Vacation Island GREG 79 7 215 233 (226 240) 2006 Vacation Island GREG 155 21 474 480 (477 483) 2005 Cypress City GREG 107 6 244 268 (259 277) 2006 Cypress City GREG 173 17 249 254 (251 256 ) 2006 Alley North GREG 1,193 2,538 (2,474 2,601) 2005 Alley North WHIB 12,750 38,275 (34,392 42,157) 2006 Alley North WHIB 13,566 29,287 (28,674 29,899) a. GREG=Great Egrets; WHIB=White Ibises. b. Number of surveys in database does no t include first survey, as all first sightings of nests were eliminated from the database before analysis (see text). absence database after the first sightings were eliminated. d. N* =weighted model averaged superpopulation estimate, calculated in Program MARK. LCI UCI are 95% confidence intervals.

PAGE 51

51 Table 2 3. All samples of Great Egret and White Ibis populations from large colony (Alley North) for which we followed nest fate through time in aerial survey photographs and calculated a superpopulation estimate. Year Sample a Species b # surveys in database c N* (LCI UCI) d Raw count for sample e Proportional difference between N* and raw count f 2006 Q1 GREG 10 56 (54 59) 24 2.33 (2.25 2.46 ) 2006 Q2 GREG 9 59 (58 60) 28 2.11 (2.07 2.14) 2006 Q3 GREG 9 58 (56 60) 20 2.90 (2.80 3.00) 2006 GE SE 1 GREG 17 43 (43 44) 22 1.95 (1.95 2.00) 2006 GE SE 2 GREG 16 73 (73 74) 57 1.28 (1.28 1.30) 2006 GE SE 3 GREG 10 54 (51 58) 19 2.84 (2.68 3.05) 2005 IT 1 WHIB 5 51 (47 54) 28 1.82 (1.68 1.93) 2005 IT 3 WHIB 6 67 (63 70) 36 1.86 (1.75 1.94) 2005 GST 1 WHIB 8 19 (16 22) 3 6.33 (5.33 7.33) 2005 GST 2 WHIB 5 30 (25 34) 14 2.14 (1.79 2.43) 2005 GST 3 WHIB 8 12 (8 15) 4 3.00 (2.00 3.75) 2006 Q5 N W HIB 5 94 (93 95) 54.5 1.72 (1.71 1.74) 2006 Q5 S WHIB 5 69 (69 70) 38 1.82 (1.82 1.84) 2006 Q6 WHIB 9 118 (115 121) 41.5 2.84 (2.77 2.92) 2006 Q6 S WHIB 8 124 (121 127) 33.5 3.70 (3.61 3.79) 2006 Q7 WHIB 3 34 (33 35) 25.5 1.33 (1.29 1.37) 2006 Q8 WHIB 5 35 (32 38) 22 1.59 (1.45 1.73) 2006 Q9 WHIB 12 31 (28 34) 13.5 2.30 (2.07 2.52) 2006 Q10 WHIB 12 98 (96 101) 50.5 1.94 (1.90 2.00) a. Alley North samples are labeled for convenience by location and nearby landmarks. b. GREG=Great Egrets; WHIB=White Ibises. c. Number of surveys in database does not include first survey, as all first sightings of nests were eliminated from the database before analysis (see text). d. N*=weighted model average superpopulation estimate, calculated in Program MARK. LC I UCI are 95% confidence intervals. e. Raw counts are the numbers of nests seen in sample areas during the survey closest to the colony peak count date. If two survey dates were equally close to the colony peak count date, the average of the two raw count s from these surveys is presented here. f. Difference calculated as N*/Raw count. Proportion CIs calculated using CIs for N*.

PAGE 52

52 Table 2 4. Subsampling of Cypress City 2006 database to determine minimum survey frequency for superpopulation method. Peak co unt for same colony during this period is 173 nests. Database Number of surveys in database (not including first; see text) Database size (# nests) N* (LCI UCI) a Inclusive (flights ~twice/week, irregular spacing) 17 249 253 (250 255) Every third survey deleted after first two (irregular spacing) 12 229 235 (232 239) Flights every 5 days 12 240 247 (243 251) Every other survey deleted after first two (flights ~once/week, irregular spacing) 10 229 237 (233 241) Weekly flights (after first two surveys) 1 0 223 232 (228 236) Flights every ten days (after first two surveys) 7 214 224 (219 229) Bimonthly flights (after first two surveys) 6 194 207 (201 213) a. N*=superpopulation estimate. Superpopulation estimates and confidence intervals are weighted mod el averages.

PAGE 53

53 Figure 2 1. Types of data that should be incorporated into estimates of breeding population size.

PAGE 54

54 Figure 2 2. Satellite photograph of the modern Everglades, with major watershed regions outlined in yellow. All research presented in this paper was conducted in Water Conservation Area 3A (WCA 3A) in the central Everglades. Photo from http://sofia.usgs.gov/ ; available at http://sofia.usgs.gov/publications/circular/1275/images/cover landcvrdataspot.jpg

PAGE 55

55 Figure 2 3A). Alley North colony was subsampled in order to develop a superpopulation estimate, as it is too large to conduct photo analyses in its entirety.

PAGE 56

56 A. B. Figure 2 4. Two types of artificial landmarks used in this study. A) Markers made of 10 cm diameter PVC pipe pounded into the ground and anchored with a steel fence post, topped with a 1.5 mete r wide X. B) Markers made of 1x1 meter white cotton cloth, tied to vegetation.

PAGE 57

57 Figure 2 5. Multi observer comparison of superpopulation estimates for Vacation Island 2006 colony, derived from photographic analyses by three independent observers. estimates.

PAGE 58

58 Figure 2 6. Peak counts and associated superpopulation estimates for three Great Egret colonies in WCA 3A (2005 2006). The percentage difference between the two estimates for each colony and year are located above the bars.

PAGE 59

59 Figure 2 7. Peak counts and extrapolated superpopulation estimates for White Ibises in 2005 and 2006 and Great Egrets in 2006 at Alley North colony (WCA 3A). Each bar is labeled with the value (in estimated number of nests) it represents; the percen tage difference between the two estimates for each colony and year are located above the bars.

PAGE 60

60 Figure 2 8. The proportional difference between superpopulation estimates and peak counts for 2005 and 2006. GREG=Grea t Egret colony; WHIB=White Ibis colony.

PAGE 61

61 CHAPTER 3 REMOTELY MEASURED EFFECTS OF RESEARCHER DISTURBANCE ON GREAT EGRETS ( Ardea alba ) AND WHITE IBISES ( Eudocimus albus ) Introduction For colonially breeding birds, the disturbance created by researchers walking very close to or within the colony may affect reproduction, especially early in the nesting cycle (Bouton et al. 2005, Giese 1996, Ell ison and Cleary 1978, Tremblay and Ellison 1979) Indeed, the effects of disturbance on nest success may be more severe for colonially breeding bird species than for solitary nesters, due to the number of nests potentially affected by a single disturbance (Carney and Sydeman 1999, Burger 1981, Kadlec and Drury 1968, Vos et al. 1985) Potential effects of disturbance may include abandonment (Cairns 1980 Safina and Bu rger 1983, Carlson and Mclean 1996) re duced clutch or brood size (Ellison and Cleary 1978 Lock and Ross 1973) increased rate of predation (Ellison and Cleary 1978, Kury and Gochfeld 1975, Hunt 1972) reduced rate of new nest initiation (Tremblay and Ellison 1 979, Safina and Burger 1983), and reduced hatching and fledging success due to exposure of eggs and chicks to extreme temperatures (Hunt 1972, Ellison and Cleary 1978, Cairns 1980) In Great Blue Herons (Vos et al. 1985, Carlson and Mclean 1996) and various species of Pelecaniformes, Ciconiiformes and Charadriiformes (Rodgers and Smith 1995) ground disturbances in which people walk through colonies have been found to have a more severe effect than boating or mechanical disturbances. However, walk in surveys are often used to measure nest success (e.g., Cairns 1980, Frederick and Collopy 1989b, Hunt 1972) or to collect biological samples or other census information (Frederick et al. 1999, Rodgers et al. 2005, Lock and Ross 1973) If ground surveys themselves significantly decrease nest success or other reproductive parameters, it may defeat the purpose of the research (Rodgers and Burger 1981) However, at least in wading birds (Ciconiiformes), there is some debate as to what level of disturbance is biologically or demographically

PAGE 62

62 significant, and wh (Carney and Sydeman 1999, Erw in 1989) The effects of human disturbance are inherently difficult to measure, since measurement itself usually requires some level of human intrusion (Shields and Parnell 1986, Frederick and Collopy 1989b) While this has been addressed through the comparison of colonies with different rate s of human disturbance (e.g., Frederick and Collopy 1989b, Piatt et al. 199 0, Davis and Parsons 1991, Tremblay and Ellison 1979, Burger 1981) it is not clear that location effects can be entirely controlled in such studies. Comparison has also been made of areas within a single colony, but this has been limited to rare situations in which birds can be observed fr om distant vantage points (Pratt 1970, Hulka and Stirling 2000, Bouton 1999, Galbraith 1987) In this paper we describe a study of human disturbance effects as inferred by observations m ade from aircraft, comparing disturbed and undisturbed areas of the same colonies. As has been indicated by other studies on ciconiiform species, we predicted that areas with walk in disturbance would experience increased nest failure rates and a decrease d rate of new nest initiation. Methods Ground Transects We monitored nesting in two breeding colonies located in Water Conservation Area 3A of the Everglades (Broward County, FL) in March May of 2006. Both colonies were located in tree islands dominated b y willow ( Salix spp.) set within an emergent marsh of primarily sawgrass ( Cladium jamaicense ) and cattail ( Typha spp). Vacation Island colony (approximately 1 ha, located at N2554.939 by W 8037.813) was composed of roughly 90% Great Egrets ( Ardea alba ), though a small number of Great Blue Herons ( Ardea herodias ) and Anhingas ( Anhinga anhinga ) nested in the taller trees. This colony was distant from other sources of human

PAGE 63

63 disturbance (12.5 km to nearest road and 600 meters to nearest airboat trail) and w e believe our visits to be the only form of human disturbance to which this colony was exposed. The second colony, Alley North (N2511.179, W 8031.431), was a much larger (45 ha) mixed species colony that included Great Egrets, Snowy Egrets ( Egretta thul a ), Little Blue Herons ( Egretta caerulea ), Black crowned Night Herons ( Nycticorax nycticorax ), Tricolored Herons ( Egretta tricolor ), White Ibises ( Eudocimus albus ), and other species. The northern end of the colony (approximately 30 ha) was exclusively co mposed of White Ibis nests, and we studied two small areas within this single researcher human disturbance (3.2 km to nearest road and 4 00 meters to nearest airboat trail). One other research team was also working in the colony during the time period in which we conducted surveys. The majority of this research was based on remotely monitoring radiotelemetry signals from outside the tree island, but researchers did occasionally enter the colony as well. We used both aerial photography and repeated walk in censuses of marked nests to in census trail encompassed the n orthern 40% of the nesting area. The southern 60% of the tree island (all nests on foot during the breeding season. Walk in censuses of Alley North were located i n the northeastern portion of the colony, and encompassed less than 1% of the area of nesting. One 100 m 2 area was censused using walk in methods only once, and then aerial methods thereafter; another 200 m 2 area was censused four times on the ground in a ddition to aerial surveys. The undisturbed (single visit) area used in this study was located approximately 200 meters south of

PAGE 64

64 the disturbed site; apart from the initial visit to erect markers to delineate the sample area, we did not come closer to this control site than 200 meters during the breeding season. We began walk in censuses after incubation was already underway for most nests. We approached the edge of the colony via airboat to within 15 meters (for Great Egrets) or 75 meters (for White Ibises ) of the closest nest. Nest checks were conducted every 5 7 days, and each visit to a colony took roughly one hour with 2 5 people present on each visit. Each nest was briefly examined for nest contents, using a mirror pole if necessary, but the contents were not removed. If abandonment was suspected, the eggs were sometimes touched to determine temperature. Nests were monitored until abandonment (all eggs in nests were cold) or failure (eggs were broken or gone from nest, or chicks were dead or gone fr om nest before fledge date). Aerial Survey Methodology We conducted aerial photographic surveys of nests that were individually identifiable from the air in order to examine nest turnover rates. Surveys were conducted twice a week at irregular intervals during the breeding season in a Cessna Skyhawk (172) aircraft, and photographs were taken using a Canon EOS 20D high resolution digital camera with a 28 135mm image stabilizing zoom lens. Camera settings were primarily on auto. We took photographs from an was removed for all survey flights so that photographs could be taken as close to vertically above nests as possible. We conducted survey flights between 08 :00 to 10:30 am or 16:00 to 18:00 pm, but flight times varied with weather, plane availability, and other factors. To allow for the identification of walking routes from the air, we erected artificial landmarks around the outer perimeter of each walk in survey route. For Vacation Island, markers were placed within 1 3 meters of the edge of the transect path at roughly 15 meter

PAGE 65

65 X at the top of the PVC markers (below) in each photograph as a metric for scaling ground distances in the colony. For the Alley North colony, we placed a marker at the four corners of both th in surveys The markers were of two types: A 3 meter tall, four inch PVC pipe with a 1.5 meter X made of PVC attached to the top of the pole (Figure 3 1A). A 1x1 meter piece of white or pale colored cotton cloth tied at the corners to vegetation (Figure 3 1B). These were placed in sheltered areas where they would be less likely to fla p in the wind. Photographic Analysis We edited photos in Adobe Photoshop Elements version 2.0. Printed copies of photos were used to identify individual nests and obtain presence absence data for each nest on each subsequent survey date. Nests in the p hotos were individually numbered and presence absence data for nests was entered into the database using the following format: Nest inactive (0) this included cases in which nest and parent/chicks were not visible; nest was visible but empty (no sign of p arents/chicks); or there was no photo of the nest area available on that date Nest active (1) parent and/or chicks were visible We compared photos of the date we were currently working on with photos of all previous dates to ensure that all nests were as signed unique and non overlapping identities. We used a conservative definition for nest failure in the photographic censuses, due to visibility constraints and because some nests were temporarily unoccupied at the moment the picture was taken. Nests tha t were apparently inactive for four consecutive survey dates were categorized as failed, and nests in the same location were thereafter treated as new nest starts. Although we used the

PAGE 66

66 presence of a large white bird as an initial indication of nesting, th ese could have been roosting birds or birds temporarily away from their nests. Before we analyzed each presence absence database, we therefore eliminated first observations of all nests from the database, and so deleted were seen only once. Although it is possible that some proportion of these single tally birds were actually short lived nest starts, this was considered to be a rare occurrence given the amount of time required for breeding and nest building in these spec ies (4 10 days for Great Egrets, according to McCrimmon et al. 2001 ; 9 10 days for White Ibises, as cited in Kushlan and Bildstein 1992) We eliminated the first sighting of nests that were seen on multiple dates (in addition to those nests seen only once) in order to avoid artificially inflating survival estimates. Analysis of Presence Absence Databases: A Mark Recapture Scenario We used aerial photographs of individually identifiable nests and a variation o n a Jolly Seber (JS) open population capture recapture model called the superpopulation method (Schwarz and Arnason 1996) in order to estimate rates of nest initiation and abandonment in large colonies of nesting birds (see also Chapter 2). JS models esti mate survival and encounter probabilities for marked animals, and the superpopulation approach also estimates the net immigration into a population. The application of the superpopulation method to nests allows for the monitoring of nests initiated in a c olony during a period of interest, such as a breeding season. In our study, nests were equivalent to individuals in a population; the number of new nests present at each survey represented immigration into the population, and the number of dividually identifiable, nests that failed between consecutive surveys indicated the level of emigration from the population. Encounter probabilities indicate the chance of seeing a nest on a particular survey date, given that it has survived and is actua lly present.

PAGE 67

67 The first survey of the colony was deleted from the database, since due to the first observation nest adjustment (described above), it contained only zeroes. The presence absence database input into MARK thus started with the second survey date. We fit regression models to the capture recapture data from colonies using the POPAN data type (Schwarz et al. 1993, Schwarz and Arnason 1996) in Program MARK version 4.3 (White and Burnham 1999) We set the time intervals to fractions of weeks between each set of consecutive surveys and assigned all nests to either the disturbed or undisturbed attribute group in the model. The four variables allowed to vary with time vary by attribute group, or some combination thereof, depending upon encounter probability, and entry probability were not allowed to vary by group or to change throu gh time. N, population size, was allowed to vary with group for all models, as the number of nests seen over the course of the season was different in each location (n=279 for disturbed area and n=194 for undisturbed area of Vacation Island colony; n=163 for disturbed area and n=201 for undisturbed area in Alley North colony). For alternative models in which survival estimable (Schwarz and Arnason 1996) We set p 1 =p 2 and p k =p k 1 for each group (if group was included in the model) so that all survival parameters were estimable (J. Nichols pers. comm.). In the mo dels in which survival was held constant or varied by group and encounter probability was time dependent, the initial p value is still inestimable, so for this model we set p 1 =p 2 for each group and allowed all other encounter probabilities to vary. We use d a sin link function to estimate survival and encounter parameters, mlogit functions to estimate entry parameters, and a log link function to estimate population size N of each group.

PAGE 68

68 peting models when testing the validity of our predictions (Williams et al. 20 02) Each model had an associated weights that expressed the relative probability of each model being the best fit for the data. We ively well supported models, while a (Williams et al. 2002) for all parameters across all models. Additionally, we quantified the model fit of the most general model in the tested suite, using chi square tests for each survey interval, to see if observed values varied from the expected number of surviving or encountered nests. These tests evaluated the following two assumptions inherent in Jolly Seber models (Pollock et al. 1990 Cooch and White 2007) : that there is no heterogeneity in cap ture probability within groups, either among individuals or among cohorts (cohort in this case meaning all nests that were seen for the first time on the same date); and that there is no heterogeneity in survival probability within groups, either among ind ividuals or cohorts. We calculated chi square values using Program RELEASE (Burnham et al. 1987) available within Program MARK. The chi square values generated for each survey interval were summed for the entire sampling period, and were divided by the degrees of freedom to obtain a variance inflation factor (c hat). Following Cooch and White (2007), we interpreted a c hat value of 1 to indicate good model fit, values of roughly 1 3 to indicate moderately good fit, and >3 to indicate probable violation of model assumptions. We adju sted model weights for the Great Egret colony using the calculated c hat for the dataset, so the information criterion values presented for these models are thus quasilikelihood (QAICc) values. This modification adjusts

PAGE 69

69 for the overestimated precision of parameter estimates caused by overdispersion in the data (Burnham et al. 1987; Burnham and Anderson 1998; Williams et al. 2002) The calculated c hat value for the White Ibis co lony was <1, which essentially means the data were underdispersed. As there is little agreement in the scientific literature as to what this means or what to do about it, we followed the recommendation of Cooch and White (2007) in using a c hat value of 1 for model weighting in this case. Results Vacation Island Great Egret Colony After adjusting the model selection used in MARK for the calculated c hat value (c hat=3.21), two models were equally supported by the data (Table 3 1). All other models in the model testing suite had less than 1% support (Appendix D for full list). Due to this relatively high c hat value, models with fewer parameters were much more highly supported; in the two best models, both encounter probability and survival probability wer e held constant over time. Time had an important effect on probability of entry into the population (new nests being initiated) in both models. Disturbance category was also retained as an important covariate in one model, though its effect was additive rather than multiplicative. The undisturbed group had slightly higher probability of entry into the population across all time periods than did the disturbed nest group (Figure 3 2). According to the most parsimonious model, this difference did not vary appreciably with the survey interval, regardless of how the entry probabilities themselves changed. Although it appears that there may have been an interaction with time on a linear scale (Figure 3 2), the interaction was not included in the best models, either because the interaction was examined on a logit scale or because including an interaction term would have added an additional 19 parameters to the model, and consequently was not supported by the QAICc model selection criteria.

PAGE 70

70 Alley North White Ibis Colony Two models were well supported by the data, and several others had some moderate level of support (Table 3 2; Appendix E for list of all tested models). Time had an important effect on all three population parameters in all supported models, b ut the interaction of time with group effect varied between models. All supported models indicated some effect of disturbance group on survival and encounter probabilities, and the majority of the models also indicated a disturbance group effect on entry probability (Table 3 2). The two well supported models both had group effect as an additive effect on survival probability. According to the best fit models, this difference did not vary appreciably with survey interval, regardless of how the survival pr obabilities themselves changed. Encounter probability varied with both group and time in an interactive fashion in the two best models. For the best fit model, entry probability was time dependent but did not vary with group; for the next best model (wit indicating some support for the model), entry probability varied with time and group in an additive manner, as for survival above. Comparison of model averaged encounter probabilities between disturbance groups for the Alley North Whi te Ibis colony showed that encounter probability did not differ consistently according to disturbance regime. In some survey intervals encounter probability was higher for the disturbed group, and for others it was higher for the undisturbed group. Encou nter probability seemed to vary widely among survey intervals and between geographic areas, regardless of level of researcher disturbance. Comparison of model averaged survival estimates indicated that survival probability was higher in the disturbed area than the undisturbed area (Figure 3 3, Table 3 3). Entry probability in the disturbed area was lower than or equal to that in the undisturbed area (Figure 3 4, Table 3 3). Although the absolute differences in probabilities

PAGE 71

71 between the two groups were co nsiderably larger for survival estimates than for entry probability estimates, entry probabilities showed a proportionally larger effect of disturbance (Table 3 3). Discussion Nest success did not appear to be negatively impacted by researcher disturbance during incubation and nestling periods for either Great Egrets or White Ibises. However, model averaged parameter estimates indicated that nest initiation was lower in the disturbed areas for both species when compared to control sites. This suggests tha t breeding birds may be sensitive to disturbance in the colony early in the nesting cycle. Violations of Model Assumptions. The c hat value for the Vacation Island (Great Egret) model suite was larger than 3, indicating potential problems with model fit. There are biologically sound reasons why model assumptions regarding encounter and survival homogeneity may be false for nesting wading birds, and the lack of model fit we saw for this dataset may be partially due to one of these factors. Fortunately, h owever, both survival and encounter probability estimates in Jolly Seber models tend to be robust to this type of heterogeneity, so long as average encounter probability is high (e.g., >0.5; Pollock et al. 1990) ; for Vacation Island, model averaged encounter probability was 0.945, a nd for Alley North, it was 0.823. Effects of Disturbance on Nest Survival According to the two best supported models for the Great Egret colony, researcher disturbance had no significant effect on nest survival or encounter probability. Researcher disturb ance after the early egg laying stage appears to cause no appreciable decrease in nest success for Great Egrets. This agrees with similar findings of lack of nest failure due to human disturbance later in the nesting cycle in Tricolored Herons, Egretta tr icolor ( Frederick and Collopy 1989b) Black crowned Night Herons, Nycticorax nycticorax ( Parsons and Burger

PAGE 72

7 2 1982) and Snowy Egrets, Egretta thula ( Davis and Parsons 1991) Although walk in surveys do affect nest success for many taxa, perhaps resistance to disturbance in later nest stages is a common characteristic for wading birds. Likewise, ou r single or multiple site visits to the White Ibis colony during incubation and early nestling stages did not appreciably decrease nest success. Moreover, nest survival was significantly higher in the disturbed area of the White Ibis colony than in the un disturbed area. This result is counter to generally accepted theory regarding the effects of walk in ground disturbance on colonially nesting birds, but may be attributable to one of two several possible causes. First, the nests in the two sites may have belonged to different cohorts; based on the single walk in survey of the undisturbed area, the average nesting stage appeared to be several days to a week behind the disturbed area. Nest initiation date has been found to be very important for nest succes s in White Ibises, presumably due to environmental variables associated with initiation date (Frederick and Collopy 1989a) A second potential cause of higher nest ano ther team of researchers studying White Ibises at the same colony, and their frequency of visitation to different areas may have affected nest survival rates. Thirdly, it is possible that researcher presence may have discouraged the presence of predators in that area of the colony, although we consider this possibility to be unlikely, as researcher disturbance has been postulated to actually facilitate predation at a White Ibis colony in North Carolina (Shields and Parnell 1986). Effects of Disturbance on Nest Initiation It makes biological sense for entry probability to be time dependent in all supported models, as for both species there tends to be a surge of initial nesting in a given area followed by lower levels of nest initiation thereafter (McCrimmon et al. 2001, Kushlan and Bildstein 1992)

PAGE 73

73 However, for both Great Egrets and White Ibises, one of the two best models also included disturbance group as a factor in entry probability. For Great Egrets, the two models were so equally weighted that for all practical purposes they were indistingui shable using QAICc. This lends some support to the hypothesis that regular researcher disturbance causes depressed levels of nest initiation as compared to areas where nests remained undisturbed. Nest initiation rates are not often measured by researcher s studying effects of disturbance (e.g., Carlson and McLean 1996, Burger et al. 1995), but decreased initiation has also been seen in Black Skimmers, Rynchops niger (Safina and Burger 1983) and Black crowned Night Herons, Nycticorax nycticorax (Tremblay and Ellison 1979) Depressed nest initiation may not be univer sal; Frederick and Collopy (1989b) presented evidence that Tricolored Heron nest initiation may not be affected by researcher disturbance, although this question was not a primary focus of the research. However, the effect of disturbance on nest initiatio n likely varies by species and frequency of visitation (Gtmark 1992) and the evidence for both Great Egrets and White Ibises indicates that further research is warranted on this topic. We have found some effect of researcher disturbance on nest initiation in wading birds, but no discernible effect on nest success This suggests that wading birds are vulnerable to disturbance early in the nesting period when they are considering nesting locations, but that they may be less sensitive once nesting has begun. It is important to note, however, that disturbance may af fect choice of nesting area in the future, a possibility that has not been examined in wading bird species. Even when birds do not overtly react to disturbance during the current breeding season, they may avoid areas in ensuing seasons where they have bee n disturbed in the past, as has been found in several raptor species (Platt 1977 as cited in Gtmark 1992 ; White and Thurow 1985) and for Adlie penguins ( Pygoscelis adeliae ) in Antarctica (Wilson et al.

PAGE 74

74 1990 Young 1990; but see Fraser and Patterson 1997) Renesting rate may also be reduced in disturbed areas (Gtmark 1992) but renesting rates in wading birds are unknown. Perhaps more important than its evidence for disturbance effects, this study serves to reinforce earlier work in using remote means to infer effects of dist urbance. We hope this new and widely applicable method of remotely measuring disturbance will prove useful for future research on this topic.

PAGE 75

75 Table 3 1. Characteristics of the two best fit, lowest AIC models describing population parameters for Great Eg ret nests in the central Everglades, Florida. Model parameters are probability of survival (phi), encounter probability (p), probability of entry into the population (pent), and size of population (N). Each parameter was held constant over the entire sur vey period (.), held constant for each group separately (group), allowed to vary with time (t), allowed to vary with both time and group (t*g) or allowed to vary with both time and disturbance group in an additive fashion (t+group). Both models are about equally supported. Model 1 does not include disturbance group as an important factor in the probability of new nests entering the population over time; the second model does include the group parameter. Only models with greater than 1% support are includ ed. Model Delta QAICc QAICc weights # parameters {p(.)phi(.)pent(t)N(group)} 0 0.50147 22 {p(.)phi(.)pent(t+group)N(group)} 0.0119 0.4985 23 Table 3 2. Characteristics of the best fit, lowest AIC models describing population parameters for White Ib is nests in the central Everglades, Florida. Model parameters are probability of survival (phi), encounter probability (p), probability of entry into the population (pent), and size of population (N). Each parameter was held constant over the entire surv ey period (.), held constant for each group separately (group), allowed to vary with time (t), allowed to vary with both time and group (t*g) or allowed to vary with both time and disturbance group in an additive fashion (t+group). Both highly supported m odels ( <2) include disturbance group as an additive effect on nest survival and a multiplicative effect on encounter probability. There is some support for the importance of disturbance group on entry probability, but the most highly supported model (55% su pport) does not include disturbance group as an important factor in entry probability. Only models with greater than 1% support are included. Model AICc weights # parameters {phi(t+group)p(t*group)pent(t)N(group)} 0.000 0.545 19 {phi(t+group)p(t*g roup)pent(t+group)N(group)} 1.697 0.233 20 {phi(t+group)p(t+group)pent(t)N(group)} 3.939 0.076 16 {phi(t*group)p(t*group)pent(t)N(group)} 4.111 0.070 23 {phi(t*group)p(t*group)pent(t+group)N(group)} 5.783 0.030 24 {phi(t+group)p(t*group)pent(t*group)N( group)} 6.108 0.026 21 {phi(t*group)p(t*group)pent(t*group)N(group)} 7.420 0.013 26

PAGE 76

76 Table 3 3. Comparison of model averaged survival and entry probabilities between disturbance groups for the Alley North White Ibis colony. Phi=survival probability; p ent=entry probability; interval number indicates survey interval (e.g., interval 1 denotes the time period between surveys 1 and 2). The percent different in daily probabilities between groups indicates the difference in the percentage of nests that survi ve or are initiated in the disturbed group in comparison to the undisturbed group; a negative difference indicates that the parameter estimate for the disturbed group was lower than that for the undisturbed group. Daily probabilities are calculated for ph i as ((disturbed weekly phi) 1/7 /(undisturbed weekly phi) 1/7 ) 1 x 100. For pent, this value was calculated as ((disturbed weekly pent) 7 /(undisturbed weekly pent) 7 ) 1 x 100. Parameter % Difference Between Disturbed and Undisturbed Groups (Daily Probabili ties) phi (interval 1) 2.72 phi (interval 2) 1.57 phi (interval 3) 2.73 phi (interval 4) 3.96 phi (interval 5) 11.00 pent (interval 1) 29.75 pent (interval 2) 48.99 pent (interval 3) 48.01 pent (interval 4) 0.00 pent (interval 5) 0.00

PAGE 77

77 A. B. Figure 3 1. Two types of artificial landmarks used in this study. A) Markers made of 10 cm diameter PVC pipe pounded into the ground and anchored with a steel fence post, topped with a 1.5 meter wide X. B) Markers made of 1x1 meter white cotton cloth, tied to vegetation.

PAGE 78

78 Figure 3 2. The difference in entry probabilities between disturbance groups at Vacation Islan d Great Egret colony as a function of time. Estimates are from second supported model, phi(.)p(.)pent(t+group)N(group). Y axis is the probability of new nests being initiated in the undisturbed area, minus the probability of new nests entering the distur bed area, over each survey interval. Positive differences indicate the disturbed group had lower probability of new nest formation than the undisturbed group. Differences between groups are slight but remain positive over all survey intervals.

PAGE 79

79 Figure 3 3. Model averaged weekly survival probabilities for both disturbance groups at Alley North White Ibis colony as a function of survey interval (time). The disturbed group had higher nest survival across all survey i ntervals than the undisturbed group.

PAGE 80

80 Figure 3 4. Model averaged weekly entry probabilities for both disturbance groups at Alley North White Ibis colony as a function of survey interval (time). The undisturbed grou p had higher probability of nest initiation for the first three survey intervals. Both groups had near zero entry probability for the last two survey intervals.

PAGE 81

81 APPENDIX A SUPERPOPULATION MODE L OUTPUT (POPAN DATA TYPE) FOR ALL COLONI ES Table A 1. For each colony or sample, we tested a set of four candidate models: a general (fully time dependent) model, a so variabl es were held constant through time, and two models in which either encounter probability or survival probability was allowed to vary with time while the other was held constant (Cooch and White 2007) Probability of entry into the population (pent) would not be expected to hold constant throughout the season, since for both species there tends to be a surge of initial nesting in an area followed by lower levels of nest initiation the reafter Thus, entry probabilities were allowed to be time dependent in probability p varied with time, not all parameters in the model were estimable. We set p 1 =p 2 and p k =p k 1 so that all survival parameters were estimable in the model. In the model in which survival was held constant and encounter probability varied, the initial p value was still inestimable, so for this model we set p 1 =p 2 and allowed all other encounter probabilities to vary. We used a sin link function to estimate survival and encounter parameters, a mlogit(1) function to estimate entry parameters, and a parameter is general (no t restricted to be equal over time); a dot notation means that the parameter is restricted to be the same over all survey intervals. Models that were completely unsupported by the data (AICc weight of zero) are not included in the table. C hat is a measu re of the goodness of fit of the most general model; when c hat was greater than 1 (indicating a lack of model fit), it was used to adjust the model based variance and covariance estimates, and thus affected the relative model weights within the tested sui te. C hat values of less than 1 are not presented in the table. Year Colony/ sample Species c hat 2005 Vulture GREG Model AICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 0.82215 581.6979 34.8814 {phi(t)p(.)pent(t)N} 3. 062 0.17785 589.5230 17.2920 2005 Vacation Island GREG 3.05 Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.87187 233.5631 6.6651 {phi(t)p(.)pent(t)N} 4.3784 0.09765 228.1620 5.9515 {phi(.)p(t)pent(t)N} 7.5307 0.0 2019 229.5506 4.9093 {phi(t)p(t)pent(t)N} 8.8803 0.01028 227.1754 5.0314 2006 Vacation Island GREG 5.90 Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 1 479.9585 3.2767

PAGE 82

82 Table A 1. Continued. Year Colony/ sample Species c hat 2005 Cypress City GREG 2.13 Model QAICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 0.54436 267.7905 6.5382 2006 Cypress City GREG 3.03 Model QAICc weights Estimated N* SE {p(.)phi(.)pent(t)N} 0 0.9909 253.648 6 2.1806 {phi(.)p(t)pent(t)N} 9.3855 0.00908 253.7785 2.1385 {phi(t)p(t)pent(t)N} 21.5337 0.00002 252.8228 2.1116 2005 Alley North/IT 1 WHIB Model AICc weights Estimated N* SE {phi(.)p(t)pent(t)N} 0 0.83334 50.6863 1.9522 { phi(t)p(t)pent(t)N} 4.4872 0.0884 50.6636 1.8088 {phi(t)p(.)pent(t)N} 4.7363 0.07805 50.4216 1.5666 {phi(.)p(.)pent(t)N} 16.5396 0.00021 52.3654 2.1018 2005 Alley North/IT 3 WHIB Model AICc weights Estimated N* SE {phi(.)p(t)pen t(t)N} 0 0.96697 66.6686 1.9037 {phi(t)p(.)pent(t)N} 7.6176 0.02144 66.6758 1.9551 {phi(t)p(t)pent(t)N} 8.8484 0.01159 66.6722 1.9652 2005 Alley North/ GST 1 WHIB Model AICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.9 9953 19.0817 1.6889 {phi(t)p(.)pent(t)N} 15.8962 0.00035 19.5700 2.1732 {phi(.)p(t)pent(t)N} 18.1392 0.00012 19.3125 1.9709 2005 Alley North/ GST 2 WHIB Model AICc weights Estimated N* SE {phi(t)p(.)pent(t)N} 0 0.41299 29.8978 2.6 669 {phi(.)p(.)pent(t)N} 0.0292 0.407 29.3304 2.2952 {phi(t)p(t)pent(t)N} 2.0753 0.14632 29.2213 1.8967 {phi(.)p(t)pent(t)N} 5.0119 0.0337 28.2547 1.8534 2005 Alley North/ GST 3 WHIB Model AICc weights Estimated N* SE {ph i(.)p(.)pent(t)N} 0 1 11.5185 1.9025

PAGE 83

83 Table A 1. Continued. Year Colony/ sample Species c hat 2006 Alley North/Q1 GREG Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.99628 55.9440 1.1204 {phi(.)p(t)pent(t)N } 11.1937 0.0037 56.1124 1.2738 {phi(t)p(t)pent(t)N} 21.4261 0.00002 56.0022 1.0206 2006 Alley North/Q2 GREG Model AICc weights Estimated N* SE {phi(.)p(t)pent(t)N} 0 0.64806 59.0446 0.3013 {phi(t)p(t)pent(t)N} 2.1356 0.22278 58 .6230 0.2728 {phi(t)p(.)pent(t)N} 3.256 0.12723 58.7090 0.9285 {phi(.)p(.)pent(t)N} 11.6317 0.00193 59.4892 1.1264 2006 Alley North/Q3 GREG 1.83 Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.9937 58.1478 1.4997 {phi(.)p(t)pent(t)N} 10.2855 0.0058 58.0003 0.6814 {phi(t)p(.)pent(t)N} 15.2309 0.00049 58.1257 1.5748 2006 Alley North/ GE SE 1 GREG 1.18 Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.99522 43.4679 0.1566 {phi(. )p(t)pent(t)N} 10.678 0.00478 43.4806 0.1608 2006 Alley North/ GE SE 2 GREG 1.53 Model AICc weights Estimated N* SE {phi(.)p(t)pent(t)N} 0 0.99995 73.4119 0.1206 {phi(t)p(t)pent(t)N} 19.8337 0.00005 73.5128 0.1368 2006 Alley North/ GE SE 3 GREG Model QAICc weights Estimated N* SE {phi(.)p(.)pent(t)N} 0 0.96247 54.2707 1.7800 {phi(.)p(t)pent(t)N} 7.8966 0.01856 54.5926 2.0213 {phi(t)p(.)pent(t)N} 7.9083 0.01846 54.2082 1.7665 {phi(t)p(t)pent(t)N} 15.085 8 0.00051 54.8939 2.1892 2006 Alley North/ Q5 N WHIB 1.46 Model QAICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 0.60102 94.3730 0.2798 {phi(t)p(.)pent(t)N} 0.9321 0.37712 94.4772 1.2067 {phi(.)p(t)pent(t)N} 6.6281 0.02186 94.2479 0.1886

PAGE 84

84 Table A 1. Continued. Year Colony/ sample Species c hat 2006 Alley North/ Q5 S WHIB Model AICc weights Estimated N* SE {phi(.)p(t)pent(t)N} 0 0.58865 69.2716 0.1492 {phi(t)p(t)pent(t)N} 0.8875 0.37769 69. 2778 0.1658 {phi(t)p(.)pent(t)N} 5.7235 0.03365 69.1884 0.1357 2006 Alley North/Q6 WHIB 2.05 Model QAICc weights Estimated N* SE {phi(t)p(.)pent(t)N} 0 0.95935 117.7637 1.9492 {phi(t)p(t)pent(t)N} 6.9465 0.02976 117.8758 1.954 8 {phi(.)p(t)pent(t)N} 8.9562 0.01089 136.0672 47.3142 2006 Alley North/Q6 S WHIB 2.12 Model QAICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 0.91789 123.7821 2.0607 {phi(.)p(t)pent(t)N} 5.5923 0.05603 123.6286 2.2899 {phi(t)p(.)pent(t)N} 7.1216 0.02608 123.3685 1.9028 2006 Alley North/Q7 WHIB Model AICc weights Estimated N* SE {p(.)phi(.)pent(t)} 0 0.44667 34.1004 0.5192 {p(.)phi(t)pent(t)} 0.2508 0.39403 34.3604 0.7027 {phi(t)p(t)pent(t) N} 2.694 0.11614 34.3604 0.7027 {phi(.)p(t)pent(t)N} 4.674 0.04316 34.1004 0.5192 2006 Alley North/Q8 WHIB 1.00 Model AICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 0.75027 34.9722 1.1473 {phi(.)p(t)pent(t)N} 3.0504 0.16 324 35.9532 2.7331 {phi(t)p(.)pent(t)N} 4.3348 0.08589 35.1331 1.9966 {phi(.)p(.)pent(t)N} 14.2487 0.0006 36.2411 2.0097 2006 Alley North/Q9 WHIB Model AICc weights Estimated N* SE {phi(.)p(t)pent(t)N} 0 0.99985 30.9664 1.5501 {phi(t)p(.)pent(t)N} 17.7047 0.00014 31.1015 1.2478 {phi(t)p(t)pent(t)N} 24.3684 0.00001 31.0495 1.1575 2006 Alley North/Q10 WHIB 1.15 Model AICc weights Estimated N* SE {phi(t)p(t)pent(t)N} 0 1 98.3859 1.2228

PAGE 85

85 APPENDIX B CALCULATION OF STAND ARD ERRORS FOR EXTRA POLATED SUPERPOPULAT ION ESTIMATES FROM ALLEY NORTH SAMPLES We conducted aerial surveys and photo analysis of sample areas from the large Alley North wading bird colony (WCA 3A) in 2005 (White Ibises) and 2 006 (White Ibises and Great Egrets). Samples were surveyed for different amounts of time, based upon where nesting was occurring and the availability of good quality photographs for each area. We calculated a gross superpopulation estimate (N*) for each POPAN data function. We took the number of nests present in each sample on the survey date closest to the colony peak count date as the raw count (the equivalent of the peak count for the individual sa mple). The superpopulation estimates with standard errors were then compared to the raw counts for all samples in order to obtain a ratio expression of the difference between the two estimates. We divided the superpopulation estimate by the raw count for each sample, to obtain each sample ratio, then averaged these ratios across all samples. We averaged individual ratios rather than finding a summed ratio across all samples, as it is unlikely that there is one underlying level of nest turnover within the colony; different nesting cohorts within the colony are likely to have different levels of nest turnover at different times within the breeding season. The average ratio of this proportional error across all samples was our estimated colony wide proport ional difference between the peak count and extrapolated superpopulation count, as follows: i /RC i ))/n = (N*colony/PCcolony) N* i is the superpopulation estimate for sample i; RC i is the raw count for sample i; n is the number of samples in the colony f i /RC i ))/n is the average ratio of the total number of nest starts to the number of nests seen in the raw count.

PAGE 86

86 extrapo lated superpopulation estimate. We also calculated the variance of the ratio estimate, in order to calculate 95% confidence intervals for the colony wide superpopulation estimate. Both the sample size and the raw counts of the samples are fixed values, and thus the only variance that remains to be calculated in the ratio estimate is the superpopulation variance for each estimate. The variance of the average ratio, var(), is equal to 1 divided by the square of the sample size n, multiplied by the sum (a cross all samples) of 1 divided by the square of the sample raw count, multiplied by the variance of the sample superpopulation estimate. This variance of the superpopulation estimate for a sample, var(N*), is equal to the square of the standard error of the estimate (which is calculated by Program MARK). Therefore it is a simple task to calculate var(), the variance of the ratio of superpopulation estimates to raw counts for all samples: i /RC i ))/n = i /RC i )/n)=(1/n 2 i /RC i )) =(1/n 2 i 2 )var(N* i )) var(N* i )= SE i 2 where RC=peak count, N*=superpopulation estimate, n=sample size, =ratio of raw counts to superpopulation estimates for all samples, and SE=stand ard error of the superpopulation estimate. This calculation ignores any covariance between the raw count and superpopulation estimate. However, judging by the comparison of peak counts to superpopulation estimates for the small Great Egret colonies exami ned in this study (Figure 2 6), we would suggest that this covariance is probably fairly small, and may be safely excluded from our calculations.

PAGE 87

87 Thus the average ratio of raw counts to superpopulation estimates for all samples (), with its associated var iance, may be used with the peak count for the entire colony in order to estimate the superpopulation estimate for the entire colony. We assumed that the average ratio between the superpopulation estimates of samples and the raw counts of samples is equiv alent to the ratio between the superpopulation estimate for the entire colony and the peak count for the colony. for the colony is considered to be a fixed variable (it does not have an associated variance estimate), these 95% CIs may be used to directly estimate the 95% CIs for th e colony superpopulation estimate. Example: The 2005 White Ibis samples. The sample raw counts, superpopulation estimates, and standard errors for the superpopulation estimates are shown below. Sample Raw count N* SE of N* IT 1 28 51 1.91 IT 3 36 67 1.9 1 GST 1 3 19 1.69 GST 2 14 30 2.38 GST 3 4 12 1.90 Substituting these values into the equations listed above, we find an average proportional difference between superpopulation estimate and raw count of 3.00198, and a variance for this value of 0.0231 8: i /RC i ))/n = ((51/28)+(67/36)+(19/3)+(30/14)+(12/4))/5 = 3.00198 var() = (1/n 2 i 2 )var(N* i )) = (1/n 2 i 2 )SE i 2 ) = (1/5 2 )(((1/(28 2 ))1.91 2 )+((1/(36 2 ))1.91 2 )+((1/(3 2 ))1.69 2 )+((1/(14 2 ))2.38 2 )+ ((1/(4 2 ))1.90 2 )) = 0.02318 Confid ence intervals for the average sample ratio may then be calculated as follows:

PAGE 88

88 The peak count for this colony in 2005 was 12,750 nests. As this colony peak count is a fixed value, we simply multiply this peak count by and its associated CIs to find the extrapolated superpopulation estimate (38,275 nests) and confidence intervals for the entire colony (34,392; 42,157): Colony N* = 127503.00198 = 38,275 Colony LCI = 127502.69743 = 34 ,392 Colony UCI = 127503.30643 = 42,157

PAGE 89

89 APPENDIX C SAMPLE BUDGET FOR DE VELOPING PEAK COUNT AND SUPERPOPULATION ESTIMATES The superpopulation method for estimating asynchrony bias is much more labor and flight intensive than many current monitoring techniques, as frequent surveys are required to follow individual nests through time from the air. For the purposes of this budgetary analysis we assumed that weekly surveys provided sufficient accuracy in superpopulation estimation (Table 2 4), but the necessary frequency of surveys will depend upon the synchrony in nesting of the species of interest and upon the desired precision of the result. For this exercise, we calculated labor and flight costs for a 3 month breeding season, surveying a Great Egret colony with a peak count of 458 nests and a su perpopulation estimate of 666 nests. The hypothetical colony was 45 km from the airport, and in this exercise it was the only colony being surveyed. Estimated labor time, flight time, and associated costs are listed in Table C 1. According to these samp le calculations, superpopulation estimation can be roughly four times as costly as peak count estimation in terms of flying and 20 times as costly in terms of labor for photographic analysis; overall, the superpopulation estimate is roughly five times more costly to produce. Costs of transportation to and from the airport, overhead costs, etc. are not included in this analysis. Clearly these relative values will vary depending upon length of breeding season, number of flights conducted per month for the superpopulation estimation, labor costs per hour, cost of plane rental per hour, and other factors. However, we hope that this example provides some guidance for managers trying to determine the proper tradeoff between monetary cost and estimation accurac y for monitoring breeding colonies. As a concluding comment, we would like to note that a cheaper survey method that gives unreliable (or uninterpretable) estimates is no bargain; cash strapped managers may want to consider surveying intensively every 2 3 years

PAGE 90

90 (and obtaining useful estimates), rather than surveying annually using a peak count or related method and failing to obtain useful information.

PAGE 91

91 Table C 1. Sample budgets for superpopulation estimation and peak count estimation of breeding populati on size for a hypothetical Great Egret colony. Superpopulation estimate Peak count estimate Hours of editing and printing photos: 10 0 Hours of photo analysis: 50 3 Cost of labor/hr: $10 $10 Total analysis costs: $600 $30 km to colony: 45 45 km/min flight time: 1.75 1.75 Flight time per survey (min): 51.48 51.48 Flights per season: 12 3 Season length (weeks): 12 12 Hours flight time for season: 10.30 2.57 Cost per hour flying: $160 $160 Cost of flights for seas on: $1,647 $412 Cost of labor flying time: $103 $26 Total flying costs: $1,750 $438 Total cost: $2,350 $468

PAGE 92

92 APPENDIX D MODEL SUITE FOR STUD Y OF EFFECTS OF GROU ND DISTURBANCE ON GR EAT EGRETS Table D notation indicates that the parameter is general (not restricted to be equal o ver time); owed to vary with time, but that the variations change in a similar way between groups over each survey interval (essentially, that the difference is additive rather than multiplicative), and a dot notation means that the parameter is restricted to be the same over all survey intervals. N, population size, is allowed to vary by group in all models, as the number of nests in each group is different. The entire model suite for these parameters included 125 potential models. Most of these models were not ru n and are not included in the table, either because it became clear during modeling that they were not going to be a good fit for the data or because the models failed to converge. Model QAICc weights # parameters {p(.)phi(.)pent(t)N(group)} 0 0.50147 22 {p(.)phi(.)pent(t+group)N(group)} 0.0119 0.4985 22 {p(t)phi(.)pent(t)N(group)} 21.4128 0.00001 40 {p(t)phi(group)pent(group+t)N(group)} 22.4091 0.00001 42 {p(t)phi(group)pent(t)N(group)} 22.4092 0.00001 42 {p(. )phi(group)pent(group*t)N(group)} 31.94 0 42 {p(t)phi(group+t)pent(group+t)N(group)} 49.3695 0 59 {p(t)phi(t)pent(t)N(group)} 53.0026 0 58 {p(group)phi(group)pent(group*t)N(group)} 53.5716 0 42 {p(t)phi(group)pent(group*t)N(grou p)} 54.2869 0 60 {p(group*t)phi(.)pent(.)N(group)} 54.29 0 44 {p(t)phi(.)pent(group*t)N(group)} 55.3114 0 59 {p(t)phi(group*t)pent(group)N(group)} 74.404 0 60 {p(group*t)phi(.)pent(group*t)N(group)} 88.5037 0 79 {p(.)phi( group)pent(.)N(group)} 14636.2824 0 5 {p(.)phi(group)pent(group)N(group)} 14640.2926 0 7 {p(group)phi(.)pent(.)N(group)} 14641.7839 0 6 {p(.)phi(group*t)pent(.)N(group)} 14688.8833 0 41 {p(t)phi(group)pent(group)N(group)} 14801.6243 0 25 {p(.)phi(.)pe nt(group)N(group)} 14820.9384 0 4 {p(group*t)phi(.)pent(group)N(group)} 14837.2927 0 44 {p(.)phi(.)pent(.)N(group)} 26347.0759 0 4

PAGE 93

93 APPENDIX E MODEL SUITE FOR STUD Y OF EFFECTS OF GROU ND DISTURBANCE ON WH ITE IBISES Table E notation indicates that the parameter is general (not restricted to be equal over t ime); to vary with time, but that the variations change in a similar way between groups over each survey interval (essentially, that the difference is additive rather than multiplicative); and a dot notation means that the parameter is restricted to be the same over all survey intervals. N, population size, is allowed to vary by group in all models, as the number of nests in each group is different. The entire model suite for these parameters included 125 potential models. Most of these models were not run and are not included in the table, because it became clear during modeling that they were not going to be a good fit for the data. Model AICc weights # parameters {phi(t+g)p(t*g)pent(t)N(g)} 0 0.54463 19 {phi(t+g)p(t*g)pent(t+g)N(g)} 1.69680 0 .23316 20 {phi(t+g)p(t+g)pent(t)N(g)} 3.93850 0.07601 16 {phi(t*g)p(t*g)pent(t)N(g)} 4.11060 0.06974 23 {phi(t*g)p(t*g)pent(t+g)N(g)} 5.78330 0.03022 24 {phi(t+g)p(t*g)pent(t*g)N(g)} 6.10840 0.02569 21 {phi(t*g)p(t*g)pent(t*g)N(g)} 7.41990 0.01333 26 {phi(t+g)p(t+g)pent(t*g)N(g)} 9.4766 0.00477 18 {phi(t*g)p(t)pent(t)N(g)} 12.8334 0.00089 19 {phi(t+g)p(t+g)pent(t+g)N(g)} 13.0595 0.00079 17 {phi(t)p(t*g)pent(t)N(g)} 14.3942 0.00041 18 {phi(g)p(t*g)pent(t)N(g)} 16.561 5 0.00014 17 {phi(t*g)p(t)pent(t*g)N(g)} 16.6314 0.00013 22 {phi(g)p(t*g)pent(t*g)N(g)} 17.5562 0.00008 19 {phi(t+g)p(t*g)pent(.)N(g)} 21.8437 0.00001 17 {phi(t)p(t)pent(t)N(g)} 32.8912 0 14 {phi(g)p(t*g)pent(g)N(g)} 43.9314 0 14 {phi(.)p(t )pent(t)N(g)} 45.552 0 11 {phi(t)p(.)pent(t)N(g)} 46.3739 0 10 {phi(t)p(t)pent(.)N(g)} 52.6684 0 14 {phi(t)p(t)pent(.)N(g)} 109.867 0 11 {phi(.)p(.)pent(t)N(g)} 325.708 0 6 {phi(.)p(.)pent(.)N(g)} 361.315 0 4

PAGE 94

94 LIST OF REFERENCES Arnason, A. N., and C. J. Schwarz. 1995. POPAN 4: enhancements to a system for the analysis of mark recapture data from open populations. Journal of Applied Statistics 22 :785 800. Barbraud, C., and G. Glinaud. 2005. Estimating the sizes of large gull colonies taking into account nest detection probability. Waterbirds 28 :53 60. Barbraud, C., Y. Kayser, D. Cohez, M. Gauthier Clerc, and H. Hafner. 2004. Detection probability of nests of Squacco Herons in southern France. Journal of Field Or nithology 75 :172 175. Bart, J., and S. Earnst. 2002. Double sampling to estimate density and population trends in birds. Auk 119 :36 45. Bayliss, P., and K. M. Yeomans. 1990. Use of low level aerial photography to correct bias in aerial survey estimates of Magpie Goose and Whistling Duck density in the Northern Territory. Australian Wildlife Research 17 :1 10. Bibby, C. J. 2000. Bird census techniques. Academic Press, London, UK. Bock, C. E., and Z. F. Jones. 2004. Avian habitat evaluation: should counting bi rds count? Frontiers in Ecology 2 :403 410. Bouton, S. N. 1999. Ecotourism in wading bird colonies in the Brazilian Pantanal: biological and socioeconomic implications. Unpublished M.S. thesis, University of Florida, Florida, USA. Bouton, S. N., P. C. Frede rick, C. Dosualdo Rocha, A. T. Barbosa Dos Santos, and T. C. Bouton. 2005. Effects of tourist disturbance on Wood Stork nesting success and breeding behavior in the Brazilian Pantanal. Waterbirds 28 :487 497. Burger, J. 1981. Effects of human disturbance on colonial species, particularly gulls. Colonial Waterbirds 4 :28 36. Burnham, K. P. and D. R. Anderson. 1998. Model selection and inference: a practical information theoretic approach. Springer Verlag, New York, New York, USA. Burnham, K. P., D. R. Anderson G. C. White, C. Brownie, and K. P. Pollock. 1987. Design and analysis of methods for fish survival experiments based on release recapture. American Fisheries Society Monograph 5 :1 437. Cairns, D. 1980. Nesting density, habitat structure, and human distur bance as factors in Black Guillemot reproduction. Wilson Bulletin 92 :352 361.

PAGE 95

95 Carlson, B. A., and E. B. Mclean. 1996. Buffer zones and disturbance types as predictors of fledging success in Great Blue Herons ( Ardea herodias ). Colonial Waterbirds 19 :124 127 Carney, K. M., and W. J. Sydeman. 1999. A review of human disturbance effects on nesting colonial waterbirds. Waterbirds 22 :68 79. Cherenkov, S.E. 1998. Accuracy of one visit censuses of forest passerine birds during a breeding season. Zoologichesky Zhur nal 77 :474 485. Cooch, E., and G. C. White. 2007. Program MARK: A gentle introduction. 5th edition. Available online at: http://www.phidot.org/software/mark/docs/book/ Davis, W. E. Jr., and K C. Parsons 1991. Effects of investigator disturbance on the survival of Snowy Egret nestlings. Journal of Field Ornithology 62 :432 435. Dodd, M. G., and T. Murphy. 1995. Accuracy and precision of techniques for counting Great Blue Heron nests. Journal of Wildlife Management 59 :667 673. Dodd, M. G., and T. M. Murphy. 1996. The status and distribution of wading birds in South Carolina, 1988 1996. Report SG9610 A. South Carolina Marine Resources, Columbia, South Carolina, USA. Earnst, S. L., R. A. Stehn, R. M. Platte, W. W. Larned, and E. J. Mallek. 2005. Population size and trend of Yellow billed Loons in Northern Alaska. Condor 107 :289 304. Ellison, L. N., and L. Cleary. 1978. Effects of human disturbance on breeding of Double Crested Cormorants. Auk 95 :51 0 517. Erwin, R. M. 1982. Observer variability in estimating numbers: an experiment. Journal of Field Ornithology 53 :159 167. Erwin, R. M. 1989. Responses to human intruders by birds nesting in colonies: Experimental results and management guidelines. Colo nial Waterbirds 12 :104 108. Erwin, R. M., and T. W. Custer. 1982. Estimating reproductive success in colonial waterbirds: an evaluation. Colonial Waterbirds 5 :49 56. Fraser, W. R., and D. L. Patterson. 1997. Human disturbance and long term changes in Adli e penguin populations: a natural experiment at Palmer Station, Antarctic Peninsula. Pages 445 452 in B. Battaglia, J. Valencia, and D. W. H. Walton, editors. Antarctic Communities: Species, Structure and Survival. Cambridge University Press, Cambridge, UK. Frederick, P. C. 2006. Recommendations for monitoring breeding populations of long legged waders (Herons, egrets, ibises, storks, spoonbills, order Ciconiiformes). Appendix 1, I1 I41 in Southeast United States Waterbird Conservation Plan. U.S. Fish and Wi ldlife Service (USFWS), Division of Migratory Bird Management.

PAGE 96

96 Frederick, P. C., and M. W. Collopy. 1989a. Nesting success of five ciconiiform species in relation to water conditions in the Florida Everglades. Auk 106 :625 634. Frederick, P. C., and M. W. C ollopy. 1989b. Researcher disturbance in colonies of wading birds: effects of frequency of visit and egg marking on reproductive parameters. Colonial Waterbirds 12 :152 157. Frederick, P. C., and M. W. Collopy. 1989c. The role of predation in determining re productive success of colonially nesting wading birds in the Florida Everglades. Condor 91 :860 867. Frederick, P. C., J.A. Heath, R. E. Bennetts, and H. Hafner. 2006. Estimating nests not present at the time of breeding surveys: an important consideration in assessing nesting populations. Journal of Field Ornithology 77 :212 219. Frederick, P. C., B. A. Hylton, J. A. Heath, and M. Ruane. 2003. Accuracy and variation in estimates of large numbers of birds by individual observers using an aerial survey simulat or. Journal of Field Ornithology 74 :281 287. Frederick, P. C., M. G. Spalding, M. S. Sepulveda, G.E. Williams Jr., L. Nico, and R. Robbins. 1999. Exposure of Great Egret nestlings to mercury through diet in the Everglades of Florida. Environmental Toxicolo gy and Chemistry 18 :1940 1947. Frederick, P. C., T. Towles, R. J. Sawicki, and G. T. Bancroft. 1996. Comparison of aerial and ground techniques for discovery and census of wading bird (Ciconiiformes) nesting colonies. Condor 98 :837 840. Galbraith, H. 1987. Marking and visiting Lapwing Vanellus vanellus nests does not affect clutch survival. Bird Study 34 :137 138. Gibbs, J. P., S. Woodward, M. L. Hunter, and A. E. Hutchinson. 1988. Comparison of techniques for censusing Great Blue Heron nests. Journal of Fie ld Ornithology 59 :130 134. Giese, M. 1996. Effects of human activity on Adlie Penguin Pygoscelis adeliae breeding success. Biological Conservation 75 :157 164. Gtmark, F. 1992. The effects of investigator disturbance on nesting birds. Pages 63 104 in D. M Power, editor. Current Ornithology, volume 9. Plenum Press, New York, New York, USA. Gratto Trevor, C. L., V. H. Johnston, and S. T. Pepper. 1998. Changes in shorebird and eider abundance in the Rasmussen Lowlands, NWT. Wilson Bulletin 110 :316 326. Herna ndez Matias, A., L. Jover, and X. Ruiz. 2003. Predation on Common Tern eggs in relation to sub colony size, nest aggregation and breeding synchrony. Waterbirds 26 :280 289. Hulka, S., and J. Stirling. 2000. A study of breeding Black throated Divers Gavia ar ctica based on observation from vantage points. Bird Study 47 :117 121.

PAGE 97

97 Hunt, G. L. 1972. Influence of food distribution and human disturbance on the reproductive success of Herring Gulls. Ecology 53 :1051 1061. James, F. C., C. E. McCulloch, and D. A. Wiede nfield. 1996. New approaches to the analysis of population trends in land birds: a comment on statistical methods. Ecology 77 :13 27. Kadlec, J. A., and W. H. Drury. 1968. Structure of the New England Herring Gull population. Ecology 49 :644 676. King, K. A. 1976. Colonial wading bird survey and census techniques. Pages 155 159 in A. Sprunt IV, J. C. Ogden, and S. Winckler, editors. Wading Birds: Research Report No.7 of the National Audubon Society. Plenum Press, New York, New York, USA. Kury, C. R., and M. G ochfeld. 1975. Human interference and gull predation in cormorant colonies. Biological Conservation 8 :23 34. Kushlan, J. A. 1979. Effects of helicopter censuses on wading bird colonies. Journal of Wildlife Management 43 :756 760. Kushlan, J. A., and K. L. B ildstein. 1992. White Ibis ( Eudocimus albus ). Volume 9 in A. Poole, P. Stettenheim, and F. Gill, editors. The Birds of North America. Academy of Natural Sciences, Philadelphia, Pennsylvania, USA. Liddle, G. M. 1994. Interannual variation in the breeding bi ology of the Antarctic Prion Pachyptila desolata at Bird Island, South Georgia. Journal of Zoology 234 :125 129. Link, W. A., and J. R. Sauer. 1998. Estimating population change from count data: application to the North American Breeding Bird Survey. Ecolog ical Applications 8 :258 268. Link, W. A., and J. R. Sauer. 2002. A hierarchical analysis of population change with application to Cerulean Warblers. Ecology 83 :2832 2840. Lock, A. R., and R. K. Ross. 1973. The nesting of the Great Cormorant ( Phalacrocorax carbo ) and the Double crested Cormorant ( Phalacrocorax auritus ) in Nova Scotia in 1971. Canadian Field Naturalist 87 :43 49. Magrath, M. J. L. 1999. Breeding ecology of the Fairy Martin. Australian Journal of Zoology 47 :463 477. Mayfield, H. F. 1975. Sugges tions for calculating nest success. Wilson Bulletin 87 :456 466. McCrimmon, D. A., J. C. Ogden, and T. Bancroft. 2001. Great Egret ( Ardea alba ). Volume 570 in A. Poole and F. Gill, editors. The Birds of North America. The Birds of North America, Inc., Phil adelphia, Pennsylvania, USA. Morrison, R. I. G., C. Downes, and B. Collins. 1994. Population trends of shorebirds on fall migration in eastern Canada 1974 1991. Wilson Bulletin 106 :431 447.

PAGE 98

98 Nagy, L. R., and R. T. Holmes. 2004. Factors influencing fecundity in migratory songbirds: is nest predation the most important? Journal of Avian Biology 35 :487 491. Nichols, J. D., J. E. Hines, J. R. Sauer, F. W. Fallon, J. E. Fallon, and P. J. Heglund. 2000. A double observer approach for estimating detection probabili ty and abundance from point counts. The Auk 117 :393 408. Ogden, J. C. 1994. A comparison of wading bird nesting dynamics, 1931 1946 and 1974 1989, as an indication of changes in ecosystem conditions in the southern Everglades. Pages 533 570 in S. Davis and J. C. Ogden, editors. Everglades: the ecosystem and its restoration. St. Lucie Press, Del Ray Beach, Florida, USA. Parsons, K. C., and J. Burger. 1982. Human disturbance and nestling behavior in Black crowned Night Herons. Condor 84 :184 187. Piatt, J. F. B. D. Roberts, W. W. Lidster, J. L. Wells, and S. A. Hatch. 1990 Effects of human disturbance in breeding Least and Crested Auklets at St. Lawrence Island, Alaska. Auk 107 :342 350. Piazza, B. P., and V. L. Wright. 2004. Within season nest persistence in large wading bird rookeries. Waterbirds 27 :362 367. Platt, J. B. 1977. The breeding behavior of wild and captive Gyrfalcons in relation to their environment and human disturbance. Unpublished M.S. thesis, Cornell University, New York, USA. Pollock, K. H., and W. L. Kendall. 1987. Visibility bias in aerial surveys: a review of estimation procedures. Journal of Wildlife Management 51 :502 510. Pollock, K. H., J. D. Nichols, C. Brownie, and J. E. Hines. 1990. Statistical inference for capture recapture experime nts. Wildlife Monographs 107 :1 97. Prater, A. J. 1979. Trends in accuracy of counting birds. Bird Study 26 :198 200. Pratt, H. M. 1970. Breeding biology of Great Blue Herons and Common Egrets in Central California. Condor 72 :407 416. Prenzlow, D. M., and J. R. Loworn. 1996. Evaluation of visibility correction factors for waterfowl surveys in Wyoming. Journal of Wildlife Management 60 :286 297. Rao, P. S. R. S. 2005. Ratio Estimators II. Pages 6978 6983 in S. Kotz, C. B. Read, N. Balakrishnan, and B. Vidakovic editors. Encyclopedia of Statistical Sciences, 10 Rodgers, J. A., and J. Burger. 1981. Concluding remarks: symposium on human disturbance and colonial waterbirds. Colonial Waterbirds 4 :69 71. Rodgers, J. A., P. S. Kubilis, and S. A. Nesbitt. 2005. Accur acy of aerial surveys of waterbird colonies. Waterbirds 28 :230 237.

PAGE 99

99 Rodgers, J. A., S. B. Linda, and S. A. Nesbitt. 1995. Comparing aerial estimates with ground counts of nests in Wood Stork colonies. Journal of Wildlife Management 59 :656 666. Rodgers, J. A., and H. T Smith. 1995. Set back distances to protect nesting bird colonies from human disturbance. Conservation Biology 9 :89 99. Rosenstock, S. S., D. R. Anderson, K. M. Giesen, T. Leukering, and M. F. Carter. 2002. Landbird counting techniques: current practices and an alternative. Auk 119 :46 53. Safina, C., and J. Burger. 1983. Effects of human disturbance on reproductive success in the Black Skimmer. Condor 85 :164 171. Sagar, P. M., and J. C. Stahl. 2005. Increases in the numbers of breeding pairs in two populations of Buller's Albatross ( Thalassarche bulleri bulleri ). Emu 105 :49 55. Schwarz, C. J., and A. N. Arnason. 1996. A general methodology for the analysis of capture recapture experiments in open populations. Biometrics 52 :860 873. Schwarz, C. J. R. E. Bailey, J. R. Irvine, and F. C. Dalziel. 1993. Estimating salmon spawning escapement using capture recapture methods. Canadian Journal of Fisheries and Aquatic Sciences 50 :1181 1197. Shields, M. A., and J. F. Parnell. 1986. Fish Crow predation on e ggs of the White Ibis at Battery Island, North Carolina. Auk 103 :531 539. Short, J., and P. Bayliss. 1985. Bias in aerial survey estimates of kangaroo density. Journal of Applied Ecology 22 :415 422. Stolen, E. D., D. R. Breininger, and P. C. Frederick. 200 4. Using waterbirds as indicators in estuarine systems: Successes and perils. Pages 409 423 in S. A. Bortone, editor. Estuarine indicators. CRC Press, Boca Raton, Florida, USA. Temple, S. A., and J. A. Wiens. 1989. Bird populations and environmental change s: can birds be bio indicators? American Birds 43 :260 270. Thompson, B. C., G. E. Knadle, D. L. Brubaker, and K. S. Brubaker. 2001. Nest success is not an adequate comparative estimate of avian reproduction. Journal of Field Ornithology 72 :527 536. Thompso n, W. L. 2002. Towards reliable bird surveys: accounting for individuals present but not detected. Auk 119 :18 25. Tomialojc, L., and J. Verner. 1990. Do point counting and spot mapping produce equivalent estimates of bird densities? Auk 107 :447 450. Torres R., and A. Mangeaud. 2006. Factors affecting the nesting success of the Cattle Egret ( Bubulcus ibis ) in Laguna Mar Chiquita, Central Argentina. Ornitologia Neotropical 17 :63 71.

PAGE 100

100 Tremblay, J., and L. N. Ellison. 1979. Effects of human disturbance on breed ing of Black crowned Night Herons. Auk 96 :364 369. Vos, D. K., R. A. Ryder, and W. D. Graul. 1985. Response of breeding Great Blue Herons to human disturbance in Northcentral Colorado. Colonial Waterbirds 8 :13 22. White, G. C., and K. P. Burnham. 1999. Pro gram MARK: Survival estimation from populations of marked animals. Bird Study 46S :120 138. White, C. M. and T. L. Thurow. 1985. Reproduction of Ferruginous Hawks exposed to controlled disturbance. Condor 87 :14 22. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations. Academic Press, New York, New York, USA. Wilson, K. J., R. H. Taylor, and K. J. Barton. 1990. The impact of man on Adlie penguins at Cape Hallett, Antarctica. Pages 183 190 in K. R. Kerry and G. Hempel, editors. Antarctic ecosystems: ecological change and conservation. Springer Verlag, Berlin, Germany. Young, E. C. 1990. Long term stability and human impact in Antarctic Skuas and Adlie Penguins. Pages 232 236 in K. R. Kerry and G. Hempel, editor s. Antarctic ecosystems: ecological change and conservation. Springer Verlag, Berlin, Germany.

PAGE 101

101 BIOGRAPHICAL SKETCH Kathryn A. Williams was born in Maine and grew up on the Scarborough Marsh. She attended Tufts University in Medford, Massachusetts from 2000 to 2004 and graduated with a double B.S degree in biology and environmental science. After b rief forays into retail and trawl fishing, she began working on the South Florida Wading Bird Project in 2005.