|
TECHNICAL REPORT NO. 19
Status Survey and Habitat Evaluation
of the
Cape Sable Seaside Sparrow
in
East Everglades, Florida
by
Timothy E. O'Meara
and
Wayne R. Marion (Principal Investigator)
Department of Wildlife and Range Sciences
University of Florida
Final Report
Submitted to:
U. S. Army Corps of Engineers
Jacksonville, Florida 32232
and
Endangered Species Field Office
U. S. Fish and Wildlife Service
Jacksonville, Florida 32250
From
Florida Cooperative Fish and Wildlife Research Unit
U. S. Fish and Wildlife Service
University of Florida
Gainesville, Florida 32611
Research Work Order No. 28
October 1985
CONTENTS
ABSTRACT...................................................... .... 2
INTRODUCTION.......................................... ...... 2
METHODS...................................... ....... .......... .. ... 4
Data Collection................................................ 4
Data Analysis................................................. 6
RESULTS AND DISCUSSION.......................................... 8
Sparrow Distribution ......... ... .................. .... 8
Habitat Analysis... ........................................... 10
Habitat Suitability........................................... 12
CONCLUSIONS.................................................... 14
ACKNOWLEDGMENTS ..................................... ..............17
Abstract: Cape Sable Seaside Sparrows were detected at 26% of 131
points sampled in East Everglades. Sparrows were found in greater
numbers, and in many cases at different points, than in a 1981 survey.
Changes in distribution could possibly be due to successional changes in
vegetation in the intervening years. Vegetation measurements describing
cover, shrub density, and distance to trees or tree islands appeared
most useful for predicting occurrence of sparrows at a point.
Vegetation variables describing the amount of clumping in grass and
grass-like vegetation appeared most useful for predicting number of
sparrows at a point. A habitat suitability index model previously
derived was significantly correlated with an index based on "expert
opinion," but apparently could be improved to better predict sparrow
occurrence. Data resulting from this survey should be useful for
refining the habitat-suitability index model to more accurately predict
occurrence of Cape Sable Seaside Sparrows.
The Cape Sable Seaside Sparrow (Ammospiza maritima mirabilis) is
federally listed as an endangered species due to its restricted
distribution and specific habitat requirements. The species was
discovered in 1918 and, as a result of the inaccessibility of its
habitat, the extent of its range was largely unknown for several
decades. Stimson (1956) described its former and current distribution
based on records and sightings. Werner (1971) reported the species'
rediscovery on Cape Sable and surveyed known populations (Werner 1975).
More recently, historical records of the species' distribution were
summarized and its range during the period 1978-1981 was systematically
surveyed (Bass and Kushlan 1982, Kushlan and Bass 1983).
Vegetation changes and fire have been identified as primary factors
influencing the distribution of the Cape Sable Seaside Sparrow (Kushlan
et al. 1982). 'In general, the species occurs in "vast brushless
graminoid seasonally flooded interior prairies" (Werner 1975:206). In a
recent census of the sparrow, muhly (Muhlenbergia filipes) prairie and
mixed prairie accounted for 96% of the habitat occupied (Bass and
Kushlan 1982, Kushlan and Bass 1983). Fire affects quality of these
habitats as a result of changes in cover and biomass of living and dead
vegetation (Taylor 1983). Schroeder and Armbruster (1985) derived a
habitat model for evaluating habitat quality based on vegetation
characteristics.
Continued developmental pressures and habitat changes necessitate
up-to-date information on the status of the Cape Sable Seaside Sparrow
for evaluation of developmental impacts and water management practices,
for assessment of critical habitat boundaries, and for recovery
planning. This report describes a survey to identify distribution of
the species in East Everglades. Data describing vegetative
characteristics also were collected to enhance our ability to predict.
habitat suitability.
Objectives:
1. To determine the current distribution of Cape Sable Seaside
Sparrows in East Everglades.
2. To describe selected vegetative characteristics at points
surveyed for Cape Sable Seaside Sparrows.
3. To identify vegetative characteristics correlated with
distribution of Cape Sable Seaside Sparrows.
4. To evaluate the utility of the habitat model developed by
Schroeder and Armbruster (1985) for predicting occurrence of
Cape Sable Seaside Sparrows.
METHODS
Data Collection
Surveys were conducted in known and suspected sparrow habitat in
East Everglades. The survey included treeless freshwater marshes and
prairies; urban and agricultural areas were excluded. Areas sampled
during the 1981 survey were inventoried as well as additional areas of
potential habitat within and around critical habitat boundaries (Figs. 1
and 2). The survey area was gridded into blocks 1 km on a side and
plotted on U. S. Geological Survey 7.5 minute quadrangle maps. Sparrows
were inventoried at locations indicated by intersection of the grid
lines between 3 April and 6 May, 1985.
Each sample consisted of landing at a point by helicopter and
listening for singing Cape Sable Seaside Sparrows for 12 minutes.
Additional sparrows detected when points were revisited for vegetation
sampling were recorded. Also, 31 points where sparrows had not been
detected on the first 2 visits were each revisited for 5 minutes to
increase the completeness of the distribution survey. Numbers of birds
seen or heard were recorded and distances to the birds identified by
sight were estimated at each point. Bird surveys were continued for 3.0
2. To describe selected vegetative characteristics at points
surveyed for Cape Sable Seaside Sparrows.
3. To identify vegetative characteristics correlated with
distribution of Cape Sable Seaside Sparrows.
4. To evaluate the utility of the habitat model developed by
Schroeder and Armbruster (1985) for predicting occurrence of
Cape Sable Seaside Sparrows.
METHODS
Data Collection
Surveys were conducted in known and suspected sparrow habitat in
East Everglades. The survey included treeless freshwater marshes and
prairies; urban and agricultural areas were excluded. Areas sampled
during the 1981 survey were inventoried as well as additional areas of
potential habitat within and around critical habitat boundaries (Figs. 1
and 2). The survey area was gridded into blocks 1 km on a side and
plotted on U. S. Geological Survey 7.5 minute quadrangle maps. Sparrows
were inventoried at locations indicated by intersection of the grid
lines between 3 April and 6 May, 1985.
Each sample consisted of landing at a point by helicopter and
listening for singing Cape Sable Seaside Sparrows for 12 minutes.
Additional sparrows detected when points were revisited for vegetation
sampling were recorded. Also, 31 points where sparrows had not been
detected on the first 2 visits were each revisited for 5 minutes to
increase the completeness of the distribution survey. Numbers of birds
seen or heard were recorded and distances to the birds identified by
sight were estimated at each point. Bird surveys were continued for 3.0
5
hours following sunrise or until increasing.wind velocity inhibited
singing by territorial males.
Vegetation sampling was conducted to describe parameters used in
the habitat model derived by Schroeder and Armbruster (1985). All
points, with the exception of 11, were revisited in late morning and
early afternoon for vegetation sampling. The 11 points that were not
revisited occurred in agricultural areas (4), on roads (1), and in areas
where tree and shrub densities precluded use by Cape Sable Seaside
Sparrows (6).
Herbaceous cover, soil depth, and woody plant densities were
described in each vegetation sample. Distance to the nearest tree
island in each quadrant was estimated to assess openness of the habitat.
Trees were defined as woody plants > 4 m tall. To assess herbaceous
cover and soil depth, a 1-m2 frame was placed at 24 sample points evenly
spaced at 8-m intervals around a square 48 m on a side and centered at
the survey point. At each placement of the frame, 5 parameters were
measured: 1) The most abundant 3 or 4 plant species were recorded in
order of abundance based on percent cover. 2) Percent cover of
herbaceous vegetation, living and dead, was estimated. 3) Number of
clumps of vegetation were counted where a clump was defined as a
bunch-grass-type growth form with a basal diameter.> 5 cm. 4) Percent
of total herbaceous vegetation cover in clumped growth form was
estimated. 5) Soil depth was measured by inserting a sharpened steel
rod (120 cm long, 5 mm in diameter, and calibrated in cm) into the soil
to bedrock at each of the 4 corners of the frame. Soil depths > 110 cm
were recorded as 120 cm.
Shrub densities were estimated by counting shrubs within the 48-m x
48-m squares. Additionally, the helicopter was flown to an elevation of
120 m (400 ft) over each survey point and vertical photos were taken in
an attempt to provide supplementary data on shrub densities. Two film
types were used, color IR and Kodachrome 64.
Subsequent to completion of all sampling, 49 points were revisited
with 0. L. Bass, Jr. (South Florida Research Center, Everglades National
Park) to get an independent assessment of habitat suitability for Cape
Sable Seaside Sparrows. Points were selected that represented a variety
of vegetation conditions over the survey areas. At each point, Bass'
recorded his estimate of relative habitat suitability on a scale of 1 to
10.
Data Analysis
Fifteen variables were derived from the soil and vegetation data
(Table 1) and their values were determined for each point for comparison
with sparrow occurrence. COVER was derived to quantify the amount of
herbaceous vegetation at each point. CLUMPNO, CLUMPCOV, and CLFREQ were
calculated as alternative indicators of the amount of vegetation in
bunch-grass-type growth form. COVERCV and COVFREQ were used as
alternative methods of assessing the "patchiness" of herbaceous
vegetative cover or the "evenness" with which cover was distributed.
Similarly, CLCOVCV and CLNOCV were used as indicators of the
distribution of clumped vegetation. SHRUBDEN was. determined from the
number of shrubs within the 48-m x 48-m square. Resolution of the
aerial photos proved inadequate for estimating shrub densities.
Three alternative indices of tree or tree-island density were used.
TREEMEAN and TREEMIN were measures of distance from the sampling point
to nearest tree or tree island while TREEDENS served as an index to the
area around .each point without trees or tree islands (i.e. density).
SOIL was recorded as a possible indicator of vegetative characteristics.
HSI was calculated from COVER, COVFREQ, CLFREQ, SHRUBDEN, and TREEMIN as
described by Schroeder and Armbruster (1985). SHSI was an index of
habitat suitability as determined by 0. L. Bass, Jr.
Correlation and regression were used to identify relationships
between the 15 variables and the number of sparrows detected at each
point. Two correlation analyses were done: one using all sample points
and one using only sample points where birds were detected. Similarly,
stepwise multiple regression was used to identify the variables most
useful for predicting the number of birds at a sample point using (1)
all sample points and (2) sample points where birds were detected.
Significance levels were set at P 0.15 for entering and retaining
variables in both regression models. Models were constructed using only
9 of the 15 habitat variables. CLUMPNO and CLUMPCOV were omitted from
this analysis because they were highly correlated with CLFREQ. When no
clumps were detected at a sample point, CLNOCY and CLCOVCV could not be
calculated. As a result, CLNOCV and CLCOVCV were omitted from the
regression analyses to maximize sample sizes. HSI and SHSI were not
included in these analyses because they did not represent vegetative
characteristics that could be used to classify sites.
T-tests and discriminant function analysis were used to identify
variables that may be useful for distinguishing between sites with and
without sparrows. Non-paired t-tests were used to compare mean values
for each of the variables between the 2 site classifications. Stepwise
discriminant function analysis was done to produce a discrimination
model that uses a subset of the variables to classify points. The same
9 variables used in stepwise multiple regression were employed in this
analysis. Significance levels for entering and retaining variables in
the model were set at P < 0.15.
Correlation and stepwise regression identify linear trends in data.
The variables measured, however, may not vary in a linear fashion with
habitat suitability for Cape Sable Seaside Sparrows (see Schroeder and
Armbruster 1985). To better examine the relationships between habitat
suitability and the variables described, sample points were grouped
based on values of each variable. The percentage of sample points in
each group at which birds were found was determined. In other words,
the range of data values for each variable was divided into 3 to 13
groupings. Sample points were then categorized into these groupings
based on the value of the respective variable. The percentage of sample
points in each grouping at which birds were detected was then plotted as
a histogram for each variable (e.g. Fig. 5).
Data-were analyzed on an Amdahl 470 V/6-11 computer at the
Northeast Regional Data Center, the University of Florida. The
Statistical Analysis System (Freund and Littell 1981, Ray 1982) was
utilized for all analyses.
RESULTS AND DISCUSSION
Sparrow Distribution
Cape Sable Seaside Sparrows were detected at 34 of 131 (26%) points
sampled (Figs. 1 and 2). Between 1 and 7 birds were detected at each of
for each of the variables between the 2 site classifications. Stepwise
discriminant function analysis was done to produce a discrimination
model that uses a subset of the variables to classify points. The same
9 variables used in stepwise multiple regression were employed in this
analysis. Significance levels for entering and retaining variables in
the model were set at P < 0.15.
Correlation and stepwise regression identify linear trends in data.
The variables measured, however, may not vary in a linear fashion with
habitat suitability for Cape Sable Seaside Sparrows (see Schroeder and
Armbruster 1985). To better examine the relationships between habitat
suitability and the variables described, sample points were grouped
based on values of each variable. The percentage of sample points in
each group at which birds were found was determined. In other words,
the range of data values for each variable was divided into 3 to 13
groupings. Sample points were then categorized into these groupings
based on the value of the respective variable. The percentage of sample
points in each grouping at which birds were detected was then plotted as
a histogram for each variable (e.g. Fig. 5).
Data-were analyzed on an Amdahl 470 V/6-11 computer at the
Northeast Regional Data Center, the University of Florida. The
Statistical Analysis System (Freund and Littell 1981, Ray 1982) was
utilized for all analyses.
RESULTS AND DISCUSSION
Sparrow Distribution
Cape Sable Seaside Sparrows were detected at 34 of 131 (26%) points
sampled (Figs. 1 and 2). Between 1 and 7 birds were detected at each of
these 34 points for a total of 68 sparrows. We found 19 birds at 9 of
59 points included on the Grossman Hammock Quadrangle Map, and 49 birds
at 25 of 72 points included on the Royal Palm Ranger Station SE
Quadrangle Map.
Bass and Kushlan (1982) reported sparrows occurring at 23 of 106
(22%) of the points we sampled (Figs. 3 and 4). If we consider only
sparrows detected during our 12 minute sampling period for comparison
with their data, the number of points at which sparrows were detected
are comparable between the 2 surveys, but number of birds detected was
greater during the 1985 survey. Bass and Kushlan (1982) found 7
sparrows at 5 of 38 points on the Grossman Hammock Quadrangle Map,
compared to 13 sparrows at 5 of the same 38 points in our survey. They
reported 25 sparrows at 18 of 68 points on the Royal Palm Ranger Station
SE Quadrangle Map, compared to 41 sparrows at 20 of the same 68 points
in our survey. Greater numbers of sparrows detected during the 1985
survey were not necessarily a result of greater densities compared to
1981. The 12 minutes we spent at each point may have averaged .greater
than the 10-15 minutes spent at each point during the 1981 survey (0. L.
Bass, pers. commun.).
Although sparrows were found at approximately the same number of
points during the 2 surveys, distributions differed. On the Grossman
Hammock Quadrangle Map, only 1 of the points at which we found sparrows
was coincident with a point where Bass and Kushlan found sparrows.
Other points with sparrows were up to 3 km from the nearest point where
sparrows were found in the 1981 survey. Sparrow distributions on the
area covered by the Royal Palm Ranger Station SE Quadrangle Map showed
considerable overlap between the 2 surveys, but sparrows were still
found at points up to 3 km from the nearest location at which sparrows
were found in the 1981 survey. Sparrows also were more widespread on
this area during the current survey than during the 1981 survey.
Habitat Analysis
Insufficient data were obtained on sparrow detection distances to
quantify detectability and to estimate densities. Detectability,
however, should not have differed appreciably among points and relative
comparisons of numbers of sparrows detected with vegetation variables
therefore could be made.
Of the 15 variables tested, only SHSI was correlated (P < 0.05)
with numbers of sparrows per point over all points (Table 2). Two other
variables, TREEMIN and CLUMPNO, had P values < 0.1. When the 15
variables were correlated with numbers of sparrows only at points where
sparrows occurred, 3 correlations with P < 0.5 were identified (Table
3). CLUMPNO, CLUMPCOV, and CLFREQ were positively correlated with
sparrow numbers across these points, indicating that amount of clumping
in the vegetation influenced habitat quality.
Stepwise regression resulted in only 1 variable being included in
each of the 2 models derived with significance levels set at P 0.15.
Coefficients of determination were low in both cases. The regression
for bird numbers at all points resulted in the model:
y=0.0026 (TREEMIN) + 0.37, (r2=0.02, P=0.143).
The regression for bird numbers only at points where birds were detected
resulted in the model:
y=0.059 (CLFREQ) + 1.34, (r2=0.16, P0.022).
Both the correlation and multiple regression analyses indicated
that some measure of vegetation clumping is a predictor of habitat
quality where sparrows occur. TREEMIN was the variable most useful for
predicting sparrow numbers, but only when all points were included.
TREEMIN may be more useful for separating sites with suitable habitat
from unsuitable sites than for predicting habitat quality. The
relatively high correlation of SHSI with bird numbers at all points
suggests that "expert opinion" can identify suitable sparrow habitat.
Values from the HSI model, however, were not significantly correlated
with sparrow numbers.
Results of the t-tests identified 5 vegetation variables that were
at least marginally significant (P < 0.1) and may be useful for
separating points with and points without sparrows (Table 4). Values
for COVFREQ, SHRUBDEN, TREEMEAN, TREEMIN, and TREEDENS differed between
the 2 point classifications. No variables describing vegetative
clumping differed between the 2 sets of points. Again, SHSI exhibited
the greatest difference between the point classifications, while HSI
values did not differ between the 2 point classifications.
Classification criteria in the discriminant function analysis were
based on the pooled covariance matrix, despite the fact that a
likelihood-ratio test indicated non-homogeneity of the within-group
covariance matrices. Although homogeneity of covariance matrices is an
underlying assumption for discriminant function analysis, violation of
this assumption is typical for ecological data and does not necessarily
negate the derivation of biologically meaningful results (Green 1971).
Stepwise discriminant analysis resulted in a significant 3 variable
model for distinguishing points with and without sparrows:
12
y=0.032 (COVFREQ) 0.025 (CLFREQ) 0.008 (TREEMIN) 2.988,
[F (3,106) = 3.55, P = 0.107]
Points with a solution to this equation < 0 were classified as "sparrows
present," points with y > 0 were classified as "sparrows absent."
The discriminant model did not serve as a very good predictor of
point classification. Forty of 124 (32%) points were misclassified as a
result of this model. Thirty points where birds were absent were
classified as "present"; 10 points where birds were present were
classified as "absent." Most (75%) of the misclassifications predicted
sparrows to occur where they did not. The poor performance of this
model may have been the result of the low utility of the variables
measured for predicting sparrow occurrence. Alternatively, birds may
have been absent from, or present but not detected at, some points with
suitable habitat.
Habitat Suitability
If the purpose of a habitat suitability model is to predict habitat
quality for a species, then some objective measure of habitat
suitability should be defined for evaluating effectiveness of the model.
The ultimate measure of habitat suitability is the ability of the
habitat to support a population that will contribute to the future gene
pool of the species (Fretwell 1972). Assessment of the productivity of
a species in a habitat often requires difficult and time consuming
measurements; most often, species density is used as an alternative
indicator of habitat suitability. If the utility of a habitat
suitability index is to predict occurrence of a species in a habitat,
then frequency of occurrence of the species in habitats with given
13
characteristics may be a suitable measure for evaluation. In this
context, a habitat suitability index with a scale of 0-1 should
represent the probability of finding the species at a site with given
characteristics, and the regression of percent occurrence of the species
on the index should have an intercept of 0 and slope of 1. Bass's
subjective index (SHSI) proved to be a good predictor of sparrow
occurrence (Fig. 6) and approached our criteria for optimal performance
of an index for predicting sparrow occurrence. Regression of percent
frequency of sparrow occurrence on SHSI resulted in the equation:
y-0.9 (SHSI) 1.9, (r=0.87).
These results suggest that the goal of the workshop to derive a habitat
suitability index model that simulated "expert opinion" of habitat
suitability was appropriate. HSI was significantly correlated with SHSI
(P < 0.01), but the low correlation coefficient (r=0.46) suggests that
the model could be improved to better mimic "expert opinion." Only 1
point had an HSI value > 0.29, while SHSI values ranged from 0.2-0.8.
Apparently, some of the suitability index curves that comprise the
habitat suitability index models need to be modified or shifted to the
left to make the resulting indices better predictors of species
occurrence.
Histograms of sparrow frequency of occurrence versus habitat
variable values were plotted (Figs. 7 to 19) to facilitate evaluation of
these models. Caution must be exercised in interpreting these charts;
sample sizes from which percentages were calculated should be
considered. In most cases, sample size declined toward one or both ends
of the x-axis, resulting in lower reliability on the extremes of the
histograms. These charts should, however, be useful for refining
habitat suitability indices.
For example, the histogram for COVER (Fig. 7) followed the pattern
of the suitability index curve for SIV1 derived by Schroeder and
Armbruster (1985). If the suitability index SIV1 of Schroeder and
Armbruster was interpreted to represent habitat suitability based on
frequency of occurrence, then maximum frequency of occurrence would be
expected to occur at between 50 and 70% cover based upon their model.
The maximum percent occurrence of sparrows, however, occurred in the
30-39% COVER class, suggesting that the curve for SIV1 may function
better if shifted to the left. Alternative interpretations are that our
subjective appraisal of percent cover was biased relative to that of
Werner (1975), which was used for deriving the models, or that percent
cover requirements may differ.between the mixed-prairies we sampled and
the muhly prairies that Werner sampled.
Although the histogram for COVER was easily interpretable, COVER
was not indicated by the statistical tests we conducted as a variable
useful for predicting sparrow occurrence or sparrow numbers, Our
analyses would be useful for identifying linear relationships, but not
curvilinear or non-linear relationships. CLUMPNO (Fig. 10), CLFREQ
(Fig. 14) and SOIL (Fig. 19) may be additional variables that are useful
for predicting sparrow occurrence, but in a non-linear fashion.
CONCLUSIONS
Results of the survey suggest that Cape Sable Seaside Sparrow
numbers have not decreased, and may have increased in the study area
since 1981. That sparrow distributions have changed on the study areas
over the 4 intervening years is not surprising since habitat suitability
depends on burning history and habitat quality can decline as early as 4
years after the most recent burn (Taylor 1983).
Since the primary objective of the study was to document current
distribution of Cape Sable Seaside Sparrows in East Everglades,
vegetation and soil sampling were conducted within the constraints of
meeting this objective. The number of points to be censused for
sparrows precluded extensive vegetation sampling. As a result, a number
of factors may have contributed to lack of correlation between
vegetation measurements and sparrow detections. 1) Habitat sampling was
restricted to a 48-n4 x 48-m square around the bird sampling point while
a number of sparrows were detected outside this square (Fig. 20).
Vegetation sampled at a point, therefore, was not necessarily
characteristic of vegetation inhabited by all nearby sparrows. 2) The
minimal number of visits to each point to census birds may not have
always resulted in sparrow counts that were related to habitat quality.
Counts may have differed as a result of weather, time of day, and period
of the season, as well as habitat quality. 3) Ahumber of sites where
sparrows were not detected during the first 2 visits were revisited for
additional sampling, resulting in an uneven sampling procedure that may
have biased the analyses. These revisits, however, should have
augmented the completeness of the survey and should have improved our
correlations between vegetation characteristics and habitat use. 4) A
one-year survey of birds may not have been of long enough duration to
accurately reflect long-term habitat characteristics to which sparrows
were responding.
Despite these sources of variation, some of the variables selected
for sampling did prove useful for predicting sparrow counts or
distinguishing points with and without sparrows. Variables describing
clumping attributes of the grass and grass-like vegetation proved most
useful for predicting sparrow counts. TREEMIN, COVFREQ, and CLFREQ were
most useful for classifying points. TREEMIN in this analysis was not
the same as the suitability index for distance to trees or tree islands
(SIV5) in Schroeder and Armbruster (1985). TREEMIN, in the correlation
analyses used here, served as an index of tree or tree island density,
whereas SIV5 represented the minimum distance from a tree or tree island
at which Cape Sable Seaside Sparrows would be expected to occur.
Although a sampling point may have been less than 50 m from a tree
island, for example, sparrows detected from that point may have been
more than 50 m from the tree island.
The lack of correlation between HSI values and sparrow numbers and
the failure of HSI values to differ between points with and without
sparrows may have been due to several factors in addition to those which
affected correlations between vegetation measurements and sparrow
occurrence. Variables measured for the HSI model may not have
represented vegetation characteristics to which the sparrows were
responding. A second possibility is that our ocular estimation of
subjective vegetation measurements, such as % cover, may have differed
from those on which the model was based. Model variables describing
herbaceous cover were derived from Werner's (1975) data. If our
subjective estimates differed from his, or if the habitats we sampled
were sufficientlydifferent from the ones from which the model was
derived, the model may have resulted in an inconsistent quantification
of habitat suitability. A final possibility is that the relationships
between the variables measured and habitat suitability may have been
misrepresented in the model.
Although the HSI model did not prove very effective for predicting
sparrow numbers or predicting sparrow occurrence, it was significantly
correlated with Bass's subjective evaluation of habitat suitability. In
this respect, it did approach accomplishing the goal for which it was
derived. Further attempts to refine the model might focus on modifying
the model to mimic Bass's indices for the 49 points he evaluated. The
refined model could then be tested through correlation with sparrow
occurrence at other points sampled in this study or at additional points
sampled in the future.
The purpose of this study was not to derive a better habitat
suitability index model for Cape Sable Seaside Sparrows than the one
previously proposed, and no attempt was made to do so herein.. Sampling
procedures employed in the study, by necessity, were not optimal for
deriving a habitat suitability index. The results, however, in
conjunction with other published data and "expert opinion,," should be
proficuous for refining the suitability index model to better predict
Cape Sable Seaside Sparrow occurrence.
ACKNOWLEDGMENTS
We are grateful to D. David and G. Pullen for their assistance in
data collection. T. Edwards provided helpful advice on data analyses.
The cooperative spirit of the helicopter pilots, D. Mitchell and J.
Gomez, was greatly appreciated. We thank 0. Bass, Jr. and H. Werner for
recommending improvements in our data collection techniques, and also 0.
Bass, Jr. for visiting sampling points to give his subjective evaluation
of habitat suitability. J. Moulding, D. Palmer, and R. Schroeder
reviewed an earlier draft of the manuscript. This project was funded by
the Jacksonville Endangered Species Office of the U.S. Fish and Wildlife
Service and the U.S. Army Corps of Engineers under Research Work Order
No. 28 with the Florida Cooperative Fish and Wildlife Research Unit,
University of Florida, Gainesville.
Literature Cited
Bass, O.L., Jr., and J.A. Kushlan. 1982. Status of the Cape Sable
Sparrow. .South Florida Research Center Report T-672. 41 pp.
Fretwell, S.D. 1972. Populations in a seasonal environment. Princeton
Univ. Press, Princeton, New Jersey.
Freund, R.J., and R.C. Littell. 1981. SAS for linear models. SAS
Institute Inc., Cary, North Carolina. 231 pp.
Green, R.H. 1971. A multivariate statistical approach to the
Hutchinsonian niche : bivalve molluscs of central Canada. Ecology
52:543-556.
Kushlan, J.A., and O.L. Bass, Jr. 1983. Habitat use and the
distribution of the Cape Sable Sparrow. Pages 139-146 in The
Seaside Sparrow, its biology and management. North Carolina Biol.
Surv. and North Carolina State Mus., Raleigh, North Carolina.
Kushlan, J.A., O.L. Bass, Jr., L.L. Loope, W.B. Robertson, Jr., P.C.
Rosendahl, and D.L. Taylor. 1982. Cape Sable Sparrow management
plan. South Florida Research Center Report M-660. 37 pp.
Ray, A.A., ed. 1982. SAS user's guide: statistics. SAS Institute
Inc., Cary, North Carolina. 584 pp.
Schroeder, R., and M. Armbruster. 1985. A habitat model for the Cape
Sable Seaside Sparrow. Review Draft (May). U.S. Fish and Wildl.
Serv., Fort Collins, Colorado. 13 pp.
Stimson, L.A. 1956. The Cape Sable Seaside Sparrow: its former and
present distribution. Auk 73:489-502.
Taylor, D.L. 1983. Fire management and the Cape Sable Sparrow. Pages
147-152 in The Seaside Sparrow, its biology and management. North
Carolina Biol. Surv. and North Carolina State Mus., Raleigh, North
Carolina.
Werner, H.W. 1971. Cape Sable Sparrows rediscovered on Cape Sable.
Auk 88:432.
Werner, H.W. 1975. The biology of the Cape Sable Sparrow. Everglades
National Park, Homestead, Florida. 215 pp.
Table 1. Variables derived from soil and vegetation measurements and
calculated for each sampling point.
Variable Definition (units)
COVER Mean % cover of herbaceous vegetation
COVERCV Coefficient of variation of COVER (%)
COVFREQ Percent of frames with COVER > 75%
CLUMPNO Mean number of clumps of bunch-grass-type growth
CLNOCV Coefficient of variation of CLUMPNO (%)
CLUMPCOV Mean % of total cover in bunch-grass-type growth
form
CLCOVCV Coefficient of variation of CLUMPCOV (%)
CLFREQ Percent of frames with CLUMPCOV > 50%
SHRUBDEN Shrub density (shrubs/ha)
TREEMEAN Mean of distances to closest tree or tree island
in each quadrant (m)
TREEMIN Distance to closest tree or tree island (m)
TREEDENS 10,000/TREEMEAN2 (tree islands/ha)
SOIL Mean soil depth (cm)
HSI -Habitat suitability index value calculated from
Schroeder and Armbruster (1985) (index)
SHSI Subjective valuation of habitat quality for Cape
Sable Seaside Sparrows (index)
Table 2. Correlations of habitat variables with numbers of Cape Sable
Seaside Sparrows detected per point for all points.
Variable r n P
COVER
COVERCV
COVFREQ
CLUMPNO
CLNOCV
CLUMPCOV
CLCOVCV
CLFREQ
SHRUBDEN
TREEMEAN
TREEMIN
TREEDENS
SOIL
HSI
SHSI
0.001
-0.011
-0.107
0.174
-0.012
0.137
-0.060
0.101
-0.080
0.118
0.166
-0.131
-0.097
0.104
0.471
120
120
120
120
79
120
79
120
121
123
124
123
111
124
49
0.991
0.905
0.243
0.057
0.918
0.135
0.599
0.273
0.385
0.194
0.066
0.148
0.309
0.248
0.001
Table 3. Correlations of habitat variables with numbers of Cape Sable
Seaside Sparrows per point for points where sparrows were detected.
Variable r n P
COVER
COVERCV
COVFREQ
CLUMPNO
CLNOCV
CLUMPCOV
CLCOVCV
CLFREQ
SHRUBDEN
TREEMEAN
TREEMIN
TREEDENS
SOIL
HSI
SHSI
0.223
-0.242
0.046
0.486
-0.314
0.449
-0.335
0.396
0.070
-0.034
-0.061
0.116
-0.133
0.101
0.306
0.205
0.167
0.797
0.004
0.103
0.008
0.081
0.020
0.700
0.848
0.734
0.515
0.462
0.570
0.268
Table 4.
with Cape
Results of t-tests comparing habitat variables between points
Sable Seaside Sparrows and points without sparrows.
Variable t n P
COVER
COVERCV
COVFREQ
CLUMPNO
CLNOCV
CLUMPCOV
CLCOVCV
CLFREQ
SHRUBDEN
TREEMEAN
TREEMIN
TREEDENS
SOIL
HSI
SHSI
0.98
-1.05
2.06
-0.48
-1.31
-0.34
-0.85
-0.13
1.71
-1.91
-2.86
3.25
1.33
-1.38
-4.06
120
120
120
120
79
120
79
120
121
123
124
123
111
124
49
0.331
0.294
0.042
0.634
0.195
0.738
0.399
0.901
0.091
0.059
0.005
0.002
0.187
0.172
0.001
Fig. 1. Locations of points sampled and points where Cape Sable Seaside
Sparrows were located (closed circles) on Grossman Hammock Quadrangle
Map., April-May, 1985. Total number of sparrows detected at each point
as well as points that were revisited (R) for sparrow censusing are
indicated. Points where sparrows were detected during vegetation
sampling (V) and number of additional sparrows added to the total (in
parentheses) also are indicated.
czY.
0.
VO) e3
S "' ; --
o *
A ,1 ,/* ", .-
-- -
--".- -r 7!: -- 4 -T-- -------.
A
-L
O -
"P~
0 0 0
0 0 0
o0
O O
I 0 0 0
Pip
0
E V-' G
O f
1*1
0
*/*
L A D
V(O)
E
0 .
L P A
O
2
R
o0
A R
070
*7 0 0
I
I 0
F'
II
HI
3.j..
*1 I
I
0 0
R
o O
Fig. 2. Locations of points sampled and points where Cape Sable Seaside
Sparrows were located (closed circles) on Royal Palm Ranger Station SE
Quadrangle Map, April-May, 1985. Total number of sparrows detected at
each point as well as points that were revisited (R) for sparrow
censusing are indicated. Points where sparrows were detected during
vegetation sampling (V) and number of additional sparrows added to the
total (in parentheses) also are indicated.
." " :,,. -
% 'A I A D IS I KIS V "
.1 1
.- *^ I'.
: -., ? *' t 9 ,lt, ..
-, "- .. .
0 ,I .r
*-'
$r *' 59
(S
* ,, **I
4, r
.4 '^ If'
- A
-4
o *- A '
,o ii '-
^*y *:
'- ' 7.. '
P 0 :-,.
9
6 9
I. "
*: ,+ _
. '. o ..o. .:
* *
o*,'' l ;." ; .**- .' ,.*
'; ^ .:* ,,* *. "-
..
9'''"
. V** .., ,. .* .-.
*i.; *- .r ^ -. ,
' */ -* .r *".-' -/
* ** :,*- ^ /... *. ,*'*..
"* "I 'i :*
S, .. .6 ; 0*
I -",-,b ,x0
.1. I~~~l e
..
.3 ~ '- A A'i*
A .'. ..
...
'** ** ^ ." 1 .V
-'"-p.
A k"
.'M.
o <' :'1
;Y ,". -1?
; -.-> i-.n rS.Vw ?.-^.r *
Ik
1i
1022)0 .6
.9 t -t. ..: '
0. I-~:o :d~
~ p
-* .r. .
S' .- *. '
"- ,' '' "* *
,, 2 f. _
*01,. *5 os ^1^. /
-,' ., ; .,"P ..:
-^ ^^.** ". j^ ^ 5:.
Fig. 3. Locations where sparrows were found and numbers detected during
the 1981 survey on the Grossman Hammock Quadrangle Map. Data are from
Bass and Kushlan (1982).
h; -" j, M .'',
, .,,., d.s
A Al 1
j t ,0 .,f l -I
'."** .*"
; -,-, .ia.-.w, .: '.-
Ar '
. .
. .," + -.
I4. -' .,.: - .
* r r ,.
. t .. ... .
* j' t-" ,. r- .Z:'" '
. .' '.. .: -.-'i
IAi A
' *.-',r
.,, ;+.,- :.;..-. .
.t *^'- f"
. o .. .
. '; : '. *' '" '' "- -" : "/' '
.. '
u *" *' *
---,'. -a -*-. rL>,,
', : ; !, .... .
~C o r0
I,: S '
At ~
,' r.,
60 30
50-
40-
12
00 -
HSI (midpoints)
20- 75
10 A
0o- 1 9 1
0.05 0.15 0.25 0.35 0.45 0.55 0.65
HSI midpointss)
Fig. 5. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by HSI values, and number of points in each grouping.
NJ'3 3.U
.T. ;
,'. 1 N "- -'. "'' -
.. .; ; _: ..
-:-: :
S- *flj _
1." 1 0"'- 0"
-: -,, f .. -
* ; --'. .-- :. o o
" :'" -. *,- 0 0
1 .
3 -'
-, A L ,-
,'' -'-.. .*. 0 0 0 "
'O O O*, O .-
O -.:O O
-.--.". :(0 o ,0 0- 0 I -.
_-. : I
-" '0 .
--. 0 0. A 0- I3 A
I..
I1 -
- -
b~n
2..3..
II I
til
i I .
~aa., ,
.3 .
Fig. 4. Locations where sparrows were found and numbers detected during
the 1981 survey on the Royal Palm Ranger Station SE Quadrangle Map.
Data are from Bass and Kushlan (1982).
wu -- -- --- -.-
80
4
70-
5 5
60-
\- 4 2
450
a. 40-
30 a
20- 6
10
0 a 7
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.55
SHSI midpointt)
Fig. 6. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by SHSI values, and number of points in each grouping.
50- 29
40- 21
C
20
10
16
4 4
0- ,. l l 0/ 1l
5 15 25 35 45 5 65 75 85
COVER midpointss)
Fig. 7. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by COVER values, and number of points in each grouping.
50-
40 -
C
*
i 30-
20-
10-
0-
. .7. i
Ir7 IZI VZ IIIV
wi r /j r Z, rz/lIfJ
/ /,, /.. .../" / / /
^ ^ ^ ^^ ^ / y
2 7 MH 0 ^
15 25 35 45 55 S 75 85 95
COVElCV midpointt)
Fig. 8. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by COVERCV values, and number of points in each grouping.
105 115
6- I
40
35- 74
30 14
25
19 5
o 20
15
10
5
/ / 4 1 2 / 1
I I
5 15 25 35 45 55 65 75
COVFREQ midpointss)
Fig. 9. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by COVFREQ values, and number of points in each grouping.
24
ii
11
I/Z
tPzo
40-
30
20-
10-
0
2 1
0.75 1.25 1.75 2.25 2.75
3.25 3.75
CLUMPNO midpointss)
Fig. 10. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by CLUMPNO values, and number of points in each grouping.
50
a'
I-- A & -I .-I._ I -a I -I I A
0.25
0.25
711-
18
(9
40 -
50 -
9
140 -
.30 ,
20 4
10
2 VA
25 75 125 175 225 275 325 375 425 47S
CLNOCV (mrdpoints)
Fig. 11. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by CLNOCV values, and number of points in each
grouping.
69
Vo
00n004
00
S
- ....I... -- r ---~ r~
CLUMPCOV midpointt)
Fig. 12. Percentage of .joints at which Cape Sable Seaside Sparrows were
detected, grouped by-CLUMPCOV values, and number of points in each
grouping.
'I
50 -
40-
C
30-
20-
10 -
0-
E L
- 1 -
23
70
s0 1
8 8
50-
5
. 40-
// 3
-0 1.
25 75 125 175 225 275 325 375 425 475
CLCOVCV midpointss)
Fig. 13. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by CLCOVCV values, and number of points in each
grouping.
16
Is
60 -
14
.. 40-
30-
75
20* 6
10 -
3 5 0 0 1
Fig. 14. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by CLFREQ values, and number of points in each
grouping.
40-
3"-
1 22
30
25
S 20 22
a.S
15-
10 10
5 15 25 345 04
SHRUBOEN midpointss)
Fig. 15. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by SHRUBDEN values, and number of points in each
grouping.
110
100-
90 -
80 -
7
7 0
40- 20
20 -
10 VJ
V IV -
4 10 204
25 75 123 175 225 275 325 375 425 475 525 575 >600
TREEMEAN midpointss)
Fig. 16. Percentage of points at which Cape Sable Seaside Sparrows were
detected, grouped by TREEMEAN values, and number of points in each
grouping.
___ ~
39
31
PA M-
29
If
0 0
-. ________________ z
I I I I ~ 1~~r
I I
12.5 37.5 75 125 175 225 275
TREEMIN midpointss)
____ m
325 375 >400
Fig. 17. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by TREEMIN values, and number of points in each
grouping.
110
100-
90-
80so
70-
80-
50-
40
30-
20 -
10 -
1
40
30
C
S31
20
9
10
015
0.25 0.75 1.2 1.7S >2.0
TREEDENS midpointss)
Fig. 18. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped-by TREEDENS values, and number of points in
each grouping.
*1
50-
40-
1 30-
20-
10-
0
37
16
14
16
I
9
//00
5 2S I 4 I i
5 15 25 35 45 SS >'0
SOIL midpointt)
Fig. 19. Percentage of points at which Cape Sable Seaside Sparrows
were detected, grouped by SOIL values, and number of points in each
grouping.
4 _
7
I:
I4
'2
2
25. 7 125 175 225 >250
Distance (midpoints)
Fig. 20. Distance from sample points versus number of sparrows detected.
Only sparrows for which accurate distance could be estimated are included.
|
RSS
HIDE
