Title: Impacts of citrus development on habitats of Southwest Florida
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00066447/00001
 Material Information
Title: Impacts of citrus development on habitats of Southwest Florida
Physical Description: Book
Creator: Pearlstine, Leonard J.
Brandt, Laura A.
Kitchens, Wiley M.
Mazzotti, Frank J,
Affiliation: University of Florida -- Florida Cooperative Fish and Wildlife Research Unit -- Department of Wildlife Ecology and Conservation
University of Florida -- Florida Cooperative Fish and Wildlife Research Unit -- Department of Wildlife Ecology and Conservation
University of Florida -- Florida Cooperative Fish and Wildlife Research Unit -- Department of Wildlife Ecology and Conservation
University of Florida -- Department of Wildlife Ecology and Conservation
Publisher: Society for Conservation Biology
Publication Date: 1995
 Notes
General Note: Drawn from Conservation Biology, Vol. 9, No. 5, pp. 1020-1032, 1995
 Record Information
Bibliographic ID: UF00066447
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.

Downloads

This item has the following downloads:

ImpactsCitrus_ConsBio_1995 ( PDF )


Full Text










impacts of Citrus Development on Habitats


of Southwest Florida


LEONARD G. PEARLSTINE,* LAURA A. i \, ;. i T,* WILEY M. KITP i ".'-:.-*
AND FRANK J. MAZZOTTIt
*National T ...;.. .i Service, Florida Cooperative Fish and Wildlife Research Unit and Department of Wildlife
Ecology and Conservation, University of Florida, Gainesville, FL 32611, U.S.A.
tDepartment of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL 32611, U.S.A.





Abstract: ." --.--", '":! resource :..... ...: .- needs at the landscape level has become critical for the conser-
vation of ecosystems and the preservation of species. Geographic information systems (GIS) that allow for the
, ..;.'...'.. of .... .". referenced databases are a powerful tool that can be used by resource managers to ex-
amine potential impacts and develop strategies for regional planning. We applied a landscape-level approach
to examine the potential impacts of citrus development on habitats and species in southwest Florida, We de-
veloped ;IS models for panthers, Sandhill Cranes, and ... '. birds that reflect changes in .- ....-. -' habitats
under a series of development scenarios. The models indicate that, under the maximum development sce-
nario, .- of potentialpanther habitat, 66% ofpotential Sandhill Crane habitat, and 67% and 33% ofpoten-
tial wading bird nesting and foraging habitats could be lost. In addition, the habitat that would remain
would be severely fragmented. Several key areas were identified that will be critical to the continued existence
of these species and to maintenance of regional : .. :. .,. The areas identified are habitats not represented
on the existing public lands concentrated in the southern portion of the study area and/or that provide con-
nections among existing natural areas.

Impacto dcl desarrollo de citricos en habitats del surocste de Florida
Resumen: El exdmen de las necesidades de manejo de los recursos a nivel del paisajese ha transformado en
un element critic para la conservaci6n de los ecosistemas y la preservacion de las species. Los Sistemas de
Informacion Geograficos (SIG), que; : .... .: la integraci6n de las base de datos registrados espacialmente,
I .: .. .una poderosa herramienta que puede ser utilizada por los administradores de recursos natu-
ralespara examiner los impacts potenciales y desarrollar estrategias de : .... : :. a nivel regional. Apli-
camos un enfoque a nivel de paisaje para examiner los impacts potenciales del desarrollo de los citricos en
species y ambientes del sudoeste de la Florida. Desarrollamos models en SIG para panteras, grullas y aves
zancudas que :7..... cambios en los habitats potenciales bajo una series de escenarios de desarrollo, Los
models indican que bajo un scenario de maximo desarrollo 63% del habitat potential de la pantera, 66%
del habitat potential de la grullay 67y 33% del habitat potencialpara anidamiento y .-. - .. ., ,. de las aves
zancudas se podrian perder. En forma adicional, el habitat que quedara estaria severamente fragmentado.
Se identificaron varias areas claves que van a ser critics para la supervivencia de estas species y el manten-
imiento de la biodiversidad a nivel regional. Las areas identificadas, constituyen habitats que no estan repre-
sentados dentro de las tierras publicas existentes, y que se encuentran concentradas en la porcion sur del
area de studio y/o proven conecciones entire dreas naturales existentes.









Paper submitted May 12, 1994; revised manuscript accepted February 16, 1995.
1020


Conrserviior Biology, Pages 1020-1032
Volume,9 No. 5, October 19)5






Impacts of Citrus Development on Florida Habitats 1021


Introduction

Protecting biological diversity will require regional and
comprehensive efforts rather than the local, site-specific
approach that has been commonly applied. Geographic
information systems (GIS) provide a tool that, combined
with other knowledge, can greatly enhance planning for
resource protection. GIS can integrate and model spa-
tially referenced data, land-cover information, future
land use scenarios, and species information in a manner
that can be easily viewed and used by resource profes-
sionals responsible for habitat and species protection.
The ability to analyze potential impacts so that actions
(land acquisition, development of plans for environmen-
tally sensitive site design, requests for more stringent-
permitting requirements, planning of restoration areas)
can be taken quickly is invaluable to the protection of all
natural areas, particularly in areas of rapid land-use
change and for the protection of regional biodiversity
such as the Everglades regional ecosystem. The ecologi-
cal integrity of the regional system is threatened by a va-
riety of factors related to urban and agricultural develop-
ment. In southwest Florida, citrus acreage has doubled
to 60,000 ha since 1980 and is projected to reach 80,000
ha by the year 2000. These developments could affect
the habitat for a diverse native flora and fauna, including
31 species listed by state and federal agencies as endan-
gered, threatened, or of special concern. We illustrate
one way that GIS models can be combined with devel-
opment scenarios to identify areas of critical habitat and
areas that can provide connections among habitats.



Methods

Study Site

The study area included 600,000 ha in southwest Florida
that encompass the Immokalee Rise (Fig. 1), an area of
predominately sandy soils (Drew & Schomer 1984) and
slightly elevated topography resulting in land suitable
for citrus production. The eastern portion was histori-
cally sawgrass marsh (Cladium jamaicense) and is now
sugar cane. The southern portion is predominately cy-
press (Taxodium distichum). The soils can be grouped
into three landscape positions: flatwoods, sloughs, and
depressions. Flatwoods are upland and support oaks
(Quercus spp.), southern slash pine (Pinus elliottii), and
saw palmetto (Serenoa repens). The sloughs occupy
transitional areas between the flatwoods and depres-
sions. Sloughs usually have an overland sheet flow of
slowly moving water through southern slash pine, wax
myrtle (Myrica cerifera), and water-tolerant grasses dur-
ing the wettest season. Depressions remain underwater
for six months or more of the year. Vegetation in depres-
sions varies from stands of pickerel weed (Pontederia


Figure 1. Location of the study area in southwest Flor-
ida and its physiographic regions (1, Caloosahatchee
Valley; 2, hnImmokalee Rise; 3, Southwestern Slope; 4,
Big Cypress Spur; 5, Everglades).



cordata) to cypress and hardwoods. The soils vary from
muck over limestone at lower elevations to continuous
layers of sand down to 200 cm at higher elevations.


Land Cover

Land-cover classification for the study area was devel-
oped from three sources of data: (1) Landsat Thematic
Mapper imagery, (2) SPOT panchromatic imagery, and
(3) wetland maps of the U.S. Fish and Wildlife Service
National Wetlands Inventory (NWI).
Landsat imagery was obtained for December 1989 and
rectified to the Universal Transverse Mercator zone 17
coordinate system by a first-order rectification to ground-
control points from U.S. Geological Survey 7.5-min topo-
graphic maps. Rectification was within one pixel (30 m)
of the true position on the ground (as determined from
the Geological Survey maps). Imagery from December
was used to classify land cover because of its availability,
but it created complications in classification. December
is a dry month, and many of the small, isolated wetlands
were difficult to separate from surrounding upland
grasslands. Furthermore, deciduous cypress does not
present its best spectral reflectance for separability at
this time of year. Using digital wetland maps from the
National Wetlands Inventory as masks during classifica-
tion improved spectral separability. We did not attempt


Conservation Biology
Volume 9, No, 5, October 1995


Pearlsline et al.






1022 Impacts of Citrus, .,'., . . on Florida Habitats


to subset the NWI classes directly into our classification.
Wetland boundaries delineated by the NWI were used to
mask defined areas for i :1:: ,:.... emergents, palustrine
forested, palustrine shrub-scrub, aquatic beds (sub-
merged aquatics), and .:r:' A.: 1. An .:: *:- ....- : classifi-
cation was performed .-.: .-. 1.. for each of these
defined areas using the ISODATA program in ERDAS im-
age :-.. : .. .. - : ... information systems soft-
ware (ERDAS 1991). The classification allowed us to
look for spectral delineation of floral communities
within the broader, predetermined types. Broad NWI
class boundaries were used to reduce the - :..:! vari-
ance within an unsupervised classification, reducing the
number of communities we had to separate at a time. Er-
rors or ...... ::: -.. in the NWI boundaries appear as
separate unsupervised classes and were identified dur-
ing labeling of the classification.
Because we were .. .: ... roughly within a type,
less than 20 classes were typically delineated. Classes
that contained only a small percentage of the total area
and/or classes that were spectrally close to other classes
were merged with larger classes based on their spectral
to the larger class. Maps of the classifications
were taken to the field for identification of classes. Ex-
tensive field truthing from the ground and from fixed-
wing aircraft and helicopter surveys were conducted for
this identification phase. During these surveys, a loran-C
was often used to fix our approximate location, but our
precise location was determined from landmarks and
matching spatial land-cover patterns. When large, homo-
geneous areas of the same class were present, they were
used for field truthing. This procedure of classifying un-
der NWI wetland boundary masks allowed us to use our
a priori knowledge of locations of land-cover types to re-
duce overall variance and confusion between types,
which significantly improved the final classification.
Boundaries of farm lands (citrus, sugar cane, and vege-
table crops) were determined by screen digitization of
farm boundaries from photo interpretation of a .: 4.. .
SPOT panchromatic image with 10-m2 spatial resolution.
Aerial photographs and water-use permits provided aux-
iliary information. The farm boundary maps were field
verified, and preliminary maps were distributed to the
local farmers for review. The areas delineated as farm-
lands were merged into the final land-cover classifica-
tion.
Lands in public or private ownership for the purpose
of conservation (Audubon Corkscrew Swamp Sanctu-
ary) were delineated as another thematic layer for the
i ... i.. of examining how much of the land modeled as
suitable for the selected wildlife species is already re-
served for conservation. Lands proposed for public ac-
quisition were also included as potential reserves.
Boundaries of holdings were obtained from Geological
Service 1:100,000 DLG files, the U.S. Fish and Wildlife
Service, the South Florida Water Management District,
and the Florida Department of Natural Resources.


Pearlstine et at


Citrus Production Feasibility Map

A feasibility map for citrus production was developed
based on land ownership and soil classification ( .
et al. 1992). The study area was divided into sections as
defined by the Tc .. -i. '' ; -Section method of land
: :.: : :: (U.S. Public Land Survey), and the final model
classified each square-mile section within the study area
into one of four feasibility i.:. excellent, good, fair,
or poor. The Public Land Survey section is the unit most
often used for the purchase and development of lands
for citrus production.
Three scenarios of land cover were projected (1) by
changing all the lands deemed excellent feasibility into
citrus i: ..:::. : .:: (DEV1), (2) ,: ;.:.:,..: all the lands
with excellent or good '. ..i .i..: into citrus production
(DEV2), and (3) in addition to the changes made in
DEV2, by changing all the land with fair feasibility into
production of vegetable crops :: -. '.). Although these
areas are marginal for citrus production, they are good
candidates for vegetables such as tomatoes. Current per-
mitting of new citrus groves, but not vegetable crop pro-
duction, requires the larger wetlands to be protected by
a wet detention system. We retained wetland-emergent
classes larger than 500 ha within lands that were mod-
eled to become citrus production.


Species Modeling

Three hundred sixty-two native vertebrates occur in
southwest Florida. Forested upland, freshwater marsh,
cypress forest, hardwood *. i' I ...,..= and pine flat-
woods are the most important habitats for a large num-
ber of species (Mazzotti et al. 1992). Species used for
modeling the effects of citrus development in these hab-
itats were chosen based on their sensitivity to change in
a variety of different habitats and their presumed utility
as ecological indicators.
Panthers .'..., concolor . w: ^: were chosen because
they depend on a varied landscape within their home
range. Extensive areas of forested uplands, hammocks,
hardwood swamp, cypress swamp, and pine flatwoods
provide habitat for more than 180 species, including
other listed and declining species such as Red-cockaded
''" '" .i. i"* (Picoides borealis), Eastern Bluebirds
,* ....'...: sialis), Eastern indigo snakes ( .' *... :..' ..* co-
rais), and black bears (Ursus americanus .:.....
Sandhill Crane (Grus canadensispratensis) is an indica-
tor species for upland-wetland linkages because it uses
temporary wetlands for nesting and forages in both up-
lands and wetlands. At least 150 additional species use
temporary i'. '.--' or the temporary pond-upland inter-
face in southwest Florida, including rice rats (*'......
palustris), round-tailed muskrats (':. ..'r... alleni), and
King Rails, (Rallus elegans) (Mazzotti et al. 1992). Wad-
ing birds (Great Egret, Casmerodius albus; Snowy


Conservation Biology
Volume 9, No. 5, October 1995






Pearlstine et al.


Egret, Egretta tbula; and Great Blue Heron, Ardea
herodias) were chosen because of their dependence on
hydrological conditions in their wetland habitats and the
importance of southwest Florida to the overall south
Florida populations of these species (Browder 1976). As
with Sandhill Cranes, over 150 species use temporary
wetlands. Over 170 species, in addition to the wading
birds, use cypress forests including river otters (Lutra
canadensis), greater (Siren lacertino) and lesser sirens
(Siren ..*. ., -....,... : Great Crested Flycatchers : .,.,:
chus crinitus), and Swallow-tailed Kites (Elanoides for-
ficatus) (:'- ;. ::: et al. 1992).
For the three species models (for Sandhill Crane, Flor-
ida Panther, and wading birds), land cover and land-
scape patterns of land cover were used to predict poten-
tial habitat. Habitat then was projected for three
scenarios of increasing agricultural development. The
three species models were created using ERDAS's file de-
scriptor (DSCEDIT) and modeling capabilities (DSCEDIT
and C-' .-':- Resources used to develop the models' pa-
rameters included literature, field sampling to relate the
literature to the :*-. ::-. study area (Depkin et al. 1994;
Mazzotti et al. 1992), and personal communication with
regional experts. Because the three modeled species are
indicators of habitat for a larger set of species, overlay-
ing the maps of potential habitat for each of the three
species together produces a spatial indication of the
habitat potential for a wide range of species. The models
were repeated for each scenario from present (1989)
conditions through DEV3.
The following information was used to define parame-
ters for Sandhill Crane foraging and nesting.

(1) Sandhill Crane are opportunistically omnivorous
and feed on vegetation, invertebrates, and small
vertebrates such as frogs and snakes (Walkinshaw
1976; Mullins & Bizeau 1978). Cranes forage in
wetlands and along edges of uplands but are rarely
found far inside a forest patch. They also forage in
pasture (Bennett 1978; Armbruster 1987).

(2) Water depth during nest initiation averages 33 cm
and ranges between 0 cm and 65 cm (C. i:. ::: :i ..
1982; Dwyer 1990). Wetlands may dry out during
incubation, but nests are rarely initiated in dry
ponds (Bennett & Bennett 1987).

(3) Estimates of the area of suitable habitat that a pair
of cranes needs around a nest site were based on a
minimum home-range size of 240 ha and an aver-
age home-range size of 500 ha : i. -. i .: & Williams
1979; Bishop 1988; Nesbitt, personal communica-
tion).

Potential foraging sites were extracted from the land-
cover map. All of the areas classified as marsh or pasture-


Impacts of Citrus .-:-.. ... on Florida Habitats 1023


Table 1. Use of water regimes by Sandhill Crane.


Water.':: :' Normal Wet Dry Area (ha)
Permanently Flooded 13,649
Intermittently : : X 511
Semipermanently flooded X X X 17,053
. ... Flooded X X X 155,232
Saturated X 0
Temporarily Flooded X X 28,348
Intermittently Flooded 0
Artificially Flooded -14
"X = may be used for nesting; = not used *.. .
bNational Wetland Inventory water regime modifiers are defined in
Crowardin et at. (1979).
cArea of each type within the study area.



range with a combined contiguous area greater than 240
ha were selected. Areas of selected land cover that were
less than 60 meters from each other were considered
contiguous. Contiguous areas that contained only pas-
ture/range and no marsh were excluded.
Because the Sandhill Crane forages along forest edges
adjacent to typical foraging habitat, upland hammocks,
hydric hammocks, and palmetto-pine-palm communities
immediately adjacent to pasture-range or marsh areas
were selected as potential habitat. Combined, the se-
lected :-.- .: .:: -range, marsh, and adjacent forests repre-
sent potential Sandhill Crane land cover. Potential nest-
ing sites are a subset of the potential land cover selected
above. Appropriate marsh hydroperiods were used to
select nesting habitat. National Wetland Inventory Water
Regime Modifiers were used to identify areas where
proper water depths were likely to occur during the ini-
tiation of nesting (Table 1). Nest sites were selected as
those marsh sites contained within foraging sites that
also had one of the selected water regimes.
Potential Sandhill Crane habitat was modeled as the
suitable land cover within the distance the cranes would
be willing to travel from their nest sites in order to for-
age. The distance used to extract this final habitat area
was 1260 meters, or the radius of the mean home range
for local Sandhill Crane, 500 ha.
The following information was used to define parame-
ters for potential panther habitat. The model displays
the relative suitability of areas as habitat based on the
model parameters and does not address issues relating
to population density.

(1) Data on panther habitat use were obtained from
the Florida Game and Fresh Water Fish Commis-
sion for 10 females and 17 males (8746 locations)
collared in southwest Florida from 2 April 1981 to
14 October 1991. Tracking periods per animal
ranged from 2 to 70 months (see Maehr et al. 1989,
McCown et al. 1990).


Conservation Biology
Volume 9, No. 5, October 1995






1024 Impacts of Citrus Development on Florida Habitats


(2) A composite home range for the tracking period
was calculated for each panther with the mini-
mum convex polygon method using McPAAL com-
puter software (Stuwe & Blohowiak 1989). Aver-
age areas used by males and females were
calculated. Average composite home range size for
males was 679 km2, with a radius of 14.7 km, as-
suming a circular shape. Average composite home
range size for females was 341 km2. The radius of a
circular home range of this size was 10.4 km. The
distance of the home-range radius was used in the
model as a reasonable distance over which a pan-
ther might be expected to travel over a period of
years to find suitable habitat.

(3) Maehr et al. (1991) found that adult panthers pre-
ferred hardwood hammock, pine flatwood, cy-
press swamp, and cabbage palm woodlands. In a
generally wetter study area, Belden et al. (1988) of-
ten found panthers using mixed swamp-forest and
hydric hammock. Spearman correlation coeffi-
cients of cover types from our land-cover map ex-
tracted from under the home-range polygons con-
firmed (p < 0.0001) that forested areas hydricc
hammock, pine-cypress, cypress, pine with nar-
row-leaf emergents, and upland pine) are the most
important cover types within a panther's home
range in our study area.


Areas of potential panther habitat were extracted
from the land-cover map developed from Landsat imag-
ery using the following model: Panther habitat value =
density value of habitat + continuity value of habitat.
Density value was determined by systematically sam-
pling at 1610-meter (1-mile) intervals to determine how
much forested area was within the area of a circular win-
dow that was centered on each cell. The window repre-
sented the maximum distance a panther would typically
travel: 14.7 km for males, 10.4 km for females. The cells
then were ranked and recorded with a value of 0-5 based
on the percentage of desirable cover.
Continuity value was determined by locating all areas
of contiguous desirable cover types in clumps of at least
500 ha (500 ha was arbitrarily chosen as a threshold for
forest fragmentation). The percentage of each 1610-m2
cell that was part of a contiguous clump was deter-
mined. If the cell had a value less than 33%, it was given
a continuity value of 0. If the cell value was 33% or more,
it was assigned a continuity value of 1. The resulting
habitat value is a six-category ranking using habitat con-
tinuity as an additive weighing factor on habitat density,
in which 1 is acceptable habitat and 6 is excellent habi-
tat. Areas with a ranking of zero are unacceptable habi-
tat. Continuity was not used as a weighting factor when
habitat density fell below 15%. Density is so sparse be-


Pearlstine et al.


low 15% that habitat continuity does not add value to
the area.
The following information was used to define parame-
ters describing habitat suitable for wading-bird foraging
and nesting:

(1) Wading birds use shallow (< 0.5 m) open-water
wetlands for foraging (Hoffman 1978). The loca-
tion and foraging success in a wetland depend on
present and past water conditions, which influ-
ence the distribution, demographics, and availabil-
ity of prey species (Kushlan 1976, 1978; Powell
1987). Suitable wading-bird foraging habitat was
determined to be those areas where birds were ob-
served foraging more than 5% of the time in our
field studies, and pasture characterized by tempo-
rary, seasonal, or semipermanent water regimes
defined using the NWI water-regime modifiers.

(2) Wading birds nest in vegetation ranging from small
shrubs to tall trees in riparian forests, swamps, and


High feasibility
Good feasibility
D Fair feasibility
S Poor feasibility

Figure 2. Feasibility of land for conversion to citrus
production in the study area, based on soil character-
istics and land ownership.


Conwn'a7uaiion Biolog
volume 9, No 5, Otohber 1995






Pearstine et at


Conservation Biology
Volume 9, No. 5, October 1995


Impacts of Citrus Development on Florida Habitats 1025


Barren or Roads
Pasture / Range
* Upland Hardwood Hammock
Upland Pine
ig' Pine / Palmetto / Palm
Upland Shrub / Scrub
Hydric Hammock
SPine / Narrow-leaf emergent under
* Pine / Saw Palmetto understory
Palmetto / Pine
Pine / Cypress
Cypress
* Shrub Cypress
Cabbage Palm / some pine & Cypress
* Wet Shrub / Scrub (willow.corkwood)
Sparse Narrow-leaf Emergents
.. Narrow-leaf Emergents
Broad-leaf Emergents
Mixed Emergents
Open Water
Submerged Aquatics
SMelaleuca
Truck Crops
Sugar Cane
SCitrus

1989































DEV 2






1026 Impacts of Citrus Development on Florida Habitats


Table 2. Classification scheme and hectares of types of land covers
in the Immokalee Rise, Florida 1989.


Classification
Barren or Roads
Pasture and Range
Upland Hardwood Hammock
Upland Pine
Pine-Palmetto-Palm
Upland Shrub-Scrub
Hydric Hammock
Pine-Narrow-Leaf Emergent
Understory
Pine-Saw Palmetto Understory
Palmetto-Pine
Pine-Cypress
Cypress
Shrub Cypress
Cabbage Palm with some Pine
and Cypress
Wet-Shrub-Scrub (willow,
corkwood)
Sparse Narrow-Leaf Emergents
Narrow-Leaf Emergents
Broad-Leaf Emergents
Mixed Emergents
Open Water
Submerged Aquatics
Melaleuca
Truck Crops
Sugar Cane
Citrus
Total


Hectares
30,058
142,938
4930
71,582
4915
15,712
8751

15,642
2250
563
2061
72,088
19,927

536

10,158
3801
28,371
3750
28,767
10,287
615
307
29,354
51,354
62,777
621,495


Total Area (%)
4.8
23.0
0.8
11.5
0.8
2.5
1.4

2.5
0.4
0.1
0.3
11.6
3.2

0.1

1.6
0.6
4.6
0.6
4.6
1.7
0.1
0.1
4.7
8.3
10.1


on islands (Chapman & Howard 1984). In Florida,
295 colonies located during aerial surveys from
1976 to 1978 were in woody or herbaceous vege-
tation over standing water or on islands sur-
rounded by open water (Nesbitt et al. 1982)

(3) Bancroft et al. (1990) found that in south Florida
foraging distance varied among years, colonies,
and species and ranged from about 5 to 11 km. For
the model, 8 km was selected as the distance wad-
ing birds would range from their nest site to locate
suitable foraging sites during the nesting season.

The habitat model developed for wading birds locates
potential foraging habitat and ranks the suitability of
nesting habitat based on (1) the presence of forest or
shrub cover in proximity to wet land covers and (2) the
density of foraging sites within 8 km of the nest site.
Suitable land covers were extracted from the land-cover
map classified from Landsat imagery.
Foraging habitat was extracted by recoding the land-
cover map for cypress, marsh (excluding broadleaf


Pearlstine et al.


marsh), and wet shrub-scrub. Areas of pasture from the
land-cover map were included that were also temporary,
seasonal, or semipermanently flooded in the hydrology
map layer.
Classes in which wading birds will nest are the pine
classes, upland shrub-scrub, hydric hammock, cypress,
and wet shrub-scrub. Potential nesting habitat within
the study area was selected as sites containing a nesting
class within one pixel (30 m) of a wet class (cypress,
wet shrub-scrub, marsh, open water, or submerged
aquatics).
The quality of the nesting sites was determined by the
density of foraging sites surrounding the nesting sites.
Sample points were selected at 1-km intervals to deter-
mine the density of foraging classes within an 8-km ra-
dius of each sample point. The density value at each
sample point was then gridded into a 1-by-l-km cell cen-
tered around each sample point. The model ranked each
nest site based on the foraging density of the grid cell
that the nest site fell into. Six ranks were assigned from
"not suitable-0% foraging density" to "nesting with
81% to 100% foraging density" in class intervals of 20%.



Results

The land cover classification is delineated in Table 2. Fig-
ure 2 illustrates the results of citrus feasibility modeling.
Most of the land with a high feasibility of conversion to
citrus is currently pasture or range. Much of the pasture
and range were historically pineland, but only approxi-
mately 12% of the historic (ca. 1940) pinelands in the
study area remain (Mazzotti et al. 1992). Uplands are
most vulnerable to citrus conversion because citrus re-
quires well-drained soils. In southwest Florida many
small, isolated wetlands are also vulnerable because of
their landscape position within the upland land-cover
types. The largest loss of natural areas would occur be-
tween DEV1 (convert all lands with excellent feasibility)
and DEV2 (convert all lands with excellent or good feasi-
bility) because most of the upland land cover has been
converted to citrus and surrounds much of the remain-
ing wetlands (see Figs. 3 and 5). Another consequence
of citrus development, in addition to loss of suitable land
cover, will be fragmentation of the remaining habitat
patches.
Eight percent of the study area (49,748 ha) is currently
in public or private ownership intended for natural re-
source conservation. These lands are predominately
south of the Immokalee Rise and are forested wetland.
Three percent is in public ownership with goals other


Conservation Biology
Volume 9, No. 5, October 1995


Figure 3. Land-cover classification for 1989 and development scenario 2 (DEV2), in which excellent and good
lands for citrus production were converted to citrus production.




Impacts of Citrus Development on Florida Habitats 1027


Pasture / grasses
Marsh
Forest edge


Least suitable


U
U
U
U


Most suitable


SNesting with
least forage density
" l


Nesting with
most forage density
Forage sites with
no nesting


Conservation Biology
Volume 9, No. 5, October 1995


1989


DEV 2


i


Pearlstine et al.






1028 Impacts of Citrus Development on Florida Habitats


Dev1989

Dev2
Dev3


Pasture Upland Cypress Wetland Crop Sugar Citrus Other
Cover type

than conservation (State Indian lands) but currently pro-
vide some protection. State and federal agencies have
proposed protection of an additional 8% of the study
area (51,132 ha).
The present conditions (1989) and DEV2 land cover
and habitats are shown as map products in Figs. 3 and 4.
The other scenarios are included as histograms of land-
cover change (Figs. 5-8). Because male and female pan-
thers were modeled separately, but show similar pat-
terns of habitat use and change, only the male results are
reported.
Under 1989 conditions, approximately 186,000 ha are
available in the study area as Sandhill Crane habitat (Fig.
4). About a third of this is marsh and therefore suitable
as nesting habitat. DEV1 results in a loss of 32,500 ha of
crane habitat (Fig. 6). The majority (28,900 ha) of this
loss is from the conversion of range and pasture to citrus
production. The greatest change in available habitat oc-
curs from DEV1 and DEV2, with another 68,600 ha lost,
again mostly from pasture and range (52,750). DEV3
causes a reduction of suitable crane habitat by an addi-
tional 21,900 ha.
Of the potential Sandhill Crane habitat, 6822 ha is lo-
cated on preserves and refuges in the study area, 3398
ha on proposed protected lands, and 3158 ha on Indian
land. Preserves and proposed preserves make up 3.7%
and 1.8% of the total crane habitat in the study area, re-
spectively.


Figure 5. Changes in the area of
selected cover types from 1989
through the alternative develop-
ment scenarios (DEV1, DEV2,
and DEV3). The scenarios repre-
sent increasing conversion of
land to citrus production based
on land suitability and land
ownership.


Under 1989 landscape conditions, approximately
375,000 ha are available in the study area as suitable pan-
ther habitat as described by our model (Fig. 4). About
half of this habitat is ranked as acceptable (1 or 2), 25%
as good but not prime habitat (3 or 4), and 25% as excel-
lent to prime habitat (5 or 6). Conversion of forest to
range and crops directly removes panther cover and
fragments the remaining forest. Combined, cover loss
and fragmentation result in consequential loss of pan-
ther habitat in the study area. Approximately 93,000 ha
of panther habitat would be lost to citrus conversion in
DEV1 (Fig. 7). With DEV1 and DEV2, another 110,000
ha would be lost, and about 32,000 ha would be lost
with DEV2 and DEV3.
Habitat quality, as reflected by the habitat rank, de-
clines as well. Development to DEV1 would mostly af-
fect the lower-ranking habitats. When DEV2 is modeled,
however, almost all of the highest quality habitat is re-
moved (Fig. 4).
Preserves and refuges in public ownership account for
approximately 18% of the potential panther habitat in
the study area, or 70,200 ha. Proposed protected lands
would add another 67,600 ha of habitat, or an additional
17% of the total habitat in the study area. Indian lands
contain 25,000 ha of habitat, or approximately 6.5% of
the total.
Approximately 169,895 ha (27% of the study area) are
currently available as wading-bird foraging habitat and


Conservation Biology
Volume 9, No. 5, October 1995


Figure 4. Modeled potential habitat for 1989 and development scenario 2 (DEV2): land-cover conditions for Sand-
hill Crane (a), Florida panther (b), and wading birds (c).


Pearlstine el al.






Impacts of Citrus Development on Florida Habitats 1029


Habitat
SRange and pasture
Marsh
Forest edge


1989 Dev1 Dev2 Dev3
Development

Figure 6. Area of Sandbill Crane habitat by cover
type, and potential changes from 1989 through the
alternative development scenarios (DEVI, DEV2, and
DEV3).


118,691 ha (20% of the study area) as nesting sites with
foraging habitat within an 8-km radius. Twenty-two per-
cent (26,656 ha) of the nesting areas have a density of
foraging habitat of 81%-100% within an 8-km radius
(Fig. 4). Under DEV1, 10,000 ha of foraging habitat
would be lost (Fig. 8).
Of the potential wading-bird habitat, 38,173 ha is lo-
cated on preserves and refuges in the study area, 40,496
ha on proposed protected lands, and 11,720 ha on In-


Habitat rank
1
2
C 3
45

6-


1989 Dev1 Dev2 Dev3
Development

Figure 7. Area of male Florida panther habitat by
habitat quality, and potential change from 1989
through the alternative development scenarios (DEV1,
DEV2, and DEV3). Habitat rankings range from 1,
acceptable, to 6, prime habitat.


1989 Devl Dev2 Dev3
Development

Figure 8. Area of wading-bird nesting habitat by habi-
tat quality and potential change from 1989 through
the alternative development scenarios (DEV1, DEV2,
and DEV3). Percentages are for foraging habitat
within a 8-km radius of the nesting habitat.

dian land. Preserves and proposed preserves make up
19% and 20% of the total wading bird habitat in the
study area, respectively.
When all three habitat models are overlain, approxi-
mately 7% (35,536 ha) of the mapped habitat supports
all three species. The majority of this habitat is in the
central wetlands core of the study area. Another 45%
(220,268 ha) of the habitat supports two of the modeled
species. Of the habitat that supports two species, 89%
includes the panther. Panther and wading birds are
found together (128,303 ha) in the mostly wetland for-
ests at the southern portion of the study area and in the
upland pine in the northern portion of the study area.
Panther and Sandhill Crane (68,165 ha) share habitat
where there is a close juxtaposition of forested and
marshland covers. The 23,800 ha of habitat for both
Sandhill Crane and wading birds is found mostly in the
rangelands to the east of the central core. Fourty-eight
percent of the modeled habitat in the study area is occu-
pied by only one of the guilds: panther, Sandhill Crane,
and wading bird habitat occur alone on 161,250, 58,832,
and 11,934 ha, respectively.

Discussion

These models provide a tool with which resource man-
agers can predict changes in potential habitat across a
landscape. They are based on good knowledge of the bi-
ology and natural history of the organisms and work
within the constraints of available geographic data. The
approach was used to project a variety of scenarios that
evaluate potential effects on habitat. These are not pro-
cess models for exploration of species dynamics. In-
stead, they operate at the level required to make re-


Conservation Biology
Volume 9, No. 5, October 1995


Pearlsline elt al.






1030 Impacts of Citrus Development on Florida Habitats


gional land-use planning decisions using generally
available information to construct species-habitat associ-
ations.
In the generated land-cover map, the central wetlands
core and Okaloacoochee Slough stand out as large, con-
tiguous, natural and low-intensity-use areas with poten-
tial for connectivity with existing public lands and the
broader region. Overlaying the spatial models confirms
the importance of this area for panthers, wading birds,
Sandhill Cranes, and other native species dependent on
contiguous forested areas and upland-wetland juxtaposi-
tion. This slough also provides a direct link between for-
ested habitats to the south and north, including the
lands now in conservation protection.
Maintaining the large contiguous areas of pine and
wetland forested land that Florida panthers depend on is
complicated by increasing pressures from crop and
range development (Maehr 1990). It is estimated that no
more than 50 panthers live in Florida. Many of these oc-
cur in southwest Florida on both public and private land
(Maehr 1990). Based on available, suitable habitat alone,
the potential panther habitat of southwest Florida could
decrease by 25% under DEV1 and by 60% under DEV3.
This decrease in the potential of southwest Florida to
support a breeding population of panthers is due to di-
rect habitat loss, habitat fragmentation, and an increase
in land covers not utilized by panthers or detrimental to
panthers, such as roads. Other species dependent on
large forested areas, especially pine flatwoods, would
experience similar threats to their habitats.
Much of the loss of foraging habitat for Sandhill Crane
is in pasture and range. Pasture and range also form the
matrix for small but numerous marshes suitable for
crane nesting and foraging. As range is converted to cit-
rus production, these small, isolated marshes will ac-
count for the majority of wetlands loss. Additional loss
of marsh habitat may result from citrus development ad-
jacent to wetlands, and lower the water table, signifi-
cantly changing the water regimes of the marshes.
Changes in water regimes could result in changes in the
community structure of amphibians that rely on tempo-
rary ponds (Moler & Franz 1987; Mazzotti et al. 1992)
and could provide habitat for nonnative species (Maz-
zotti et al. 1992).
The conversion of wetlands to other land uses and re-
sulting changes in hydrology in adjacent wetlands also
influences the distribution and abundance of wading-
bird foraging sites. This directly affects the availability of
suitable and prime nesting sites (those surrounded by a
high density of foraging sites) within the region. DEV2
and DEV3 would result in less loss of habitat than DEV1
but would continue to fragment the remaining habitat,
resulting in a loss of nesting sites in proximity to high-
density foraging habitat. In addition to directly affecting
wading birds that nest in this area, conversion of marsh
could influence overall South Florida wading-bird popu-


lations by eliminating regionally important foraging habi-
tat (Browder 1976).
Eight percent of the study area is protected. These
lands are concentrated in the southern portion of the
study area and are predominately forested wetlands.
They provide good habitat for the Florida panther, but
they do not provide connections with suitable habitat
on the Immokalee Rise and in the region. The protected
lands in the south also provide good wading-bird habi-
tat, but only 19% of the potential habitat identified in
our models is within conservation areas. Little of the
protected lands is suitable as Sandhill Crane habitat be-
cause of the lack of open grasslands associated with tem-
porary wetlands. To preserve habitat for these species
and the regional biodiversity associated with those habi-
tats, we need to provide protection for a wider range of
land-cover types that will connect existing conservation
lands with other suitable habitats within the region.
These models illustrate several areas that should be in-
cluded in a regional conservation plan. The wading-bird
model identifies nest sites with high densities of forag-
ing habitat that could be used as a nucleus for a pro-
tected area. In addition, Okaloacoochee Slough runs di-
rectly up the middle of the study area and is an area of
high species utilization in the models. Regionally, this
slough provides a direct link between upland pine and
grassland ecosystems to the north, forested wetland and
everglades ecosystems to the south, and the Corkscrew
Regional Ecosystem to the west. Conservation strategies
also should include the forest, pasture-range, and associ-
ated isolated wetlands to the east of the slough that are
important habitat for all three species. The pasture-range
and wetlands further east are particularly utilized by the
Sandhill Crane.
The biases and limitations of the models must be con-
sidered when examining their outputs. We have not
done a sensitivity analysis on these models (Openshaw
1989; Lodwick et al. 1990; Stoms et al. 1992). It would
be useful to quantify the effects on output of testing the
range of probable values for both model parameters
(such as search distance) and data layers (such as land-
cover misclassification [Lyon et al. 1987]). Some of the
issues include the resolution (scale) of the base maps in
relation to the subject of the model, the accuracy of the
base map, the influences of accuracy on model outputs,
and selection of parameters to include in the models.
Scale, or the spatial resolution at which the analysis is
conducted, can greatly affect the results of the habitat
modeling we performed (Laymon & Reid 1986; Turner
et al. 1989). As resolution decreases, cover types are ag-
gregated, and preferred habitat present at a finer resolu-
tion can be swamped by other cover types, reducing
overall apparent habitat. Just as likely, depending on
cover densities, is that the preferred habitat will pre-
dominate in a grid, but, because of the large grids, habi-
tat will be over represented. As resolution increases be-


Conservation Biology
Volume 9, No. 5, October 1995


Pearlstine et al.






Pearlstine et al.


yond the scale at which the species perceives its
environment, the areal extent of habitat should remain
largely unchanged; it is simply represented by more and
more grids. Specific studies of the scales at which spe-
cies perceive the landscape are typically lacking because
of the difficulty of knowing how different species react
to their surroundings. Different species have different
ecological neighborhoods that can be related to vari-
ables such as body size, home range, and habitat utiliza-
tion (Addicott et al. 1987). The ideal scale for the spe-
cies represented here is not known, but we believe that
30 m2 is below the resolution at which these species se-
lect habitat; thus, areal extent as a result of scale is not
overly generalized in our models.
When developing land-cover maps, it is always desir-
able to conduct an accuracy assessment-and in some
cases absolutely necessary. But accuracy assessments re-
quire additional time and funding and may be perceived
as unnecessary once the map is "complete." We were
unable to conduct an accuracy assessment for this study.
This should be considered when examining the outputs.
In our opinion, because land-cover classes were aggre-
gated to broad habitat classes used in model develop-
ment (for example, forested areas for panthers), and be-
cause these models were developed to illustrate changes
on a regional scale and to direct researchers and re-
source managers to areas that needed attention, the lack
of accuracy assessment does not detract from the in-
tended use of the outputs.
Although no assessment of accuracy was performed in
this study, there have been other efforts to identify re-
gionally significant ecological resources. These efforts
were designed to identify habitats and areas important
for panthers (Logan et al. 1993) and to identify areas im-
portant for South Florida restoration efforts (Weaver et
al. 1993). The major conclusions of these efforts were
identical to those of this study. Although they do not val-
idate the predictions of our models, the similar conclu-
sions of other efforts do make our conclusions more ro-
bust.
Among the parameters considered by our models was
patch size. Patch shape was not considered. For exam-
ple, Sandhill Crane foraging habitat excludes patches
smaller than 240 ha. It is certainly possible that a nar-
row, elongated patch would meet our patch-size
requirements without providing attractive foraging habi-
tat to the crane. Consideration of patch shape may
strengthen the model if appropriate criteria can be for-
mulated. The panther model also considers patch size
but not shape in the exclusion of some otherwise ac-
ceptable land cover. In this model, however, the weighing
of desirable habitat by density of patches in a broader
area mitigates the impact of extended patches on suit-
ability.
These limitations, with one caveat independent of
model quality, minimally influence our, ability to apply


Impacts of Citrus Development on Florida Habitats 1031


the models for the stated purpose. The GIS models inte-
grate spatially referenced information on land cover,
wildlife habitat, and future land-use scenarios to predict
the effect of citrus development on the ecological integ-
rity of the region. The purpose of the models is to direct
the focus of natural resource planners to potentially crit-
ical areas for habitat conservation. The models as ap-
plied strongly suggest that unconstrained development
causes loss and fragmentation of wildlife habitat for the
surrogate species. The models identify specific habitats
(especially uplands) and locations (the central wetlands
core, known locally as the Okaloacoochee Slough) as
potentially significant for the regional conservation of
ecological resources. The model outputs provide a basis
for assessing the degree of protection offered by exist-
ing and proposed protected areas.
Regardless of the quality of a landscape model, there
will always be a degree of uncertainty regarding its pre-
dictions. Models used for planning encounter real time
and budget constraints. Model predictions should be re-
garded as hypotheses of ecosystem response, and moni-
toring programs should be designed as experiments to
test the hypotheses. With this caveat in mind, GIS and
spatial models are excellent tools for forecasting the ef-
fects of land-use changes.



Acknowledgments

This study was funded in part by the South Florida Wa-
ter Management District. Many thanks to Dave Black,
project coordinator at the District, and Calvin Arnold,
Director of the Southwest Florida Research and Educa-
tion Center of the University of Florida, for their contin-
uous support. Programs used to generate areas of cover
types under a specified polygon were written by Brad
Stith and John Richardson. Jim Cox, Dave Maehr, John
Ogden, and Steve Nesbitt all provided helpful comments
on the habitat models. We also thank the participants in
the oversight committees and workshops. This paper
represents Florida Agricultural Experiment Station, Jour-
nal Series No. R-04477.


Literature Cited

Addicott, J. F., J. M. Aho, M. F. Antolin, D. K. Padilla, J. S. Richardson,
and D. A. Soluk. 1987. Ecological neighborhoods: Scaling environ-
mental patterns. Oikos 49:340-346.
Armbruster, M. J. 1987. Habitat suitability index models: Greater Sand-
hill Crane. Biological report 82(10.140). U.S. Fish and Wildlife Ser-
vice, Washington, D.C.
Bancroft, G. T., S. D. Jewell, and A. M. Strong. 1990. Foraging and nest-
ing ecology of herons in the lower Everglades relative to water con-
ditions. Final report. South Florida Water Management District,
West Palm Beach, Florida.
Belden, R. C., W. B. Frankenberger, R. T. McBride, and S. T. Schwikert.
1988. Panther habitat use in southern Florida. Journal of Wildlife


Conservation Biology
Volume 9, No. 5, October 1995







1032 Impacts of Citrus Development on Florida Habitats


Management 52:660-663.
Bennett, A. J. 1978. Ecology and status of Greater Sandhill Cranes in
southeast Wisconsin. M. S. thesis. University of Wisconsin, Stevens
Point.
Bennett, A. J., and L. A. Bennett. 1987. Evaluation of the Okefenokee
Swamp as a site for the development of a nonmigratory flock of
Whooping Cranes. Final report for cooperative research agreement
no. 14-16-009-1551. University of Georgia, Athens.
Bishop, M. A. 1988. Factors affecting productivity and habitat use of
Florida sandhill cranes: An evaluation of three areas in central Flor-
ida for a nonmigratory population of Whooping Cranes. Ph.D. the-
sis. University of Florida, Gainesville.
Browder, J. A. 1976. Water, wetlands, and woodstorks in southwest
Florida. Ph.D. thesis. University of Florida, Gainesville.
Chapman, B. R., and R.J. Howard. 1984. Habitat suitability index mod-
els: Great Egret. FWS/OBS-82/I0.78. U.S. Fish and Wildlife Service,
Washington, D.C.
Cowardin, L. M., V. Carter, F. C. Golet, and E. T. LaRoe. 1979. Classifi-
cation of wetlands and deepwater habitats of the United States.
FWS/OBS/-79/31. U. S. Fish and Wildlife Service, Washington, D.C.
Depkin, F. C., L. A. Brandt, and F. J. Mazzotti. 1994. Nest sites of Sand-
hill Crane in southwest Florida. Florida Field Naturalist 22:39-47.
Drew, R. D., and N. S. Schomer. 1984. An ecological characterization
of the Caloosahatchee River/Big Cypress watershed. FWS/OBS-82/
58.2. U.S. Fish and Wildlife Service, Washington, D.C.
Dwyer, N. 1990. Nesting ecology and nest-site selection of Florida
sandhill cranes. M.S. thesis. University of Florida, Gainesville.
ERDAS. 1991. Image processing and multivariate image analysis. ER-
DAS, Atlanta.
Hoffman, R. D. 1978. The diets of herons and egrets in southwestern
Lake Erie. Pages 365-370 in A. Sprtmt, J. C. Ogden, and S. Winck-
ler, editors. Wading birds. Research report no. 7. National Audu-
bon Society, Tavernier, Florida.
Kushlan, J. A. 1976. Wading bird predation in a seasonally flooded
pond. The Auk 93:464-476.
Kushlan, J. A. 1978. Feeding ecology of wading birds. Pages 249-297
in A. Sprunt, J. C. Odgen, and S. Winckler, editors. Wading birds.
Research report no. 7. National Audubon Society, Tavernier, Flor-
ida.
Laymon, S. A., andJ. A. Reid. 1986. Effects of grid-cell size on tests of a
Spotted Owl HSI model. Pages 93-96 in J. Verner, M. L. Morrison,
and C. J. Ralph. editors. Wildlife 2000: Modeling habitat relation-
ships of terrestrial vertebrates. The University of Wisconsin Press,
Madison.
Lodwick, W. A., W. Monson, and L. Svoboda. 1990. Attribute error and
sensitivity analysis of map operations in geographical information
systems. International Journal of Geographical Information Sys-
tems. 4:413-428.
Logan, T., A. C. Eller, Jr., R. Morrell, D. Ruffner, and J. Sewell. 1993.
Florida panther habitat preservation plan: south Florida population
1993. Florida Panther Interagency Committee, Tallahassee.
Lyon, J. G., J. T. Heinen, R. A. Mead, and N. E. G. Roller. 1987. Spatial
data for modeling wildlife habitat. Journal of Surveying Engineering
113:88-100.
Maehr, D. S. 1990. The Florida panther and private lands. Conservation
Biology 4:167-170
Maehr, D. S., E. D. Land, J. C. Roof, andJ. W. McCown. 1989. Early ma-


Pearlstine et al.


eternal behavior in the Florida panther (Felix concolor coyi). Amer-
ican Midland Naturalist 122:34-43.
Maehr, D. S., E. D. Land, and J. C. Roof. 1991. Florida panthers. Na-
tional Geographic Research and Exploration 7:414-431.
Mazzotti, F. J., L. A. Brandt, L. G. Pearlstine, W. M. Kitchens, T. A.
Obreza, F. C. Depkin, N. E. Morris, and C. E. Arnold. 1992. An eval-
uation of the regional effects of new citrus development on the
ecological integrity of wildlife resources in southwest Florida. Final
report. South Florida Water Management District, West Palm
Beach, Florida.
McCown, J. W., D. S. Maehr, and J. Robhoski. 1990. A portable cush-
ion as a wildlife capture aid. Wildlife Society Bulletin 18:34-36.
Moler, P. E., and R. Franz. 1987. Wildlife values of small, isolated wet-
lands in the southeastern coastal plain. Pages 234-238 in R. C.
Szaro, K. E. Severson, and D. R. Patton, editors. Proceedings of the
third symposium on southeastern nongame and endangered spe-
cies wildlife. General technical report RM-166. U.S. Forest Service,
Fort Collins, Colorado.
Mullins, W. H., and E. G. Bizeau. 1978. Summer foods of Sandhill
Cranes in Idaho. The Auk 95:175-178.
Ncsbitt, S. A., and L. E. Williams, Jr. 1979. Summer range and migration
routes of Florida wintering Greater Sandhill Cranes. Journal of
Wildlife Management 54:92-96.
Nesbitt, S. A., J. C. Ogden, H. W. Kale, II, B. W. Patty, and L. A. Rowse.
1982. Florida atlas of breeding bird sites for herons and their allies
1976-1978. FWS/BOS-81/49. U.S. Fish and Wildlife Service, Wash-
ington, D.C.
Obreza, T. A., H. Yamataki, and L. G. Pearlstine. 1992. Classification of
land suitability for citrus production using DRAINMOD. Journal of
Soil and Water Conservation 48:58-64.
Openshaw, S. 1989. Learning to live with errors in spatial databases.
Pages 263-276 in M. Goodchild and S. Gopal, editors. The accuracy
of spatial data bases. Taylor and Francis, London.
Powell, G. V. N. 1987. Habitat use by wading birds in a subtropical es-
tuary: implications of hydrology. The Auk 104:740-749.
Stoms, D. M., F. W. Davis, and C. B. Cogan. 1992. Sensitivity of wildlife
habitat models to uncertainties in GIS data. Photogrammetric Engi-
neering and Remote Sensing 58:843-850.
Stuwe, M., and C. W. Blohowiak. 1989. McPAAL-Microcomputer pro-
grams for the analysis of animal locations, version 1.2. National
Zoological Park Conservation and Research Center, Washington,
D.C.
Turner, M. G., R. V. O'Neill, R. H. Gardner, and B. T. Milne. 1989. Ef-
fects of changing spatial scale on the analysis of landscape pattern.
Landscape Ecology 3:153-162.
Walkinshaw, L. H. 1976. The Sandhill Crane on and near the Kissim-
mec Prairie, Florida. Pages 1-18 in J. C. Lewis, editor. Proceedings
of the International Crane Workshop. Oklahoma State University,
Stillwater.
Walkinshaw, L. H. 1982. Nesting of the Florida Sandhill Crane in cen-
tral Florida. Pages 53-62 in J. C. Lewis, editor. Proceedings of the
1981 Crane Workshop. National Audubon Society, Tavernier, Flor-
ida.
Weaver, J., et al. 1993. Federal Objectives for the South Florida restora-
tion. Science sub-group report. South Florida Management and Co-
ordination Working Group, Gainesville.


Conservation Biology
Volume 9, No. 5, October 1995




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs