Site by Bennett Web & Design Co.
2006 SOUTHEASTERN NATURALIST 5(3):425–434
Microhabitat Use by Introduced Hemidactylus turcicus
(Mediterranean Geckos) in North Central Florida
Patricia Gomez-Zlatar1, Michael P. Moulton1,*, and Richard Franz2
Abstract - We examined the relationship between seven wall-microhabitat features
and the occurrence of the nonindigenous Hemidactylus turcicus (Mediterranean
Gecko) in north-central Florida using a repeatable technique. We characterized 160
one-story walls by age of the building, cardinal orientation, color, length, presence or
absence of a light source, building material, and vegetation level, and recorded the
presence or absence of H. turcicus for each wall during two separate nocturnal visits.
The occurrence of H. turcicus was only dependent on wall surface color and length.
Both the lack of significance of the majority of the microhabitat variables investigated
and the fact that H. turcicus was found on all wall types suggest that this gecko
is capable of inhabiting a wide variety of wall environments. This habitat flexibility
may be a key factor in the prolific expansion of this gecko’s nonnative range.
Since its initial introduction to Florida in 1915 (Stejneger 1922),
Hemidactylus turcicus (Mediterranean Gecko) has become widely distributed
throughout the southern United States (Conant and Collins 1991). A
native of Eurasia and Africa, this nocturnal gecko exhibits a discontinuous
nonnative range that suggests dispersal via human movement, particularly
along highways (Davis 1974, Meshaka 1995). Although frequently seen on
buildings in both its native and nonnative range, detailed microhabitat studies
for this species are not available. The information available consists
mainly of incidental observations made during other studies (Conant 1955,
Davis 1974, King 1959, Meshaka 1995, Punzo 2001, Rose and Barbour
1968, Selcer 1986) or for populations that coexist with other gecko species
(Luiselli and Capizzi 1999, Meshaka 1995, Saenz 1996). The few results
supported by quantitative data (Luiselli and Capizzi 1999, Nelson and Carey
1993) often lack a rigorous framework making such studies difficult to
compare and conclusions regarding habitat use difficult to draw. Further, the
data-collection techniques in these few studies are vague and impossible to
reproduce. Thus, we embarked on a systematic exploratory study of microhabitat
in a locality where H. turcicus exists in isolation from potential
competing gecko species in an attempt to contribute quantitative baseline
data. A secondary goal of our work was to devise a technique for rapidly
assessing microhabitat use that could be repeated easily by future workers.
1Department of Wildlife Ecology and Conservation, Box 110430, University of
Florida, Gainesville, FL 32611-0430. 2Florida Museum of Natural History, Box
117800, University of Florida, Gainesville, FL 32611-7800. *Corresponding author -
426 Southeastern Naturalist Vol. 5, No. 3
Methods and Materials
This study was conducted on the University of Florida campus and the
adjacent Gainesville Veterans Affairs Medical Hospital Center in
Gainesville, Alachua County, FL, between March and June 2002. We located
160 one-story buildings, and then randomly selected one wall from
each. Walls were restricted to one story in height to facilitate the detection of
geckos. Only one wall per building was used in an attempt to reduce the
potential bias from sampling the same gecko more than once. We considered
a wall to represent accurately the microhabitat use in this species, as H.
turcicus has been shown to possess a small home range (Rose and Barbour
1968, Selcer 1986, Trout and Schwaner 1994). We characterized each selected
wall by age of building, cardinal orientation, color, length, presence
or absence of a light source, building material, and vegetation level.
We obtained building-age information (in years) from literature provided
by the University of Florida (UF Physical Plant Division 2000) and an
unofficial list created specifically for this study by the Veterans Administration
Hospital Engineering Department. We ignored the possibility of any
current renovations, as none of the buildings in our study had any renovations
since 1998 and there were no construction crews present during our
study. We used the age of a building as a rough estimate of the number of
daytime retreats (cracks and/or crevices), as Luiselli and Capizzi (1999)
found the age of a building to be highly positively correlated with the
dilapidation condition of the walls. Although the Luiselli and Capizzi (1999)
study included 2000- to 25-year-old buildings whose construction materials
were not specified, the necessity of using this surrogate for the number of
retreats arose when it became apparent that estimating the number of potential
retreats with the naked eye was highly unreliable. It is important to note
that this retreat approximation needs to be used with caution, as building
disintegration is a complex mechanism whereby age is but one factor in a
long list that includes but is not limited to material and climate.
We determined the cardinal orientation of walls using the 2000 Building
Information List for the University of Florida and official maps of the VA
Hospital Engineering Department. We classified walls as north, south, west, or
east. For walls not clearly oriented in one of these directions, we allowed a 45°
angle of leeway on each side of the cardinal direction. The motivation behind
including this variable derives from the disparity in sun exposure among the
four orientations, which may result in differences in night wall temperatures.
Specifically, west and south walls would be expected to retain greater warmth
into the night as these directions receive sunlight later in the day.
We categorized wall color as dark if we could distinguish a 3-inch by 5-
inch index card fastened to the wall at a perpendicular distance of 10 feet
(3.048 meters) in daylight. Walls where the index card could not be perceived
at a perpendicular distance of 10 feet were appointed to the light level. All the
walls we surveyed were uniform in color across their entire surface.
Length was measured along the base of the wall in meters. Length was
used as a general measure for size, as all walls were one-story high and thus
roughly the same height.
2006 P. Gomez-Zlatar, M.P. Moulton, and R. Franz 427
We measured light occurrence with respect to the entire wall in order to
account for gecko movement. We classified walls as high light if they possessed
at least one light source (any brightness), whereas walls containing no
light source were assigned to the low-light category. It is important to mention
that we initially attempted to measure light intensity in lumens via a light meter.
However, study walls displayed intricate light-intensity mosaics that were too
complex to be quantified as one measurement, which would make repeatability
of the technique problematic. Further, inconsistency in management increased
the variability in light intensity as bulb replacement was never immediate, and
when new bulbs were installed, the wattage was often different. Thus, we
consider our measurement to be an adequate representation of the basic lighting
conditions available to H. turcicus at our study site.
Construction material was confined to four types: aluminum, brick,
cement, and wood. We restricted the study to walls that featured a predominate
material (> 60% of wall).
We classified walls into three vegetation levels that were based on the
cement/vegetation ratio bordering the wall. The cement level referred to a
wall where at least 60% of the length was bordered by cement, the mixed
level to a wall whose length was bordered more than 40% but less than 60%
by either cement or vegetation, and the vegetated level to a wall whose
length was bordered at least 60% by vegetation. We quantified the vegetation
in this way, as opposed to plant species diversity or height, to account
for the highly unpredictable management program (e.g., pruning or mowing)
encountered throughout the study. Additionally, since habitat selection in
reptiles is believed to be most effective when controlled by reliable environmental
cues that are independent of daily and/or seasonal fluctuations
(Heatwole 1977), we considered a wall’s cement/vegetation ratio to be an
acceptable basis for vegetation classification as it is constant and therefore
evident in all situations. A summary of the variables we used and the
different levels is presented in Table 1.
Our sampling regime consisted of visiting 10 walls per night on two
nights per week. Each visit occurred approximately two hours after sunset,
a period of high activity for H. turcicus in Gainesville (Gomez-Zlatar and
Moulton 2005, King 1959). The four months we selected for our study
coincided with part of the reproductive season of H. turcicus, and thus
further ensured gecko activity (Selcer 1986). Sampling duration was kept
under two hours each night in an effort to homogenize environmental
conditions among walls. We examined each of the 160 walls twice for
completeness, once during March/April and once during May/June. We
sampled each wall by passing a flashlight systematically across the entire
surface, going from right to left, top to bottom. We recorded the presence
or absence of H. turcicus for each wall. Data from the two visits were
pooled, and we considered a wall to be inhabited by a gecko if at least one
gecko was present during at least one of the two visits.
We used chi-square analyses to test for differences in frequencies of
gecko occurrence versus cardinal orientation, wall color, construction material,
presence of a light source, and vegetation level. We evaluated building
428 Southeastern Naturalist Vol. 5, No. 3
age and wall length with respect to gecko presence with one-way ANOVAs.
Means were compared with Duncan’s multiple-range tests.
We conducted a preliminary study to elucidate wall-temperature patterns
over seven nights during the months of July and August 2001. Each
survey night began approximately one to two hours after sunset, and lasted
two hours. The number of walls visited per night was dictated solely by the
two-hour sampling time permitted, and thus varied as some buildings were
more isolated than others. All walls surveyed were located on the University
of Florida campus and the Gainesville Veterans Affairs Medical
Hospital Center. We characterized walls with respect to construction material,
and cardinal orientation as detailed above. We recorded a total of four
temperature readings from each sampled wall with a Raytek Raynger ST
model temperature gun. The readings were measured from the upper
middle, lower middle, right middle, and left middle of each wall. All walls
of a building were sampled whenever possible.
We compared average wall temperatures across construction material
and cardinal orientation with a two-way ANOVA, and used Duncan’s
multiple-range tests to compare means. All statistical analyses had a significance
level of 5%, and were performed using the SAS system, Version 9.
Table 1. Description of wall-characterization variables for microhabitat study
# of walls
Variables Levels sampled Criteria
Age - 158 Age in years*
North 43 Location of wall on official maps**
South 49 Location of wall on official maps**
West 34 Location of wall on official maps**
East 34 Location of wall on official maps**
Color Light 95 The inability to perceive a 3"x 5" white index card at a
perpendicular distance of 10ft (3.048m) from the wall;
during the day***
Dark 65 The ability to perceive a 3"x 5" white index card at a
perpendicular distance of 10ft (3.048m) from the wall;
during the day***
Length - 160 Measure along base of wall in meters
Light High 38 Presence of at least one light source on the wall
Low 122 No light source present on the wall
Material Aluminum 46 Physical observation; > 50% of wall surface
Brick 33 Physical observation; > 50% of wall surface
Cement 59 Physical observation; > 50% of wall surface
Wood 22 Physical observation; > 50% of wall surface
Vegetation Cement 30 60% of wall length bordered by cement
Mix 31 > 40% to less than 60% of wall length bordered by cement or
Vegetation 99 60% of wall length bordered by any type or height of
*Sources used: 2000 Building Information List for the University of Florida prepared by the UF
Physical Plant Division and unofficial list prepared by the VA Hospital Engineering Department.
**Sources used: 2000 Building Information List for the University of Florida prepared by the
UF Physical Plant Division and official VA Hospital Engineering maps.
*** Index cards were provided by AMPAQ, Dallas, TX 75252.
2006 P. Gomez-Zlatar, M.P. Moulton, and R. Franz 429
Of the 160 walls that we surveyed, 26 walls had at least one gecko on the
first visit only, 20 walls had geckos on just the second visit, and 39 walls had
geckos during both sampling visits. Seventy-five walls had no geckos on
either visit. Geckos were observed on walls of all categorical variable-level
types. The proportion of walls recording at least one H. turcicus observation
is listed for each categorical variable level in Table 2. The presence and
absence of H. turcicus by wall frequency is displayed for each categorical
variable level in Figure 1.
We observed geckos more often than expected by chance on dark
colored walls, whereas fewer than expected were observed on light-colored
walls (p = 0.0023).
The results of our chi-square tests indicated that gecko presence was
independent of cardinal orientation, construction material, presence of a light
source, and vegetation level. These results are presented in Table 3. Chisquare
tables detailing observed and expected frequencies, along with cell
chi-square values for each of these variables are presented in Appendix 1.
Although insignificant, general trends gleaned from the chi-square tables
Table 2. Proportion of walls of each categorical variable level at which at least one H. turcicus
observation was recorded.
Percentage of walls at which at least one
Variables Levels H. turcicus observation was recorded
Cardinal orientation North 39.5%
Color Light 43.2%
Light High 65.8%
Material Aluminum 41.3%
Vegetation Cement 50.0%
Table 3. Chi-square statistics and corresponding p-values for all categorical microhabitat
Variables Degrees of freedom Chi-square value P-value
Cardinal orientation 3 4.9238 0.1775
Color 1 9.3288 0.0023*
Light 1 3.2098 0.0732
Material 3 4.3282 0.2281
Vegetation 2 0.1609 0.9227
*Significant at the 5% level.
430 Southeastern Naturalist Vol. 5, No. 3
included lower gecko occurrence than expected by chance on both northern
and aluminum walls. Conversely, geckos were present more than expected by
chance on walls containing at least one light source. Vegetation demonstrated
no discernable trend.
Building age was not a factor in determining the presence of H. turcicus on a
wall (p = 0.3183). Wall length appeared to influence H. turcicus presence (p less than
0.0001), as walls recording a gecko observation were an average length of 19.6
meters whereas walls with no gecko observations were 13.1 meters long.
We also compared average temperatures of a total of 16 brick, 52 cement,
38 wood, and 78 aluminum walls. The average wall temperatures of the four
types of material were all significantly different from each other (p less than
0.0001). Brick walls had the highest mean temperatures with an average of
26.8 °C, followed by cement walls at 26.1 °C, wood walls at 25.1 °C, and
aluminum walls at 23.8 °C. Average temperatures of 47 southern, 46 northern,
46 eastern, and 45 western walls were assessed. No significance was
found among the average temperatures of walls oriented in different cardinal
Figure 1. The presence and absence of H. turcicus by wall frequency for cardinal
orientation, color, material, presence of a light source, and vegetation.
2006 P. Gomez-Zlatar, M.P. Moulton, and R. Franz 431
directions (p = 0.499). The two-way interaction between material and cardinal
orientation also was not significant (p = 0.9998).
Of the seven variables we examined, two were significant. The first was
wall color and the second was wall size. We discuss these in turn below. In
addition, we address the unexpected insignificance of light sources, and also
the potential trends that emerged from the insignificant variables.
A greater number of dark walls than expected by a hypothesis of independence
were occupied by H. turcicus. This gecko species possesses both
light and dark dorsal markings (Bartlett and Bartlett 1999) that allow it to
conceal itself against a variety of substrate (i.e., wall) colors and patterns,
making camouflage an unlikely motive for this choice. Further study will be
necessary to determine the basis for the result.
Walls that recorded H. turcicus observations were longer than those with
no geckos. Additional space may be an important component of this gecko’s
microhabitat, as it is likely to provide a greater number of retreats, a larger
foraging area, and an increase in social interactions crucial for reproduction.
The occurrence of H. turcicus on a wall was independent of the presence
of a light source. This result was unanticipated, as investigators have argued
that H. turcicus was always found high on a wall in proximity to a light
source, to facilitate the capture of their insect prey (Conant 1955, Conant
and Collins 1991, Davis 1974, Nelson and Carey 1993, Punzo 2001). The
idea that H. turcicus might not require a light source for prey capture is
alluded to in two dietary studies (Capula and Luiselli 1994, Saenz 1996),
which found that a large proportion of H. turcicus’ diet consisted of grounddwelling
prey rather than flying prey. Further evidence against the need for a
light source for prey detection came from our observations of geckos on
walls with a wide gamut of light intensities, including complete darkness.
Nonetheless, it is prudent to keep in mind the contradicting general trend
whereby walls with at least one light source tended to have geckos present
more often than would be expected under independence. Although insignificant,
this trend emphasizes that the role of light as a microhabitat cue for H.
turcicus remains unclear. Future studies, particularly controlled experiments,
may be the key to elucidating this relationship.
Another general trend we ascertained was that a smaller proportion of
north walls contained geckos than all other directions. It is unlikely that
thermoregulation is an underlying factor as north walls had an average
temperature comparable to all other cardinal directions. Explanations for
this trend are difficult to formulate at this time.
The last general trend detected was that a smaller proportion of aluminum
walls contained geckos. The average summer temperature for
aluminum walls was at least 1.3 °C lower than all other materials surveyed,
and may explain this trend. With an average summer temperature of 23.8 °C,
aluminum is warmer than the average adult substrate for fall (23.2 °C) but
cooler than the average for spring (24.9 °C), which is the beginning of the
reproductive season (Gomez-Zlatar and Moulton 2005). Thus, it is possible
432 Southeastern Naturalist Vol. 5, No. 3
that lower substrate temperatures deter H. turcicus from aluminum walls,
especially during the reproductive season. An alternative explanation for
this trend, which was not studied here, is the possibility that aluminum
disintegrates at a slower rate than other materials and provides fewer cracks
and crevices for H. turcicus to use as retreats. Further study into the thermal
field preferences of this species and into the complexities of material decay
is necessary before any conclusions can be made.
The lack of a significant relationship between building age, material,
cardinal orientations, presence of light, and vegetative structure suggests a
variety of scenarios. Although the variables we selected were mostly those
measured by other observers, it is possible that the variables of this study are
simply not important to these geckos. Alternatively, there could be subtle
microclimatic factors that we did not measure. However, since H. turcicus
was consistently found on all wall types, a more likely interpretation of our
results is that these geckos are highly plastic with respect to their microhabitat
needs relative to the range of physical parameters present on the walls
of buildings. This ultimately permits this species to thrive on a variety of
substrates, and thus facilitates its ability to extend its nonnative range.
As with all scientific studies, our data and conclusions have limitations.
For instance, it is impossible to determine if the absence of H. turcicus on a
wall is due to a microhabitat cue or if it is an artifact of this species’ dispersal
via human movement. Also, shortcomings in sampling technique, such as not
observing a wall throughout the entire night and frightening geckos as we
approached a wall, could have resulted in an underestimation of walls containing
geckos. In addition, it should be noted that performing seven separate
statistical tests resulted in the inflation of the experiment-wise error rate.
Although unfortunate, the absence of certain variable combinations in our
study site (resulting in zeros in the matrix of all possible variable combinations)
greatly limited the number of statistical tests that could be used on this
dataset (Gomez-Zlatar 2003). We encourage future studies to consider this
problem when selecting a study site so as to avoid a similar predicament.
As many as 11 species of introduced geckos have established reproductive
populations in south Florida, whereas an additional five species have
occasionally been sighted and as such are considered to be of uncertain
status (Meshaka et al. 2004). The probability is high that one or more of
these species will ultimately expand its range into our study area. Thus, we
believe that our results not only provide important baseline data for H.
turcicus, but for all future studies of the ecology of invading gecko species.
We have made a concerted effort to develop a simple technique for
scoring microhabitat use, and we hope that future studies will not only use
these methods, but also ultimately refine them so that investigators anywhere
in the world can compare results among sites in a meaningful way.
We thank Camilo Gomez, Elza Kephart, Chuck Knapp, Esther Langan, Alex
Martin, and Robin Sternberg for their help in the field. We especially thank the
University of Florida and the staff at the Gainesville Veterans Affairs Medical Center
2006 P. Gomez-Zlatar, M.P. Moulton, and R. Franz 433
for allowing us access for our study. Finally, we thank two reviewers for their
insightful comments, which allowed this manuscript to reach its full potential.
Bartlett, R.D., and P.P. Bartlett. 1999. A Field Guide to Florida Reptiles and Amphibians.
Gulf Publishing Company, Houston, TX.
Capula, M., and L. Luiselli. 1994. Trophic niche overlap in sympatric Tarentola
mauritanica and Hemidactylus turcicus: A preliminary study. Herpetolological
Conant, R. 1955. Notes on three Texas reptiles, including an addition to the fauna of
the state. American Museum Novitates 1726:1–6.
Conant, R., and J.T. Collins. 1991. A Field Guide to Reptiles and Amphibians of
Eastern and Central North America. Houghton Mifflin Company, Boston, MA.
Davis, W.K. 1974. The Mediterranean Gecko, Hemidactylus turcicus in Texas.
Journal of Herpetology 8:77–80.
Gomez-Zlatar, P. 2003. Microhabitat preference of the introduced gecko,
Hemidactylus turcicus in an urban environment. M.Sc. Thesis. University of
Florida, Gainesville, FL.
Gomez-Zlatar, P., and M.P. Moulton. 2005. Habitat use by the nonindigenous Mediterranean
Gecko (Hemidactylus turcicus) in north central Florida. Florida Scientist
Heatwole, H. 1977. Habitat Selection in reptiles. Pp. 137–155, In C. Gans and D.W.
Tinkle (Ed.). Biology of the Reptilia, Vol. 7 (Ecology and Behavior A). Academic
Press Inc., New York, NY.
King, W. 1959. Observations on the ecology of a new population of the Mediterranean
Gecko, Hemidactylus turcicus, in Florida. Quarterly Journal of the Florida
Acadamy of Science 21:317–318.
Luiselli, L., and D. Capizzi. 1999. Ecological distribution of the geckos Tarentola
mauritanica and Hemidactylus turcicus in the urban area of Rome in relation to
age of buildings and condition of the walls. Journal of Herpetology 33:316–319.
Meshaka, Jr., W.E. 1995. Reproductive cycle and colonization ability of the Mediterranean
Gecko (Hemidactylus turcicus) in south-central Florida. Florida Scientist
Meshaka, Jr., W.E., B.P. Butterfield, and J.B. Hauge. 2004. The exotic amphibians
and reptiles of Florida. Krieger Publishing Company, Malabar, FL.
Nelson, D.H., and D. Carey. 1993. Range extension of the Mediterranean Gecko
(Hemidactylus turcicus) along the northeastern gulf coast of the Unites States.
Northeast Gulf Science 13:53–58.
Punzo, F. 2001. The Mediterranean Gecko, Hemidactylus turcicus: Life in an urban
landscape. Florida Scientist 64:56–66.
Rose, F.L., and C.D. Barbour. 1968. Ecology and reproductive cycles of the introduced
gecko, Hemidactylus turcicus, in the southern United States. American
Midland Naturalist 79:159–168.
Saenz, D. 1996. Dietary overview of Hemidactylus turcicus with possible implications
of food partitioning. Journal of Herpetology 30:461–466.
Selcer, K.W. 1986. Life history of a successful colonizer: The Mediterranean Gecko,
Hemidactylus turcicus, in southern Texas. Copeia 1986:956–962.
Stejneger, L. Two geckos new to the fauna of the United States. Copeia 1922:56.
Trout, L., and T.D. Schwaner. 1994. Allozyme evidence for insularity in exotic
populations of the Mediterranean Gecko (Hemidactylus turcicus). Journal of
434 Southeastern Naturalist Vol. 5, No. 3
University of Florida Physical Plant Division, Architecture/Engineering Department.
2000. 2000 Building information list for the University of Florida.
Appendix 1. Observed and expected frequency followed by individual cell chisquare
values for the chi-square test of each categorical variable. Cells that have a
large influence on the chi-square table are indicated by an asterisk.
A. Cardinal orientation
H. turcicus 2 parameters East North South West
No Observed 16 26 19 14
Expected 15.938 20.156 22.969 15.938
Cell 2 0.0002 1.6492* 0.6858 0.2355
Yes Observed 18 17 30 20
Expected 18.063 22.844 26.031 18.063
Cell 2 0.0002 1.4949* 0.6051 0.2078
H. turcicus 2 parameters Light No light
No Observed 13 62
Expected 17.813 57.188
Cell 2 1.3002* 0.405
Yes Observed 25 60
Expected 20.188 64.813
Cell 2 1.1473* 0.3573
H. turcicus 2 parameters Aluminum Brick Cement Wood
No Observed 27 13 27 8
Expected 21.563 15.469 27.656 10.313
Cell 2 1.3712* 0.394 0.0156 0.5186
Yes Observed 19 20 32 14
Expected 24.438 17.531 31.344 11.688
Cell 2 1.2099* 0.3476 0.0137 0.4576
H. turcicus 2 parameters Cement Mix Vegetation
No Observed 15 14 46
Expected 14.063 14.531 46.406
Cell 2 0.0625 0.0194 0.0036
Yes Observed 15 17 53
Expected 15.938 16.469 52.594
Cell 2 0.0551 0.0171 0.0031