Nesting Ecology of Snail Kites

in Water Conservation Area 3A

Robert E. Bennetts, Michael W. Collopy
and Steven R. Beissinger

.T'

Prepared for the
United States Army Corps of Engineers ;, -
and the ..
South Florida Water Management District
through the '' .
Florida Cooperative Fish and Wildlife Research Unit *a,'A."mI- ,

NESTING ECOLOGY OF SNAIL KITES IN
WATER CONSERVATION AREA 3A

Final Report

by

Robert E. Bennetts, Michael W. Collopy
and Steven R. Beissinger

Department of Wildlife and Range Sciences
118 Newins-Ziegler Hall
University of Florida
Gainesville, FL 32611-0301

prepared for

U.S. Army Corps of Engineers
400 West Bay Street, P.O. Box 4970
Jacksonville, FL 32232

and

South Florida Water Management District
3301 Gun Club Road, P.O. Box V
West Palm Beach, FL 33402

through the

Florida Cooperative Fish and Wildlife Research Unit
Unit Cooperative Agreement No. 14-16-0009-1544
Research Work Order No. 40

July 1988

Citation should read: Bennetts, R.E., M.W. Collopy, and S.R. Beissinger. 1988. Nesting
ecology of Snail Kites in Water Conservation Area 3A. Dept. Wildl. and Range Sci., Univ. of
Florida, Gainesville. Florida Coop. Fish and Wildl. Res. Unit, Tech. Rept. No. 31. 174 pp.

SUMMARY

During 1986 and 1987, we studied the nesting success and nest-site selection of Snail

Kites (Rostrhamus sociabilis) in Water Conservation Area 3A (WCA-3A) of the Florida

Everglades. Particular emphasis was placed on evaluating the influence of hydrologic

conditions. Water conditions in both 1986 and 1987 were higher than the long-term average

during the primary nesting season (January through July). A lack of rainfall resulted in

drier than average conditions by mid-June of 1987; however, most nesting efforts had been

completed by this time.

We monitored 148 active nests (i.e. nests in which at least one egg was laid) during

1986 and 227 during 1987. The distribution of nesting kites in WCA-3A during both years

was restricted to a relatively narrow range of ground elevations. These elevations

corresponded to areas in which hydrologic conditions were suitable for nesting. Most nests

(94%) were initiated at sites where water depths ranged from 20 to 80 cm. Water depths at

adjacent foraging areas (open sloughs) generally were 10 cm or more deeper than at nest

sites. The areas in WCA-3A where nesting occurred on average dried out once every 1.9 to

3.8 years (for the 19-year period of record). A large portion of WCA-3A dries out more

frequently than once every 1.9 years and was not used by nesting kites. A smaller portion

of WCA-3A dries out less frequently than every 3.8 years. No extensive nesting occurred in

this region during this study; however, it may be important to kites during drier years.

Areas where Snail Kites nested had a greater ratio of open water to sawgrass than did

regions that were not used for nesting. Snail Kites tended to select nest substrates that

offered sturdy structural support and were located over water. Willow was the most

commonly used nest substrate but was used less than expected based on its high

availability. Pond apple and melaleuca were selected in higher proportion to their

availability. Dry hammocks contained numerous sturdy substrates, but were avoided,

presumably because of high predator densities.

The environmental correlates of nesting success were not consistent between 1986 and

1987. Our results indicate that the relative importance of factors influencing nesting

success varies among years and that predictability of nesting success, based on

environmental conditions above or below threshold levels may be low. Step-wise logistic

regression indicated that the date of nest inititation was the most important correlate of

nesting success in 1986 and 1987; although previous studies found no such relationship.

Nesting success was not influenced by water levels, which were relatively high, during this

study; however, previous studies have shown clearly that success often decreases to zero

when areas dry out completely. This suggests that the influence of water level is a

threshold response.

ACKNOWLEDGMENTS

This project could not have been accomplished without the help of the many individuals

who contributed greatly to its success. It is with our deepest appreciation that we

acknowledge their contributions.

Our field assistants, Elaine Caton, Hugh Dinkler, and Nancy Dwyer worked many long

hours under conditions that were often well beyond what is normally expected of a field

assistant, and did so willingly, cheerfully, and without hesitation. Elaine Caton was an

integral part of the initial project design and her suggestions undoubtedly contributed to

the quality of the research. Nancy Dwyer provided relentless assistance with the data

entry, analyses, and all phases of this report. Their efforts are sincerely appreciated.

Peter Frederick and his assistants, Reed Bowman and Susan Fitzgerald, often provided

assistance in the field, and were a constant source of logistic support, stimulating

discussions, and friendship. Peter also came to the rescue on numerous occasions, with

tools in hand, when we were stranded by broken boats or trucks.

We appreciate the helpful comments on drafts of this report provided by Walt Dineen,

Nancy Dwyer, Lewis Hornung, Jon Moulding, Jim Rodgers, and Noel Snyder. We carefully

considered all criticisms; however not surprisingly for a report of this magnitude, we are

not in complete agreement on all interpretations. We have tried wherever possible to

identify alternative viewpoints to our interpretations.

Many people assisted with various aspects of the field work. Alan Gillespie, Patrick

Railey, and Paul Stone contributed numerous hours of effort dredging the marsh in search of

apple snails. We thank them for their efforts and for keeping their sense of humor in

spite of being stranded more than once in the Everglades. For assisting with foraging

observations we thank Mike Green, Rose-Marie Etemad-Green, and Dave Westneat. We also

thank Irma Caton for assisting us with our nest searches and checks. Marilyn Spaulding

helped us to identify nest parasites.

iii

We are deeply indebted to James Wyatt for keeping our airboats running. Without his

help we would have lost many hours of field work. We greatly appreciated his expertise and

the numerous repairs that frequently extended well into the night and weekends.

Ken Portier provided considerable statistical advice. Ken Portier and Lori Pfahler

also set up and ran the logistic regression analyses. We received additional advice on

statistical analyses and/or applications of the Mayfield Method from Barbara Black, Tom

Edwards, Doug Johnson, Jim Nichols, and Karen Steenhof. We appreciate their help. We

thank Reed Bowman and Peter Frederick for the many thought provoking discussions of

Mayfield analyses and survivorship.

Jon Moulding and Lewis Hornung of the U.S. Army Corps of Engineers were a constant

source of logistic support. During various aspects of the project they provided everything

from airboats to hydrologic data. We are also grateful to the Florida Cooperative Fish and

Wildlife Research Unit; particularly Wiley Kitchens, John Richardson, and Leonard

Pearlstein, for their patience and many hours of assistance with the ERDAS computer system.

Their knowledge of geographic information systems (GIS) were both an asset to the project

and a source of inspiration.

For sharing their knowledge of apple snails and use of the suction dredge, we thank

Earl Rich and Oscar Owre. For sharing their knowledge of snake lore we thank George

Dalrymple, Dick Franz, and Paul Andreatis. George Dalrymple spent several hours in the

field with us an generously shared his knowledge of the Everglades herpetofauna. For their

insight of the Everglades and for providing us with living conditions which greatly

enhanced our experience in South Florida we are indebted to Fred and Sandy Dayhoff. We

also thank Fred Dayhoff for his emergency welding which kept our airboat going.

Joyce Kleen kept us informed of Snail Kite sightings through the sighting program at

Loxahatchee National Wildlife Refuge. Betty Wargo and Noel Chandler informed us of kite

nesting activity on Lakes Kissimmee and Okeechobee during 1986. James Rodgers, Jr. kept us

informed of nesting activity and assisted in banding throughout South Florida during 1987.

Mark Spier kept us informed of kite activity in Everglades National Park and helped us

obtain weather data. Mike Brown of the Florida Game and Fresh Water Fish Commission

informed us of several nests in WCA-3 that we had missed.

We thank the Florida Game and Fresh Water Fish Commission, particularly James Rodgers,

Jr., for allowing us to use unpublished data from the annual Snail Kite Surveys. The

Florida State Division of Forestry generously allowed us to park our airboats at their

Trail Glades facility.

For their secretarial support, forwarding mail, and assisting in numerous other ways,

we are grateful to Barbara Fesler, Cathy Ritchie, and Lonell Segars. We thank Kathy

Engstrom-Solheim for the cover artwork and Margy Summers for the cover design. We are

grateful to Adele Karolik and Rick Yates for their help during the frantic final hours of

preparing this report.

TABLE OF CONTENTS

SUMMARY .................... ........ i

ACKNOWLEDGMENTS ................... .... iii

LISTOF TABLES.................... .. ..... .viii

LISTOFFIGURES .................. ....... ix

INTRODUCTION ................... .......... 1

PURPOSE AND OBJECTIVES OF THE STUDY. ... .. . .. 2

STUDYAREA ....... ..... ..... ............ 5
Vegetation . . . . . . 5
Hydrologic and Weather Conditions .. .. . ... 8

METHODS ....................... .......... 16
Nest Searches and Monitoring. .. . . . . 16
Nesting Terminology ... .. .. ... .. .. ..... .. 18
Habitat Selection . . . . . . 18
Nest-site Selection ... ... .. ... ... ..... 26
Nest Success and Productivity. . . . . 27

RESULTS . . . . ..... . 30
Distribution of Nesting Snail Kites ... ................ 30
Habitat Selection... ............ .. ........... 34
Water depth . . . . . .34
Dry-down interval ......... ............ 37
Proportion of open water . . . ... 37
Apple snail abundance ............. ...... 41
Nest-site Selection ....................... .46
Nest substrate. . . . . . 46
Nest height. . . . . . .. 46
Stand size . . . . . . 46
Size of the Breeding Population . . . . ... 50
Nesting Success . . . . . . .53
Survivorship and Age-specific Nest Failure ... . . .. 53
Productivity .............. ......... ........56
Causes of Nest Failure . . . .. . . 60
Predators .. .. .... .. .. .. ..... .. ... ..60
Structural collapse .. ...................... 64
Abandonment. ...................... 64
Parasites . . . . . . 65
Human disturbance .................... 65
Influences of Nesting Success ... .. . . . 65
Date of initiation . . . . . 65
Nesting substrate . . . . . 65
Nest height. ........... ............ .68
Distance from land .. . . . . 68
Water depth . . . . . . 68
Rainfall . . . .. . ... 68
Wind speed . . . . . . .68

Table of Contents (Cont'd.)

Temperature ....... .......
Relative importance of factors influencing
nesting success . . .

DISCUSSION......................
Habitat Selection . . . ...
Water depth ...... ..........
Dry-down interval and hydroperiod . .
Proportion of open water . . .
Apple snail abundance . .. . .
Nest-site Selection . . . .
Size of the Breeding Population... . .
Comparison of Nesting Success During 1986 and 1987
With Previous Years . . . .
Evaluation of the Mayfield Method . .
Reproductive Success and Productivity .. .
Causes of Nest Failure . . . .
Predation . . ... .
Structural collapse . . . .
Abandonment . . . .
Parasites . . ... .. . ..
Human disturbance .. . . .
Influences of Nesting Success . . ..
Date of initiation . . . .
Nest substrate. .. . . .
Nest height. ................
Distance to land . . . .
Water level. .... . . ...
Inter-relationships and relative importance of
factors influencing nesting success ..
Predicted Influence of Alternative Water Delivery Plans on
Nesting Snail Kites and Mitigation Alternatives .

. . 74

S . 74

.79
.79
.79
.82
.90
.92
.94
.95

.96
.98
101
104
105
109
109
110
110
110
111
113
113
114
114

. . 115

. . 118

LITERATURE CITED. ..........

APPENDICES . . .

. . . 125

vii

LIST OF TABLES

Table 1. Clutch sizes reported in Florida since 1880 .. . . .

. .59

Table 2. Hatching success from nests with 2 and 3-egg clutches in WCA-3A during
1986and 1987 ................... ......... .61

Table 3. Estimates of Snail Kite productivity in south Florida from 1968 through
1987 . . . . . ... . .. 62

Table 4. Condition of Snail Kite nests when found ..... ......... 63

Table 5. Results from stepwise logistic regression analyses for discriminating
between successful and unsuccessful nests with respect to influencing
variables. Variables are listed in the order they were entered into the
model (by order of the highest initial Chi-square value; see Harrell
1980) .............. ..... .............. ... .77

Table 6. Ground elevations and dry-down intervals for Snail Kite nesting areas in
the water conservation areas. ........... . .

Table 7. Traditional estimates of nesting success of Snail Kites in south Florida
from 1968 to 1987. Includes nests found before eggs were laid (i.e. all
occupied nests) . . . . . ..

Table 8. Comparison of nesting success during 1986 and 1987 calculated using
traditional and Mayfield methods. . . . .

Table 9. Predicted hydroperiods (days/hr) for the alternative water delivery plans
currently being considered under the General Design Memorandum. Base
condition is the "no action" alternative .. ... . . .

* 84

S. .97

S. 99

. 121

viii

LIST OF FIGURES

Figure 1. Minimum monthly water stage in WCA-3A (average of 3-4 and 3-28 stations)
from 1968-1986 shown in relation to annual Snail Kite surveys from WCA-3A
for the same time period. Surveys from 1968-1980 were conducted by the
USFWS (Sykes 1983a, 1983b), and surveys from 1981-1986 were conducted by
FGFWFC (FGFWFC unpubl. data). . . . . 3

Figure 2. Percent of the total annual Snail Kite Surveys that are accounted for by
kites in WCA-3A. Surveys from 1968-1980 were conducted by the USFWS
(Sykes 1983a, 1983b) and surveys from 1981-1986 were conducted by FGFWFC
(FGFWFC unpubl. data).................. ...... 4

Figure 3. Location of primary study area (shaded) within WCA-3A. . . 6

Figure 4. Water Conservation Area 3A showing levees, continuous water recording
stations (3-4 and 3-28) and elevation in feet (m) . . . 7

Figure 5. Stage at 3-4 water gauging station from December 1985 through August 1987,
comparing actual stage with mean monthly maximums and minimums, calculated
for the period of record at gauge 3-4. . . . . 9

Figure 6. Mean daily water stage at 3-4 and 3-28 stations for 1 January through 31
July 1986 (-) and 1987(--) ..................... 11

Figure 7. Daily rainfall totals at Tamiami Ranger Station in 1986 and 1987
winter/spring seasons . . . . . .12

Figure 8. Monthly rainfall totals at Tamiami Ranger Station, expressed as a percent
deviation from the monthly average, calculated for the period of record
(1949-1987) ................... ........ 13

Figure 9. Daily maximum and minimum temperatures during winter/spring of 1986 and
1987, measured at Tamiami Ranger Station . . . .... 14

Figure 10. Daily mean wind speed during spring of 1986 and 1987, measured at Tamiami
Ranger Station.................... .........15

Figure 11. Diagram of the four regions of WCA-3A that we systematically searched for
Snail Kite nests . . . . . . 17

Figure 12. Boundaries of WCA-3A, showing the portion which we classified satellite
imagery for habitat analyses ................... ... 20

Figure 13. Location of nesting areas in WCA-3A in which we sampled apple snails.
Shown within the circles are the estimated number of Snail Kite nests that
occurred in the area during 1987 . . . .. .. 22

Figure 14. Mean apple snail egg cluster counts ( SD) along the sawgrass/open water
edge and within the interior of the sawgrass stand. Sample sizes
(i.e. number of plots) are shown. . . . .... 25

Figure 15.

Distribution of nesting Snail Kites in WCA-3A during 1986 (N-148) and 1987
(N-227). Major tree islands within the nesting distribution are shown for
reference . . . . . . .

Figure 16. Distribution of nesting Snail Kites in WCA-3A during 1986 and 1987 in
relation to topography (1 ft [m] contour intervals). . . .

Figure 17. Distribution of nesting Snail Kites in WCA-3A during 1986 and 1987 in
relation to the nesting areas reported by Sykes (1984) and Beissinger
(1983a; successful nests only) . . . . . .

Figure 18. Frequency distribution of the number of nests found in each water depth
class (10 cm intervals) at the time of breeding initiation The depth
value shown is the midpoint of the 10 cm class (e.g. 15 cm represents
water depths ranging from 10.1 to 20.0 cm inclusive) . ..

Figure 19. Mean daily water depths within the elevation range used by nesting Snail
Kites (shaded). Values were derived by subtracting the ground elevation
(i.e. 2.1 2.5 m) from the mean daily water stage at gauges 3-4 (higher
elevation) and 3-26 (lower elevation). Note that these gauges are located
within open slough communities and therefore represent depths of 10 cm or
deeper than would be expected at actual nest sites . . .

Figure 20. Minimum monthly water stage in WCA-3A (average of gauges 3-4 and 2-28)
from 1968-1986. The shaded area represents the elevation range for which we
found breeding Snail Kites. . . . . . .

Figure 21. Proportion of open water to sawgrass for areas within the nesting range of
Snail Kites in WCA-3A that were used and not used for nesting. Also shown
is the proportion of open water from systematically sampled areas outside
of the nesting range (i.e. above 2.5 m elevation) . . . .

Figure 22.

Figure 23.

S.31

S32

.33

.35

. 36

.38

. 39

Classified GIS Map of southern WCA-3A showing distribution of nesting Snail
Kites (shaded). Note the amount of open water within the distribution
compared with outside the nesting distribution . . . .. 40

Mean capture times ( SD) of foraging Snail Kites in areas of low (-)
and high (--) nesting densities. Areas 1-2, 3-4, 5-6, and 7-8 were
paired samples observed observed during approximately the same period
of time . . .. .. . . .. .

. 42

Figure 24. Egg clusters/10 m2 in areas of low (-) and high (--) nesting densities.
Areas 1-2, 3-4, 5-6, and 7-8 were paired samples observed during
approximately the same period of time. . . . . 43

Figure 25. Egg cluster indices (see methods) in areas of low (a) and high (0)
nesting densities. Areas 1-2, 3-4, 5-6, and 7-8 were paired samples
observed during approximately the same period of time . . .. 44

Figure 26. Scattergram showing the relationship between capture times and egg
cluster indices for the eight nesting areas sampled. . . . 45

Figure 27. The relative use of nest substrates in WCA-3A during 1986 and
1987. . . . . ... . . 47

x

Figure 28. Percent departures from expected frequencies of use of nesting substrates
compared with their abundance in WCA-3A. . . ..... 48

Figure 29. Frequency distribution of Snail Kite nests in WCA-3A during 1986 and 1987
by nest height (1 m increment) . . . . .. 49

Figure 30. Lengthwise cross section of typical tree island in WCA-3A showing the zone
usually selected by nesting Snail Kites. The hardwood hammock portion is
usually on the northern end of the islands. . . . 51

Figure 31. Cross section of typical small (< 500 m2 ) willow head showing the zone
usually selected by nesting Snail Kites. ... . . 52

Figure 32. Mayfield estimates of overall nesting success of Snail Kite nests in WCA-3A
during 1986 and 1987. Ninety-five percent confidence intervals about the
estimate and sample sizes are shown . . . . .. 54

Figure 33. Mayfield estimates of nesting success during the incubation and nestling
stages of Snail Kite nests in WCA-3A during 1986 and 1987. Ninety-five
percent confidence intervals about the estimate and sample sizes are
shown . . ..... . . . .. 55

Figure 34. Daily survival of Snail Kite nests in WCA-3A of four consecutive 6-day
nestling periods during 1986 and 1987. Ninety-five percent confidence
intervals about the estimates and sample sizes are shown . . .. 57

Figure 35. Survivorship of Snail Kite nests through the nesting cycle, calculated
as the percent of nests found that survived daily (-). Survivorship
from Mayfield estimates (--) are shown for comparison . . .. 58

Figure 36. Mayfield estimates of overall nesting success of Snail Kite nests in
WCA-3A that were initiated early, middle, and late season (see methods).
Ninety-five percent confidence intervals about the estimates and sample
sizes are shown . . . . . ... 66

Figure 37. Mayfield estimates of overall nesting success of Snail Kite nests in
WCA-3A in nesting substrates willow, pond apple, cypress, and melaleuca.
Ninety-five percent confidence intervals about the estimates and sample
sizes are shown ................... ........ 67

Figure 38. Mayfield estimates of overall nesting success of Snail Kite nests in
WCA-3A at nest heights <200 cm, 200-300 cm, and >300 cm. Ninety-five
percent confidence intervals about the estimates and sample sizes are
shown . . . . . .69

Figure 39. Mayfield estimates of overall nesting success of Snail Kite nests in
WCA-3A in which the distance to land was <100 m, 100-500 m, >500 m.
Ninety-five percent confidence intervals about the estimates and sample
sizes are shown ......................... .. ... 70

Figure 40. Mayfield estimates of daily nest survival of Snail Kite nests in WCA-3A
at water depths 0-25 cm, 25-50 cm, and 50-75 cm. Ninety-five percent
confidence intervals about the estimates and sample sizes are
shown . . . . . . . .71

Figure 41. Mayfield estimates of daily nest survival of Snail Kite nests in WCA-3A
when average daily rainfall between nest visits was 0 cm, 0-0.5 cm,
and >0.5 cm. Ninety-five percent confidence intervals about the estimates
and sample sizes are shown. . . . . .. 72

Figure 42.

Mayfield estimates of daily nest survival of Snail Kite nests in WCA-3A
in which average daily wind speeds were 0-1.5 kph, 1.5-3.0 kph, and
>3.0 kph. Ninety-five percent confidence intervals about the estimates
and sample sizes are shown. ... . . . . 73

Figure 43. Mayfield estimates of daily nest survival of Snail Kite nests in WCA-3A
Sin which the minimum temperatures were 0-100C, 10-200C, 20-300C.
Ninety-five percent confidence intervals about the estimates and sample
sizes areshown . . . . . .75

Figure 44. Mayfield estimates of daily nest survival of Snail Kite nests in WCA-3A
with maximum temperatures 10-200C, 20-30C, 30-400C. Ninety-five percent
confidence intervals about the estimates and sample sizes are shown.
Insufficient sample size precluded calculation for nests during 1986 in
which the maximum temperature was from 10-20C. . . . 76

Figure 45. Habitat suitability of Snail Kite nesting habitat in relation to water
depth. Suitability of 0 implies that the habitat is not suitable and
suitability of 1 implies that the habitat is completely suitable with
respect to water depth. Becuase of the uncertainty regarding the nature
of the upper limit, we present three possible scenarios: A) a rapid
linear decline; B) a slower linear decline (with uncertain intercept);
and C) a threshold decline .....................

Figure 46. Snail Kite populations in WCA-1 and WCA-2A from 1969 through 1986, as
determined from annual Snail Kite Surveys. Surveys from 1968-1980 were
conducted by the USFWS (Sykes 1983a, 1983b), and surveys from 1981-1986
were conducted by FGFWFC (FGFWFC unpubl. data). . . .

Figure 47.

. 81

. .86

Snail Kite populations for south Florida and WCA-3A from 1969 through
1986, as determined from annual Snail Kite Surveys. Surveys from
1968-1980 were conducted by the USFWS (Sykes 1983a, 1983b), and surveys
from 1981-1986 were conducted by FGFWFC (FGFWFC unpubl. data) .

. .87

Figure 48. Snail Kite populations in WCA-3B and WCA-2B from 1969 through 1986, as
determined from annual Snail Kite Surveys. Surveys from 1968-1980 were
conducted by the USFWS (Sykes 1983a, 1983b), and surveys from 1981-1986
were conducted by FGFWFC (FGFWFC unpubl. data) .. . .

.88

Figure 49. Habitat suitability of Snail Kite nesting habitat in relation to dry down
interval. Suitability of 0 implies that the habitat is not suitable and
suitability of 1 implies that the habitat is completely suitable with
respect to dry-down interval. The effects of a higher dry-down interval
(--) are unclear, they may be negative on woody vegetation and positive
on apple snail populations ....................... 91
/

xii

Figure 50. Snake activity in the Everglades, as determined from capture frequency of
systematic monthly effort from 1984-1986 (after Dalrymple 1986);
shown in conjunction with the primary period of Snail Kite nesting
activity . . . . . .. 112

xiii

INTRODUCTION

The Snail Kite (Rostrhamus sociabilis) is a medium-sized raptor of the Neotropics.

Although Snail Kites may be locally common in South and Central America, Mexico, and Cuba

(Sykes 1984), the Florida Snail Kite (R. I. olumbeus) is listed as endangered both

federally and by the State of Florida (U.S. Fish and Wildlife Service 1986).

Historically, the numbers of Snail Kites in Florida have not been well documented

(Nichols et al. 1980), and estimates prior to the early 1900's are lacking. Howell (1932)

gives a general indication of kite numbers during the early 1900's by describing that

"scattered flocks of a hundred or more birds" were frequently found within a limited area.

It is impossible to assess, however, whether these flocks were widespread or local

concentrations during times of food shortage. As recently as 1985, over 350 Snail Kites

have been reported using a single roost during a period when Water Conservation Area 3A had

dried out (J. Takekawa, pers. comm.). At any rate, we can reasonably assume that kite

numbers in Florida, up until the 1930's, were at least 100 and probably numbered in the

hundreds or even thousands. During the mid-1900's, estimates of the Snail Kite population

in Florida were consistently under 100 (see Sprunt 1954, Steiglitz and Thompson 1967, Sykes

1979). In recent years (1970's-1980's), Snail Kite numbers generally have been increasing

with population estimates of at least 668 birds during 1984 (Florida Game and Fresh Water

Fish Commission, unpubl. data).

Declines in the kite population from the early to mid 1900's generally have been

attributed to widespread drainage of Florida's marsh habitats (e.g. the Everglades) (see

Bent 1937, Steiglitz and Thompson 1967, Sykes 1979, 1983b, Beissinger 1986). There seems

little doubt that Snail Kite populations are influenced by the hydrologic conditions (see

Sykes 1983b, 1987b, Beissinger and Takekawa 1983, Beissinger 1986).

Although it is impossible at this point to determine cause and effect, the kite

population increase beginning in the 1970's is likely, in part, a response to the

impounding of WCA-3A. The impounding of WCA-3A lengthened the hydroperiod, which resulted

in increased apple snail populations (Kushlan 1975) and vegetation changes (e.g. opening of

sawgrass stands) that enhanced Snail Kite habitat (see Sykes 1987b). WCA-3A was completed

in 1962 (Zaffke 1983). There was a subsequent period of over 10 years before Snail Kite

populations began increasing. This lag period probably was related to the time it took for

WCA-3A to fill, snail populations and vegetation to respond to the increased hydroperiod,

and kites to colonize the area. After this lag period, however, Snail Kite populations

began increasing and closely tracked the hydrologic conditions (Fig. 1).

Most of Florida's Snail Kites currently are found in WCA-3A (FGFWFC, unpubl. data) and

the relative importance of WCA-3A to the total Florida Snail Kite population has been

increasing since kite populations first began increasing in WCA-3A (Fig. 2). The

proportion of the total population (from annual surveys) of Snail Kites in WCA-3A has

reached as high as 92.2% during 1983 (J. Rodgers, pers. comm.). There is little question

that WCA-3A has become an increasingly important area for the Snail Kite in Florida.

PURPOSE AND OBJECTIVES OF THE STUDY

In 1983, the U.S. Army Corps of Engineers was authorized by Congress to conduct an

experimental program of water deliveries to Everglades National Park (ENP) (U.S. Army Corps

of Engineers 1985). This experimental program was in response to requests from ENP for

water deliveries that were more timely and better suited to their management needs. In

1983, a "flow-through" system was employed which left three of the four S-12 structures

(gates which allow water to flow from WCA-3A to ENP) open. This system resulted in very

low water levels in WCA-3A and reduced the water supply storage function of WCA-3A.

Beginning with the resumption of the summer wet season in 1985, a "rainfall-driven" system

was employed which incorporated current rainfall into a formula for determining flow rates

(see U.S. Army Corps of Engineers 1985 for details of plan).

The potential for hydrologic changes resulting from this phase of the experimental

release program led the Corps of Engineers and the South Florida Water Management District

. 3 -400

I-
S2. I 0 1

2.2 I I

18- -100

1 68' O70' 72' '74' '176 '78' '80' 82' '84 '86 0

YEAR

Figure 1. Minimum monthly water stage in WCA-3A (average of 3-4 and 3-28 stations) from 1968 1986
shown in relation to annual Snail Kite surveys from WCA-3A for the same time period. Surveys from
1968 1980 were conducted by the USFWS (Sykes 19831, 1983b), and surveys from 1981 1986 were
conducted by FOFWFC (FGFWFC uapubl. data).

20-
0
I-,

S40

IYEAR
Z
W

'S 20

68 70 72 74 76 78 80 82 84 86
YEAR
Figure 2. Percent of the total annual Snail Kite Surveys that are accounted for by kites in WCA-3A.
Surveys from 1968 1980 were conducted by the USFWS (Sykes 1983a, 1983b) and surveys from 1981 1986
were conducted by FGFWFC (FGFWFC unpubL data).

to fund this study to determine the influence of these changes on nesting Snail Kites.

This report summarizes the results from that study.

The objectives of this study were to evaluate the nesting success and nest-site

selection of Snail Kites during the experimental "rain-driven" water release program and to

evaluate the potential impacts of the water delivery program on future nesting populations.

Emphasis was placed on the effects of hydrologic conditions in an effort to evaluate the

influence of the experimental release program; however, the scope of the project included

examining several potential influences on the reproductive ecology of Snail Kites. We also

intended for this project to provide a comprehensive data base that would assist in future

management decisions related to the Snail Kite within the Everglades.

STUDY AREA

Water Conservation Area 3A is an approximately 237,000 ha impoundment that lies 25 km

west of Miami and immediately north of Everglades National Park. Our primary study area

was located in the portion of WCA-3A that lies south of Alligator Alley (Hwy 84) (Fig. 3)

because most Snail Kite use in WCA-3A in recent years has occurred in this region (Sykes

1984).

The primary study area is dissected by the Dade and Broward County lines; the northern

portion is in Broward County and the southern portion is in Dade County. The primary study

area was bordered on the north by Alligator Alley, the south by Tamiami Trail (Hwy 41), the

west by the L-28 levee, and on the east by the L-67A (southeast) and L-68A (northeast)

levees (Fig. 4). The area slopes gradually from the northwest to southeast and ranges in

elevation from approximately 2.0 m (6.7 ft) to 3.0 m (11 ft).

Vegetation

The vegetative communities of WCA-3A have been described in considerable detail by

Loveless (1959), McPherson (1973), Zaffke (1983), and Tanner et al. (1987). The southern

to fund this study to determine the influence of these changes on nesting Snail Kites.

This report summarizes the results from that study.

The objectives of this study were to evaluate the nesting success and nest-site

selection of Snail Kites during the experimental "rain-driven" water release program and to

evaluate the potential impacts of the water delivery program on future nesting populations.

Emphasis was placed on the effects of hydrologic conditions in an effort to evaluate the

influence of the experimental release program; however, the scope of the project included

examining several potential influences on the reproductive ecology of Snail Kites. We also

intended for this project to provide a comprehensive data base that would assist in future

management decisions related to the Snail Kite within the Everglades.

STUDY AREA

Water Conservation Area 3A is an approximately 237,000 ha impoundment that lies 25 km

west of Miami and immediately north of Everglades National Park. Our primary study area

was located in the portion of WCA-3A that lies south of Alligator Alley (Hwy 84) (Fig. 3)

because most Snail Kite use in WCA-3A in recent years has occurred in this region (Sykes

1984).

The primary study area is dissected by the Dade and Broward County lines; the northern

portion is in Broward County and the southern portion is in Dade County. The primary study

area was bordered on the north by Alligator Alley, the south by Tamiami Trail (Hwy 41), the

west by the L-28 levee, and on the east by the L-67A (southeast) and L-68A (northeast)

levees (Fig. 4). The area slopes gradually from the northwest to southeast and ranges in

elevation from approximately 2.0 m (6.7 ft) to 3.0 m (11 ft).

Vegetation

The vegetative communities of WCA-3A have been described in considerable detail by

Loveless (1959), McPherson (1973), Zaffke (1983), and Tanner et al. (1987). The southern

Lake
Ikeechobee

I.

'I

g I
lades

Everglades

National

0 5 10 15km
2 H t

Park

I.
L.

.I

**'

"S W "60

Figure 3. Location of primary study are (shaded) within WCA-3A.

Figure 4. Water Conservation Area 3A showing levees, continuous water recording stations
(3-4 and 3-28) and elevation in feet (m).

7

portion (Dade Co.) of the WCA-3A, except for the extreme western edge, is comprised of open

sloughs (see Loveless 1959) interspersed with stands of sawgrass (Cladium iamaicensis).

Tree islands of all size classes are relatively common throughout the area. Tree islands

in this region are comprised predominantly of willow (Sai caroliniana), but the

relatively dry northern ends of the larger islands often were of mixed species. Single

shrubs of willow and pond apple (Annona glabra) are common throughout the area, but a

variety of other species also occurred.

Coco Plum (Chrysobalanus icaco and cypress (Taxodium spp) become increasingly

abundant in the tree islands along western edge of southern WCA-3A; with an approximately

0.5 1.0 km wide strip of predominately cypress occurring along the L-28 levee. The

slough communities along this western portion are largely replaced by wet prairie (see

Loveless 1959) with spikerush (Eleocharis spp) becoming the predominate emergent graminoid.

Cattail (Tvyha latifolia) interspersed with sawgrass and open sloughs occur throughout

the northeast region and the wet prairie communities are largely absent. The northeastern

portion of the study area has relatively few tree islands; those few tree islands present

are predominantly willow. The northwest region of the study area has relatively high

proportion of wet prairie communities and numerous tree islands of mixed species.

Hvdroloaic and Weather Conditions

Site specific water levels may vary greatly with local features (e.g. local topography

and vegetation); however, the general trend in WCA-3A is for water depths in WCA-3A to

increase from northwest to southeast following the topographic contours.

The general hydrologic trend in WCA-3A is for water levels to decrease through the

spring months and to increase again with the onset of daily thunderstorms, usually

beginning in late May or early June (Fig. 5). The nesting seasons of 1986 and 1987 can be

characterized as relatively wet compared to the long-term average.

Hydropattern in WCA 3A
January 1986 July 1987

3.0-

E

CO
\oI

20-

17-

'L I

1, 4

* a
a a

a

*v
*
A

Dec-BS

Jul-86

* Maximum. 1968-1987

Jon-87

A Minimum, 1968-1987

Figure 5. Stage at 3-4 water gauging station from December 1985 through August 1987, comparing actual
stage with mean monthly maximums and minimums, calculated for the period of record at gauge 3-4.

.9

-8
I-e

2.

During' 1986 and 1987, water levels in WCA-3A declined rapidly beginning in mid-March

(Fig. 6). Water levels during both years reached their lowest levels in June. As

expected, water levels during 1986 increased sharply in late June and July with the onset

of daily thunderstorms. In 1987, however, below normal rainfall in May and June resulted

in low water levels persisting through July.

The rainfall pattern in WCA-3A generally consists of periodic storm systems through

the winter and spring followed by near daily localized thunderstorms. The most notable

deviation from the normal rainfall pattern (i.e. long-term average) was a storm system in

March 1987 (see Fig. 7) which contributed to the month of March having over 300% more rain

than the long-term average (Fig. 8).

The 1986 nesting season in general was slightly cooler than 1987, with a greater

number of cold fronts occurring in March when nesting was being initiated (Fig. 9). Wind

patterns generally were similar in 1986 and 1987, but a few storm systems resulted in

relatively high winds in January and early February 1987 (Fig. 10).

STATION 3-4

DATE
STATION 3-28

DATE

Figure 6. Mean daily water stage at 3-4 and 3-28 stations for January through 31 July
1986 (-) and 1987 (--).

Rainfall
TImwnl Rangw s#an, 10

11 -

10-

9-

a
3-

7-

6-

S-

4-

3-

1-

.a I

-Jaon 09-Feb 20-Mar 29-Apr 08-lJu 1-Jul

Rainfall
Tomiomf Range Station. 1987

D

9-

B-

7

6-

4-

3-

2-

ILrlbLiiJidi

00-Jan

09-Feb

20-Mor

29-Apr

oe-Jun

Figure 7. Daily rainfall totals at Tamiami Ranger Station in 1916 and 1987 winter/spring
seasons.

0

14

I I

a,

a,
0I

E
V-

0

x
u
C
a-

C
0
a
I.
U-

Tamlami Ranger Station
350

300-

250-

200-

150

100-

50-7/,

-50- Z

-100 I I I I I i
Jan-86 Mar-86 May-86 Jul-86 Sep-86 Nov-86 Jan-87 Mar-87 May-87 Jul-87

Month

Figure 8. Monthly rainfall totals at Tamiami Ranger Station, expressed as a percent deviation from the
monthly average, calculated for the period of record (1949 1987).

1986 Temperatures
Taoioml Roer Station

09-Feb 20-Mor 29-Apr 08-Jun 18-Jul

1987 Temperatures
Toiaomi Ronger Station

09-Feb

29-eAp

Figure 9. Daily maximum and minimum temperatures during winter/spring of 1986 and 1987,
measured at Tamiami Ranger Station.

05-
00-Jon

0 ---
00-Jon

Wind Speed
- Temomi MRnge Station. 1987

13
12
11
10

9
I

S
7

S
4
3-
2
1
0
00-Jon

29-Apr

Wind Speed
Taniom Ranger Station, 1986

7
6
5
4
3-
2
1
0
00-Jon

1 1 lllIjlluBD h 1l

o9-Feb 20-Mor 29BAi Ji-Ji 18-J*ll

Figure 10. Daily mean wind speed during spring of 1986 and 1987, measured at Tamiami
Ranger Station.

20-Mor

.......... t.. i i.... ............... ..............."

09-Fob

06-Jun

18-Ju

I&

--

qR_

Il1lffflfff

I ti
lJnHkb6& im

METHODS

Nest Searches and Monitoring

We searched the primary study area for Snail Kite nests by systematically traversing

four regions (Fig. 11). The regions were based on the location of access points and our

ability to search the area within a 12-hour day. The western half of the north region

dried out for much of each nesting season, and was consequently not accessible during these

periods. Because of the relatively few nests found, we decreased our search effort in the

entire north region during 1987 to periodic searches intended to determine any major

changes in distribution.

We located nests primarily through the behavior of adult Snail Kites. When adult

kites were flushed from a nest they tended to circle upward, whereas non-nesting birds that

were flushed, flew more horizontally away from the boat. This behavior allowed us to find

nests with relative ease by intensively searching the area from which birds exhibiting this

flight pattern had departed. In addition to this flight pattern we also were able to find

nests by: 1) observation of kites carrying sticks; 2) adult kites bringing apple snails to

females (courtship) or young; 3) aerial courtship displays (see Steiglitz and Thompson

1967, Sykes 1987b, Beissinger 1988); 4) vocalizations of the adults or begging calls of the

young (see Beissinger 1988); and 5) nest searches after repeated observations of adult

birds at the same location.

The latitude and longitude of nest locations were recorded using a LORAN-C

navigational unit. We found through repeated visits to the same locations that the unit

was consistently accurate to within 20-30 m; however, this precision may not extend to the

actual latitude and longitude values. Locations were entered into a geographical

information system (GIS) data base and overlaid on to a geo-referenced satellite image from

SPOT Image Corporation using the ERDAS computer system (ERDAS Inc. 1987).

North

I I
I I
/ 9

Q 4 1/
a 0 ,W
q.

IQ) (i

U! *
I I
I I te l Tral1

Figure 11. Diagram of the four regions of WCA-3A that we systematically earched for Snal
Kite aets.

METHODS

Nest Searches and Monitoring

We searched the primary study area for Snail Kite nests by systematically traversing

four regions (Fig. 11). The regions were based on the location of access points and our

ability to search the area within a 12-hour day. The western half of the north region

dried out for much of each nesting season, and was consequently not accessible during these

periods. Because of the relatively few nests found, we decreased our search effort in the

entire north region during 1987 to periodic searches intended to determine any major

changes in distribution.

We located nests primarily through the behavior of adult Snail Kites. When adult

kites were flushed from a nest they tended to circle upward, whereas non-nesting birds that

were flushed, flew more horizontally away from the boat. This behavior allowed us to find

nests with relative ease by intensively searching the area from which birds exhibiting this

flight pattern had departed. In addition to this flight pattern we also were able to find

nests by: 1) observation of kites carrying sticks; 2) adult kites bringing apple snails to

females (courtship) or young; 3) aerial courtship displays (see Steiglitz and Thompson

1967, Sykes 1987b, Beissinger 1988); 4) vocalizations of the adults or begging calls of the

young (see Beissinger 1988); and 5) nest searches after repeated observations of adult

birds at the same location.

The latitude and longitude of nest locations were recorded using a LORAN-C

navigational unit. We found through repeated visits to the same locations that the unit

was consistently accurate to within 20-30 m; however, this precision may not extend to the

actual latitude and longitude values. Locations were entered into a geographical

information system (GIS) data base and overlaid on to a geo-referenced satellite image from

SPOT Image Corporation using the ERDAS computer system (ERDAS Inc. 1987).

Nesting Terminolovg

For the purposes of this report, we generally followed the terminology suggested by

Steenhof (1987). A breeding attempt was considered to begin with the laying of the first

egg. An occupied nest was any nest which was actively being attended by adult Snail Kites,

regardless of whether breeding had been initiated. A breeding pair was a mated pair of

Snail Kites in which the female had laid at least one egg. Because Snail Kites are

sequentially polygamous (Beissinger and Snyder 1987) and iteroparous (Beissinger 1986), an

individual adult kite may have been a member of more than one breeding pair during one

nesting season. When referring to a nest (rather than the breeding pair), we considered a

nest active if breeding had been initiated (i.e. at least one egg had been laid)

(Postupalsky 1974).

Nesting success was defined as the proportion of active nests from which at least one

young survived to fledging age. Because of the difficulties in determining if a nest was

successful after the young began flight, we considered fledging age to be 80% of the

average age at first flight (Steenhof and Kochert 1982). Although the reported age of

first flight is variable (see Nicholson 1926, Steiglitz and Thompson 1967, Chandler and

Anderson 1974, Beissinger 1988) we found that Snail Kites generally were capable of first

flight and often left the nest at 30 days. We therefore considered a nest successful when

at least one young reached 24 days old (80% of 30 days). This approach assumes that

nestling mortality between 24 and 30 days is negligible, but reduces the potential to

mistakenly identify a nest that fledged young as having failed. We made no attempt to

assess nesting success prior to the laying of the first egg or after the young reached

fledging age.

Habitat Selection

At each nest site we placed a water gauge which was read on each subsequent visit to

the nest (approximately every 7 to 10 days). Estimates of nest-site water depth at the

Nesting Terminolovg

For the purposes of this report, we generally followed the terminology suggested by

Steenhof (1987). A breeding attempt was considered to begin with the laying of the first

egg. An occupied nest was any nest which was actively being attended by adult Snail Kites,

regardless of whether breeding had been initiated. A breeding pair was a mated pair of

Snail Kites in which the female had laid at least one egg. Because Snail Kites are

sequentially polygamous (Beissinger and Snyder 1987) and iteroparous (Beissinger 1986), an

individual adult kite may have been a member of more than one breeding pair during one

nesting season. When referring to a nest (rather than the breeding pair), we considered a

nest active if breeding had been initiated (i.e. at least one egg had been laid)

(Postupalsky 1974).

Nesting success was defined as the proportion of active nests from which at least one

young survived to fledging age. Because of the difficulties in determining if a nest was

successful after the young began flight, we considered fledging age to be 80% of the

average age at first flight (Steenhof and Kochert 1982). Although the reported age of

first flight is variable (see Nicholson 1926, Steiglitz and Thompson 1967, Chandler and

Anderson 1974, Beissinger 1988) we found that Snail Kites generally were capable of first

flight and often left the nest at 30 days. We therefore considered a nest successful when

at least one young reached 24 days old (80% of 30 days). This approach assumes that

nestling mortality between 24 and 30 days is negligible, but reduces the potential to

mistakenly identify a nest that fledged young as having failed. We made no attempt to

assess nesting success prior to the laying of the first egg or after the young reached

fledging age.

Habitat Selection

At each nest site we placed a water gauge which was read on each subsequent visit to

the nest (approximately every 7 to 10 days). Estimates of nest-site water depth at the

time breeding was initiated was obtained directly from nests located during that period;

nest-site water depths for those nests found after breeding had been initiated were

estimated using a regression equation relating water depths at the nest-site and the

nearest continuous gauging station. P. Frederick (pers. comm.) used this method and found

that the correlation between sites in southern WCA-3A and these continuous recording

stations usually had r2 values greater than 0.90. We made no attempt to estimate the water

depth at time of breeding initiation for nests that we could not estimate the time that the

first egg was laid.

The average frequency at which nesting areas dry out was estimated by the number of

years that the minimum water stage recorded at gauges 3-4 and 3-28 fell below the elevation

range within which we found Snail Kite nests. This provided an average interval between

dry downs expressed in years. A similar approach was used to estimated the dry down

interval of other areas within the Everglades. In these other areas, however, we estimated

the elevation range from reported nesting distributions and used the closest continuous

water recording station to determine an approximate dry down interval.

The proportion of open water in areas that were used and not used for nesting was

determined using a satellite image from SPOT Image Corporation. The image used was a

composite of multi-spectral and panchromatic bands with pixel resolution of 10 m2. The

image was classified using training fields of known habitat types (Jansen 1986) and

compared with low level (300 m) aerial photographs of known areas for accuracy. We

concentrated our classification on distinguishing sawgrass from open water. Habitats that

that were functionally similar to Snail Kites (e.g. sloughs with different species of

floating vegetation) were combined in the final classification. We did not attempt to

distinguish the species composition of tree islands.

Due to available imagery and our current computer capabilities we were able to

classify only a portion of WCA-3A (Fig. 12). The portion we classified, however, contained

(i ert fe ger h it a s

aain rl

Fiue1.Budre fWA3,soigte oto hc ecasfe sn aelt
ma erfohbttanls.

approximately 80% of the nest locations in WCA-3A. We therefore believe it reasonably

represents the nesting habitat of kites in WCA-3A.

Based on approximately 105 hrs of observation including 184 prey captures by nesting

Snail Kites, we estimated the average foraging range extended 1 km from each nest site.

We used the search function of the ERDAS computer system (ERDAS 1987) to delineate a 1 km

radius around each nest site. The proportion of open water to sawgrass was then assessed

using the BSTATS function of ERDAS (ERDAS 1987) within polygons of areas that were used and

not used by nesting kites. Areas of overlapping (i.e. < 2 km from the closest nest) use

were included within each polygon. Nesting areas that were a minimum of 2 km from the

closest nest (i.e. did not overlap) were considered a separate polygon. The proportion of

open water to sawgrass was compared between areas that were used for nesting (i.e. use

polygons) and continuous areas that were not used for nesting by Snail Kites (non-use

polygons), but were within the overall distribution of kites in WCA-3A. We also compared

the proportion of open water to sawgrass between areas that were used for nesting and areas

that were outside of the distribution (i.e. above the elevation range within which we found

all nesting kites) of nesting kites.

We estimated apple snail abundance during 1987 using three separate measures. We used

two indices (capture time and egg cluster counts) at each of eight nesting areas with a 1

km radius that were centered around nests or colonies; four in areas of high kite nesting

density and four in areas of low nesting density (Fig. 13). An area was considered high

nesting density if it had an accumulative total of at least 10 occupied Snail Kite nests

for the season. An area was considered low nesting density if it had an accumulative total

of no more than 5 occupied nests for the season. Some nests in these areas probably went

undetected; however, because our search effort and observation time in these areas was

extensive, it was unlikely that we had overlooked enough nests to mis-classify an area.

One area of high and low nesting density each were observed simultaneously by two

observers over a three to four-day period. These simultaneous observations were intended

Area 8
Area 7
Area 6 Area 2
0

Area 1

Area 5
D Area 3/

0

* $gm

Figure 13. Location of nesting areas in WCA-3A in which we sampled apple snails. Shown
within the circles are the estimated number of Snail Kite nests that occurred in the area
during 1987.

to minimize the influence of seasonal environmental changes that could influence kite

foraging behavior or egg laying by apple snails. Foraging observations were conducted

between the hours of 0900 and 1200 to minimize the influence of daily temperature changes

on foraging behavior.

Capture time was measured as the interval (in seconds) from when a Snail Kite left a

perch and commenced foraging until a snail was captured. In order to maintain comparable

samples, we did not sample any areas in cypress habitat where still-hunting was a common

foraging method. In all of the areas we sampled, kites primarily foraged by flying low

over the marsh until apple snails were detected (see Sykes 1987a, Beissinger in press).

Consequently, only actual flight time was included in the total time to capture a snail.

In the event that a kite perched before capturing a snail, the time was stopped and

continued when the kite resumed foraging.

Egg cluster counts were conducted by traversing the high and low density nesting areas

along east/west transects. The first transect began approximately 1 km north of the nest

or colony. At the end of each transect, we looked at the second hand of our watch and

moved to the south 10 times the number of seconds displayed by the second hand to begin our

next transect. We repeated this procedure until the nesting area had been completely

traversed. Because the slough systems are oriented north/south, our transects frequently

crossed sawgrass/open water edges. Each time we crossed a sawgrass/open water edge we

counted the number of egg clusters using a 1 x 2.5 m PVC frame that was flipped end over

end four times. This resulted in sampling a strip that was 1 x 10 m. Because we suspected

that the number of egg clusters present was influenced by the proximity to the sawgrass

edge, at each edge we sampled a strip on the edge, 7.5 m into the sawgrass from the edge,

and 15 m into the sawgrass from the edge.

We developed an egg cluster index based on the number of egg clusters on the edge and

within the interior sawgrass that accounted for how much edge habitat was within the

nesting area. Based on the distribution of egg clusters in relation to the sawgrass edge

(Fig. 14), we calculated an egg cluster index as:

ECI CePe + CiPi

where Ce is the mean number of egg clusters per 10 m from the edge samples, Pe is the

proportion of the sawgrass area that is along a sawgrass/open water edge, Ci is the mean

egg cluster count for the interior sawgrass samples, and Pi is the proportion of the

sawgrass area that is not along an open water edge. This procedure weights the egg cluster

counts by the amount of edge habitat within each nesting area. Our sampling showed that

the number of egg clusters was higher along the sawgrass/open water edge, but that there

was little difference between the samples taken at 7.5 m and 15 m from the edge.

We estimated the proportion of edge to interior sawgrass from satellite imagery using

the BOUNDRY program of the ERDAS computer system (ERDAS Inc. 1987). This program

identifies when pixels classified as sawgrass are adjacent to pixels of open water. The

relative areas could then be calculated using the BSTATS program (ERDAS Inc 1987). Because

of the resolution of the image, only areas of sawgrass or open water of at least 10 m2

would have been included in this analysis. We believe this level of resolution was

acceptable since most kite foraging occurred in sloughs considerably larger than 10 m2.

In addition to the two indices of snail abundance we used one direct measure of snail

abundance. This method was a modified version of a technique described in detail by Brook

(1979) and adapted for sampling apple snails by Owre and Rich (1987). The technique

involves the use of a portable suction dredge which was powered by a Honda 3.5 hp pump (see

Owre and Rich 1987 for details of the pump and its operation). Water and the substrate

(e.g. peat) are sucked via a probe into a large (6mm) meshed collecting bag. Snails were

then sorted from the substrate and counted.

Whereas Owre and Rich (1987) estimated that 100 probes into the substrate covered an

area of 0.5 m (based on the diameter of the probe), we used a 1 m2 wire mesh frame that

extended vertically to above the water surface and had steel prongs (made from a barbecue

grill) that extended approximately 10 cm into the substrate. This enabled us to sample a

10-

206

8
04
E

Co
W

3-
62-

65- 206

1189
0, 4
O
ce
w
LU
W 3
z
Z
2

1-

0 I
0 7.5 15

DISTANCE FROM EDGE (m)

Filure 14. Mean apple snail egg cluster counts (t SD) along the awgranss/open water edge
nd within the interior c the awgrass stand. Sample sizes (ie. no. of plots) are shown.

completely contained 1 m vertical column, which we sampled down to a depth of 8 cm into the

substrate. The emergent vegetation within the 1 m column was first removed and searched by

hand. This enabled us to suck the entire substrate down to 8 cm without interference from

vegetation.

Because this method is relatively labor intensive (a 3-person crew could sample a

maximum of approximately twenty, 1 m2 plots per day), we only were able to sample one area

of high nesting density (area 6) and one of low nesting density (area 5). In order to

control for the influence of vegetation, all of our samples were take approximately I m

into the slough from the sawgrass edge. The plot locations were selected randomly within

the nesting area by observing the second hand on a watch and traveling along the sawgrass

edge via airboat for 10 times the number of seconds shown on the watch.

Nest Site Selection

We compared the relative use of nesting substrates to their availability for nest

sites in stands smaller than 100 m2. We hope to include larger stands in this analyses at

a later date using satellite imagery; however, without extensive ground-truthing, we are

currently unable to classify tree islands by species. We measured nest site availability

of stands smaller than 100 m2 in southern WCA-3A below 25.90 latitude. This area includes

approximately 60% of the nests we observed. We did not measure availability in the entire

area because of the extensive sampling time required and because much of northern WCA-3A

was dry (i.e. inaccessible) at the time of our sampling.

Using a LORAN C navigational unit, we traversed southern WCA-3A and counted the number

of stands less than 100 m2 of each species within a 100 m radius of the intersection of

each minute of latitude and longitude. We estimated the 100 m radius using a Leitz

rangefinder. We did not attempt to measure surface area of the nesting stands.

Consequently, the availability of species occurring in larger stands (e.g. willow) would be

under represented.

Nest Success and Productivity

The Mayfield Method (Mayfield 1961, 1975, Miller and Johnson 1978, Johnson 1979,

Hensler and Nichols 1981, Hensler 1985) was used to calculate nest success because it has

several advantages over traditional measures (i.e. no. successful nests/no, nests

observed). Essentially we chose this method for two reasons; first, unless all nests are

found on the first day of the nesting period, traditional success estimates are biased

(Hensler 1985) and tend to overestimate success (Mayfield 1975). Secondly, the Mayfield

estimate of success is better suited to statistical comparisons than traditional methods

(Miller and Johnson 1978).

The Mayfield Method requires that nests be checked at intervals throughout the nesting

cycle. We visited nests at approximately 7 to 10 day intervals. The failure date for

nests that failed between intervals was assumed to be the midpoint between the last two

nest visits. Johnson (1979) found this assumption was reasonable when intervals between

nest visits did not exceed 15 days.

An inherent assumption of the Mayfield Method is that nests fail at a constant rate

throughout the nesting period (Hensler and Nichols 1981, Hensler 1985). This assumption

may not always be valid (Green 1977). We assessed the assumption of constant failure in

two ways. The first was to test for differences in the failure rate of nests between the

egg and nestling stages during each year. The second method of assessing the assumption of

constant failure was to construct survivorship curves from nests found during egg laying.

These curves were calculated as the proportion of observed nests surviving each day. As

with our Mayfield calculations, the midpoint between the last visit when the nest was

viable and the first visit after the nest had failed was assumed to be the day of failure.

Based on our results, we used separate estimates for the incubation and nestling

stages, but believed that differences within stages were slight and did not warrant further

separation. Our overall success estimates were derived by combining the separate

incubation and nestling estimates in accordance with the procedures described by Hensler

(1985).

Hensler and Nichols (1981) demonstrated that the Mayfield estimate is a maximum

likelihood estimator (m.l.e.) for which statistical analyses for the asymptotic

distribution are appropriate. Formulae for calculating nest success and the corresponding

statistical analyses are provided in Appendix 1. Tests of significance between groups were

performed using a standard normal test.

We compared nest success in relation to several environmental and nest-site variables.

Nest success for the entire nesting period (overall success) was used for comparisons among

variables having one value per nest (e.g. nest substrate and nest height). We used daily

nest survivorship in comparisons of nest success among variables having values that changed

throughout the season (e.g. water level and weather).

We partitioned the nesting season into three equal 36-day periods (early, middle, and

late season) based on the range of dates in which nests were initiated. We then compared

nesting success between nests in which the first egg was laid within each of these 36-day

periods. The early period was from 31 January through 6 March. The middle period was from

7 March through 12 April. The late period was from 13 April through 19 May. Nests for

which we could not estimate the date of initiation were excluded from these analyses.

We also compared nesting success between nests that were located in each of the four

major substrates (willow, pond apple, cypress, and melaleuca) and of varying nest height

and distance to land. Sample sizes were insufficient for analyses among the lesser-used

nest substrates.

We compared daily nest survivorship of nests while they were in each of three water

level classes (< 25.0 cm, 25.1-50.0 cm, 50.1-75.0 cm). We considered a nest to have been

in a given water level class for an observation interval if the water depth at the nest

remained within that class on the nest visits at the beginning and end of the observation

interval and if the continuous water recording stations did not show water level changes

that would indicate that the water depth could have crossed into a different depth class.

Observation intervals in which water depth crossed from one depth class to another were not

used in the analyses. This approached eliminated the possibility that a given observation

interval was arbitrarily assigned to a depth class at the cost of a reduction in sample

size.

Because we did not have weather instruments at each nest site, the influence of

weather variables was compared based on averages recorded at the nearest continuous station

(an average of gauges 3-4 and 3-28 for rainfall, and Tamiami Ranger Station for temperature

and wind). We compared daily survival of nests based on daily averages (e.g. of rainfall)

recorded between each nest visit.

The relative importance of how environmental and nest-site characteristics influenced

nest success was assessed using stepwise logistic regression. We used the LOGIST procedure

of SAS (Harrell 1980) to develop a model for each year (and one for combined years) that

best discriminated successful from unsuccessful nests. The variables that were entered

into the analyses are summarized in Appendix 2, and those having an initial

Chi-square value of < 0.05 were entered stepwise into the model by order of highest initial

Chi-square value (Harrell 1980).

Comparisons between years of clutch size and the number of young fledged were made

using Chi-square contingency tests. Hatchability within clutch sizes was compared using

Mann-Whitney tests.

We considered a clutch complete only after the maximum number of eggs observed was

maintained for at least one nest visit after the maximum number was reached (i.e. no egg

loss was observed). Nests in which we detected egg loss before our second visit,

regardless of the number of eggs, were not presumed to have been complete.

RESULTS

Distribution of Nestina Snail Kites

Snail Kites were distributed throughout the south and western portions of WCA-3A (Fig.

15). The overall nesting range in WCA-3A did not markedly differ between 1986 and 1987;

however, the distribution of nests within that range was patchy and varied between years.

Some areas where kites nested in 1987 were not used in 1986 and some areas used in 1986

were not used in 1987. The most notable area that was used in 1986 but not 1987 was a

large willow head along the northern portion of the L-67 levee. We monitored six nests in

this area during 1986 and we suspect that several more were undetected. During 1987, we

observed one kite in the area, but found no indication of nesting activity. Several areas,

particularly within the south-central portion of WCA-3A were used in 1987, but not in 1986.

Because we traversed the entire area searching for nests (see methods) during both years,

we do not believe that differences in distribution were attributable to sampling bias.

The range of nesting within WCA-3A tended to occur within 2.1 m (6.8 ft) and 2.5 m

(8.2 ft) elevation (Fig. 16). Steiglitz and Thompson (1967) also reported that Snail Kite

nesting distribution corresponded with an elevational gradient at Loxahatchee National

Wildlife Refuge. The nesting range tended to extend toward slightly lower elevations (i.e.

deeper water) in 1986 and slightly higher elevations (i.e. shallower water) in 1987.

The distribution of nesting Snail Kites during 1986 and 1987 differed from the

historic distribution reported by Sykes (1984) for 1968 through 1980 and the more recent

distribution of successful nests reported by Beissinger (1983a) for 1983 (Fig. 17). We

found considerably more nesting activity in the south-western and south-central regions of

WCA-3A than has previously been reported.

RESULTS

Distribution of Nestina Snail Kites

Snail Kites were distributed throughout the south and western portions of WCA-3A (Fig.

15). The overall nesting range in WCA-3A did not markedly differ between 1986 and 1987;

however, the distribution of nests within that range was patchy and varied between years.

Some areas where kites nested in 1987 were not used in 1986 and some areas used in 1986

were not used in 1987. The most notable area that was used in 1986 but not 1987 was a

large willow head along the northern portion of the L-67 levee. We monitored six nests in

this area during 1986 and we suspect that several more were undetected. During 1987, we

observed one kite in the area, but found no indication of nesting activity. Several areas,

particularly within the south-central portion of WCA-3A were used in 1987, but not in 1986.

Because we traversed the entire area searching for nests (see methods) during both years,

we do not believe that differences in distribution were attributable to sampling bias.

The range of nesting within WCA-3A tended to occur within 2.1 m (6.8 ft) and 2.5 m

(8.2 ft) elevation (Fig. 16). Steiglitz and Thompson (1967) also reported that Snail Kite

nesting distribution corresponded with an elevational gradient at Loxahatchee National

Wildlife Refuge. The nesting range tended to extend toward slightly lower elevations (i.e.

deeper water) in 1986 and slightly higher elevations (i.e. shallower water) in 1987.

The distribution of nesting Snail Kites during 1986 and 1987 differed from the

historic distribution reported by Sykes (1984) for 1968 through 1980 and the more recent

distribution of successful nests reported by Beissinger (1983a) for 1983 (Fig. 17). We

found considerably more nesting activity in the south-western and south-central regions of

WCA-3A than has previously been reported.

WC A

3A

0
0

*.

* S

0

0
:. -*.

* *

0

* -"
"**

SNAIL KITlE ESTS
* 1t

Fiure 15. Distribution of nesting Snail Kites in WCA-3A during 1986 (N-148) and 1987
(N-227). Major tree islands within the nesting distribution are shown for reference.

WCA

3A

9 (2.7)

-. 00

0
0. 04.
0

S .. # .7./.
** C
*, '. *

I ** ** M
-KIT r ///
,6 0 *

S ** i
0 .p

00
0 %A
". :/

to topography (1 ft Iml contour intervals).

WCA

3A

.*1

0* 0 .
0
.8 0* .
I .
to ** 00 .
i *o ON 0 0

00 *a
So8. 1 .

*-sAIL KiT *EST IENb eer) e.
r -asItmN AREtA Is "t96d1)
m(eeot e oS6 4"*a 5 og4) S

Figure 17. Distribution of nesting Snail Kites in WCA-3A during 1986 and 1987 in relation
to the nesting areas reported by Sykes (1984) and Beissinger (1983a, successful nests
eely).

Habitat Selection

Water detth.-- We were able to estimate the water depth for the time that breeding

was initiated (i.e. when the first egg was laid) for 281 nests. At this stage of breeding,

almost all nests were over water (280 of 281 nests), and most sites (94% of 281 nests) had

water depths ranging from 20 80 cm. In only one case during 1986 and 1987 did we

encounter a nest that was built over dry land. This nest was built within an ongoing kite

colony that recently had dried out directly under the nest trees. The surrounding sloughs

of this colony, however, were completely inundated. Water depth at nest sites during the

initiation of breeding ranged from 0 to 75 cm in 1986 and from 21 to 105 cm in 1987 (Fig.

18). Mean depth at the time of initiation was lower in 1986 (x 41.22, n 96) than in

1987 (x 49.63, n = 185) (t 4.69, P < 0.01).

Although most nests were initiated in areas with water depths ranging from 20 80 cm,

considerable fluctuation in depth occurred throughout the season. A prolonged drying trend

occurred during both years from March through May (Fig. 19). A strong increase in water

depth resulting from daily thunderstorms occurred in June of 1986. In 1987, however, below

normal rainfall during early summer resulted in low water levels persisting through July.

Water depths at nest sites usually were shallower, by 10 cm or more, than depths in

the surrounding open water sloughs where the kites often foraged. This resulted from kites

nesting within inundated tree islands or sawgrass stands which often were 10 cm or more

higher elevation than the surrounding sloughs (see also U.S.D.I. 1972, McPherson 1973,

Worth 1983). As a consequence of higher elevation, some nest sites dried out (i.e. nests

that were built initially over water); as the seasons progressed; however, we did not

observe any nests in which the surrounding sloughs dried completely.

The permanent water gauging stations in this region (i.e. 3-4 and 3-28) are located in

open sloughs. These gauges therefore indicated depths of the foraging habitat as opposed to

nest sites. Although we had water depth gauges at each nest site, some general trends are

illustrated (see Fig. 19) by these continuous water depth recording stations. It should be

1986

40

30-

20-

0 I A V I '11 / I ^'- r \ I.
5 15 25 35 45 55 65 75 85 95 105
WATER DEPTH AT INTIATION (cm)

1987
701

60

40

30 0

20

10

6 15 25 35 45 55 65 75
WATER DEPTH AT IITIATION (cm)

85 95 105

Fitgur 1. Frequency distribution of the number of nests found in each water depth class
(10 cm intervals) at the time of breeding initiation The depth value shown is the
midpoint of the 10 cm class (e.g. 15 cm represents water depths ranging from 10.1 to 20.0
Cem elusive). q,

i//

LI -- __ -

I ~ mn w m |- =

_ i

1986

0E 4 0-
OA ....... .... .
0.3 1.0

00.5
S 0 0
1 15 1 1 1 1 15 1I5 1 15 1 15
SJAN FEB MAR APR MAY JUN JU 0
2 DATE
1987

115 1 15 1 15 1 15 1 15 1 15 1 15
JAN FEB MAR APR MAY JUN JUL

DATE

Figure 19. Mean daily water depths within the elevation range used by nesting Snail Kites
(shaded). Values were derived by subtracting the ground elevation (i.e. 2.1 2.5 m) from
the mean daily water stage at gauges 3-4 (higher elevation) and 3-26 (lower elevation).
Note that these gauges are located within open slough communities and therefore represent
depths of 10 cm or deeper than would be expected at actual nest sites.
z -.0 Uj

0.2

1 1 1 1 15 15 15
JAN FEB MAR APR MAY JUN AUL
DATE

Figure 19. Mean daily water depths within the elevation range used by nesting Snail Kites
(shaded). Values were derived by subtracting the ground elevation (i.e. 2.1 2.5 m) from
the mean daily water stage at gauges 3-4 (higher elevation) and 3-26 (lower elevation).
Note that these gauges ar located within open slough communities and therefore represent
depths of 10 cm or deeper than would be expected at actual neat aites.

Habitat Selection

Water detth.-- We were able to estimate the water depth for the time that breeding

was initiated (i.e. when the first egg was laid) for 281 nests. At this stage of breeding,

almost all nests were over water (280 of 281 nests), and most sites (94% of 281 nests) had

water depths ranging from 20 80 cm. In only one case during 1986 and 1987 did we

encounter a nest that was built over dry land. This nest was built within an ongoing kite

colony that recently had dried out directly under the nest trees. The surrounding sloughs

of this colony, however, were completely inundated. Water depth at nest sites during the

initiation of breeding ranged from 0 to 75 cm in 1986 and from 21 to 105 cm in 1987 (Fig.

18). Mean depth at the time of initiation was lower in 1986 (x 41.22, n 96) than in

1987 (x 49.63, n = 185) (t 4.69, P < 0.01).

Although most nests were initiated in areas with water depths ranging from 20 80 cm,

considerable fluctuation in depth occurred throughout the season. A prolonged drying trend

occurred during both years from March through May (Fig. 19). A strong increase in water

depth resulting from daily thunderstorms occurred in June of 1986. In 1987, however, below

normal rainfall during early summer resulted in low water levels persisting through July.

Water depths at nest sites usually were shallower, by 10 cm or more, than depths in

the surrounding open water sloughs where the kites often foraged. This resulted from kites

nesting within inundated tree islands or sawgrass stands which often were 10 cm or more

higher elevation than the surrounding sloughs (see also U.S.D.I. 1972, McPherson 1973,

Worth 1983). As a consequence of higher elevation, some nest sites dried out (i.e. nests

that were built initially over water); as the seasons progressed; however, we did not

observe any nests in which the surrounding sloughs dried completely.

The permanent water gauging stations in this region (i.e. 3-4 and 3-28) are located in

open sloughs. These gauges therefore indicated depths of the foraging habitat as opposed to

nest sites. Although we had water depth gauges at each nest site, some general trends are

illustrated (see Fig. 19) by these continuous water depth recording stations. It should be

noted therefore that these gauges often indicate depths of 10 cm or more than would be

expected at nest sites.

Dry-down interval.-- Snail Kites in WCA-3A were distributed throughout the area

between 2.1 m and 2.5 m elevation (see Distribution of Nesting Snail Kites). Based on the

minimum water levels recorded at gauges 3-4 and 3-28 this elevation range dries out to

ground level or below approximately every 1.9 to 3.8 years (Fig. 20).

A large portion of WCA-3A dries out more frequently than the areas in which we found

nesting kites (i.e. areas above 2.5 m elevation). We observed Snail Kites foraging in the

wet prairie communities of these higher elevation areas during times of high water, but

found no indication (e.g. courtship displays and stick carrying) that any nesting occurred

in these areas.

A region of elevation lower than 2.1 m (i.e. an area that dries out less frequently

than every 3.8 years) occurs along the northern portion of the L-67 levee, but relatively

few kites were observed in this area. Snail Kites nested in one large willow head (N 6)

on the edge of this wetter area during 1986, but only one kite was observed in the area

during 1987 and no nests were found.

Proportion of open water.-- The habitat in which Snail Kites nested had a ratio of

open water to sawgrass ranging from 12 67% (x 32.4 0.13 (SD) for nesting areas of

1986 and 1987 combined). Within the general area of nesting distribution (i.e. between 2.1

and 2.5 m elevation) the ratio of open water to sawgrass in areas used and not used by

nesting kites did not differ significantly in either 1986 or 1987 (Mann-Whitney Tests, P >

0.05) (Fig. 21). Differences in the proportion of open water of areas used for nesting

aLso did not differ significantly between 1986 and 1987 (Mann-Whitney Test, P > 0.05).

Systematically sampled areas sampled above 2.5 m elevation (where we found no nesting
kites) had a significantly lower proportion of open water than areas below 2.5 m that were

sed for nesting in 1986 (Mann-Whitney Test, P < 0.05) or 1987 (Mann-Whitney Test, P <

0 05) (Fig. 22).

noted therefore that these gauges often indicate depths of 10 cm or more than would be

expected at nest sites.

Dry-down interval.-- Snail Kites in WCA-3A were distributed throughout the area

between 2.1 m and 2.5 m elevation (see Distribution of Nesting Snail Kites). Based on the

minimum water levels recorded at gauges 3-4 and 3-28 this elevation range dries out to

ground level or below approximately every 1.9 to 3.8 years (Fig. 20).

A large portion of WCA-3A dries out more frequently than the areas in which we found

nesting kites (i.e. areas above 2.5 m elevation). We observed Snail Kites foraging in the

wet prairie communities of these higher elevation areas during times of high water, but

found no indication (e.g. courtship displays and stick carrying) that any nesting occurred

in these areas.

A region of elevation lower than 2.1 m (i.e. an area that dries out less frequently

than every 3.8 years) occurs along the northern portion of the L-67 levee, but relatively

few kites were observed in this area. Snail Kites nested in one large willow head (N 6)

on the edge of this wetter area during 1986, but only one kite was observed in the area

during 1987 and no nests were found.

Proportion of open water.-- The habitat in which Snail Kites nested had a ratio of

open water to sawgrass ranging from 12 67% (x 32.4 0.13 (SD) for nesting areas of

1986 and 1987 combined). Within the general area of nesting distribution (i.e. between 2.1

and 2.5 m elevation) the ratio of open water to sawgrass in areas used and not used by

nesting kites did not differ significantly in either 1986 or 1987 (Mann-Whitney Tests, P >

0.05) (Fig. 21). Differences in the proportion of open water of areas used for nesting

aLso did not differ significantly between 1986 and 1987 (Mann-Whitney Test, P > 0.05).

Systematically sampled areas sampled above 2.5 m elevation (where we found no nesting
kites) had a significantly lower proportion of open water than areas below 2.5 m that were

sed for nesting in 1986 (Mann-Whitney Test, P < 0.05) or 1987 (Mann-Whitney Test, P <

0 05) (Fig. 22).

4
t*

0

z
O

2

YEAR

Figure 20.
The shaded

Minimum monthly water stage in WCA-3A (average of gaes 3-4 and 2-2) from 196 196.
area represents the elevation range for which we found breedinS Snai Kit..

2 11
S8

= 0.4-
z
0.
0a

oI
S 0.3

o 0.2

0.1

0.0
1986 1986 1987 1987 IOTSIDE
NESTING NON NESTINI NON SF THE
AREA NESTINO AREA NESTING NESTING
AREA AREA RANGE

Figure 21. Proportion of open water to sawgrass for areas within the nesting range of Snail Kites in
WCA-3A that were used and not used for nesting. Also shown is the proportion of open water from
systematically sampled areas outside of the nesting range (i.e. above 2.5 m elevation).

- N-

0 5km
&- mr-

M Sawgrass E Open Water M Tree Islands

] Snail Kite Nesting Distribution

Figure 22. Classified GIS Map of southern WCA-3A showing distribution of nesting Snail
Kites (shaded). Note the amount of open water within the distribution compared with
outside the nesting distribution.

Aoole snail abundance.-- If snail abundance is an important determinant of habitat

selection for Snail Kites, then it should be greater in areas of higher nesting density.

We hypothesized therefore that areas of higher nesting density would have lower capture

times, higher apple snail egg cluster indices, and higher snail counts from suction

dredging.

There was no significant difference in the time that it took foraging birds to capture

snails in areas of high nesting density compared with areas of low nesting density

(Wilcoxon paired-sample test, P 0.43); however, in three of four paired comparisons

capture time was greater in areas of high nesting density (Fig. 23). This result was not

consistent with our prediction that capture times would be lower in areas of high nesting

density.

In contrast, apple snail egg cluster densities and the egg cluster indices were

significantly higher in areas of higher kite nesting density for all of the paired samples

(Wilcoxon paired-sample test, P = 0.05) (Figs. 24 and 25). These results were consistent

with our predictions.

Because of time and logistical constraints we were only able to sample two areas using

the portable suction dredge (one each of high and low nesting density). Twenty plots each

were dredged at areas 5 and 6. We found no significant difference in snail abundance

between the two areas (Mann-Whitney test, P 0.46); however, our snail density estimates

were higher in the area of high nesting density (x 0.65 per m2 1.04 [SD]) than in the

area of low nesting density (x = 0.45 per m2 0.60 [SD]).

Capture time and snail egg cluster indices were not significantly correlated (r2 =

0.18, P > 0.05) (Fig. 26). We did not have a sufficiently large sample size to
tatistically assess the correlation between suction dredging and the two snail abundance

ndices however, results were consistent between the suction dredging and egg cluster

counts. but not between suction dredging and capture times.

900-

800-

700-

600-

500-

400-

300-

200 -

100 -

0-

Figure 23. Mean capture times ( SD) of foraging Snail Kites in areas of low (-) and high (--) nesting
densities. Areas 1-2, 3-4, 5-6, and 7-8 were paired samples observed observed during approximately the
period of time.

23 31 14 26
S23
J I4

to I I I I I I
1 2 3 4 5 6 7 8
NESTING AREA

E

ca

0

WU

14

cm 12

29

I
I
I
I
I
I
30
4.
I
I
I
I
I
I
I
I

27
i
23 I
I
I
I
I

7 8

NESTING AREA

Figure 24. Egg clusters/10 m2 in eas of low (-) and high (--) nesting densities. Areas 1-2, 3-4, 5-6
and 7-8 were paired samples observed during approximately the same period of time.

26
.
I
I
I
28 I
I
*
I
I
I
I

10

8

6

4

23 20

t i

4-

3.5 -

S -
3-

1 2.3 5 -

Figure 25. Egg cluster indices (see methods) in areas of low (ai) and high (*) nesting densities. Areas
-2, 3-4, d 7- er p d maple served during pproimtely t me period of time.
o

0;-

1 2 3 4 5 6 7 8
HESTIHG AREA

Fige 25. Egg cluster indices (see methods) in areas of low (0) and high (4) nesting densities. Areas
1-2, 3-4, 5-6, sad 7-8 were paired samples observed during approximately the same period of time.

S00 -

400

300-

200-

a O a
D

100- 0 0

0 --.... I ...I.. ......- I ... ...-I "--"---
0 1 2 3
EGO CLUSTER INDEX

Figure 26. Scattergram showing the relationship between capture times and egg cluster indices for the
eight nesting areas sampled.

Nest-site Selection

Nest substrate.-- The relative use of nest substrates did not differ between 1986 and

1987 (X2 6.58, P > 0.25, df 5); however, the use of nest substrates did differ from

what was available (X2 79.21, P < 0.001, df 6). Willow was the most frequently used

substrate in both 1986 and 1987, followed in decreasing order of use by pond apple,

cypress, melaleuca, and wax myrtle (Fig. 27). Snail Kites also rarely used coco plum (<

4%), sweetbay (< 2%), sawgrass (< 2%), buttonbush (< 1%), and cattail (< 1%) as nest

substrates.

Although willow was the most frequently used substrate, it was used less than expected

compared to its relative abundance in WCA-3A (Fig. 28). Sweet bay also was used less than

expected from its relative abundance, but the departure from expected was not as

pronounced. Pond apple was used considerably more than expected based on its relative

abundance. Because we measured the availability of nest substrates by the number of

available clumps < 100 m2, rather than by total area (see methods), we did not assess the

use of sawgrass or cattail compared to their relative abundance; however, because sawgrass

was extremely abundant and cattail common, both were undoubtedly used less than would have

been expected. The remaining substrates were used at nearly expected frequencies.

Nest heinht.-- Nests ranged in height from 0.9 m (3 ft) to 12.4 m (41 ft) above ground

level in 1986, and from 0.9 m to 8.6 m (28 ft) in 1987. Nest height in 1986 (x 2.24 m [7

ft], SE 0.06; outliers removed [Sokal and Rohlf 1969]) did not differ significantly from

1987 (x 2.36 m [8 ft], SE 0.08; outliers removed) (t = 1.10, P 0.27) (Fig. 29).

Although we did not measure what heights were potentially availability to nesting

kites, the heights selected appeared to correspond with the height of the stand (or part of

the stand) within which the birds were nesting. Higher nest sites were potentially

available at sites that were not selected for nesting (e.g. hardwood hammocks).

Stand size.-- Although we estimated the stand size within which each kite nest was

located, we did not measure each stand. There were, however, no obvious preferences shown

100

90,

80,

70

S60

a. 40-

30

20

10

0
1986 1987
YEAR
J willow melaleuca cypress

pond apple E wax myrtle other

Figure 27. The relative use of met substrates in WCA-3A during 1986 and 1987.

> 4UU
z
w

Sb 300

200-
a

- 100

-50- .-.
WILLOW CYPRESS POND WAX COCO SWEET
MELALEUCA APPLE MYRTLE PLUM BAY

Figure 28. Percent departures from expected frequencies of use of nesting substrates compared with
their abundance in WCA-3A.

1986

to
o -

2 3 A 6 7 a. 10 11 12 is
IoST Hmor (m)

1987

120-

110-

100-

S*//

40
EsT hIHO (m)
SO -

10 -
Ito

to

1 2 S 4 S 6 to as Ia Is
d-go

srn toinnr Cm)

Figure 29. Frequency distribution of Snail Kite nests in WCA-3A during 1986 and 1987 by
est height (1 m increment).

Nest-site Selection

Nest substrate.-- The relative use of nest substrates did not differ between 1986 and

1987 (X2 6.58, P > 0.25, df 5); however, the use of nest substrates did differ from

what was available (X2 79.21, P < 0.001, df 6). Willow was the most frequently used

substrate in both 1986 and 1987, followed in decreasing order of use by pond apple,

cypress, melaleuca, and wax myrtle (Fig. 27). Snail Kites also rarely used coco plum (<

4%), sweetbay (< 2%), sawgrass (< 2%), buttonbush (< 1%), and cattail (< 1%) as nest

substrates.

Although willow was the most frequently used substrate, it was used less than expected

compared to its relative abundance in WCA-3A (Fig. 28). Sweet bay also was used less than

expected from its relative abundance, but the departure from expected was not as

pronounced. Pond apple was used considerably more than expected based on its relative

abundance. Because we measured the availability of nest substrates by the number of

available clumps < 100 m2, rather than by total area (see methods), we did not assess the

use of sawgrass or cattail compared to their relative abundance; however, because sawgrass

was extremely abundant and cattail common, both were undoubtedly used less than would have

been expected. The remaining substrates were used at nearly expected frequencies.

Nest heinht.-- Nests ranged in height from 0.9 m (3 ft) to 12.4 m (41 ft) above ground

level in 1986, and from 0.9 m to 8.6 m (28 ft) in 1987. Nest height in 1986 (x 2.24 m [7

ft], SE 0.06; outliers removed [Sokal and Rohlf 1969]) did not differ significantly from

1987 (x 2.36 m [8 ft], SE 0.08; outliers removed) (t = 1.10, P 0.27) (Fig. 29).

Although we did not measure what heights were potentially availability to nesting

kites, the heights selected appeared to correspond with the height of the stand (or part of

the stand) within which the birds were nesting. Higher nest sites were potentially

available at sites that were not selected for nesting (e.g. hardwood hammocks).

Stand size.-- Although we estimated the stand size within which each kite nest was

located, we did not measure each stand. There were, however, no obvious preferences shown

Nest-site Selection

Nest substrate.-- The relative use of nest substrates did not differ between 1986 and

1987 (X2 6.58, P > 0.25, df 5); however, the use of nest substrates did differ from

what was available (X2 79.21, P < 0.001, df 6). Willow was the most frequently used

substrate in both 1986 and 1987, followed in decreasing order of use by pond apple,

cypress, melaleuca, and wax myrtle (Fig. 27). Snail Kites also rarely used coco plum (<

4%), sweetbay (< 2%), sawgrass (< 2%), buttonbush (< 1%), and cattail (< 1%) as nest

substrates.

Although willow was the most frequently used substrate, it was used less than expected

compared to its relative abundance in WCA-3A (Fig. 28). Sweet bay also was used less than

expected from its relative abundance, but the departure from expected was not as

pronounced. Pond apple was used considerably more than expected based on its relative

abundance. Because we measured the availability of nest substrates by the number of

available clumps < 100 m2, rather than by total area (see methods), we did not assess the

use of sawgrass or cattail compared to their relative abundance; however, because sawgrass

was extremely abundant and cattail common, both were undoubtedly used less than would have

been expected. The remaining substrates were used at nearly expected frequencies.

Nest heinht.-- Nests ranged in height from 0.9 m (3 ft) to 12.4 m (41 ft) above ground

level in 1986, and from 0.9 m to 8.6 m (28 ft) in 1987. Nest height in 1986 (x 2.24 m [7

ft], SE 0.06; outliers removed [Sokal and Rohlf 1969]) did not differ significantly from

1987 (x 2.36 m [8 ft], SE 0.08; outliers removed) (t = 1.10, P 0.27) (Fig. 29).

Although we did not measure what heights were potentially availability to nesting

kites, the heights selected appeared to correspond with the height of the stand (or part of

the stand) within which the birds were nesting. Higher nest sites were potentially

available at sites that were not selected for nesting (e.g. hardwood hammocks).

Stand size.-- Although we estimated the stand size within which each kite nest was

located, we did not measure each stand. There were, however, no obvious preferences shown

Nest-site Selection

Nest substrate.-- The relative use of nest substrates did not differ between 1986 and

1987 (X2 6.58, P > 0.25, df 5); however, the use of nest substrates did differ from

what was available (X2 79.21, P < 0.001, df 6). Willow was the most frequently used

substrate in both 1986 and 1987, followed in decreasing order of use by pond apple,

cypress, melaleuca, and wax myrtle (Fig. 27). Snail Kites also rarely used coco plum (<

4%), sweetbay (< 2%), sawgrass (< 2%), buttonbush (< 1%), and cattail (< 1%) as nest

substrates.

Although willow was the most frequently used substrate, it was used less than expected

compared to its relative abundance in WCA-3A (Fig. 28). Sweet bay also was used less than

expected from its relative abundance, but the departure from expected was not as

pronounced. Pond apple was used considerably more than expected based on its relative

abundance. Because we measured the availability of nest substrates by the number of

available clumps < 100 m2, rather than by total area (see methods), we did not assess the

use of sawgrass or cattail compared to their relative abundance; however, because sawgrass

was extremely abundant and cattail common, both were undoubtedly used less than would have

been expected. The remaining substrates were used at nearly expected frequencies.

Nest heinht.-- Nests ranged in height from 0.9 m (3 ft) to 12.4 m (41 ft) above ground

level in 1986, and from 0.9 m to 8.6 m (28 ft) in 1987. Nest height in 1986 (x 2.24 m [7

ft], SE 0.06; outliers removed [Sokal and Rohlf 1969]) did not differ significantly from

1987 (x 2.36 m [8 ft], SE 0.08; outliers removed) (t = 1.10, P 0.27) (Fig. 29).

Although we did not measure what heights were potentially availability to nesting

kites, the heights selected appeared to correspond with the height of the stand (or part of

the stand) within which the birds were nesting. Higher nest sites were potentially

available at sites that were not selected for nesting (e.g. hardwood hammocks).

Stand size.-- Although we estimated the stand size within which each kite nest was

located, we did not measure each stand. There were, however, no obvious preferences shown

for any.particular .stand size, with the possible exception of a slight tendency towards the

use of single shrubs or small patches of woody vegetation.

There were, however, obvious preferences for the placement of nests within stands.

For example, even though nests frequently were located in large tree islands, we never

found a nest within the dry hammock portion of any tree island. Nests within large tree

islands usually were located within the trailing southern portion that was in deeper water

(Fig. 30). These trailing portions usually were comprised predominantly of willow, but

kites frequently selected single pond apples or other species as nest substrates when they

were available.

When nests were located in larger tree islands they also were usually placed in

isolated shrubs adjacent to the main body of the stand or in the outermost edges of the

canopy (Fig. 31). Unlike many of the wading birds (with which kites often nested), nests

seldom were placed far within a dense canopy.

Size of the Breeding Population

We found 148 nests in which breeding (i.e. at least I egg was laid) occurred in 1986;

227 nests were found in 1987. An improved estimate of the number of breeding attempts in

the area was obtained by calculating the number of breeding attempts that would have to

have had been initiated in order to observe the number of successful nests that were found,

given the probability that a nest would be successful (Miller and Johnson 1978). This

calculation yielded an estimate of 196 and 284 breeding attempts during 1986 and 1987,

respectively (based on our Mayfield estimates of nesting success). These estimates also

may be low because of the assumption that all successful nests are found, a condition

probably not true for this study. In several cases we felt that the time required to

locate all nests within a colony could have been detrimental to the eggs or young of those

adults that were kept off their nest during the search. In such cases we restricted the

time of our search, regardless of whether all nests had been found. We also did not study

zone usually selected by nesting Snail Kites

WILLOW

HARDWOOD
HAMMOCK

Figure 30. Legnthwise cross section of ypicl tree island in WCA-3A showing the zone usually selected
by nesting Snail Kites. The hardwood hammock portion is usually on the northern end of the islds.

jFPRIMIAY NESTINI ZINE

Figure 31. Cross section of typical small ( 500 m2) willow head showing th one usually selected by
nesting Snail Kites.

'451

- r

one colony (Miccosukee) of nesting kites because it was regularly under observation by the

public. We suggest that reasonable adjusted estimates of the number of breeding attempts

in WCA-3A are 200-250 during 1986 and 300-350 during 1987. These estimates, however, would

not include nest initiation in which breeding (i.e. eggs laid) did not occur.

We found 167 occupied nests (i.e. nests with actively attending adults but not

necessarily having initiated breeding) in 1986, and 237 in 1987. This estimate, however,

is undoubtedly low because most nests were found after breeding had been initiated and many

failures before eggs laying were probably missed in our surveys.

Nesting Success

Our overall estimates (Mayfield) of nesting success in WCA-3A was 23% in 1986, and 36%

in 1987 (Appendices 3). Success was significantly greater in 1987 than in 1986 (Z 2.86,

P < 0.05) (Fig. 32).

Survivorship and Ate-Svecific Nest Failure

Daily nest survival differed significantly between the egg and nestling stages during

1986 (Z 3.40, P < 0.001), but not during 1987 (Z 1.53, P 0.13). Contrary to the

findings of Beissinger (1986) and Sykes (1987b), daily nest survival during the nestling

stage of 1986 was lower than during the egg stage. This may in part result from our

inclusion of nests that failed after the predicted hatch date (i.e. day 27) in the nestling

stage. Except in cases of hatching failure or when we had evidence that the eggs had not

hatched, we included nests that failed after day 27 to have failed in the nestling stage

(i.e. we assumed that they had hatched). There was no significant difference in overall

success between the incubation periods of 1986 and 1987 (Z 0.73, P 0.23) (Fig. 33);

however, success through the nestling period was significantly greater (Z 4.38, P <

0.001) in 1987 than in 1986.

one colony (Miccosukee) of nesting kites because it was regularly under observation by the

public. We suggest that reasonable adjusted estimates of the number of breeding attempts

in WCA-3A are 200-250 during 1986 and 300-350 during 1987. These estimates, however, would

not include nest initiation in which breeding (i.e. eggs laid) did not occur.

We found 167 occupied nests (i.e. nests with actively attending adults but not

necessarily having initiated breeding) in 1986, and 237 in 1987. This estimate, however,

is undoubtedly low because most nests were found after breeding had been initiated and many

failures before eggs laying were probably missed in our surveys.

Nesting Success

Our overall estimates (Mayfield) of nesting success in WCA-3A was 23% in 1986, and 36%

in 1987 (Appendices 3). Success was significantly greater in 1987 than in 1986 (Z 2.86,

P < 0.05) (Fig. 32).

Survivorship and Ate-Svecific Nest Failure

Daily nest survival differed significantly between the egg and nestling stages during

1986 (Z 3.40, P < 0.001), but not during 1987 (Z 1.53, P 0.13). Contrary to the

findings of Beissinger (1986) and Sykes (1987b), daily nest survival during the nestling

stage of 1986 was lower than during the egg stage. This may in part result from our

inclusion of nests that failed after the predicted hatch date (i.e. day 27) in the nestling

stage. Except in cases of hatching failure or when we had evidence that the eggs had not

hatched, we included nests that failed after day 27 to have failed in the nestling stage

(i.e. we assumed that they had hatched). There was no significant difference in overall

success between the incubation periods of 1986 and 1987 (Z 0.73, P 0.23) (Fig. 33);

however, success through the nestling period was significantly greater (Z 4.38, P <

0.001) in 1987 than in 1986.

100

80-

.j -

223
S40

144

20-

0 I
1986 1987
YEAR

Figure 32. Mayfield estimates of overall nesting success of Snail Kite nests in WCA-3A
during 1986 and 1987. Ninety-five percent confidence intervals about the estimate and
sample sizes are shown.

100-

80-

60-

40-

20-

m

E.I ~I

&

148

128

INC 1986

201

I 198
NEST 1986

INC 1987 NEST 1987

NESTING STAGE

Figure 33. Mayfield estimates of nesting success during the incubation and nestling stages
of Snail Kite nests in WCA-3A during 1986 and 1987. Ninety-five percent confidence
intervals about the estimate and sample sizes are shown.

n

Mayfield estimates of daily survival during successive 6-day intervals after hatching

showed that survival was lowest during the first 6-day interval after hatching for both

1986 and 1987 (Fig. 34). There was a significant difference (Z = 2.93, P 0.002) in 1986

between the first and last 6-day interval after hatching (Appendix 4); the first interval

also was significantly different from the third (Z 3.10, P 0.001) and fourth (Z = 2.09,

P 0.18) 6-day periods during 1987.

Another method for illustrating age-specific survivorship is plotting the proportion

of nests with a known date of initiation that survive each day. This approach revealed

that failure during the egg stage of 1986 tended to occur late (Fig. 35). This result

probably was due both to hatching failure (11 cases in 1986) and to nests that might have

failed after hatching but were mis-classified as egg-stage failures. This latter result

arises from our procedure of estimating the failure date as the midpoint between the last.

nest visit when the nest was viable and the first visit after failure. This pattern of

late failure during the egg stage was not as dramatic during 1987. Although not as

pronounced as the Mayfield comparison using 6-day intervals, this approach to age-specific

survivorship also showed a tendency for failure during the nestling stage to occur earlier

(i.e. concave curve between day 27 and day 51). The less pronounced change in survivorship

shown from this method probably is the result of presenting failures on a daily basis

rather than lumping by 6-day intervals.

Productivity

The frequency distributions of different-sized clutches were not different between

1986 and 1987 (X2 2.62, P > 0.95, df 3) (Table 1). Clutch sizes ranged from one to

three in 1986, and from one to four in 1987; modal clutch size was three during both years.

The mean ( S.D.) clutch sizes of 2.59 ( 0.61) and 2.53 ( 0.64) for 1986 and 1987,

respectively, were slightly lower than those reported by Beissinger (1986) or Sykes

1986

100-

98-

96-

94-

1987

100-

98-

92
I

9g-

94-

92-

92-

90 1 i.
slet 2nd 3rd 4th

121

I

141

111

I

107

*

vu 2 r
1st 2nd 3rd 4th

SIX DAY INTERVAL AFTER HATCHING

Figure 34. Daily survival of Snail Kite nests in WCA-3A of four consecutive 6-day nestling periods
during 1986 and 1987. Ninety-five percent confidence intervals about the estimates and sample sizes are
shown.

1986 1987
100- 100

90 % 90

080 *0 7 V
1 600 0 060 00

5 50- 50*
0 hatchhig hatch

30 *3

20' nuo iu, l ,,, m..." i|u u'" "| 'u 'uh '1 'h 'u | 20 i"""'hhiii ; gI5""N" i"i'i"i"""
10 20 30 40 50 10 20 30 4
DAY IN NEST CYCLE DAY IN NEST CYCLE

Figure 35. Survivorship of Snail Kite nets through the nesting cycle, calculated s the percent of nests found
that survived daily (-). Survivorship from Mayfield estimates (--) are shown for comparison.

Table 1. Clutch sizes reported in Florida since 1880.

1-Egga 2-Egg 3-Egg 4-Egg 5-Egg 6-Egg
Clutches Clutches Clutches Clutches Clutches Clutches

Year(s) N X No. % No. % No. % No. % No. % No. % Source

1880-1925 91 3.23a -- 12 13 54 59 17 19 8 9 0 0 Beissinger (1986)

1925-1959 57 2.96a -- 9 16 43 75 4 7 0 0 1 2 Beissinger (1986)

1968-1978 --- --- 2 2 23 17 101 80 1 1 0 0 0 0 Sykes (1987b)

1979-1983 48 2.71a -- 48 31 105 67 3 2 0 0 0 0 Beissinger (1986)

1986 109 2.59 7 6 31 28 71 65 0 0 0 0 0 0 This study

1987 161 2.53 11 6 55 34 93 58 2 1 0 0 0 0 This study

a Beissinger (1986) excluded one-egg clutches as being incomplete.

(1987b); however, they were within the range of variability found by Snyder et al. (in

review).

Hatchability was not significantly different between 1986 and 1987 for 2-egg clutches

(Mann-Whitney, P 0.41) or for 3-egg clutches (Mann-Whitney, P 0.54) (Table 2).

Hatchability differed, however, between 2 and. 3-egg clutches during 1986 (Mann-Whitney, P =

0.04); differences in hatchability from 2 and 3-egg clutches were nearly significant for

1987 (Mann-Whitney, P = 0.07). These results indicate that on a per-egg basis,

productivity was greater for 2-egg clutches than for 3-egg clutches.

Although there was a disproportionately higher number of nests that fledged two and

three (rather than one) young in 1987 compared with 1986, this difference was not

statistically significant (X2 5.15, 0.05
1987 than 1986, whether expressed on a per breeding attempt, occupied nest, or successful

nest basis (Table 3).

We observed 65 young reach fledging age in 1986, and 172 in 1987. Because it is

unlikely that we found all successful nests, these figures should be considered minimum

estimates of production. Based on our estimates of the number of nests, their nesting

success, and the number of young fledged per successful nest, we estimate that 68-83 young

reached fledging age in WCA-3A during 1986, and 178-208 during 1987. These are production

values, however, and should not be assumed to represent recruitment estimates since we have

no measure of juvenile survival during 1986 or 1987.

Causes of Nest Failure

Predators.-- The most common situation we observed when visiting a failed nest was to

find all eggs or young missing, with no indication that the nest structure had been

disturbed (Table 4). When this occurred at nests built on a sturdy substrate we suspected

that the probable cause of failure was predation. The contents of these empty nests could

possibly have been scavenged subsequent to mortality due to other causes; however, the

(1987b); however, they were within the range of variability found by Snyder et al. (in

review).

Hatchability was not significantly different between 1986 and 1987 for 2-egg clutches

(Mann-Whitney, P 0.41) or for 3-egg clutches (Mann-Whitney, P 0.54) (Table 2).

Hatchability differed, however, between 2 and. 3-egg clutches during 1986 (Mann-Whitney, P =

0.04); differences in hatchability from 2 and 3-egg clutches were nearly significant for

1987 (Mann-Whitney, P = 0.07). These results indicate that on a per-egg basis,

productivity was greater for 2-egg clutches than for 3-egg clutches.

Although there was a disproportionately higher number of nests that fledged two and

three (rather than one) young in 1987 compared with 1986, this difference was not

statistically significant (X2 5.15, 0.05
1987 than 1986, whether expressed on a per breeding attempt, occupied nest, or successful

nest basis (Table 3).

We observed 65 young reach fledging age in 1986, and 172 in 1987. Because it is

unlikely that we found all successful nests, these figures should be considered minimum

estimates of production. Based on our estimates of the number of nests, their nesting

success, and the number of young fledged per successful nest, we estimate that 68-83 young

reached fledging age in WCA-3A during 1986, and 178-208 during 1987. These are production

values, however, and should not be assumed to represent recruitment estimates since we have

no measure of juvenile survival during 1986 or 1987.

Causes of Nest Failure

Predators.-- The most common situation we observed when visiting a failed nest was to

find all eggs or young missing, with no indication that the nest structure had been

disturbed (Table 4). When this occurred at nests built on a sturdy substrate we suspected

that the probable cause of failure was predation. The contents of these empty nests could

possibly have been scavenged subsequent to mortality due to other causes; however, the

Table 2. Hatching success from nests with 2 and 3-egg clutches in WCA-3A during 1986 and 1987.

2-Egg clutches 3-Egg clutches

Total Xno. Total Xno.
no. Percent young no. Percent young
No. young eggs hatched No. young eggs hatched
Year nests hatched hatched per nest nests hatched hatched per nest

1986 16 30 94 1.88 43 103 80 2.40

1987 37 64 86 1.73 67 155 77 2.31

Table 3. Estimates of Snail Kite productivity in south Florida from 1968 through 1987.

No. No. No.
Total no. Total no. Total no. fledged fledged fledged
occupied active successful Total no.. per occ. per act. per succ.
Year nests nests nests fledged nest nest nest Source

-2.18

-1.63

1.50

2.33

2.42

-1.83

-2.50

-1.36

-2.50

-- 1.82

.16. 2.00

1968

1969

1970

1971

1972

1973

1974

1975

1976

1977

1978

1979

1981

1982

1983

1986

1987

2.00

2.00

1.44

1.65

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Sykes (1987b)

Beissinger (1986)

Beissinger (1986)

Beissinger (1986)

Beissinger (1986)

This Study

This Study

a No nesting activity was detected during this year.

1.85

1.00

0.63

0.00

0.13

0.63

0.44

0.76

1.17

0.85

0.31

1.21

0.88

1.33

1.43

0.82

0.00

0.10

0.42

0.38

0.73

1

93

8

19

32

148

227

Table 4. Condition of unsuccessful Snail Kite nests when found.

1986 1987

% of % of
total total
Condition of failed nest when found No. failures No. failures

Empty and intact 58 63 69 64

Structure tilted >150 and eggs or
young missing 10 11 10 9

Structure tilted >15 and broken
eggs or dead young present 0 0 4 4

Broken eggs but structure intact 11 12 7 7

Dead young but structure intact 8 9 6 6

Structure intact with eggs or young,
but no evidence of adult attendance 5 5 10 9

Predation observed 0 0 1 1

remains (e.g. feathers and bones) of young known to have died from other causes usually

were detectable for at least one visit after the death occurred. It was therefore unlikely

that widespread mortality with subsequent scavenging went undetected.

In an attempt to assess which predators were responsible for nest losses, we applied a

layer of synthetic grease over an approximately 60 cm portion of the trunk of 12 shrubs

used to support kite nests. We raked the grease surface with a linoleum comb so that any

animal crossing the surface would leave identifiable tracks. Only three such nests failed

while this grease was in place: one trunk had positively been climbed by a snake, one had

been climbed by what appeared to be a snake, and one showed no apparent signs of having

been climbed (although it could have been possible for a snake to have bypassed the

grease). Although this evidence certainly is not conclusive as to the cause of these nest

failures, we believe that two of the three nests probably were preyed upon by snakes. In

addition, two actual observations of attempted snake predation have been reported.

Bennetts and Caton (1988) observed a rat snake (Elaphe obsoleta) prey upon a nestling Snail

Kite chick; and J. Kern (Toner 1984) photographed a rat snake attempting to swallow a kite

egg.

Structural collapse.-- Ten nests (11% of failures) during 1986 and 14 nests (13% of

failures) during 1987 experienced some degree (>15% tilted) of structural collapse. Some

of these failures, however, may have been caused by other factors.

Abandonment.-- We found a total of 15 cases of apparent abandonment in 1986 and 1987

(see Table 4); however, three of these nests previously had undergone a partial egg or

young loss (probably due to predation). In five cases, incubation had extended well beyond

the normal incubation period suggesting that the eggs were not viable; in five additional

cases the nest had only one egg, suggesting that a partial loss might have occurred prior

to our discovery of the nest. In only 2 of 375 nests did we observe what we believed to be

abandonment of viable eggs or young in which the nest had not had at least partial

predation. Even when eggs were not viable, kites appeared reluctant to abandon their nest.

remains (e.g. feathers and bones) of young known to have died from other causes usually

were detectable for at least one visit after the death occurred. It was therefore unlikely

that widespread mortality with subsequent scavenging went undetected.

In an attempt to assess which predators were responsible for nest losses, we applied a

layer of synthetic grease over an approximately 60 cm portion of the trunk of 12 shrubs

used to support kite nests. We raked the grease surface with a linoleum comb so that any

animal crossing the surface would leave identifiable tracks. Only three such nests failed

while this grease was in place: one trunk had positively been climbed by a snake, one had

been climbed by what appeared to be a snake, and one showed no apparent signs of having

been climbed (although it could have been possible for a snake to have bypassed the

grease). Although this evidence certainly is not conclusive as to the cause of these nest

failures, we believe that two of the three nests probably were preyed upon by snakes. In

addition, two actual observations of attempted snake predation have been reported.

Bennetts and Caton (1988) observed a rat snake (Elaphe obsoleta) prey upon a nestling Snail

Kite chick; and J. Kern (Toner 1984) photographed a rat snake attempting to swallow a kite

egg.

Structural collapse.-- Ten nests (11% of failures) during 1986 and 14 nests (13% of

failures) during 1987 experienced some degree (>15% tilted) of structural collapse. Some

of these failures, however, may have been caused by other factors.

Abandonment.-- We found a total of 15 cases of apparent abandonment in 1986 and 1987

(see Table 4); however, three of these nests previously had undergone a partial egg or

young loss (probably due to predation). In five cases, incubation had extended well beyond

the normal incubation period suggesting that the eggs were not viable; in five additional

cases the nest had only one egg, suggesting that a partial loss might have occurred prior

to our discovery of the nest. In only 2 of 375 nests did we observe what we believed to be

abandonment of viable eggs or young in which the nest had not had at least partial

predation. Even when eggs were not viable, kites appeared reluctant to abandon their nest.

One pair incubated eggs for at least 84 days (over three times the normal incubation

period).

Parasites.-- We frequently found kite chicks with mite infestations, but did not

suspect that any failures were attributable directly to these infestations. We found only

two nests which had dermestid beetle larvae and, although each had lesions similar to those

described by Snyder et al. (1984), both successfully fledged one young.

Human disturbance.-- We observed adult kites that were flushed from nests close to

airboat trails (i.e. < 75 m) when airboats passed; however, we did not observe any

prolonged disturbance, and the adults usually returned to the nests immediately after the

boat had passed.

Our own research effort was another potential cause of disturbance. To assess this

impact, we randomly selected 10 nests during incubation in 1986; five nests were not

visited until the young were approximately two weeks old and five nests were visited at our

regular interval (7-10 days). In each case, two of the five nests failed. Although these

sample sizes were too small to draw definitive conclusions, these results suggest that our

visitation frequency was not causing increased nest failure.

Influences of Nesting Success

Date of Initiation.-- Nesting success was lowest during late season in both 1986 and

1987 (Fig. 36). Differences were significant for 1987 (standard normal test, P < 0.05,

Appendix 5), but not for 1986. Overall nesting success was highest during the early period

of 1986 and the middle period of 1987.

Nesting substrate.-- Nesting success did not differ significantly among nests built on

the four major substrates in either 1986 or 1987 (standard normal test, P > 0.05, Appendix

6). The ranking of nesting success in relation to substrate was the same in each year,

nests in melaleuca were most successful, followed in decreasing order of success by willow,

cypress, and pond apple (Fig. 37).

One pair incubated eggs for at least 84 days (over three times the normal incubation

period).

Parasites.-- We frequently found kite chicks with mite infestations, but did not

suspect that any failures were attributable directly to these infestations. We found only

two nests which had dermestid beetle larvae and, although each had lesions similar to those

described by Snyder et al. (1984), both successfully fledged one young.

Human disturbance.-- We observed adult kites that were flushed from nests close to

airboat trails (i.e. < 75 m) when airboats passed; however, we did not observe any

prolonged disturbance, and the adults usually returned to the nests immediately after the

boat had passed.

Our own research effort was another potential cause of disturbance. To assess this

impact, we randomly selected 10 nests during incubation in 1986; five nests were not

visited until the young were approximately two weeks old and five nests were visited at our

regular interval (7-10 days). In each case, two of the five nests failed. Although these

sample sizes were too small to draw definitive conclusions, these results suggest that our

visitation frequency was not causing increased nest failure.

Influences of Nesting Success

Date of Initiation.-- Nesting success was lowest during late season in both 1986 and

1987 (Fig. 36). Differences were significant for 1987 (standard normal test, P < 0.05,

Appendix 5), but not for 1986. Overall nesting success was highest during the early period

of 1986 and the middle period of 1987.

Nesting substrate.-- Nesting success did not differ significantly among nests built on

the four major substrates in either 1986 or 1987 (standard normal test, P > 0.05, Appendix

6). The ranking of nesting success in relation to substrate was the same in each year,

nests in melaleuca were most successful, followed in decreasing order of success by willow,

cypress, and pond apple (Fig. 37).

One pair incubated eggs for at least 84 days (over three times the normal incubation

period).

Parasites.-- We frequently found kite chicks with mite infestations, but did not

suspect that any failures were attributable directly to these infestations. We found only

two nests which had dermestid beetle larvae and, although each had lesions similar to those

described by Snyder et al. (1984), both successfully fledged one young.

Human disturbance.-- We observed adult kites that were flushed from nests close to

airboat trails (i.e. < 75 m) when airboats passed; however, we did not observe any

prolonged disturbance, and the adults usually returned to the nests immediately after the

boat had passed.

Our own research effort was another potential cause of disturbance. To assess this

impact, we randomly selected 10 nests during incubation in 1986; five nests were not

visited until the young were approximately two weeks old and five nests were visited at our

regular interval (7-10 days). In each case, two of the five nests failed. Although these

sample sizes were too small to draw definitive conclusions, these results suggest that our

visitation frequency was not causing increased nest failure.

Influences of Nesting Success

Date of Initiation.-- Nesting success was lowest during late season in both 1986 and

1987 (Fig. 36). Differences were significant for 1987 (standard normal test, P < 0.05,

Appendix 5), but not for 1986. Overall nesting success was highest during the early period

of 1986 and the middle period of 1987.

Nesting substrate.-- Nesting success did not differ significantly among nests built on

the four major substrates in either 1986 or 1987 (standard normal test, P > 0.05, Appendix

6). The ranking of nesting success in relation to substrate was the same in each year,

nests in melaleuca were most successful, followed in decreasing order of success by willow,

cypress, and pond apple (Fig. 37).

1986

80-

29

a
80 -

&so-

40-

us 20
Z

Eary

Mid

60

40

26

1

20

4~
U

Late

106

I

Ea~y

76

*

Mid

41

DATE OF INITIATION

Figure 36. Mayfield estimates of overall nesting success of Snail Kite nests in WCA-3A that were
initiated'early, middle, and late season (see methods). Ninety-five percent confidence intervals about
the estimates and sample sizes are shown.

63

+

-

Late

1987

1986

100-

80-

60-

40-

12

1i

107

+

1987

36

'

1J

SP C M

NEST SUBSTRATE

Figure 37. Mayfield estimates of overall nesting success of Snail Kite nests in WCA-3A in nesting
substrates willow, pond apple, cypress, and melaleuca. Ninety-five percent confidence intervals about
the estimates and sample sizes are shown.

100-

6so

820
w4-

z 20-

20-

One pair incubated eggs for at least 84 days (over three times the normal incubation

period).

Parasites.-- We frequently found kite chicks with mite infestations, but did not

suspect that any failures were attributable directly to these infestations. We found only

two nests which had dermestid beetle larvae and, although each had lesions similar to those

described by Snyder et al. (1984), both successfully fledged one young.

Human disturbance.-- We observed adult kites that were flushed from nests close to

airboat trails (i.e. < 75 m) when airboats passed; however, we did not observe any

prolonged disturbance, and the adults usually returned to the nests immediately after the

boat had passed.

Our own research effort was another potential cause of disturbance. To assess this

impact, we randomly selected 10 nests during incubation in 1986; five nests were not

visited until the young were approximately two weeks old and five nests were visited at our

regular interval (7-10 days). In each case, two of the five nests failed. Although these

sample sizes were too small to draw definitive conclusions, these results suggest that our

visitation frequency was not causing increased nest failure.

Influences of Nesting Success

Date of Initiation.-- Nesting success was lowest during late season in both 1986 and

1987 (Fig. 36). Differences were significant for 1987 (standard normal test, P < 0.05,

Appendix 5), but not for 1986. Overall nesting success was highest during the early period

of 1986 and the middle period of 1987.

Nesting substrate.-- Nesting success did not differ significantly among nests built on

the four major substrates in either 1986 or 1987 (standard normal test, P > 0.05, Appendix

6). The ranking of nesting success in relation to substrate was the same in each year,

nests in melaleuca were most successful, followed in decreasing order of success by willow,

cypress, and pond apple (Fig. 37).

One pair incubated eggs for at least 84 days (over three times the normal incubation

period).

Parasites.-- We frequently found kite chicks with mite infestations, but did not

suspect that any failures were attributable directly to these infestations. We found only

two nests which had dermestid beetle larvae and, although each had lesions similar to those

described by Snyder et al. (1984), both successfully fledged one young.

Human disturbance.-- We observed adult kites that were flushed from nests close to

airboat trails (i.e. < 75 m) when airboats passed; however, we did not observe any

prolonged disturbance, and the adults usually returned to the nests immediately after the

boat had passed.

Our own research effort was another potential cause of disturbance. To assess this

impact, we randomly selected 10 nests during incubation in 1986; five nests were not

visited until the young were approximately two weeks old and five nests were visited at our

regular interval (7-10 days). In each case, two of the five nests failed. Although these

sample sizes were too small to draw definitive conclusions, these results suggest that our

visitation frequency was not causing increased nest failure.

Influences of Nesting Success

Date of Initiation.-- Nesting success was lowest during late season in both 1986 and

1987 (Fig. 36). Differences were significant for 1987 (standard normal test, P < 0.05,

Appendix 5), but not for 1986. Overall nesting success was highest during the early period

of 1986 and the middle period of 1987.

Nesting substrate.-- Nesting success did not differ significantly among nests built on

the four major substrates in either 1986 or 1987 (standard normal test, P > 0.05, Appendix

6). The ranking of nesting success in relation to substrate was the same in each year,

nests in melaleuca were most successful, followed in decreasing order of success by willow,

cypress, and pond apple (Fig. 37).

Nest height.-- Nesting success did not differ significantly with respect to nest

height during 1986 (standard normal tests, P > 0.05, Appendix 7). Although not

statistically significant, nesting success increased with height during 1986 (Fig. 38).

During 1987, the highest nests (>3 m) were most successful, but nesting success was lowest

for nests of intermediate height (2-3 m).

Distance from land.-- Nesting success during 1986 was significantly higher for nests

that were greater than 500 m from uplands compared with nests that were from 100 500 m

(Z 3.09, P 0.002). Nests that were less than 100 m from uplands had the highest

success, but did not differ significantly from those of greater distance (standard normal

test, P > 0.05, Appendix 8). Nesting success during 1987 tended to increase with

increasing distance from upland habitat (Fig. 39), but differences were not significant

(standard normal test, P > 0.05).

Water depth.-- As with the other environmental variables (e.g. weather), water depth

changes throughout the season at each nest. For this reason, we compared daily nest

survival (as opposed to overall nesting success) among nests while they were within a given

water depth class (see Methods). Daily nest survival during 1986 or 1987 did not differ

significantly among nests that were in shallow (<25 cm), intermediate (25-50 cm), or deep

(50-75 cm) water (standard normal test, P > 0.05, Appendix 9) (Fig. 40).

Rainfall.-- Daily nest survivorship was highest when the average daily rainfall was

lowest in both 1986 and 1987 (Fig. 41); however, differences were not significant in either

year (standard normal test, P > 0.05, Appendix 10). During 1986, daily nest survivorship

was lowest when average daily rainfall was highest. In 1987, survivorship was lowest when

rainfall was intermediate.

Wind soeed.-- We detected no significant differences in daily nest survivorship with

varying levels of average daily wind speed (standard normal test, P > 0.05, Appendix 11).

In 1986, daily nest survivorship was highest when the average daily wind speed was highest

(Fig. 42); however, in 1987 daily survivorship was lowest when wind speed was highest.

Nest height.-- Nesting success did not differ significantly with respect to nest

height during 1986 (standard normal tests, P > 0.05, Appendix 7). Although not

statistically significant, nesting success increased with height during 1986 (Fig. 38).

During 1987, the highest nests (>3 m) were most successful, but nesting success was lowest

for nests of intermediate height (2-3 m).

Distance from land.-- Nesting success during 1986 was significantly higher for nests

that were greater than 500 m from uplands compared with nests that were from 100 500 m

(Z 3.09, P 0.002). Nests that were less than 100 m from uplands had the highest

success, but did not differ significantly from those of greater distance (standard normal

test, P > 0.05, Appendix 8). Nesting success during 1987 tended to increase with

increasing distance from upland habitat (Fig. 39), but differences were not significant

(standard normal test, P > 0.05).

Water depth.-- As with the other environmental variables (e.g. weather), water depth

changes throughout the season at each nest. For this reason, we compared daily nest

survival (as opposed to overall nesting success) among nests while they were within a given

water depth class (see Methods). Daily nest survival during 1986 or 1987 did not differ

significantly among nests that were in shallow (<25 cm), intermediate (25-50 cm), or deep

(50-75 cm) water (standard normal test, P > 0.05, Appendix 9) (Fig. 40).

Rainfall.-- Daily nest survivorship was highest when the average daily rainfall was

lowest in both 1986 and 1987 (Fig. 41); however, differences were not significant in either

year (standard normal test, P > 0.05, Appendix 10). During 1986, daily nest survivorship

was lowest when average daily rainfall was highest. In 1987, survivorship was lowest when

rainfall was intermediate.

Wind soeed.-- We detected no significant differences in daily nest survivorship with

varying levels of average daily wind speed (standard normal test, P > 0.05, Appendix 11).

In 1986, daily nest survivorship was highest when the average daily wind speed was highest

(Fig. 42); however, in 1987 daily survivorship was lowest when wind speed was highest.