|
TECHNICAL REPORT NO. 1
Identification of Feeding Habitat
Determinants for Gulf Coastal
Colonial Birds
Prepared by:
Barbara Black, Graduate Assistant
School of Forest Resources and Conservation
University of Florida
Gainesville, FL 32611
Submitted to:
Project Officer
Cooperative Fish and Wildlife Research Unit
117 Newins-Ziegler Hall
University of Florida
Gainesville, FL 32611
Period Covered: January March 1980
Contract No.: 14-16-0009-79-063
ACKNOWLEDGMENTS
I would like to thank my committee chairman, Dr. Larry D. Harris,
for his assistance throughout this project. I also appreciate the
advice and guidance of my committee members Drs. John W. Hardy,
Richard W. Gregory and Michael W. Collopy.
I would like to thank Dr. David H. Hirth for encouragement
early in my program and Dr. Randy L. Carter for statistical advice
throughout the study. I would also like to express my gratitude
to Dr. R. Michael Erwin for both suggestions and professional
inspiration. To three persons, Dr. Carter R. Gilbert,Lourdes
Bielsa and Jeff Williams, I am grateful for assistance in identifying
fish and shrimp species. Also I would like to thank the staff of
St. Vincent's National Wildlife Refuge for their kindness and
cooperation and my fellow graduate students for their field
assistance, encouragement and friendship. To many other persons
who assisted me in this study, I extend my gratitude.
I am especially grateful to my parents for their constant
support on my endeavors and to my husband, David, whose love and
unique perspective have been invaluable.
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS.
LIST OF TABLES .
LIST OF FIGURES.
ABSTRACT .
INTRODUCTION
APPROACH AND METHODS .
Study Area.
Variables
Observation Periods
Phases
RESULTS
Feeding Habitat .
Habitat Description
Foraging Zones
Canals .
Habitat Comparison
Testing
Prey Resource
Distribution and Composition
Consumption.
Foraging Patterns
Tidal and Flurry
Nocturnal
DISCUSSION
Spatial Aspects
Tidal Feeding Habitat
Structure
Location
Wind
Canals and Channelization
Dietary Aspects
Temporal Aspects.
Social Aspects
S19
S19
S23
S23
. 26
S26
S31
S31
. 33
S37
. 37
S46
S 54
S 54
S 54
S 54
S 56
S 56
S 57
S 57
S 59
S 61
Svii
*viii
19
Page
MANAGEMENT RECOMMENDATIONS 64
SUMMARY 66
LITERATURE CITED 68
APPENDICES 72
A PRINCIPAL FACTOR ANALYSIS. 73
B CORRELATION MATRIX CREATED FROM ONE FEEDING-SPECIFIC
AND TWELVE SITE-SPECIFIC VARIABLES .75
C STATISTICAL ANALYSES COMPARING SITE-SPECIFIC
VARIABLES IN AREA TYPES AT CEDAR KEY AND AT ST.
VINCENT'S NWR. 77
D FISH AND SHRIMP SPECIMENS RECOVERED FROM THIRTEEN
BLACK SKIMMER STOMACHS 80
BIOGRAPHICAL SKETCH 81
LIST OF TABLES
Table Page
1 Discrepancies in Black Skimmer feeding times
found in the literature 5
2 Variables recorded during ten-minute observation
periods 11
3 Black Skimmer utilization of study areas during
Phases la and Ib as shown by Duncan's multiple
range test on transformed data 20
4 The first three factors, structure, distance to
mainland and wind speed, created by the factor
analysis .. 21
5 Chi-square goodness of fit test comparing Black
Skimmer utilization versus availability of
foraging zones 24
6 Chi-square goodness of fit test comparing Black
Skimmer utilization of study areas with and
without canals 25
7 Results of three 2 x 2 Analyses of Variance comparing
structure, distance to the mainland and wind speed
within location and between locations 28
8 Chi-square goodness of fit test comparing Black
Skimmer utilization of "presumed suitable" and
"presumed unsuitable" areas at St. Vincent's NWR 30
9 Means, standard errors and results of Duncan's
multiple range test for fish samples collected
in two highly utilized and two non-utilized
areas at Cedar Key 32
10 Percent distribution of weight, volume and frequency
of fish, shrimp and digested material in thirteen
Black Skimmer stomachs 36
11 Prey species described in Black Skimmer diets as
recorded in the literature 38
Table -Page
12 Means, standard errors and results of statistical
analyses comparing the nine variables recorded
during flurry and tidal feeding 40
LIST OF FIGURES
Fi aure Page
1 Locations of the study areas at Cedar Key,. Levy
County, Florida 9
2 Locations of study areas at St. Vincent's National
Wildlife Refuge, Franklin County, Florida 18
3 Frequency histograms of fish and shrimp sampled in
256 seine hauls in two highly utilized and two
non-utilized areas at Cedar Key 35
4 Foraging group sizes in tidal and flurry feeding 43
5 Percentage of total forages occurring throughout
the day for tidal and flurry feeding 45
6 Percehtage of total forages occurring at twelve
tidal stages for tidal and flurry feeding 48
7 Percentage of total forages followed by flight to
a loafing site 50
8 Percentage of total forages occurring in 500 m
increments from the mainland 52
Abstract of Thesis Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
WINTER FEEDING ECOLOGY OF BLACK SKIMMERS
ON THE FLORIDA GULF COAST
By
Barbara Buttram Black
March 1981
Chairman: Lawrence D. Harris
Major Department: Forest Resources and Conservation
The feeding ecology of Black Skimmers (Rynchops niger) was
studied at Cedar Key and St. Vincent's National Wildlife Refuge
on the Florida Gulf coast during the winters of 1979 and 1980,
respectively. Objectives of the study were to: 1) delineate
characteristics of Black Skimmer winter feeding habitat, 2)
monitor the prey resource and prey selection by wintering Black
Skimmers, 3) describe social, temporal and spatial aspects of
winter foraging, and 4) develop recommendations for Black
Skimmer feeding habitat management.
In the first winter observations were made to identify and
characterize preferred feeding habitat. Food availability and
prey selection by skimmers were sampled. In the second winter
feeding habitat selection was tested in a second locality.
Black Skimmers at Cedar Key typically fed in study areas
with shallow water (10-20 cm) interspersed between oyster bars
or mudflat. Study areas with deep (> 30 cm), uninterrupted bodies
viii
of water typically were not used by the birds. Feeding habitat
selection was similar at St. Vincent's NWR and was predictable from
water depth, land-water interspersion and the proportions of open
water and mudflat.
Skimmers foraged 71 percent of the time within 2 m of a land-
water interface and preferred zones next to mudflat and oyster bar.
Feeding birds used areas with small canals in proportion to their
occurrence but avoided large channels. The lack of preferred
structural features (land-water interface zones) in these channelized
areas may explain the absence of skimmer feeding.
Food abundance alone did not explain the birds' preference for
feeding areas since utilized and non-utilized areas did not differ
significantly (P > 0.05) in prey abundance. Prey composition was
similar among areas sampled, but prey composition in the environment
differed from that of skimmer diets. In the diet Fundulus spp. and
Mugil sp. were predominant by weight and volume while shrimp were
most frequent. In the environment Fundulus spp. and Mugil sp. were
scarce while shrimp were abundant
Two winter foraging patterns were observed. Tidal feeding was
characterized by few birds conducting many forages over a large
area during flow tide. Flurry feeding occurred at evening ebb tide
and was spatially restricted and socially intense. In flurry feeding,
forages increased logarithmically (r2 = 0.82) with increasing bird
numbers. Evidence identifying flurry feeding as social stimulation
or a foraging tactic is inconclusive.
Chairman
INTRODUCTION
In order to survive each organism must acquire from its environ-
ment the resources necessary for its maintenance. Investigations
through theoretical, experimental and descriptive modes have focused
on social (Ward 1965, Krebs et al. 1974), temporal (Emlen 1966,
Schoener 1971), spatial (Recher 1966, MacArthur and Pianka 1966,
Bobisud and Voxman 1979, Hamilton 1971), and dietary (Ashmole 1968,
Pearson 1968) aspects of resource acquisition. Such studies of feed-
ing biology have provided insight into theoretical as well as practical
ecological aspects of avian habitat utilization.
Practical considerations of habitat availability are important
as the nation's coastal land and water resources undergo increasing
utilization by man. While approximately half of all Americans reside,
and are employed, within 80 km of a coastline (OTA 1976), this area
comprises less than 10 percent of the nation's land. In Florida, the
coastal zone contains 25 percent of the state's land area while
supporting over 75 percent of the state's population (DER 1978). In
1972 approximately 63 percent of the state's proposed developments
of regional impact (DRI's) fell within the coastal zone (DSP 1974).
In addition, the U.S. Army Corps of Engineers' National Shoreline
Study noted that 340 km of'Florida's beaches are in a state of
critical erosion. Florida's beaches annually receive 30 million
tourists (Barada.and Partington 1970). Thus, knowledge of coastal
2
species' requirements as well as coastal ecosystem mechanics is
required for future decision making.
Florida serves as a wintering ground for many North American
birds including a large number of coastal shorebirds and colonial
waterbirds. The state provides unique opportunities for biological
and ecological research on wintering species, many-of which occupy
Florida habitats for over half of every year. The present research
should contribute to knowledge of one of these species and elicit
interesting comparisons for similar studies on the species' northern
breeding grounds.
Florida's coastal development facilitates investigation regard-
ing the effects of environmental alterations on avian populations.
The feeding ecology of colonial waterbirds and the resulting manage-
ment implications involve at least two considerations. Firstly,
the effects of dredging are potentially detrimental to coastal
bird communities. Gilmore and Trent (1974) found that by volume,
macro-invertebrates were twice as' abundant in a natural marsh area
in Texas as in an adjacent marsh altered by channelization. Working
in Florida, Hall and Lindall (1974) found a reduction .in the abundance
and species of benthic organisms while Lindall et al. (1975) found a
reduction in the abundance of fishes and crustaceans in estuarine
zones altered by development canals. Thus, as a result of dredging,
bird populations may be indirectly affected by a reduction in food.
Secondly, alterations in coastal nesting habitat may cause
changes in feeding ecology. Since the initiation of the Atlantic
and Gulf Intracoastal Waterway System in the 1930's, 2000 man-made
islands have been created (Lunz et al. 1978). Studies to determine
the extent of dredge island use by colonial birds indicate an
apparent shift by coastal colonial species from historic nesting
sites to spoil islands (Landin and Soots 1977). In Florida an
estimated 70 percent of colonial seabird colonies occur on man-made
as opposed to natural sites (S. Nesbitt, H. Kale pers. comm.). The
reasons for such shifts are speculative but may involve either
increasing pressure on former nesting sites or the creation of
preferred nesting habitat in the form of spoil islands. In either
case, information concerning comparable changes in feeding habitat
is unavailable.
Inherent in the question of habitat alteration is the distance
of newly created spoil island colonies from former or suitable
feeding areas. Erwin (1977a) applied Horn's optimal spacing
model (coloniality vs. territoriality) to three colonial seabird
species and concluded that Black Skimmers (Rynchops niger) flew
shorter distances between nesting and feeding areas than did the
tern species. Erwin's results also indicate that the Black Skimmer
has a more restricted niche breadth on a habitat dimension than the
otner species studied. This apparent restriction in foraging site-
colony distance in conjunction with the more restricted feeding
habitat identifies the Black Skimmer as a likely species to be
impacted by changing environments and, thus, a valuable candidate
for study.
The.skimmer is pantropical in distribution, with two species,
Rynchops albicollis (India to Indo-China) and Rynchops flavirostris
(tropical Africa) occurring in the Eas'tern HemispKere. Three races
of Black Skimmer occur in the Western Hemisphere; these include
R. n. niqer (coastal North America), R. n. cinerescens (Brazil west
to Argentina) and R. n. intercedens (Argentina, Uruguay, Brazil).
The North American race is primarily a coastal species while the
South American subspecies also inhabit large river systems.
Rynchops n. niger breeds from New York to the Gulf coast of
Mexico with the greatest northern breeding concentrations along the
Virginia coast (2687 breeding adults) (Erwin and Korschgen 1979)
and the greatest southern breeding concentrations along the Louisiana-
Mississippi-Alabama coasts (29,980 breeding adults) (Portnoy 1978).
The birds typically leave the northern states in November (M.
Gochfeld pers. comm.) to winter in Florida, along the Gulf coast
and in Central America. Since very few skimmers seen along the north
Florida Gulf coast are banded (compared with up .to 60% of some east
coast populations, W. Robertson pers. comm.) these birds must be
either residents or migrants from states other than New York,-New
Jersey, Virginia or the Carolinas where banding is conducted.
Typical of the Charadriiformes, the Black Skimmer has-undergone
adaptive specialization in its feeding. While the morphological
adaptations of skimmer feeding have been definitively analyzed by
Zusi (1959), the species' ecological adaptations are less well
understood. In fact, the literature concerning several aspects
of the species' feeding ecology is quite varied and disparate (Table
1). For instance, six authors (Arthur, 1921, Bales 1919, Erwin 1977a,
a (
c0
. / 1 30
4-'
vi ix U
X X X XX
c)
4-J
tu C) (a
* ) x x x
4-1
co U .
ea
S4-
5" xx x x x
4. ro &-'
= 4 X
5.. 0 .
(-:) < C
,-l o 0-. !-
U-\J 01 *r- C 0 I1 S.. 0i
1r** I-C OS C C '-C 0 '
..a 4I. a. 0 Q
cc^
I L. I- *
E- 0 < C I .
^-. (M i'- '^ s o' *^ '"
CT a T C '"' -r -t 0r O r r
^- 'e ^ fg et*- is'- '**- (I o F o in
.., ; U-*- r r : i : o < n*
^_3 -a (/ '-'i ,~ *r e *'" *- i i nj / D i
ai -.- e n r r i c -- *r /
0) 0) 3i 3 <^ s- e i o o --3 -
L_ C O < i t La- .-*- ~ 4'0 ) *-t! 0
.3 333 vr : n> S UI* *-^ 5 ^ "^
Nicholson 1948, Tomkins 1951, Zusi 1959) have suggested that skimmers
are somewhat nocturnal in their feeding habits while three investi-
gators (Chapman 1908, Pettingill 1937, Zusi 1959) have suggested that
skimmers respond to diurnal influences. Chapman (1908), Tomkins
(1951) and Zusi (1959) each state that these birds are-strongly
crepuscular while Bent (1921), Erwin (1977a), Pettingill (1937)
and Tomkins (1951) suggest that tidal stage primarily governs skimmer
foraging activity. Although season, location, and moon phase may
influence perceptions of primary foraging period, the relative
importance of these factors is ambiguous. Also, discrepancies in
feeding habitat selection must be clarified to understand the
species' ecological adaptations.
Previous studies concerning the breeding and feeding biology
of the Black Skimmer have been conducted primarily in the spring
and summer (Chapman 1908, Bales 1919, Pettingill 1937, Chamberlain
1959, Zusi 1959, Erwin 1977a, 1977b, Gochfeld 1977). Only two
published works (Arthur 1921, Tomkins 1951) deal with fall
activities and only one study (Nicholson 1948) has described
ecological or biological aspects of wintering skimmers (Table 1).
The principal objective of this study is to derive information
pertaining to the feeding ecology of wintering Black Skimmers. The
specific objectives are to: 1) delineate characteristics of Black
Skimmer winter feeding habitat, 2) monitor the prey resource and
prey selection by wintering Black Skimmers, 3) describe social,
temporal and spatial aspects of winter foraging, and 4) develop
recommendations for Black Skimmer feeding habitat management.
APPROACH AND METHODS
Study Area
The Cedar Keys are a group of small islands on the west coast
of peninsular Florida at 290 080 30" N and 830 02' 30" W, roughly
midway between the Suwannee and Wacassassa Rivers. The islands are
surrounded by soft-bottomed flats, oyster bars and salt marshes
extending toward the mainland. Three vegetational zones characterize
the salt marshes. From the water's edge they are: a Spartina alter-
niflora zone, a Juncus roemerianus midzone and a third zone comprised
of Salicornia virginiana, Batis maritima and Spartina dystichum.
Tidal fluctuations in this area of the Gulf rarely exceed a meter.
Due to the low energy nature of these tides, little tidal flushing
occurs. This limited turbulence in conjunction with the detritus
from annual turtle grass (Thalassia testudinum) dieoffs results in a
highly organic sediment. Th-is organic component in proximity to fresh
water creates a rich estuarine environment.
The specific study area was comprised of a 16.0 km2 area located
within the key system (Fig. 1). Eighteen similar tidal flat areas were
selected as study sites. The 6.0 ha areas were of variable dimensions
because of particular shoreline features of the tidal flat. The
minimum distance between study areas was 100 m if separated by land
and 400 m if separated by water. Each area was investigated at some
point in the study but not all areas were observed in all phases of
the study.
APPROACH AND METHODS
Study Area
The Cedar Keys are a group of small islands on the west coast
of peninsular Florida at 290 080 30" N and 830 02' 30" W, roughly
midway between the Suwannee and Wacassassa Rivers. The islands are
surrounded by soft-bottomed flats, oyster bars and salt marshes
extending toward the mainland. Three vegetational zones characterize
the salt marshes. From the water's edge they are: a Spartina alter-
niflora zone, a Juncus roemerianus midzone and a third zone comprised
of Salicornia virginiana, Batis maritima and Spartina dystichum.
Tidal fluctuations in this area of the Gulf rarely exceed a meter.
Due to the low energy nature of these tides, little tidal flushing
occurs. This limited turbulence in conjunction with the detritus
from annual turtle grass (Thalassia testudinum) dieoffs results in a
highly organic sediment. Th-is organic component in proximity to fresh
water creates a rich estuarine environment.
The specific study area was comprised of a 16.0 km2 area located
within the key system (Fig. 1). Eighteen similar tidal flat areas were
selected as study sites. The 6.0 ha areas were of variable dimensions
because of particular shoreline features of the tidal flat. The
minimum distance between study areas was 100 m if separated by land
and 400 m if separated by water. Each area was investigated at some
point in the study but not all areas were observed in all phases of
the study.
Figure 1. *Locations of the study areas at Cedar Key, Levy County,
Florida.
LEVY COUNTY
j .r~
s -
SFLdRIDA
CEDAR KEY
I KM
Zl
- ell
Variables
Seven feeding-specific and twenty site-specific variables (Table
2) were defined and measured in the following manner. A forage was
defined as a feeding effort made by a skimmer during which the bird
was actively skimming and did not alter the wingbeat in the manner
necessary to change direction or otherwise lift itself from the
water's surface. By this definition, the bill may be removed
briefly from the water and re-immersed without indicating the
initiation of a.new foraging effort. Due to the difficulty of.
monitoring more than one bird of a group, a single bird was chosen
and assumed to be representative of the group. The number of key
bird forages per group was obtained by recording the number of
forages made by this single bird, termed the key bird of a group.
The number of key bird forages per observation period was obtained
by summing the number of key bird forages per group for all groups
seen in an observation period. Similarly, the number of birds per
observation period was the number of birds per group summed over
groups within an observation period. Thus, whereas the number of
birds per observation period measures the individuals utilizing an
area, the number of key bird forages per observation period
measures the intensity of this utilization. The number of birds
and the number of key bird forages were multiplied by group and
summed within an observation period to derive total forages, an
approximation of total foraging effort per observation period.
The distance to the post-feeding loafing site was classified
as: 1 = loafing site in area, 2 = loafing site within 250 m of
Table 2. Variables recorded during ten-minute observation periods.
Variables Mneumonic Unit
Feeding-specific
Social
Number of birds per group
Number of birds per observation
period
Behavioral
Number of key bird forages per
group
Number of key bird forages per
observation period
Number of total forages per
observation period
Distance to loafing site
Temporal
Minutes to sunset
bird
bird
(NUF)
forage
forage
forage
classified
(MTS)
minute
Site-specific
Physical
Light intensity
Cloud cover
Wind direction
Wind speed
Air temperature
(WI D)
(WIS)
footcandle
percentage
degree
kph
degree
Social
Number of other feeders
Locational
Distance from loafing site
Distance to mainland
Physiognomic
Tidal stage
Water depth
Patchiness
Water roughness
Shelter by land
Number of obstructions
Mudflat
Sandbar
Oyster bar
Spartina
Open water
Land
(FED)
(DFL)
(DTM)
(TIS)
(WAD)
(PAT)
(SBL)
(OPW)
(LND)
bird
meter
meter
hour
centimeter
classified
percentage
meter
obstruction
percentage
percentage
percentage
percentage
percentage
percentage
area, 3 = loafing site over 250 m from area, 4 = loafing site unknown.
This is a behavioral variable and represents the distance to the
loafing site selected by the bird after feeding. Minutes to sunset
were determined using astronomical data for the local latitude and
longitude.
The 20 site-specific variables were estimated or measured as
described below. The physical variables, light intensity, wind
direction, wind speed and air temperature were measured and recorded
directly while cloud cover was est-imated in ten-percent increments.
The number of other feeders in the area was obtained by counting
only those birds in the area that were foraging. The locational
variable, distance from loafing site, represented the distance from
the center of a study area to the nearest available loafing site at
the time of observation. The variable, distance to mainland, was
classified by 500 m increments from the mainland with one representing
the least and eight representing the greatest distance from the main-
land-. Twelve physiognomic variables also were defined and measured.
The tidal stage was classified as one of 12 one-hour increments
beginning with the previous high tide. Water depth was measured in
centimeters using a water depth marker at the deepest non-channel
location in each area. Patchiness was designated by one of four
classes ranging from 1 = minimal land-water interspersion to 4 =
maximal land-water interspersion. Water roughness was estimated
by taking ten readings of the water surface through an ocular
scope (Winkworth and Goodall 1962) to estimate the percentage of
rough water as compared to smooth water surface. For each study
area a "shelter by land" classification was established for each
of the directions, N, NE, E, SE, S, SW, W, NW. Shelter by land
was classified by 50 m increments of land within 500 m of an area
in each designated direction, thus, resulting in eight readings per
area. A single classification was then assigned to an area depending
on the wind direction at the time of observation. The percentages
of mudflat, sandbar, oyster bar, spartina and open water in an area
were estimated by taking 25 point readings in a standardized pattern
with a cross-wire ocular scope and converting to percentage values.
The percentage of land was a value derived from summing the mudflat
and sandbar percentages.
Observation Periods
A ten minute interval was chosen as the basic time unit of
observation (the observation period). Such periods were assigned
randomly to the areas being sampled in a particular phase of the
study. During observation periods in which skimmers were seen
foraging, all variables (Table 2) were recorded. Only the site-
specific variables were recorded during observation periods in
non-utilized areas.
Phases
The objectives of the study were divided into a number of
components. Habitat delineation, the first objective, involved: 1)
identifying and 2) describing preferred foraging areas, 3) comparing
utilized and non-utilized feeding areas, and 4) testing these
findings in an independent location. Also, the use of topographic
features within an area and the use of areas with and without canals
area a "shelter by land" classification was established for each
of the directions, N, NE, E, SE, S, SW, W, NW. Shelter by land
was classified by 50 m increments of land within 500 m of an area
in each designated direction, thus, resulting in eight readings per
area. A single classification was then assigned to an area depending
on the wind direction at the time of observation. The percentages
of mudflat, sandbar, oyster bar, spartina and open water in an area
were estimated by taking 25 point readings in a standardized pattern
with a cross-wire ocular scope and converting to percentage values.
The percentage of land was a value derived from summing the mudflat
and sandbar percentages.
Observation Periods
A ten minute interval was chosen as the basic time unit of
observation (the observation period). Such periods were assigned
randomly to the areas being sampled in a particular phase of the
study. During observation periods in which skimmers were seen
foraging, all variables (Table 2) were recorded. Only the site-
specific variables were recorded during observation periods in
non-utilized areas.
Phases
The objectives of the study were divided into a number of
components. Habitat delineation, the first objective, involved: 1)
identifying and 2) describing preferred foraging areas, 3) comparing
utilized and non-utilized feeding areas, and 4) testing these
findings in an independent location. Also, the use of topographic
features within an area and the use of areas with and without canals
were considered as a portion of habitat delineation. The second
objective dealt with the prey resource and prey selection by
skimmers. Food abundance was sampled in two utilized and two
non-utilized areas while prey composition was compared between
areas and with skimmer diets. The third objective was to describe
feeding patterns.
To accomplish these objectives four phases of research were
outlined. During the first phase (July-October 1979) observations
to determine area use-were made according to a randomized block
design in 12 areas (Phase la, A, B, C, D, E, F, G, H, I, J, K, J,
M) during low tide. Data collection.for Phase Ib was conducted
in the same manner but differed in the addition of new areas (M,
N, 0, P) and the deletion of other areas (G, H, I, J) outside
of the birds' observed range (Fig. 1).
During the second phase of study (November 1979-March 1980)
observations were conducted to describe feeding habitat. Phase
one results revealed that skimmer foraging was not restricted
to low tide stages. Therefore, observation periods were assigned
to 6 utilized areas for 20 hours in each of 12 tidal stages. In
this manner 89 skimmer-specific observations were recorded with
accompanying environmental measurements. A principal factor
analysis was used to analyze these data (Appendix A).
In order to determine Black Skimmer response to topographic
features within an area, a number of foraging zones were defined
as: edge mudflat, edge mudflat spartina, edge mudflat oyster bar
and non-edge open water. The "edge" was defined as a 2 m zone
to either side of a land-water interface. The relative abundance
of these foraging zones was estimated by taking 25 readings with
a cross-wire ocular scope and converting to percentage values. The
number of skimmer forages in the various zones was recorded to evaluate
use. All results are reported using the standard probability levels:
n.s. P > 0.05, P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.
The prey resource was sampled using a 3 cm delta mesh seine
(3.1 m) in a.6.0 m sweep immediately adjacent and parallel to the
land-water interface. Samples from 64 seine hauls in each of 4
areas (A, C, D, 0) were characterized by the number of individuals
per seine haul and by the number of individuals per family. Prey
selection by Black Skimmers was estimated by collecting 13 birds
during feeding in April 1980 and analyzing their stomach contents.
During the third phase of study (September-October 1980)
observations to compare utilized and non-utilized feeding habitat
were made from low tide to flow tide, the time frame identified
as the primary foraging period. Ten days of observations were
structured such that low tide occurred in each of the 10 hours
from 0700-1700. Nocturnal observations were also made during this
study phase in two highly utilized areas, A and C, and two inter-
mittently utilized areas, B and M. Three nights of observations
were conducted in each of the periods from 1900-2200, 2300-0200
and 0300-0600.
While the results of the Cedar Key study may accurately portray
skimmer feeding ecology, the results also may reflect the unique
characteristics of that area's geography. A final test phase was,
therefore, initiatedin order to account for the effects of location
while holding tides, season and time of day constant. Observations
were conducted using the same procedures described for the third
phase habitat comparison in Cedar Key. The study site, St. Vincent's
National Wildlife Refuge, was selected because it has tidal similarities
to Cedar Key (cycles (6 hrs) and fluxes (Cedar Key x = 0.8 m, St.
Vincent's x = 0.5 m)) as well as geographic similarities to the
Louisiana-Mississippi-Alabama coast, the center of Black Skimmer
breeding in North America. These geographic similarities include
barrier beaches with associated protected tidal-flats.
Candidate study sites were first judged in terms of selected
parameter values and assigned to a "presumed suitable" or "presumed
unsuitable" status (Fig. 2). Observations of skimmer use were then
conducted to ascertain if the null hypothesis of equal utilization
by feeding skimmers should be rejected.
Figure 2. Locations of the study areas at St. Vincent's National
Wildlife Refuge, Franklin County, Florida. Areas V
and X were "presumed suitable" and areas T and U were
"presumed unsuitable."
FRANKLIN COUNTY
INDIAN PASS LAGOON
ST. VINCENT'S
GULF OF MEXICO
ST. VINCENT
I KM
GULF COUNTY
RESULTS
Feeding Habitat
Habitat Description
The first step in identifying Black Skimmer feeding habitat
determinants in Cedar Key was the establishment of preference
classes for various feeding areas. In Phases la and Ib skimmers
utilized some study areas more frequently (ANOVA, P < 0.05) than
others: In both phases, areas A and C were used significantly
more (P < 0.05) than the other areas (Table 3). Areas D, 0, N,
H, J and I were consistently non-utilized; however, since areas
H, I, J, and N were not within the birds' observed range, these
areas were deleted from further observation. Thus, D and 0 were
the only areas within feeding range that were consistently non-
utilized. All other areas (B, E, F, G, K, L, M, P) were classified
as low utilization areas since they were intermittently used by
foraging skimmers.
Measurements of environmental variables recorded during
active skimmer feeding were used to distinguish the key factors
of intermittently and frequently used habitat (Table 4). The
first three factors created by the factor analysis accounted for
83 percent of the variation in the original data. Since the four
variables with high loadings (Appendix A) on the first factor are
variables that describe some structural feature of skimmer feeding
habitat, I refer to the first factor as structure. The variables
RESULTS
Feeding Habitat
Habitat Description
The first step in identifying Black Skimmer feeding habitat
determinants in Cedar Key was the establishment of preference
classes for various feeding areas. In Phases la and Ib skimmers
utilized some study areas more frequently (ANOVA, P < 0.05) than
others: In both phases, areas A and C were used significantly
more (P < 0.05) than the other areas (Table 3). Areas D, 0, N,
H, J and I were consistently non-utilized; however, since areas
H, I, J, and N were not within the birds' observed range, these
areas were deleted from further observation. Thus, D and 0 were
the only areas within feeding range that were consistently non-
utilized. All other areas (B, E, F, G, K, L, M, P) were classified
as low utilization areas since they were intermittently used by
foraging skimmers.
Measurements of environmental variables recorded during
active skimmer feeding were used to distinguish the key factors
of intermittently and frequently used habitat (Table 4). The
first three factors created by the factor analysis accounted for
83 percent of the variation in the original data. Since the four
variables with high loadings (Appendix A) on the first factor are
variables that describe some structural feature of skimmer feeding
habitat, I refer to the first factor as structure. The variables
RESULTS
Feeding Habitat
Habitat Description
The first step in identifying Black Skimmer feeding habitat
determinants in Cedar Key was the establishment of preference
classes for various feeding areas. In Phases la and Ib skimmers
utilized some study areas more frequently (ANOVA, P < 0.05) than
others: In both phases, areas A and C were used significantly
more (P < 0.05) than the other areas (Table 3). Areas D, 0, N,
H, J and I were consistently non-utilized; however, since areas
H, I, J, and N were not within the birds' observed range, these
areas were deleted from further observation. Thus, D and 0 were
the only areas within feeding range that were consistently non-
utilized. All other areas (B, E, F, G, K, L, M, P) were classified
as low utilization areas since they were intermittently used by
foraging skimmers.
Measurements of environmental variables recorded during
active skimmer feeding were used to distinguish the key factors
of intermittently and frequently used habitat (Table 4). The
first three factors created by the factor analysis accounted for
83 percent of the variation in the original data. Since the four
variables with high loadings (Appendix A) on the first factor are
variables that describe some structural feature of skimmer feeding
habitat, I refer to the first factor as structure. The variables
w
c
e 1
v(,
u
a0
.- ,
I
- -
3: i
a
i I
C 0
C
r-CL
i- e
* ^ i-
10
F- s
o -
II yo i. tX
a U
0c5
0
9-
0
9-
0
-A
m
C4
co
uL
I-.
9 -
LL
0
lo
0
U-
0
=
0
9-
0
0
*
4.J
>1
3
'U
u
4-
C
0,
e
9.-
U
S.-
4-
4-
4-
0
e
0
a.,
C
3
I
-a
T;
3
r:~
1/1
ig
!
a
c
m
u
c
f.
<-
in
0
0
.4-
o
4-1
V,
C
'I
e
'a
0,
'iU
-a
Table 4. The first three factors, structure, distance to mainland,
and wind speed, created by the factor analysis.a These factors
accounted for 83 percent of the variation in the original data
describing skimmer feeding habitat.
Factor Pattern
Factor 1 Factor 2 Factor 3
Variablesbc "Structure" "Distance to Mainland" "Windspeed"
TIS -0.24 -0.30 -0.16
WAD 1 -0.85 -0.16 0.05
PAT 2 0.90 -0.09 0.04
DFL -0.36 0.48 0.25
DTM -0.25 5 -0.52 .-0.53
MTS 0.58 -0.38 0.09
SBL 0.24 0.46 -0.11
WID -0.10 0.45 .-0.29
WIS 0.32 -0.26 6 0.65
FED 0.05 0.49 -0.03
P 3 -0.85 -0.22 0.19
L 4 0.85 -0.06 -0.21
values represent loadings for each variable on the respective
factor and indicate relative importance. Signs indicate direction-
ality and do not indicate positive or negative status.
P represents the arcsine transformation of the percentage of open
water.
cL represents the arcsine transformation of the percentage of land
(mudflat or sandbar).
and their loadings (Table 4, first column, values 1-4, respectively)
were water depth (0.85), patchiness (0.90), amount of open water
(0.85) and amount of land including mudflat and sandbar (0.85).
Of these variables, water depth and the amount of open water were
negatively correlated with the number of forages while patchiness
and the amount of land were positively correlated with the number
of forages (Column 1, values 1-4 respectively of Appendix B). In
other words, as the water depth or amount of open water decreased,
the number of skimmer forages increased. Conversely, as the
patchiness (amount of land-water interspersion) or land (mudflat
or sandbar) increased, the number of forages increased.
The second factor had no variables with loadings greater than
0.60. However, the highest variable loading (0.52) (Table .4, value
5) was associated with the distance to the mainland and, thus, the
factor was identified as distance. The number of forages was
negatively correlated with the distance to the mainland indicating
that foraging increased closer to the mainland (Appendix B, column
1, value 5).
The third factor was designated as-wind speed since that variable
had the highest loading (0.65) (Table 4, value 6) on the third factor.
Although the number of forages was positively correlated with wind
speed (Appendix B, column 1, value 6), the interpretation that
skimmers prefer stronger winds for feeding is probably not correct.
In fact, increasing wind speed may reduce feeding success and,
thus, increase the amount of effort needed to attain a satisfactory
diet. Therefore,-increased foraging may be a bird's response to
increased wind speed, rather than a reflection of a bird's preference
for feeding in stronger winds.
Foraging Zones
In order to further evaluate the skimmer's utilization of feed-
ing habitat, several foraging zones were defined and monitored for
their availability to and utilization by foraging skimmers. These
foraging zones, edge mudflat, edge mudflat spartina, edge mudflat
oyster bar and non-edge open water, were not found to.be utilized
by feeding skimmers according to their relative occurrence within
an area (X2 = 339, P < 0.05). Thus, the birds in this study were
not only selecting areas for foraging but also preferentially
foraging near certain topographic features within an area. Specifically,
the edge mudflat (X2 = 23.8, P < 0.05) and edge mudflat oyster bar
(X2 = 187.5, P < 0.05) zones were 'utilized more frequently than
would be expected by their relative occurrence in the environment
(Table 5). Edge mudflat spartina (X2 = 172.3, P < 0.05) and non-
edge open water (X2 = 40.8, P <. 0.05) were utilized less than would
be expected if foraging were non-selective (Table 5). These results
further support the hypothesis that the structure of an area may
casually determine utilization by foraging skimmers.
Canals
A chi-square goodness of fit test comparing areas with and
without canals indicated no significant difference between their
availability and use by feeding skimmers (Table 6). While areas
with canals were used non-selectively by skimmers, some topographic
features in the areas were preferred by feeding birds. Of the
increased wind speed, rather than a reflection of a bird's preference
for feeding in stronger winds.
Foraging Zones
In order to further evaluate the skimmer's utilization of feed-
ing habitat, several foraging zones were defined and monitored for
their availability to and utilization by foraging skimmers. These
foraging zones, edge mudflat, edge mudflat spartina, edge mudflat
oyster bar and non-edge open water, were not found to.be utilized
by feeding skimmers according to their relative occurrence within
an area (X2 = 339, P < 0.05). Thus, the birds in this study were
not only selecting areas for foraging but also preferentially
foraging near certain topographic features within an area. Specifically,
the edge mudflat (X2 = 23.8, P < 0.05) and edge mudflat oyster bar
(X2 = 187.5, P < 0.05) zones were 'utilized more frequently than
would be expected by their relative occurrence in the environment
(Table 5). Edge mudflat spartina (X2 = 172.3, P < 0.05) and non-
edge open water (X2 = 40.8, P <. 0.05) were utilized less than would
be expected if foraging were non-selective (Table 5). These results
further support the hypothesis that the structure of an area may
casually determine utilization by foraging skimmers.
Canals
A chi-square goodness of fit test comparing areas with and
without canals indicated no significant difference between their
availability and use by feeding skimmers (Table 6). While areas
with canals were used non-selectively by skimmers, some topographic
features in the areas were preferred by feeding birds. Of the
Table 5. Chi-square-goodness of fit test comparing Black Skimmer
utilization versus availability of foraging zones.
Forages N = 5460
Forage Zone Observed Expecteda Chi-square
For all zones
Edge mudflat 1826 1660
Edge mudflat spartina 491 841 338.6****
Edge. mudflat oyster bar 1522 1114
Non-edge open water 1622 1845
By zone
Edge mudflat 1826 1660
Non-edge mudflat 3635 3800 23.8**
Edge mudflat spartina 491 841
Non-edge mudflat spartina 4970 4619 172.3**
Edge mudflat oyster bar 1522 1114
Non-edge mudflat oyster 3939 4346 187.5***
bar
Non-edge open water 1622 1845
Non-non-edge open water 3839 3615 40.8**
aExpected values were derived from the number of forages which should
occur in each zone based only on the zone's availability. The zone's
availability was derived in an area at the time of feeding by taking
25 readings with an ocular point sampling scope and converting to a
percentage value.
Table 6. Chi-square goodness of fit test comparing Black Skimmer
utilization of study areas with and without canals.
Forages
Feature Observed Expecteda Chi-square
Phase la
(N = 1163 forages)
With canals
N = 6 areas 605 582
n.s.
1.9
Without canals
N = 6 areas 558 582
Phase Ib
(N = 778 forages)
With canals
N = 4 areas 259 257
O.03n.s.
Without canals
N = 8 areas 519 521
aExpected values are
without canals.
based on the percentage of areas with and
total forages recorded in the second phase (N = 5460) 71 percent were
within 2 m of a land-water interface. Skimmers also preferred
interfaces next to mudflat and oyster bar. While the creation of
canals may reduce land-water interface zones in some portions of
an area, their spoil deposited nearby may result in loafing sites,
land-water interface zones, and shallow water,, features. used by
feeding skimmers. Thus, while the canal, itself, is not used for
feeding it may indirectly result in preferred habitat features in
an area.
Habitat Comparison
When utilized and non-utilized feeding areas at Cedar Key were
compared, no significant differences were found for any climatic
variables (cloud cover, wind speed, wind direction), temporal
variables (minutes to sunset, minutes from sunrise, tidal stage)
or the percentage of spartina (Appendix C1). Significant differences
were found for a number of social, locational and physiognomic
variables. Utilized areas were characterized by more feeding
birds of all species as well as proximity to loafing sites and the
mainland. Physiognomically, utilized areas were typified by
shallower, smoother water with fewer obstructions and more shelter
provided by surrounding land. There was less open water, less
oyster bar, more land-water interspersion and more mudflat in the
utilized areas.
Testing
At St. Vincent's-NWR three days of unstructured observations
indicated that the marine bird community utilized the area in a
total forages recorded in the second phase (N = 5460) 71 percent were
within 2 m of a land-water interface. Skimmers also preferred
interfaces next to mudflat and oyster bar. While the creation of
canals may reduce land-water interface zones in some portions of
an area, their spoil deposited nearby may result in loafing sites,
land-water interface zones, and shallow water,, features. used by
feeding skimmers. Thus, while the canal, itself, is not used for
feeding it may indirectly result in preferred habitat features in
an area.
Habitat Comparison
When utilized and non-utilized feeding areas at Cedar Key were
compared, no significant differences were found for any climatic
variables (cloud cover, wind speed, wind direction), temporal
variables (minutes to sunset, minutes from sunrise, tidal stage)
or the percentage of spartina (Appendix C1). Significant differences
were found for a number of social, locational and physiognomic
variables. Utilized areas were characterized by more feeding
birds of all species as well as proximity to loafing sites and the
mainland. Physiognomically, utilized areas were typified by
shallower, smoother water with fewer obstructions and more shelter
provided by surrounding land. There was less open water, less
oyster bar, more land-water interspersion and more mudflat in the
utilized areas.
Testing
At St. Vincent's-NWR three days of unstructured observations
indicated that the marine bird community utilized the area in a
manner similar to the Cedar Key avian community. Specifically, the
marine birds arrive in the study area shortly after sunrise. Loaf-
ing site utilization varies as the tides wax and wane. Typically
as the tide recedes, movement toward the mainland occurs, as birds
spread out over the tidal-flats to feed. As the tide advances and
covers these low areas the birds are again concentrated on higher
loafing sites, beaches, progressively farther from the mainland
(Cedar Key 4000 m, St. Vincent's 2200 m). In both localities the
entire community (except for Black-crowned Night Herons (Nycticorax
nycticdrax) and Great Blue Herons (Ardea herodius)) abandons these
inland sheltered areas at dusk to roost on isolated reefs (Cedar
Key Corrigan's Reef) or islands (St. Vincent's Pelican Island).
several kilometers from the mainland.
Study areas were selected such that significant differences
(P < 0.05) in structure existed between the-two sets of areas at
St. Vincent's but notbetween the Cedar Key and St. Vincent's sites
(Table 7). In other words,-two sets of two replicates were selected
at the second site such that their similarity to the Cedar Key sites
was maximized. Selection based on the second factor, distance to
the mainland, was more.difficult since the only areas with structural
features similar to the non-utilized habitat were located close to
the mainland. The third factor, wind speed, varied significantly
(P < 0.01) both between area and location (Table 7). However, since
wind speed did not vary significantly (P > 0.05) between utilized
and non-utilized areas at Cedar Key (Appendix C), it was not heavily
considered in study area selection at St. Vincent's. Thus, those
Table 7. Results of three 2 x 2 Analyses of Variance comparing
structure, distance to the mainland and wind speed within
location and between locations. Locations are Cedar Key,
Florida and St. Vincent's NWR.
Factor 1 Factor 2 Factor 3
Distance to
Effect Structure Mainland Wind speed
Within Location
(utilized & "presumed suitable" **** **** *
vs non-utilized and "presumed
unsuitable")
Between Locations
(Cedar Key vs St. Vincent's) n.s. **** ***
Interaction n.s. n.s. n.s.
variables important in the identification of structure (water depth,
patchiness, percentage of land and percentage of open water) were
the principal criteria for selection of the St. Vincent's sites.
The second major consideration in study site selection was the
distance to the mainland.
When "presumed suitable" and "presumed unsuitable" areas at
St. Vincent's were compared, no significant differences (P > 0.05)
were found in climatic variables (cloud cover, wind speed, wind
direction), temporal variables (minutes to sunset, minutes from
sunrise, tidal stage) or the percentage of spartina (Appendix C).
As expected, significant differences were found.for all structural
variables (water depth, patchiness, percentage of mudflat, percentage
of open water) by which the area types were selected. In addition,
"presumed suitable" areas had more feeders of other species, less
water roughness, more obstructions, less shelter by land, more
oyster bar, and were farther from the mainland than "presumed
unsuitable" areas. Thus, St. Vincent's "presumed suitable" sites
were farther from the mainland, had more oyster bar, less shelter
by land and more obstructions than did Cedar Key's utilized sites.
Allowing for these differences, the area types at Cedar Key and
St. Vincent's were similar for all other variables tested.
Black Skimmers used "presumed suitable" areas 2.5 times more
frequently than they utilized "presumed unsuitable"areas. A
significantly unequal distribution of forages (X2 = 15.1, P < 0.05)
was recorded in the two area types (Table 8). Two factors prevented
this difference from being even greater. First, the area most
Table 8. Chi-square goodness of fit test comparing Black Skimmer
utilization of "presumed suitable" and "presumed unsuitable" areas
at St. Vincent's NWR.
Forages N = 144
Areas Observed Expected Chi-square
Utilized 105- 72
.15.1*
Non-utilized 39 72
representative of utilized habitat was located so close to the only
"presumed unsuitable" areas that it could not be selected as a study
area. Secondly, feeding in this area resulted in some feeding in
the adjacent "presumed unsuitable" area. Considering these factors,
the results of this study indicated that feeding habitat selection
of St. Vincent's Black Skimmers was consistent with that of Cedar
Key Black Skimmers and predictable from the factor analysis of Cedar
Key habitats.
Prey Resource
Distribution and Composition
The importance of Black Skimmer feeding habitat features was
confounded by questions regarding the distribution and abundance
of the prey resource. The importance of structural features should
only be evaluated in conjunction with information regarding the
prey base. In the Cedar Key study no significant difference
(Duncan's P > 0.05) in prey abundance was found between utilized
area C and non-utilized area D (Table 9). In addition, no signi-
ficant difference was found in non-utilized area D, utilized area
A and non-utilized area 0. If the currently used index of prey
abundance were the only criterion used for discriminating between
foraging areas, then areas 0 and 0 should not differ in utilization
from areas A and C. This relation had previously been shown not
to be the case since areas A and C were utilized significantly
more (P < 0.05) than any other study areas and areas D and 0
were consistently non-utilized.
representative of utilized habitat was located so close to the only
"presumed unsuitable" areas that it could not be selected as a study
area. Secondly, feeding in this area resulted in some feeding in
the adjacent "presumed unsuitable" area. Considering these factors,
the results of this study indicated that feeding habitat selection
of St. Vincent's Black Skimmers was consistent with that of Cedar
Key Black Skimmers and predictable from the factor analysis of Cedar
Key habitats.
Prey Resource
Distribution and Composition
The importance of Black Skimmer feeding habitat features was
confounded by questions regarding the distribution and abundance
of the prey resource. The importance of structural features should
only be evaluated in conjunction with information regarding the
prey base. In the Cedar Key study no significant difference
(Duncan's P > 0.05) in prey abundance was found between utilized
area C and non-utilized area D (Table 9). In addition, no signi-
ficant difference was found in non-utilized area D, utilized area
A and non-utilized area 0. If the currently used index of prey
abundance were the only criterion used for discriminating between
foraging areas, then areas 0 and 0 should not differ in utilization
from areas A and C. This relation had previously been shown not
to be the case since areas A and C were utilized significantly
more (P < 0.05) than any other study areas and areas D and 0
were consistently non-utilized.
Table 9. Means, standard errors and results
range test for fish samples collected in two
two non-utilized areas at Cedar Key.
of Duncan's multiple
highly utilized and
Preference Duncan'sa
Area Class Mean & S.E. Test
C high 293.5 + 44.2 A
A
D non 195.2 + 73.7 A B
B
A high 124.1 + 38.6 B
B
0 non 66.4 + 8.9 B
aMeans with the same letter are not significantly different _P = .05)
# No difference in prey resource composition was apparent between
the two utilized areas, A and C, and two non-utilized areas, D and
0, (Fig. 3). lhile fish were classified by family as Sciaenidae
(including spot, drum and perch), Mugilidae (including mullet,
Mugil spp.), Cyprinodontidae (including killifish, Fundulus spp.)
Engraudidae (including bay anchovy, Anchoa spp.), Atherinidae
(including silversides, Menidia spp.) and "others", all shrimp
were grouped together. Shrimp and members of the family Sciaenidae
comprised at least 87 percent of the total catch in each of the
four sampled areas.
Consumption
In order to compare prey abundance with the prey captured by
skimmers, thirteen birds were collected and their stomach contents
analyzed (Appendix D). A total of 63 items were recovered of which
60 percent were shrimp and 40 percent were fish (Table 10). However,
by both weight and volume fish were the predominant (81% and 77%,
respectively) food acquired. In order of abundance the fish
recovered were: longnose killifish (Fundulus similis) 8; striped
mullet (Mugil cephalus) 7; Gulf killifish (F. grandis) 3; larva
(Sciaenidae) 2; diamond killifish (Adenia xenica) 1; Atlantic thread
herring (Opisthonema oglinam) 1; tidewater silvers-ide (Menidia
beryllina) 1; needlenose fish (family Belonidae) 1; larva (Clupeiformes)
1. Longnose killifish, mullet and Gulf killifish were the largest
and most frequently acquired fish species.
Although shrimp were the most abundant prey item recovered,
they represented onTy 7 percent of the weight and 9 percent of the
Frequency histograms of fish and shrimp sampled in
256 seine hauls in two highly utilized areas, A and
C, and two non-utilized areas, D and 0. SC =
Sciaenidae, MU = Mugilidae, CY = Cyprinodontidae,
EN = Engraudidae, AT = Atherinidae, SH = Shrimp,
0 = other.
Figure 3.
UTILIZED
N = 7942
100
90
80
70
60
50
40
301
20
10
SC MU CY EN SH AT 0
CLASSES
N 18,785
SC MU CY EN SH AT 0
CLASSES
NON-UTILIZED
1001
FLcm
SC MU CY EN SH AT 0
CLASSES
100,
90
80
N 12,498
100
90
80
70
60
50
40
30
20
10
N= 4248
w
C,
-LL
0
0
LJ
z
CU
------- I -_
SC MU CY EN SH AT 0
CLASSES
L
Table 10. Percent distribution of weight, volume and-frequency
of fish, shrimp and digested material in thirteen Black Skimmer
stomachs.
Category Weight (%) Volume (%) Frequency (%)a
Fish 81 77 40
Shri-mp 7 9 60
Digested Material 12 .14
aBased on partial or whole specimens.
volume of the analyzed stomach contents. The mean length and weight
of whole specimens were 24.6 + 1.2 mm and 0.11 + 0.2 gm, respectively.
Shrimp species identified from whole specimens included Palaemonetes
pugio 4; P. intermedius 2; and P. vulgaris 1. Finely digested material
accounted for 12 percent of the weight and 14 percent of the volume
of the stomach contents analyzed.
Prey species found in the diets of Cedar Key skimmers did not
differ strikingly from those previously described (Table 11). How-
ever, 2 new fish and 3 new shrimp species were added: Adenia xenica,
Opisthonema oglinum, P. pugio, P. intermedius, and P. vulgaris.
The species composition of fish taken by foraging skimmers
differed markedly compared to the species available within the
foraging zones frequented by the birds. While Fundulus spp. were
by volume the primary prey item acquired by skimmers, they comprised
only one percent of the total sample in each of the areas seined.
Mugil spp., second in importance to skimmers, comprised.from 2 to 10
percent of the prey sampled in the four areas: Shrimp on the other
hand were the most frequent item acquired by skimmers as well as
the most frequent prey item in three of the four areas sampled.
Foraging Patterns
Tidal and Flurry
During the second phase of this study two types of feeding
were apparent. The first type was characterized by several birds
concentrating their efforts in a small area for an extended period
of time. This behavior occurred in the evening and was associated
with mid-tide stages. The second type of feeding behavior was
volume of the analyzed stomach contents. The mean length and weight
of whole specimens were 24.6 + 1.2 mm and 0.11 + 0.2 gm, respectively.
Shrimp species identified from whole specimens included Palaemonetes
pugio 4; P. intermedius 2; and P. vulgaris 1. Finely digested material
accounted for 12 percent of the weight and 14 percent of the volume
of the stomach contents analyzed.
Prey species found in the diets of Cedar Key skimmers did not
differ strikingly from those previously described (Table 11). How-
ever, 2 new fish and 3 new shrimp species were added: Adenia xenica,
Opisthonema oglinum, P. pugio, P. intermedius, and P. vulgaris.
The species composition of fish taken by foraging skimmers
differed markedly compared to the species available within the
foraging zones frequented by the birds. While Fundulus spp. were
by volume the primary prey item acquired by skimmers, they comprised
only one percent of the total sample in each of the areas seined.
Mugil spp., second in importance to skimmers, comprised.from 2 to 10
percent of the prey sampled in the four areas: Shrimp on the other
hand were the most frequent item acquired by skimmers as well as
the most frequent prey item in three of the four areas sampled.
Foraging Patterns
Tidal and Flurry
During the second phase of this study two types of feeding
were apparent. The first type was characterized by several birds
concentrating their efforts in a small area for an extended period
of time. This behavior occurred in the evening and was associated
with mid-tide stages. The second type of feeding behavior was
Table 11. Prey species described in Black Skimmer diets as
recorded in the literature.
Author
Species
Erwin (1977b)
Virginia
Arthur (1921)
Louisiana
Leavitt (1.957)
Fl ori da
Loftin (pers. comm.)
Florida
Menidia sp., Fundulus sp., Anchoa
mitchilli, Mugil sp., Leiostromus
zanthurus, Pomontomus saltatrix
Cynoscion nothus, Aterina sp., C.
nebulosus, Sciaenops ocellatus,
Trachi-notus carolinus, Carangus hiopos,
M. cephalus, Scomberomorus maculatus,
P. saltatrix
Tylosurus sp., Lutjanus sp., Fundulus
sp., Palaemonetes sp.
M. cephalus, F. heteroclitus, Brevoortia
tryannus, Paralichthys sp., Elops saurus,
Echeneis naucrates
characterized by fewer birds foraging in an area for a short period
before moving on to another feeding area. This type of feeding
occurred throughout the day and was associated with low tide. A
bimodal frequency distribution of the total forages confirmed a
temporal difference between feeding times. The frequency distri-
bution of the number of birds per observation period also was bimodal.
Based on these preliminary observations a working definition of the
first foraging type was established. Flurry feeding was defined
as seven or more groups feeding in a single area for at least 20
minutes. All other feeding was considered tidal-flat or tidal
feeding. Of the total forages recorded during the winter 1979-80,
80 percent were tidal forages and 20 percent were flurry forages.
Although the mean number of birds per group did not differ
significantly (P > 0.05) between tidal and flurry feeding, the
key bird forages per group were significantly (P < 0.05) greater
in tidal feeding than in flurry feeding (Table 12). Within an
observation period, the number of birds in tidal feeding was
significantly (P < 0.05) less than in flurry feeding. Additionally,
there were significantly more (P < 0.05) tidal key bird forages
than flurry key bird forages for the same time interval. Thus,
within an observation period tidal feeding was characterized by a
few birds conducting many forages while flurry feeding was typified
by many birds executing fewer forages.
In flurry feeding observations the number of key bird forages
increased logarithmically (r2 = 0.82) with increasing bird numbers.
No similar pattern was found in tidal feeding. Despite this difference
Table 12. Means, standard errors and results of statistical
analyses comparing the nine variables recorded during flurry
and tidal feeding. S = student's t test (homogeneous variance),
A = approximate t test (heterogeneous variance)a W = Wilcoxon's
sum rank test.
Flurry Tidal Statistical
Variable N = 30 N = 89 Test
n.s.
Number of birds pern. 2.5 + 0.2
group
Number of birds per** 11.4 + 2.7
observation period
Number of key bird **** 4.1 + 0.5
forages per group
Number of key bird* 16.5 + 2.5
forages per observation
period
Total forages pern.s. 133.4 + 37.3
observation period
Minutes to sunset**** 59.9 + 16.1
Tidal stage**** 4.8 + 0.6
Distance to loafing*b 1.5
site
2.2 + 0.2
2..3 + 0.3
24.4 + 3.5
29.2 + 4.4
87.1 + 20.2
315.5 + 16.8
8.3 + 0.2
2.5
Distance to mainland**
8.0 + 0.0
4.9 + 0.3
aApproximate t test (Stee
1 and Torrie 1960, p. 81)
Median value replaces mean
and significant differences in key bird forages and number of birds,
the total forages per observation period were not significantly
(P > 0.05) different for tidal and flurry feeding.
Flurry feeding occurred significantly (P < 0.05) closer to
sunset than did tidal feeding (Table 12). In addition, flurry
feeding occurred closer to the loafing site utilized after
feeding (Table 12). A median value of 1.5 indicated that flurry
feeding birds most often used the loafing site within the foraging
area while tidal feeding birds differed significantly (P < 0.05)
in that-they frequently used loafing sites (median 2.5) outside of
the foraging area. Also, flurry feeding occurred at significantly
greater (P < 0.05) distances from the mainland than did tidal
feeding. Thus, flurry feeding occurred closer to sunset, closer
to the utilized loafing site and farther from the mainland than did
tidal feeding.
In both foraging regimes birds fed typically alone or in
pairs with less than 10 percent of the sightings containing more
than five birds (Fig. 4). These results are consistent with
those of Erwin (1977b) who reported skimmers foraging predominantly
singly or in pairs in-Virginia.
Relative frequency histograms of total forages are shown in
Figures 5-8.. During flurry feeding 78.3 percent of the forages
occurred within 100 minutes of sunset (Fig. 5). This contrasts
with only 4.5 percent of tidal feeding during the same interval
since tidal feeding occurred primarily from midmorning until
midday. Increased sampling at preferred tidal stages occurring
at midday confirmed the lack of foraging effort during this period.
Figure 4. Foraging group sizes in tidal and flurry feeding.
40-
30
20'
IO
50-
401
TI DAL
N- 156
in-rn,
U I
I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17
GROUP SIZE
FLURRY
N 132
I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17
GROUP SIZE
507-
Figure 5. Percentage of total forages occurring throughout
the day for tidal and flurry feeding.
TIDAL
N=5460
NOON SUNRISE
MINUTES TO SUNSET
I Im ~
0
SUNSET
FLURRY
N= 1334
0 "
0 i
SUNSET
100 150 200 250 300 350 400 450 500 550 600
NOON SUNRISE
MINUTES TO SUNSET
The percentage of tidal forages occOrring at various tidal
stages is shown in Figure 6. The peak of tidal feeding was in
tidal state 8; 2 hours after low tide. A single long feeding
effort during the winter's lowest tide (26 February 1980) resulted
in an atypical tidal feeding peak 1 hour after high tide. If this
peak is deleted, 69.3 percent of tidal feeding is seen to occur 2
to 3 hours after low tide. Tidal feeding was typically initiated
at low tide, was most frequent during flow tide and was concluded
by high tide. Approximately 70.4 percent of flurry forages occurred
3 to 4 hours after high tide. Flurry feeding was typically initiated
at ebb tide and occurred from ebb tide to early low tide stages.
Flurry feeding and tidal feeding are not mutually exclusive and
both were recorded at evening low tide on -11 and 14 November 1979.
One hundred percent of flurry forages were recorded within
250 m of a subsequently utilized loafing site (Fig. 7). Only
57.5 percent of tidal forages occurred within this distance of the
selected loafing site. Additionally, one hundred percent of flurry
forages occurred over 3500 m from the mainland, while tidal forages
were dispersed from 1 to 4000 m from the mainland (Fig. 8).
Nocturnal
During nine nights of nocturnal observation, skimmers were
recorded utilizing diurnal feeding habitat on only one occasion.
This feeding occurred in early evening (1900-2200) during a full
moon phase and differed from the typical diurnal feeding in the
rapidity with which groups (x = 6.3 birds) moved from one study
area to another. In addition, the degree of vocalization between
Figure 6. Percentage of total forages occurring at twelve tidal
stages for tidal and flurry feeding.' Stage 1 = high
tide, stage 6 = low tide.
48
c5 50
0 40
Li
-1
30
LL
o 20
LU
a. 0 '- O
I2 3 4 5 6 7 8
TIDAL STAGE
LUJ
o 50
o 40
Lu
-
30
Ll
20
LU
z 10
Sr
LU
0
LULr7
0
TIDAL
N=5460
9 10 II 12
FLURRY
N= 1334
TIDAL STAGE
1 2 3 4. 5 6 7 8 9 10 1 12
Figure 7. Percentage of total forages followed by flight to
a loafing site: 1) within area, 2) within 250 m
of the area, 3) over 250 m from the area, and
4) unknown.
100
90:
80
1 2 3 4
lOOr
- I
I I -f
DISTANCE TO LOAFING SITE
FLURRY
N=1334
DISTANCE TO LOAFING
TI DAL
N-5460
20
10
90
80
70
60
50
40
30,
20
10
0
SITE
Figure 8. Percentage of total forages occurring in 500 m
increments from the mainland.
100
90
80
70
60
50
40
3 4 5. 6.
DISTANCE TO MAINLAND
7 8
FLURRY
Nx1334
I -
2 3 4
DISTANCE
5 6
TO MAINLAND
TIDAL
Nx5460
100
901
801
70
60
50
40
30
20
10
0
7 8
53
group members was noticeably greater in nocturnal feeding than in
diurnal feeding. Since nocturnal feeding, however, was recorded
on only this single occasion, evidence is lacking for typical
features of nocturnal feeding.
DISCUSSION
The feeding habits and habitat of wintering Black Skimmers on the
Florida Gulf coast were described first for one locality and then
tested in a second locality. Feeding habitat utilization was similar
in both localities and other aspects of feeding ecology were parallel.
In the following discussion spatial, dietary, temporal and social
facets of Black Skimmer feeding ecology will be discussed as they
apply to the localities studied. However, since most of the infor-
mation derives from Cedar Key that study will serve as the focal
.point.
Spatial Aspects
Tidal Feeding Habitat
Structure
A premise underlying this study is that a species having evolved
a specialized feeding behavior will demonstrate specificity rather
than plasticity in feeding habitat selection. If such specificity
exists then the identification and delineation of such feeding
habitat should be possible. The results of this study indicate
that in tidal feeding such specificity does exist.
Areas utilized most frequently by foraging skimmers had more
mudflat and more land-water interspersion than did non-utilized
areas. While a skimmer may "key into" the amount of mudflat in
an area, the amount of edge created by land-water interspersion also
54
DISCUSSION
The feeding habits and habitat of wintering Black Skimmers on the
Florida Gulf coast were described first for one locality and then
tested in a second locality. Feeding habitat utilization was similar
in both localities and other aspects of feeding ecology were parallel.
In the following discussion spatial, dietary, temporal and social
facets of Black Skimmer feeding ecology will be discussed as they
apply to the localities studied. However, since most of the infor-
mation derives from Cedar Key that study will serve as the focal
.point.
Spatial Aspects
Tidal Feeding Habitat
Structure
A premise underlying this study is that a species having evolved
a specialized feeding behavior will demonstrate specificity rather
than plasticity in feeding habitat selection. If such specificity
exists then the identification and delineation of such feeding
habitat should be possible. The results of this study indicate
that in tidal feeding such specificity does exist.
Areas utilized most frequently by foraging skimmers had more
mudflat and more land-water interspersion than did non-utilized
areas. While a skimmer may "key into" the amount of mudflat in
an area, the amount of edge created by land-water interspersion also
54
DISCUSSION
The feeding habits and habitat of wintering Black Skimmers on the
Florida Gulf coast were described first for one locality and then
tested in a second locality. Feeding habitat utilization was similar
in both localities and other aspects of feeding ecology were parallel.
In the following discussion spatial, dietary, temporal and social
facets of Black Skimmer feeding ecology will be discussed as they
apply to the localities studied. However, since most of the infor-
mation derives from Cedar Key that study will serve as the focal
.point.
Spatial Aspects
Tidal Feeding Habitat
Structure
A premise underlying this study is that a species having evolved
a specialized feeding behavior will demonstrate specificity rather
than plasticity in feeding habitat selection. If such specificity
exists then the identification and delineation of such feeding
habitat should be possible. The results of this study indicate
that in tidal feeding such specificity does exist.
Areas utilized most frequently by foraging skimmers had more
mudflat and more land-water interspersion than did non-utilized
areas. While a skimmer may "key into" the amount of mudflat in
an area, the amount of edge created by land-water interspersion also
54
may be important. Thus, the amount of mudflat in itself may or
may not be critical. At St. Vincent's, for instance, oyster bar
appeared to be a topographic substitute for mudflat. Whether or
not these two features influence a skimmer's selection of an area,
they are important in a bird's utilization of an area. Both edge
mudflat and edge mudflat oyster bar were utilized significantly
more by skimmers than would be expected by their occurrence in the
study areas. The reason for the avoidance of the edge mudflat
spartina zone is unknown.
Water depth and the amount of open water were significantly
less in highly utilized areas than in non-utilized areas. Prey
abundance must be considered with the possible importance of these
habitat features. For instance, non-utilized area D had the second
highest prey yield while at the same time having the greatest water
depth and most open water of the areas studied. While prey abundance
was relatively good the structure of the area was such that the area
was not selected for feeding by skimmers. The highly utilized
areas A and C had relatively high prey abundance, shallow water
and less open water. Finally, area 0 while having shallower water
and less open water than area D, was also non-utilized. Prey
abundance in this area was the lowest of any area.sampled. Thus,
while certain structural features (shallow water, amount of mudflat,
lack of large expanses of open water, interspersion of land and water)
of an area are important criteria for utilization by foraging skimmers,
the value of these features is not independent of prey resource
concentrations and the importance of prey concentrations is not
independent of an area's structure.
Location
Tidal feeding areas typically were located close either to the
mainland or to large land masses. The influence of location is
illustrated when considered conjointly with the speed and direction
of the prevailing winds.
Wind
At Cedar Key the water roughness was significantly less in
utilized areas although the wind speed did not differ significantly
between utilized and non-utilized areas. Two factors may account
for this discrepancy: 1) wind measurements made 2.0 m above the
water may differ from those occurring at the water's surface,
2) significantly greater patchiness in the utilized areas may
influence water roughness by creating calm microhabitats along
land-water interface zones where wind movement at the water's
surface is broken by exposed terrain.
Utilized areas are significantly more sheltered from large
bodies of water than are the non-utilized areas. This location
influences the water roughness and the water depth. Since the
non-utilized areas are unprotected from large bodies of water in
the direction of the prevailing westerly winds, water is typically
blown into these areas. This results in surface water movement,
increased water roughness and deeper water in these areas. This
deeper water covers those features (oyster bar and mudflat) that
both interrupt water movement and alter wind flow adjacent to the
water's surface. Thus, the non-utilized areas due to their location
and lack of shelter from prevailing winds had significantly deeper
57
water, more water roughness and less land-water interspersion. These
features were associated with decreased utilization by foraging
skimmers at both Cedar Key and St. Vincent's NWR.
Canals and Channelization
In this study feeding birds used areas with canals in proportion
to their occurrence but .avoided large channels (Fig. 1, No. 3 & 4
channels) (Fig. 8, distances 3, 5, 6). The lack of preferred
structural features (land-water interface zones) in large channels
may explain the. absence of skimmer feeding. Since traveling distance
between patches varies linearly with the linear dimension of the
patch and the hunting area within a patch varies as its square,
then larger patches offer smaller travel time per unit hunting
area (MacArthur and Pianka 1966). An alternative to larger patches
in reducing traveling time between areas is "connectedness" (MacArthur
1968). These concepts may apply to Cedar Key skimmers that use a
sequence of feeding areas (M, L, E, C, P, 8, A) that allows almost
continuous hunting except in large channels (Fig. 1, No..3 & 4
channels). Large channels may lower feeding efficiency by inter-
rupting the connectedness of feeding areas with preferred structural
features. Canals on the other hand, are not often used by skimmers
but may indirectly result in shallow water, loafing sites and land-
water interface zones within the same area. Canals also are not
large enough to disrupt'the connectedness of feeding areas.
Dietary Aspects
While fish were the predominant prey item by volume in the
stomachs analyzed, empirical observations indicated that shrimp
were more abundant by volume in the environment. In addition, the
fish species captured by skimmers comprised a very small percentage
of the prey resources acquired by seining. Such discrepancies may
have been caused by a number of factors. Firstly, skimmers may
select particular prey species using visual acuity during feeding
tn shallow water. However, the speed of skimmer flight while feed-
ing (26-29 kph) in conjunction with the length and straightness of
this flight, indicates that skimmers do not typically swerve and
dart to obtain prey. A highly innervated bill (Zusi 1962) implies
a tactile feeding method by which the bird could non-selectively
acquire prey as they were encountered in skimming. Also skimmers
exhibit broad food preferences as suggested by the number of prey
species listed previously. However, this lack of specificity in
prey acquisition does not preclude the possibility that skimmers
utilize visual cues in pursuing large prey while exercising tactile
skills to randomly acquire smaller, less visible prey.
Secondly, foraging skimmers are clearly biased in prey acquisition
since they sample only the population subset near the water's surface.
Thirdly, certain fish, i.e. Fundulus spp., occur near the water's
surface and are more accessible to feeding skimmers than those
species that occupy lower strata. Fourthly, seining is inadequate
for accurately sampling the prey available to skimmers because
fast swimming species may escape the seine, and the seine samples
both the surface species and the species at greater depths. Thus,
some other sampling method, e.g. Wegener ring (Wegener et al. 1974),
or a combination of methods may be preferable for sampling prey
availability for Black Skimmers.
The selective nature of the skimmer diet as compared to prey
availability in the Cedar Key study areas may result, therefore,
from predator selectivity, prey behavior and/or sampling inadequacies.
The greatest disparity probably is associated with the logistics of
accurately sampling the prey in the foraging zones used by skimmers.
Temporal Aspects
Previous studies of Black Skimmer feeding biology have indicated
diel (Arthur 1921, Zusi 1959) influences as well as tidal influences
(Erwin 1.977b, Tomkins 1951) on skimmer feeding activity. The feeding
activity of wintering skimmers at Cedar Key was influenced by both
.of these factors. Tidal-flat feeding occurred in midmorning and late
afternoon at low to incoming tide. Flurry feeding occurred typically
within one and a half hours of sunset and at ebb tide stages. Thus,
both feeding patterns are apparently governed by tidal and diel
influences.
Shifts in optimal foraging period may result from tidal and
diel changes and from seasonal and geographic changes. If the
assumption is made that animals expand their foraging time (Schoener
1971) when faced with increased energy requirements or decreased
energy resources then seasonal variation in foraging periods would
seemingly occur. In addition, Erwin (1977b) reported skimmers
foraging primarily at low tide (+ 1 hr) while the results of this
study show that feeding was initiated at low tide but occurred
primarily during the flow tide. Topographic features of the area
in conjunction with prey abundance may reconcile and explain these
differences. Erwin suggested that skimmers acquired fish trapped
in shallow mudflat pools and as the tide came in the birds shifted
to the utilization of tidal streams. Similar patterns were noted
in this study during tidal feeding but isolated tidal pools are
infrequent at Cedar Key due to the very gradual sloping of the Gulf
in this area. Additionally, areas exist which have very gentle
shoreward slopes, thus, allowing the birds up to an hour foraging
time on the tide's leading edge where fish activity is augmented.
Thus, optimal foraging periods may not onlybe tide-determined but
may be indirectly determined by the geographic features of the area.
The degree of nocturnal or crepuscular feeding performed by
Cedar Key Black Skimmers must be greater than the diurnal counter-
part. Of approximately 350 wintering birds, 30 to 40 birds will
typically leave the loafing area during low tide to feed in the
tidal flats. Observations of flight-lines to and from the loafing
site throughout the day confirmed this pattern. Thus, on any given
day the majority of the birds present are not believed to forage
diurnally. While this pattern does fluctuate in relation to moon-
phase, the primary foraging period is believed to occur between
sunset and sunrise. Further investigations will require radio and/or
luminescent tracking techniques. However, some aspects of nocturnal
feeding may be inferred from patterns of diurnal feeding.
If diurnal activity is analyzed in relation to moonchase, an
interesting pattern emerges. Since tidal feeding is during low
tide and mid to late morning (0900-1200) or mid to late afternoon
(1400-1700) (Fiq. 6), then tidal feeding must be associated with
the waxing and waning crescent moon and the waxing and waning
gibbous moon. These moon phases are associated with tides of inter-
mediate volume in the monthly tidal cycle. Conversely, the two
troughs in diurnal tidal feeding activity coincide with the full
and new moons (low tide- 0700-0800, sunrise) and both quarter moons
(low tide- 1230-1330, after noon) (Fig. 6). In other words, diurnal
feeding in the described habitats is more frequent at tides other
than neap tides (monthly low tides associated with quarter moon
phases) and spring tides (monthly high tides associated with full
and new moon phases). If the assumption is made that skimmers main-
tain a consistent dietary intake during a month, then nocturnal
feeding should increase during full moon, new moon and first and
second quarter moons.
Social Aspects
Colonial breeding and group foraging have been interpreted as
adaptations to exploit patchily distributed food resources (Ward
1965, Crook 1965, Krebs 1974). As a colonial breeding bird, however,
the Black Skimmer has been shown to utilize a more uniform food
resource. The species also exhibits less strongly the colonial
tendencies attributed to tern species (Sterna hirundo, Sterna
maxima) that exploit very patchy food resources (Erwin 1977a).
Foraging group size, one indication of social cohesion, is low
in 31ack Skimmers. In Virginia, Erwin (1977b) reported skirmners
hunting singly or in pairs while the mean group size in this study
was 2.2 + 0.2. Reduced foraging group size may be a response to a
fairly uniform prey resource and also a result of difficulties
incurred by skimming in large numbers. As an edge feeding species
(71% of observed feeding was within 2 m of a land-water interface)
skimmers are restricted to a 2-dimensional habitat, as compared to
other terns (i.e. Forster's Tern, Sterna forsterii) which feed in
large numbers and move freely in 3-dimensional space over the water's
surface. Augmented numbers of foraging skimmers in a 2-dimensional
habitat could increase the interference between feeding individuals
and result in reduced foraging efficiency.
One method by which skimmers could increase foraging efficiency
in a 2-dimensional habitat when prey is increased would be the
temporal packing of foraging groups. In flurry feeding group size
did not differ from the mean group size in tidal feeding but the
number of birds feeding per observation period increased significantly.
In addition, 83 percent of flurry forages were within 2 m of a land-
water interface. If prey concentrations are greater then flurry
feeding could provide increased foraging efficiency by allowing
more birds to feed.in an area, in a given period of time, without
violating the restrictions imposed in exploiting a 2-dimensional
habitat. Ward (1965) described the 3-dimensional "roller feeding"
or "fly over feeding" in Ouelea quelea as a mechanism to increase
foraging efficiency in response to increased food abundance. Simi-lar-
ly, feeding is a temporal "roller feeding" in a 2-dimensional habitat.
A selective advantage to gregariousness also may explain this
activity. Flurry feeding is frequently associated with increased
social behavior such as aerial flights and vocalizations. In addition,
the relationship of increased forages with increasing bird numbers,
as found in flurry but not tidal feeding, suggests social stimulation.
The identification of flurry feeding as a social phenomenon or a
foraging tactic requires further investigation concerning changes
in the prey abundance concurrent with flurry feeding.
A second method by which feeders could efficiently exploit a
2-dimensional habitat is through morphological divergence. Recher
(1966) proposed that shorebird species utilizing a 2-dimensional
habitat, the littoral zone, have occupied different segments of
the environment and diverged morphologically to use these segments
with greater efficiency. Although the Black Skimmer exhibits sexual
dimorphism (males are one quarter larger in most dimensions), the
selection pressures causing this divergence could be associated
with either breeding or feeding biology. However, the dimorphic
character of bill size may be related to feeding since some evidence
does exist that male-male and female-female foraging groups pre-
dominate over male-female groups (M. Gochfeld per. comm.). Further
study is needed to confirm or reject this hypothesis and to
investigate sexual stratification in relation to the water's
edge.
MANAGEMENT RECOMMENDATIONS
The management of habitat for coastal birds is necessary in
states such as Florida that have escalating coastal growth. Jahn
(1979) suggests a 4-step planning and evaluation approach to the
conservation of critical habitat of fish and wildlife. These are:
1).identification, 2) delineation, 3) maintenance, and 4) manage-
ment. The primary objective of this study has been the identification
and'delineationof Black Skimmer feeding habitat. I would like to
address further some implications these results have for the
maintenance and management of the species feeding habitat.
First, natural tidal-flat areas must be conserved if the species
is to persist in Florida. These tidal-flats should be characterized
by shallow water (10-20 cm at low tide) with a large degree of land-
water interspersion. Land in this case refers to mudflat, oyster
bar or sandbar. These structural features in conjunction with
shelter provided by nearby land masses can be used to identify
typical habitat. Some species, i.e., the Common Tern (Sterna
hirundo), forage both in tidal-flats and bay/inlet habitats. Because
of the Black Skimmers dependence on land-water interspersion and
shallow water, the species cannot efficiently exhibit such plasticity
in feeding habitat selection. In other words, the restricted niche
breadth described by Erwin (1977a) for Virginia Black Skimmers also
is seen in Florida Black Skimmers. This species is expected to have
limited flexibility in adapting to alternative feeding habitats.
64
65
Small canals are acceptable alterations to skimmer feeding
habitat while large channels are detrimental. Large channels
create large areas of open water with little land-water inter-
spersion. These areas are avoided by feeding skimmers. In addition,
if skimmers do rely on the connectedness of foraging areas for
efficient feeding, then channelization may interrupt this connected-
ness by making some areas unprofitable for skimming.
Natural tidal-flat areas should be maintained for other species
in addition to skimmers. In this study there were significantly
more feeders of other species in the areas characterized by shallow
water, areas of mudflat and land-water interspersion. These features
provided shallow foraging zones for many tidal-flat feeders.
Future research should investigate aspects of colony site-
feeding habitat distances. The shift by coastal colonial species
such as the skimmer td dredge deposit islands may have removed
then from proximity to traditional feeding habitats. The energy
budget of breeding birds may be stressed by these environmental
alterations. A study is needed to investigate .colony site-feeding
habitat distances, adult foraging distances and foraging time, and
colony productivity in altered and traditional habitats.
SUMMARY
Based on 599 structured and 520 unstructured hours of observation
the following findings concerning Black Skimmer winter feeding ecology
on the Florida Gulf coast were noted. The structure of feeding areas
selected by skimmers was typified by shallow water (10-20 cm) inter-
spersed between oyster bar or mudflat. Study areas with deep (> 30 cm),
uninterrupted bodies of water typically were not used by the birds.
Feeding habitat selection was similar at an independent location
and was predictable from water depth, land-water interspersion
and the proportion. of open water and mudflat.
Feeding birds used areas with small canals in proportion to their
occurrence but avoided large channels. The lack of land-water inter-
face zones in these channelized areas may explain the absence of
skimmer feeding. In the geographic and topographic setting studied,
skimmers foraged 71 percent of the time within 2 m of a land.-water
interface and preferred zones next to mudflat and oyster bar. These
preferred structural features are present in areas with small canals
but lacking in areas with large channels.
Skimmers used study areas that were close to the mainland or
other large land masses. This location provided protection from
the prevailing wind, and reduced water roughness. Non-utilized
areas had less shelter provided by surrounding land and greater
water roughness.
The skimmer's preference for particular habitat structure is
not explained by food abundance alone since utilized and non-
utilized areas did not differ significantly (P > 0.05) in prey
abundance. Prey composition was similar among the areas sampled
with shrimp and members of the family Sciaenidae comprising at
least 87 percent of the total catch in each of the four sampled
areas. Prey composition in the environment differed from that of
skimmers' diets. In the diet Fundulus spp. and Mugil sp. were pre-
dominant by weight and volume while shrimp were most frequent. In
the environment Fundulus spp. and Mugil sp. represented less than
10 percent of the prey sampled while shrimp were abundant (> 30%).
The skewed nature of the skimmer diet as compared to prey availability
in this study may be explained by predator selectivity, prey behavior,
and/or sampling inadequacies.
Two winter feeding patterns were observed. Tidal feeding was
characterized by few birds conducting many forages over a large
area during flow tide. Flurry feeding was spatially restricted
and socially intense and generally occurred at evening ebb tide.
In flurry feeding, forages increased logarithmically (r2 = 0.82)
with increasing bird numbers. While this relation suggests social
stimulation, further prey sampling is needed to identify flurry
feeding as a social phenomenon or a foraging tactic.
LITERATURE CITED
Arthur, S. 1921. The feeding habitats of the Black Skimmer. Auk
38:566-574.
Ashmole, N. P. 1968. Body size, prey size and ecological segregation
in five sympatric tropical terns (Aves:Laridae). Syst. Zool.
17:292-304.
Bales, B. R. 1919. Twenty-four hours in a Black Skimmer colony.
Wilson Bull. 31:186-193.
Barada, W., and W. M. Partington, Jr. 1970. Report of an investi-
gation of the environmental effects of private waterfront canals.
Prepared for the State of Florida. 63pp.
Bent, A. C. 1921. Life histories of North American gulls and terns.
U.S. Nat. Mus. Bull. 113. 337pp.
Bobisud, L. E., and W. L. Voxman. 1979. Predator response to
variation of prey density in a patchy environment: A
model. Am. Natur. 114:63-75.
Chamberlain, J. L. 1959. A nesting colony of
Black Skimmers in southwestern Louisiana.
283-284.
Chapman, F. M. 1908. Camps and cruises of an
Appleton and Co., New York. 432pp.
Cooley, W. W;, and P. R. Lohnes. 1971
Wiley and Sons, Inc., New York.
Forster's Terns and
Wilson Bull. 71:
ornithologist.
. Multivariate data analysis.
364pp.
Crook, J. H. 1965. The adaptive significance of avian social
organization. Symp. Zool. Soc. London 14:181-218.
Davis, L. 1951.
53:259.
Fishing-efficiency of the Black Skimmer. Condor
Department of Environmental Regulation. 1978. The Florida
zone management program. Legislative draft. Available
Florida Office of Coastal Zone Manage. 168pp.
coastal
from
Division of State Planning. 1974. Developments of regional impacts:
a summary report for the first year. DSP-BLWM-19-74. Avail-
able from Florida Dept. of Admin. 19pp.
Emlen, J. M. 1966. The role of time and energy in food preference.
Am. Natur. 100:611-617.
Erwin, R. M. 1977a. Foraging and breeding adaptations to different
food regimes in three seabirds: the Common Tern, Sterna hirundo,
Royal Tern, Sterna maxima, and Black Skimmer, Rynchops niger.
Ecology 58:389-397.
1977b. Black Skimmer breeding ecology and behavior. Auk
94:709-717.
and C. E. Korschgen. 1979. Coastal waterbird colonies:
Maine to Virginia, 1977. U.S.D.I. FWS/OBS-79-08. 647pp.
Gilmore, G., and L. Trent. 1974. Abundance of benthic macro-
invertebrates in natural and altered estuarine areas. NOAA
Tech. Rep. .NMFS-SSRF-677.
Gochfeld, M. 1977. Colony and nest site selection by Black Skimmers.
Proc. Colonial Waterbird Group 2:78-90.
Hall, J. R., and W. N. Lindall,.Jr. 1974. Benthic macroinvertebrates
and sediments from upland canals in Tampa Bay, Florida. NOAA
Data Rep. NMFS-94. 221pp.
Hamilton, W. D. 1971. Geometry for the selfish herd. J. Theor.
Biol. 31:295-311.
Jahn, L. R. 1977. Habitat data needs for managing wildlife. Pages
5-12 in Classification, inventory and analysis of fish and
wildlife habitat. U.S.D.I. FWS/OBS-78-76. 604pp.
Krebs, J. R. 1974. Colonial nesting and social feeding strategies
for exploiting food resources in the Great Blue Heron (Ardea
herodius). Behaviour 51:99-134.
J. Ryan and E. L. Charnov. 1974. Hunting by expectation
or optimal foraging? A study of patch use by chickadees.
Anim. Behav. 22:953-964.
Landin, M. C., and. R. F. Soots. 1977. Colonial bird use of dredged
material islands; a national perspective. Proc. Colonial Water-
bird Group 2:62-72.
Leavitt, B. B. 1957. Food of the Black Skimmer (Rynchops nigra).
Auk 74:394.
Lindall, W. N., W. A. Fable, and L. A. Collins. 1975. Additional
studies of the fishes, macroinvertebrates and hydrological
conditions of upland canals in Tampa Bay, Florida. Fish.
Bull. 73:81-85.
Lunz, J. D., R. J. Diaz, and R. A. Cole. 1978. Upland and wetland
habitat development with dredged material: Ecological consider-
ations. Tech. Rep. DS-78-15. U.S. Army Engineers Waterways
Exp. Sta., Vicksburg, MS.
MacArthur, R. H. 1968. The theory of the niche. Pages 159-176
in R. C. Lewontin ed. Population Biology and Evolution.
Syracuse Univ. Press, Syracuse.
and E. R. Pianka. 1966. On optimal use of a patchy
environment. Am. Natur. 100:603-609.
Nicholson, D. J. 1948. Nocturnal fresh-water wandering of the
Black Skimmer. Auk 65:299-300.
Office of the Technology Assessment. 1976. Coastal effects of
offshore energy systems. Prepared for the U.S. Congress.
278pp.
Pearson, T. H. 1968. The feeding biology of sea-bird species
breeding on the Farne Islands, Northumberland. J. Anim.
Ecol. 37:521-552.
Pettingill, 0. S. 1937. Behavior of Black Skimmers at Cardwell
Island, Virginia. Auk 54:237-244.
Portnoy, J. W. 1978. Black.Skimmer abundance on the Louisiana-
Mississippi-Alabama coast. Wilson Bull. 90:438-441.
Recher, H. F. 1966. Some aspects of the ecology of migrant shore-
birds. Ecology 47:393-407.
Schoener, T. W. 1971. Theory of feeding strategies. Ann. Rev.
Ecol. Syst. 11:369-404.
Tomkins, I. 1951. Methods of feeding of the Black Skimmer. Auk
68:236-239.
Ward, P. 1965. Feeding ecology of the Black-faced Dioch, Quelea
quelea, in Nigeria. Ibis 107:173-214.
Wegener, IV., D. Holcomb, and V. Williams. 1974. Sampling shallow
water fish populations using the Wegener ring. Proc. Annual
Conf. S.E. Assoc. Game and Fish Comm. 27:663-673.
Wilson, A. 1813. American Ornithology. Vol. 7. Philadelphia.
Winkworth and Goodall. 1962. A crosswire sighting tube for
point quadrat analysis. Ecology 43:342-343.
Zusi, R.. 1959. Fishing rates in the Black Skimmer. Condor 61:
298.
71
S1962. Structural adaptations of the head and neck in the
Black Skimmer, Rhynchops nigra. Linnaeus Publ. Nuttall Ornithol.
Club. No. 3:1-101.
APPENDICES
APPENDIX A
Principal Factor Analysis
The principal factor procedure uses a correlation matrix of the
measured variables to create factors that are linear combinations of
the original variables and that are independent of one another. The
number of factors created.equals the number of original variables
measured. While these factors account for all of the variation in
the original data, usually the first few factors explain a large
proportion of the variation. Using a principal axis method, the
first factor is that linear combination of the original variables
that accounts for the most variance in the original data. The
second factor is that combination of the original variables that
accounts for the second greatest amount of variance and is independent
of the first factor and so on. Thus, the many variables recorded
are reduced to a manageable number of factors and a more concise
description of feeding habitat is provided.
The factor analysis results in a factor pattern which is comprised
of the loadings of each original variable on each factor. The loading
of a variable is the correlation of that variable with the factor
(Cooley and Lohnes 1971) and indicates the variable's relative im-
portance as part of the linear combination of variables comprising
that factor. Thus, variables with high loadings (e.g. > 0.60) on a
factor are considered important in the identification of that factor.
APPENDIX B
CORRELATION MATRIX CREATED FROM ONE FEEDING-SPECIFIC AND
TWELVE SITE-SPECIFIC VARIABLES
U 0NtOOCO O OU,(e
r NOvi00000COO
Lo -y a M 'fj 0 0 L0 CNJ MT w n
Ovi %oo oo ccioc"Lo
dd-ocd~ddoddd
0 OO -00 00 00
a- t m 3- -0- c'j 0N M- LO ,z r
S idddoodopddo
t m 0 o 9 i 0 0 r-oofM cM yM o
0 r Ln -O'
0OOr- 0000000
-I CO LO C)O:rCOD.O
% t. r Z; 0 U, W Lw LO 0 r0 0 -
C cM nnCM OO0- cM 0 I
0 00 00 00 0 00 000
I I I I I I
0 C) U, 0 CD 0 0 C -=
N- ^COO N CMOC
r. a rU, ).00 Uo o .v U, Lo CM C 4CM
r- cD ; co0 Cm U, 0M1.0 LO
00-00 C000 00
I I I I I I
M m 1cN a i
CD C 0 C, 0C 0 U a 0 0 aa
0< CD r0 CDi-naOU IO
CO00M0CNO a0CO
I I I I I I I I
O.0 C 0 U, U 1
i O MO.0C 00o0-
-- -- -- c~o m-o cct-- CM-0i
0000000000
- C LOU
CD C%.0 I C) U,- 0i 00 0
OO0 'V-- O-CO
000000o00000-
I I f I I I
rL U 0oCD' U, O CM %0 M 00 CD
a d d do a o. dV d a D a 0
LU mr %VU.0 -V CO qr~ I 0O C CO CM
Ll. CM CM M" CD 0- m 0OCM
0 orcomO oCOOCM
000000000000
I I I I I I I
n T 0M r- O C0 C -m
O 1 C ('0 C\ 1ONO (O. CUCO U
0 O-1.0 O O -
a D V 0 d VC CD CD 0 dd 0 0
-- -O O cM ni o- CI) 0 C aq
SMoCOO ooalaconc L
0 000 0 0000-000
I I I I Ia I
0 O OMLO.M Uf, M
.00000 0I00-0
I I I I I I I
Om C 0- CO-CM
U '.00U C 0 00
0 0000 00-00000
aI I I I I
'. m 'r
U- (A M U -j Z on -j t0
- a < U- M -- u .0
E 3100203211
a
-o
0
vV
IV
as
c c
CL
0
4- 4-
) *O
0 0
C C
. o
0 0
4- 4-
0 0
1) (V
O 4
00
SU
Ca a
0 0
0 0
0 0
0 0-
0 0
1 4-,
C C
U (
= c
GJ 0a
Il IA
a a
*a Ca
APPENDIX C
STATISTICAL ANALYSES COMPARING SITE-SPECIFIC VARIABLES IN
AREA TYPES AT CEDAR KEY AND AT ST. VINCENT'S NWR
Table Cl. Means, standard errors and results of statistical analyses
comparing site-specific variables in utilized and non-utilized feed-
ing habitat in the Cedar Key area. S = student's t test (homogeneous
variance), A = approximate t test (heterogeneous variance)a, W =
Wilcoxon's sum rank test.
Non-
Utilized utilized Statistical
Variable N = 49 N = 70 Test
Physical
Minutes to sunset'
Minutes to sunrisen.s.
Cloud Covern.s.
Wind directionn.s.
Wind speedn-s.
266.7
414.0
54.4
240.1
7.6
26.8
26.1
4.8
14.3
0.7
321.8
380.2
47.1
201.6
6.6
24.1
24.3
3.6
16.0
0.5
Social
Number of other feeders*
18.7 + 6.5
2.1 + 0.3
Locational
Distance from loafing site****
Distance to the mainland**
5.9 + 0.5
2.8 + 0.2
13.4 + 0.2
3.5 + 0.1
Physiognomic
Tidal stagen.s -
Water depth****
Patchiness****b
Water roughness**
Shelter by land****
Number of obstructions**
Mudflat or Land****c
Oyster bar****
Spartinan.s.
Open water****
7.8 + 0.1
13.4 + 1.1
4.0
69.6 + 3.5
5.8 + 0.5
5.0 + 0.3
43.8 + 2.8
11.6 + 0.9
17.1 + 1.6
22.7 + 2.6
7.8 + 0.1
31.9 + 1.5.
2.0
80.0 + 2.6
1.9 + 0.3
6.4 + 0.2
8.2 + 1.5
16.3 + 0.6-
13.6 0.6
62.2 + 1.6
aApproximate t test (Steel and Torrie 1960, p. 81).
bMedian value replaces mean.
cSince no sandbar was recorded, mudflat and land are synonomous.
Table C2. Means, standard errors, and results of statistical analyses
comparing site-specific variables in utilized and non-utilized feed-
ing habitat in the St. Vincent's area. S = student's t test (homo-
geneous variance), A = approximate t test (heterogeneous variance)a,
W = Wilcoxon's sum rank test.
Presumed Presumed
Suitable Unsuitable Statistical
Variable N = 41 N = 39 Test
Physical
Minutes to sunsetn.s.
Minutes from sunrisen.s"
Cloud covern.s.
Wind directionn.s.
Wind speedn.s.
293.6
339.3
36.3
104.1
8.2
22.6
22.1
6.9
15.1
0.6
310.6
322.8
53.5
108.9
7.6
20.8
20.6
7.1
17.5
0.5
S
S
S
S
S
Social
Number of other feeders****
Locational
Distance from loafing site****
Distance to the mainland****
Physiognomic
n.s.
Tidal stage
Water depth****
Patchiness****b
Water roughness****
Shelter by land**
Number of obstructions*
Mudflat or land****c
Oyster bar****
Spartinan.s.
Open water****
10.2 + 1.3
4.1 + 0.4
1.4 + 0.1
7.7 + 0.1
21.0 + 2.2
4.0
73.2 + 4.2
4.4 + 0.4
4.5 + 0.3
33.6 + 3.2
28.2 + 1.8
17.9 + 1.4
19.7 + 2.0
2.7 + 0.4
19.1 + 0.2
1.0 + 0.0
7.8 + O.T
43.6 + 2.5
92.3
7.2
3.7
0.7
17.4
16.6
65.9
aApproximate t test (Steel and Torrie 1960, p
81).
Median value replaces mean.
CSince no sandbar was recorded, mudflat and land are synonomous.
APPENDIX D
APPENDIX D
and shrimp specimens recovered from thirteen Black
Weights and lengths of whole specimens are given.
Skimmer
Number of Whole Length Weight
Specimens Specimens (mm) (gm)
FISH
Species
Fundulus similis
Mugil cephalus
F. grandis
Adenia xenica.
Opisthonema oglinum
Menidia beryllina
.Family
Sciaenidae
Belonidae
Order
Clupeiformes
Total
69.9 + 6.9
51.3 + 6.7
60.7
S38.2
20.4
18.4
25
3.71 + 1.09
1:44 + 0.65
2.31
0.85
0.06
0.01
13
SHRIMP
Species
Palaemonetes pugio
P. intermedius
P. vulgaris
Unidentified
Total
24.6 + 1.2a 0.11 + 0.02a
aValues for whole shrimp
Fish
stomachs.
BIOGRAPHICAL SKETCH
Barbara Buttram Black was born on February 18, 1951, and was
raised in Chattanooga, Tennessee. She graduated in 1969 from
Girl's Preparatory School and entered Vanderbilt University where
she received a B.S. degree in 1973 with a major in general biology
and a minor in chemistry. Upon graduation she entered the field
of nuclear medicine and supervised two departments in the Knoxville,
Tennessee region for three years until she enrolled in supplemental
ecology courses at the University of Tennessee. In 1978 she
married David Joseph Black and in the following year moved to
Gainesville, Florida,where he joined the faculty of the College of
Veterinary Medicine at the University of Florida. She entered
the School of Forest Resources and Conservation from which she
received a Master of Science with a specialization in wildlife
ecology.
|
RSS
HIDE
