Austral Ecology (2020) , –

Microhabitat selection and niche partitioning in two

syntopic geckos
MELISSA ANNE PETFORD* AND GRAHAM JOHN ALEXANDER
School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, P.O. Wits,
Johannesburg, 2050, South Africa (Email: Melissa.petford@outlook.com)

Abstract Syntopic species often exhibit evolutionary mechanisms that reduce competition. A common mecha-
nism facilitating coexistence is niche separation, which may manifest through spatial, temporal or trophic dimen-
sions. Species that are morphologically similar, such as congeners, are likely to compete directly and thus
separate their niche spatially. The microhabitat selection and partitioning of two endemic geckos of the Sout-
pansberg Mountains, Lygodactylus incognitus (Jacobsen, 1992; Squamata: Gekkonidae) and L. soutpansbergensis
(Jacobsen, 1994; Squamata: Gekkonidae), was investigated by recording fine- and broad-scale habitat variables.
Results reveal that L. incognitus is restricted to high elevations above 1100 m a.s.l. and is associated with moist
microclimates. Although primarily saxicolous, they also utilise tree trunks, branches and stems. Lygodactylus sout-
pansbergensis occurs above 800 m a.s.l and is restricted to rocky outcrops, open woodlands and rocky grasslands,
often found on loose rocks. Perch height for L. soutpansbergensis is close to ground level, and they are limited to
drier microclimates. Microhabitat partitioning does not appear to be due to interspecific exclusion as the smaller
L. soutpansbergensis occupies the same niche dimensions in areas where L. incognitus is not present. Sites where
L. incognitus occurs in the absence of L. soutpansbergensis are forested with high canopy cover, likely preventing
the colonisation of L. soutpansbergensis. This suggests that morphological and physiological constraints define
microhabitat limits, although this still requires testing. Anthropogenic activities are likely to threaten the persis-
tence of L. incognitus and L. soutpansbergensis on the Soutpansberg in the future. The results from this study may
aid the interpretation and understanding on the syntopy of morphologically similar species that inhabit the same
macrohabitats in other areas.

Key words: microhabitat, niche separation, syntopic, coexistence, conservation.

INTRODUCTION (Heller 1971; Myrberg & Thresher 1974; Toft 1985;

Wherever a species exists, there is always likely to be
other species with overlapping ecological require-
ments with which it must compete. Niche separation
is one of the potential mechanisms which enables this
coexistence and enables high diversity and syntopy of
species. Syntopy (Rivas 1964) and its underlying evo-
lutionary mechanisms have been studied in detail in
order to gain a better understanding of communities
(Hamilton 1962; Adolph 1990; Kitchen et al. 1999;
Luiselli 2008; Lisi
cic et al. 2012). Syntopic species
frequently come into contact with each other and
thus exhibit behavioural, morphological or ecological
modes of separation in order to reduce interspecific
competition (Pianka 1973; Snyder 1979; Melville
2002; Steinberg et al. 2007; Luiselli 2008). This is
also known as resource partitioning (Pianka 1973).
Resource partitioning in syntopy may be due to inter-
specific aggression, with a dominant species forcing
others to occupy different niches, or due to ecologi-
cal, physiological and morphological differences
*Corresponding author.
Accepted for publication December 2019.
Robinson & Terborgh 1995; Bay et al. 2001; Langk-
ilde et al. 2005; Lisi
cic et al. 2012).
Lizards are ideal organisms to examine and identify
coexistence mechanisms as they are often locally
abundant, and several species are commonly found
inhabiting the same macrohabitats. As such, niche
separation has been well documented for several syn-
topic lizard species (Huey et al. 1974; Hess & Losos
1991; Akani et al. 2002; Noble et al. 2010; Lisi
cic
et al. 2012). Spatial niche separation is one of the
most common methods of partitioning for syntopic
lizards, particularly as most species are generalist,
opportunistic feeders and are thus unlikely to parti-
tion food resources (Luiselli 2008; Noble et al. 2010;
Lisi
cic et al. 2012). Resource partitioning and micro-
habitat selection in lizards can be attributed to physi-
ological and morphological constraints (Snyder 1979;


Steinberg et al. 2007; Zagar et al. 2015; Ferreira et al.
2017). Species which are morphologically similar,
such as congeners, are likely to compete directly as
morphology, ecology and behaviour are linked, and
thus, these species are likely to have similar ecologi-
cal requirements (Huey et al. 1974; Huey & Pianka
1977; Pianka 1986; Losos 1990).

© 2020 Ecological Society of Australia doi:10.1111/aec.12857

2 M . A . P E T F O R D ET AL.
Investigating the mechanisms behind niche separa-
tion in closely related species can provide knowledge
on fine-scale ecological interactions, particularly for
restricted species whose basic ecology is poorly
known. Understanding a species requirements and
interactions on a deeper level can strengthen conser-
vation planning (Filippi & Luiselli 2006; Triska et al.
2017), particularly as habitat destruction through
both direct and indirect anthropogenic pressures is
increasingly an issue in conservation biology, where
generalist species are likely to benefit from reduced
competition from specialist species (Nordberg & Sch-
warzkopf 2019).
We measured the microhabitat parameters of the
Cryptic Dwarf Gecko (Lygodactylus incognitus; Jacob-
sen, 1992; Squamata: Gekkonidae) and the Sout-
pansberg Dwarf Gecko (L. soutpansbergensis;
Jacobsen, 1994; Squamata: Gekkonidae), two syn-
topic geckos endemic to the Soutpansberg Mountains
in South Africa, in an attempt to understand their
microhabitat requirements, whether niche partition-
ing occurs on a spatial scale and whether either spe-
cies occurs in disturbed areas. We recorded fine- and
broad-scale habitat variables for individual gecko
sightings along random transects. A comparison of
niche limits between areas where the species were
and were not syntopic were used to assess whether
niche separation was facilitated by interspecific
aggression.
MATERIALS AND METHODS
Study area
The study area is located in the western part of the Sout-
pansberg Mountain range, Limpopo Province, South
Africa. For the purpose of the study, the western Soutpans-
berg is defined as the 70 km stretch between Vivo in the
west and Louis Trichardt in the east. Most of the fieldwork
took place at the Lajuma Research Centre ( 23.04, 29.44)
and Medike Mountain Reserve ( 23.00, 29.61) with
additional surveying taking place at Goro Game Reserve
( 22.97, 29.43) and Hanglip Forest Reserve ( 22.99,
29.89). The western Soutpansberg ranges in elevation from
800 to 1748 m a.s.l., is roughly 25 km wide at its widest
Fig. 1. Lygodactylus incognitus (left) and Lygodactylus soutpansbergensis (right)
doi:10.1111/aec.12857
point and is situated within a rain shadow from the Drak-
ensberg escarpment. It therefore receives low annual rainfall
(average of 618 mm at Louis Trichardt; Hahn 2006). In
spite of the mountain range being relatively narrow, there
are still large differences in climate between the northern
and southern sides. The northern side of the mountain is
much hotter and arid with areas on the northern slopes
receiving an average annual rainfall of only 300–400 mm.
The higher elevations of southern and eastern facing slopes
are classified as being in a mist-belt due to prevailing winds
bringing moisture laden air from the Indian Ocean; this is
also facilitated by the east–west axis of the mountain
(Kabanda 2003; Hahn 2006; Kirchhoff et al. 2010).
Vegetation classifications follow Mucina and Rutherford
(2006), Mostert (2006) and Mostert et al. (2008). The
main vegetation type present in the western Soutpansberg
is Soutpansberg Mountain Bushveld, which is further sub-
divided into the following: Subtropical Moist Thickets;
Mist-belt Bush Clumps; Open Savanna Sandveld; and Arid
Mountain Bushveld (Mucina & Rutherford 2006; Mostert
2006; Mostert et al. 2008). Additional vegetation types in
the Soutpansberg are Northern Mist-belt Forest and Sout-
pansberg Summit Sourveld. Disturbed areas in the western
Soutpansberg include silviculture and previous agricultural
areas.
Study species
Lygodactylus incognitus and L. soutpansbergensis (Fig. 1) are
endemic to the Soutpansberg Mountain range and thus
have very restricted distributions with an area of occu-
pancy (AOO) of 675 and 1250 km2, respectively (Bates
et al. 2014). The known distribution for L. soutpansbergen-
sis extends from the far western region of the Soutpans-
berg to Thohoyandou and Thathe Vondo areas in the
east. Lygodactylus incognitus has a smaller known distribu-
tion of only 10 locations from the far western Soutpans-
berg to the Thathe Vondo area and falls completely
within the distribution of L. soutpansbergensis. Lygodactylus
incognitus has an average SVL of 33.9 mm  1.7, with
adult males reaching 36 mm and adult females 37 mm
(Jacobsen 1992). Lygodactylus soutpansbergensis is smaller
with an average SVL of 29.2 mm  1.8, with adult males
reaching 31 mm and adult females 32.5 mm (Jacobsen
1994).
Both species have been defined as saxicolous, with previ-
ous studies on the reptiles of the Soutpansberg concluding
that L. incognitus occurs on rocky outcrops in woodland,
© 2020 Ecological Society of Australia
 NICHE PARTITIONING IN TWO SYNTOPIC GECKOS 3

grassland and bush clumps between 1282 and 1748 m Table 1. Microhabitat variables recorded when L. incogni-
a.s.l., whilst L. soutpansbergensis occurs on rocky outcrops in tus and L. soutpansbergensis were observed
grassland and open woodland between 850 and 1550 m
a.s.l (Jacobsen 1992, 1994; Kirchhoff et al. 2010; Bates Variable Radius Description
et al. 2014).
Rock Type – For example, sandstone and
quartzite
Methods Rock – Type of formation, for example
Structure scree slope, cliff and ridge
Sampling took place between January 2017 and November Micro-aspect – Aspect to which the gecko is
2017. Transects were randomly walked across study sites facing to the nearest ordinal
through all the vegetation types present in the study area, direction
including disturbed areas. A total of 250 transects were Macro-aspect – Aspect of the slope on which the
gecko is located to the nearest
selected using a simple random design, with start points
ordinal direction
randomly generated in each of the field locations, excluding
Height off – Height from the ground to where
inaccessible areas (e.g. steep cliff faces). Each transect was
Ground the gecko was first observed in
roughly 500 m long. The number of transects in each vege- metres (to the nearest cm)
tation type was size representative. Each transect was Substrate – Soil type where the gecko was
repeated in summer and winter. On each transect, active Type first observed (sandy soil,
searching techniques (turning rocks; looking in crevices) humus soil, clay soil)
were used to detect the geckos. If either species was not % Tree Cover 1 m2 % of tree cover when looking up
found in a particular habitat type, transects were repeated directly from where the gecko
and additional intensive surveys were conducted. As the was first observed to the
geckos are diurnal, sampling took place during daylight nearest 5%
hours between 7 am and 5 pm in winter and until 6pm in % Lichen 1 m2 % of lichen cover to the nearest
summer. Incidental records were also used from across the Cover 5%
study sites. % of Leaf 1 m2 % of leaf litter to the nearest 5%
When the geckos were observed, microhabitat data were Litter
collected using a personalised CyberTracker v3.386 appli- % of Grass 1 m2 % of grass to the nearest 5%
cation (Table 1). Macro-aspect, micro-aspect, substrate % of Loose 1 m2 % of loose rocks to the nearest
type and whether the gecko was in direct sunlight or shade Rocks 5%
were recorded. Preferred perch height was assessed by mea- % of Wood 1 m2 % of wood debris to the nearest
suring height off ground to the nearest cm with a measur- Debris 5%
Distance to 5 m2 Distance to the nearest tree taller
ing tape. Heights above 3 m were estimated. Percentage
Nearest than 2 m measured in m
canopy cover, lichen cover, leaf litter, grass, loose rocks
Tree
and wood debris were recorded within a 1 m2 radius to the
% Ground 5 m2 % of ground covered by woody
nearest 5% using a quadrat. Within 5 m2, the distance to Cover of vegetation to the nearest 5%
nearest tree >2 m (to the nearest 1 m), percentage ground Trees
cover of trees, exposed rock, grass and bare soil (to the % Exposed 5 m2 % of ground which is covered by
nearest 5%) were recorded by visual estimation as adapted Rock exposed rock to the nearest
from Quirt et al. (2006). If a gecko was observed within a 5%
refuge, the type of refuge was noted, for example under a % Grass 5 m2 % of ground which is covered by
rock or in a crevice. Records of each gecko were separated grass or non-woody vegetation
by season. As the study site is located in a subtropical area, (e.g. sedge) to the nearest 5%
autumn and spring are not prominent, and thus, the sea- % of Bare Soil 5 m2 % of ground which is bare soil to
sons were split into summer (October–March) and winter the nearest 5%
(April–September). Type of – If found in refuge: under a rock;
Refuge in a rock crevice; under bark

Statistical analyses
The resulting factor scores were assessed using a multi-
As the dataset consisted of mixed variable types, an opti-
variable linear regression model in order to assess whether
mal scaling approach was used to reduce the dimension-
there were any significant differences between the micro-
ality of the 19 microhabitat selection variables.
habitat selection of L. incognitus and L. soutpansbergensis.
Categorical principal component analyses (CATPCA)
The regression scores were the dependent variables, with
were used to first monotonically transform the variables
season as a covariate and species as the fixed factor. Fol-
in order to maximise the variability explained. The corre-
low-up univariate regression tests with a Bonferroni correc-
lation matrix from the CATPCA output was inspected
tion were performed. With the Bonferroni correction, only
and any variables which did not have at least one corre-
P-values <0.0167 were considered statistically significant.
lation coefficient >0.3 were removed. The transformed
In addition to this, a further two multivariable linear regres-
variables were then used to perform factor analyses (FA)
sions were performed with factor scores as the dependent
(Starkweather 2010).
variable, season as a covariate and syntopy (yes/no) as a

© 2020 Ecological Society of Australia doi:10.1111/aec.12857

4 M . A . P E T F O R D ET AL.

fixed factor to compare the habitat selection of each species
between areas where they were syntopic, and where they
were not. Again, follow-up univariate regressions tests with
a Bonferroni correction were then performed; only P-values
<0.0167 were statistically significant. All statistics were con-
ducted in IBM SPSS Statistics 20, and all relevant assump-
tions were assessed prior to analyses.
RESULTS
Habitat measures were recorded for 167 individual
L. incognitus: 83 of these were recorded in winter and
84 in summer. For L. soutpansbergensis, measures for
124 individuals were recorded with 58 in winter and
66 in summer. The elevation range for L. incognitus
was 1100–1748 m a.s.l. and for L. soutpansbergensis
800–1748 m a.s.l. Both species were found in Sout-
pansberg Summit Sourveld and Soutpansberg Moun-
tain Bushveld (Table 2), in which they co-occurred
extensively above 1100 m a.s.l, these habitats are also
the most abundant at these elevations. In addition,
Table 2. Presence of Lygodactylus incognitus and Lygo-
dactylus soutpansbergensis in the vegetation types located in
the western Soutpansberg. Vegetation types follow Mucina
and Rutherford (2006); Mostert (2006); and Mostert et al.
(2008). In cases where both species occur in a particular
habitat, it is indicated whether these species occur in
syntopy
Does
Lygodactylus Lygodactylus syntopy
Habitat Type incognitus soutpansbergensis occur?
L. incognitus was found in Mist-belt Bush Clumps
and Northern Mist-belt Forest, whilst L. soutpansber-
gensis was also found in Arid Mountain Bushveld and
Open Savanna Sandveld. Neither species was located
in silivicultural areas. Lygodactylus soutpansbergensis
was absent from sites where rocks had been histori-
cally cleared for agriculture. Lygodactylus incognitus
was common on house walls, whereas no L. soutpans-
bergensis were observed on buildings. Average perch
height of L. incognitus was 1.75 m (0–15 m [min-
max height]) and for L. soutpansbergensis 0.27 m (0–
1.2 m [min–max height]). More detailed information
regarding the habitat variables for each species is pre-
sented in Tables 3 and 4, and images of typical
microhabitat of the two species are presented in Figs
2 and 3. During the surveys, it was observed that
L. incognitus may potentially be the dominant species,
chasing L. soutpansbergensis away from high quality
areas and performing aggressive displays.
Factor analyses on the microhabitat variables for
both species of gecko retained three components
explaining 65.5% of the total variance and with a
Kaiser–Meyer–Olkin (KMO) of 0.76. To increase the
interpretability of the model, a varimax orthogonal
rotation was used. Component loadings can be found
in Table 5. Component 1 showed high positive load-
ings on rock structure and exposed rock (within
Table 3. Average values of each microhabitat variable for
Lygodactylus incognitus and Lygodactylus soutpansbergensis in
summer and winter
Lygodactylus Lygodactylus
incognitus soutpansbergensis

Arid No Yes N/A Habitat variables Winter Summer Winter Summer

Mountain
Bushveld Macro-aspect N S N N
Open Savanna No Yes N/A Micro-aspect N N N N
Sandveld Leaf Litter 11.45% 7.86% 5.45% 5.11%
Mist-belt Yes No N/A (1-m2 radius)
Bush Grass/Sedge 15.18% 13.93% 17.95% 17.02%
Clumps (1-m2 radius)
Northern Yes No N/A Loose Rocks 1.33% 8.45% 28.3% 30.43%
Mist-belt (1-m2 radius)
Forest Wood Debris 6.20% 7.08% 3.41% 3.63%
Soutpansberg Yes Yes Yes (1-m2 radius)
Summit Lichen Cover 51.14% 50.12% 27.95% 21.95%
Sourveld (1-m2 radius)
Soutpansberg Yes Yes Yes Canopy Cover 24.34% 23.93% 5.23% 5.74%
Mountain (1-m2 radius)
Bushveld Distance to Nearest 2– 0–0.1 m >5 m >5 m
Subtropical No No N/A Tree 2.5 m
Moist Ground Cover of 29.64% 23.57% 17.27% 14.26%
Thickets Trees
Silviculture No No N/A (5-m2 radius)
Historical No No N/A Exposed Rock 48.31% 52.02% 52.04% 53.19%
Agricultural (5 m2 radius)
Land Grass (5-m2 radius) 18.07% 20.34% 27.27% 26.17%
(Rocks Bare Soil 5.06% 4.64% 3.40% 6.38%
Cleared) (5-m2 radius)

doi:10.1111/aec.12857 © 2020 Ecological Society of Australia

 NICHE PARTITIONING IN TWO SYNTOPIC GECKOS 5

Table 4. Percentage of Lygodactylus incognitus and Lygo- differences between the two species with P-values of
dactylus soutpansbergensis found in the open, in a crevice or 0.006, ˂0.001 and ˂0.001, respectively. Indeed,
under a rock and out of all gecko observations the percent- L. incognitus was often found utilising areas with high
age of those found on trees. Standard deviation is shown in
tree cover and moist microclimates whereas L. sout-
brackets
pansbergensis was more likely to be found in areas
Lygodactylus Lygodactylus with more rock, less trees and in drier microclimates.
First observed incognitus soutpansbergensis A multivariable linear regression indicated that
there was a statistically significant difference between
Found in the open 80% (11.15) 57% (4.87) habitat selection where L. soutpansbergensis occurs
Found in a crevice 20% (4.27) 12% (1.70)
with or without L. incognitus after accounting for sea-
Found under a rock 0% (1.13) 31% (3.36)
Found on trees 25% (10.61) 1% (0.32) son, F123 = 94.0, P = ˂0.001, Wilks’ Λ = 0.02. Fol-
low-up univariate regression tests with a Bonferroni
correction indicated that Component 1 and 3 did not
5 m2) and a negative loading on ground cover of show significant differences of L. soutpansbergensis
trees (5 m2). Thus, this component is indicative of habitat selection between sites where they occurred
areas with high amount of rock and low ground tree with L. incognitus and where they occurred alone,
cover. Component 2 displayed high positive loadings with P-values of 0.316 and 0.051, respectively. Com-
on lichen cover (1 m2) and elevation, suggesting that ponent 2 did show a significant difference in habitat
component 2 is indicative of moist areas at high ele- selection between sites where L. soutpansbergensis
vations. Component 3 had high positive loadings on occurred with L. incognitus and those where they
canopy cover (1 m2), wood debris (1 m2) and leaf lit- were found alone (P < 0.001). This suggests that
ter (1 m2) and negatively on distance to nearest tree there may be niche displacement on Component 2;
(>2 m). Therefore, component 3 is associated with however, this Component is indicative of high eleva-
areas of high tree cover and thus small distances to tion areas with high lichen cover and it is only these
the nearest trees. areas where L. incognitus occurs, thus also the only
A multivariable linear regression model found a areas where the two geckos occur syntopically.
statistically significant difference between the two A multivariable linear regression comparing habitat
Lygodactylus species on the factor scores after selection of L. incognitus between sites where
accounting for season, F 290 = 36.4, P ˂ 0.001, L. incognitus occurred with L. soutpansbergensis to
Wilks’ Λ = 0.7. Follow-up univariate regression tests sites where they occurred alone also found a statisti-
with a Bonferroni correction indicated that Compo- cally significant difference, F166 = 82.3, P = <0.001,
nents 1, 2 and 3 showed statistically significant Wilks’ Λ = 0.39. Follow-up univariate regression tests

Fig. 2. Images depicting typical microhabitat for Lygodactylus incognitus

© 2020 Ecological Society of Australia doi:10.1111/aec.12857

6 M . A . P E T F O R D ET AL.

Fig. 3. Images depicting typical microhabitat for Lygodactylus soutpansbergensis

Table 5. Component loadings of a factor analyses on the DISCUSSION

microhabitat selection of Lygodactylus incognitus and Lygo-
dactylus soutpansbergensis after a varimax orthogonal rota- Lygodactylus incognitus and L. soutpansbergensis are syn-
tion. Loadings greater than 0.35 are in bold to highlight the
topic, but separate their niche by utilising different
variables that are the most significant for each component
microhabitats, as shown by the highly significant dif-
Rotated factor matrix† ferences in the FA loadings. Whilst both species can
be found in Soutpansberg Summit Sourveld and
Factor Soutpansberg Mountain Bushveld, only L. incognitus
was found in Mist-belt Bush Clumps and Northern
Habitat variables 1 2 3
Mist-belt Forest and only L. soutpansbergensis was
Rock structure 0.922 0.090 0.184 found in Arid Mountain Bushveld and Open Savanna
Ground cover of trees 0.884 0.021 0.332 Sandveld. Additionally, L. incognitus was only found
Exposed rock 0.806 0.002 0.290 above 1100 m a.s.l. whilst L. soutpansbergensis was
Location 0.058 0.970 0.070 found from the lowest to highest elevations of the west-
Elevation 0.007 0.944 0.033 ern Soutpansberg; the latter was also found on the hot
Lichen cover 0.200 0.386 0.302
and arid northern slopes. On a finer resolution,
Canopy cover 0.421 0.004 0.760
Distance to nearest tree 0.436 0.028 0.735 L. incognitus occurs in microhabitats with high lichen
Leaf litter 0.244 0.050 0.453 cover, often higher tree cover and is generally associ-
Wood debris 0.015 0.029 0.389 ated with moist microclimates whilst L. soutpansber-
gensis occurs in drier microclimates and sites with
Extraction Method: Maximum Likelihood. Rotation more exposed rock. The two species also differ in their
Method: Varimax with Kaiser Normalisation.
† perch height, with L. soutpansbergensis generally staying
Rotation converged in 4 iterations.
much closer to the ground than L. incognitus. Dis-
turbed areas appear to be unsuitable for both these
with Bonferroni correction found that all components species. In areas where loose rocks had been cleared
were statistically different with P < 0.001. This sug- for agricultural purposes, no L. soutpansbergensis was
gests that L. incognitus might exhibit niche displace- found, indicating that the availability of suitable rocky
ment on all Components. The areas where structures for L. soutpansbergensis appears to be an
L. incognitus occurs without the presence of L. sout- important determinant of whether this species will
pansbergensis were in Mist-belt Forest and Mist-belt occur in an area. Neither species was recorded in silivi-
Bush Clumps, habitat types with high canopy cover culture areas, despite them occurring in adjacent areas.
and high moisture levels which L. soutpansbergensis Although L. incognitus was found on buildings, these
does not occur in. structures were surrounded by natural habitat.
doi:10.1111/aec.12857 © 2020 Ecological Society of Australia
 NICHE PARTITIONING IN TWO SYNTOPIC GECKOS
Lygodactylus soutpansbergensis was recorded on small
to medium loose rocks and boulders as well as rocky
outcrops at low perch heights, whilst L. incognitus
was often recorded utilising trees and tall rocky struc-
tures at higher perch levels. The fact that L. incogni-
tus was found utilising trees is notable as prior to this
analysis, L. incognitus was considered exclusively saxi-
colous (Jacobsen 1992; Kirchhoff et al. 2010). This
species was frequently observed utilising trees for
shelter, basking and mating, including in areas with
sparse rock cover and high canopy cover such as
Northern Mist-belt Forest.
Although L. incognitus can be found in areas with
high canopy cover, they can also be found in more
open areas such as in Soutpansberg Summit Sour-
veld. Lygodactylus soutpansbergensis does not occur in
areas with high canopy cover, even in areas with
exposed rocks, and this is most likely due to physio-
logical constraints. Areas with high tree density and
canopy cover reduce sunlight penetration, resulting
in cooler environments, with high elevation forests
being particularly cold (Chen et al. 1999; Pringle
et al. 2003; Huang et al. 2014). As L. soutpansbergen-
sis has shown an affinity for warmer areas, it is possi-
ble that L. soutpansbergensis requires a higher Tb for
efficient physiological process and consequently areas
with high canopy cover may be thermally unsuitable
(Huang et al. 2014).
Microhabitat separation between syntopic lizards
has been shown to be influenced by interspecific
aggression, or competition, for several different coex-
isting species (Hess & Losos 1991; Langkilde &
Shine 2004; Langkilde et al. 2005; Lisi
cic et al.
2012). There was evidence of interspecific aggression
during the study with L. incognitus appearing to be
the dominant species, chasing L. soutpansbergensis off
rocks and showing aggression through display.
Despite this, L. soutpansbergensis was observed going
back to the same rock multiple times after having
been chased away.
If interspecific aggression was forcing L. soutpans-
bergensis to occupy a different spatial niche, it would
be expected that L. soutpansbergensis would occupy a
different microhabitat in areas where they occur with
L. incognitus. Analysis on habitat selection of L. sout-
pansbergensis in sites where they occur with L. incogni-
tus and alone showed that there was no significant
difference on Components 1 and 3. Whilst there was
a significant difference on Component 2, this Com-
ponent is indicative of high elevations and high lichen
cover and as L. incognitus is restricted to high eleva-
tion areas, it is only in these locations that the two
species are frequently found together. Thus, L. sout-
pansbergensis used similar microhabitats and occupied
the same perch heights regardless of whether
L. incognitus was also recorded at the site. This indi-
cates that although interspecific aggression was
© 2020 Ecological Society of Australia
7
present, this was not significantly influencing the
habitat selection of L. soutpansbergensis.
Lygodactylus incognitus showed differences in habitat
selection on all components between sites where
L. soutpansbergensis occurred and sites where L. incog-
nitus was found alone. This is unlikely to be due to
interspecific aggression and is more likely to be due
to the differences in habitat selection of the two spe-
cies. Only L. incognitus was found in Mist-belt Forest
and Mist-belt Bush Clumps, habitats which have
high density of trees, high canopy cover and moist
microclimates; features which L. soutpansbergensis
appeared to avoid. As a result of this, it appears that
interspecific aggression is not driving the microhabi-
tat selection of these two Lygodactylus species. We
hypothesise that it is rather their physiological and
morphological constraints that are influencing their
microhabitat preferences.
As L. incognitus is restricted to the cooler, high ele-
vations and selects wetter microhabitats than L. sout-
pansbergensis, there are likely differences between the
two species physiology, which currently remain
untested. For example, L. incognitus may select a
lower body temperature (Tb) for physiological func-
tion or have a greater evaporative water loss than
L. soutpansbergensis. Microhabitat selection is com-
monly associated with thermoregulatory requirements
in reptiles (Huey et al. 1989; Adolph 1990; Huey
1991; Piantoni et al. 2019), and relationships between
evaporative water loss and habitat use have been iden-
tified in the past (Snyder 1979; Neilson 2002; Stein-
berg et al. 2007). Of particular interest is that species
living at high and moist elevations have been shown to
have higher evaporative water loss than their counter-
parts occupying more arid areas (Snyder 1979; Stein-
berg et al. 2007). Thus, investigation of rates of
evaporative water loss and Tb of the two Lygodactylus
species may provide some insightful results.
The fact that L. incognitus is restricted to high ele-
vations and moist microclimates means that climate
change could be a threat to this species. The moist
microclimates of the mountain are most likely sus-
tained by the mist-belt, which occurs at high eleva-
tions. Current global climate change projections
suggest that the frequency of the mist may be
reduced in the future, in addition to predicted rising
temperatures (Still et al. 1999; Sanford et al. 2014).
This may negatively affect this species by reducing
the number of suitable microhabitats, particularly if
L. incognitus does have a high evaporative water loss
and low thermal tolerance as hypothesised.
The two species displayed differences in their levels
of arborealism; one potential explanation for this
could be due to the scansors and toe length (Russell
1979). Although statistical verification of these differ-
ences is still required, visual inspection indicates that
L. incognitus has longer toes than L. soutpansbergensis
doi:10.1111/aec.12857
8 M . A . P E T F O R D ET AL.
as well as different scansor arrangements and claw
lengths. Claw and scansor morphology have been
shown to differ between arboreal and terrestrial
lizards (Russell & Bauer 1989; Tulli et al. 2009;
Crandell et al. 2014; Russell & Delaugerre 2017).
Arboreal lizards generally have relatively larger toe
pads and longer claws that are positioned higher on
the toe than their terrestrial counterparts (Elstrott &
Irschick 2003; Macrini et al. 2003; Tulli et al. 2009;
Crandell et al. 2014). Differences in toe morphology
between the two species may also explain why
L. soutpansbergensis is not found on the walls of build-
ings as they may not possess the necessary foot struc-
ture to hold on in this microhabitat.
One disadvantage to the analyses conducted here is
that differences in detectability between the different
habitat types were not accounted for, potentially pro-
ducing sampling bias and affecting overall interpreta-
tion of results (Ruiz-Gutierrez & Zipkin 2011).
Nevertheless, survey methods were robust, intensive
and were repeated multiple times across all habitat and
microhabitat types. In addition, habitats that did not
produce one of the species was intensively targeted to
ensure that the species was absent. Furthermore, as
the Lygodactylus are heliotherms, they are relatively
easy to locate, as they are often in sunny patches in the
early morning and late afternoon. Because of this, we
consider that differences in detectability would not
have introduced a significant bias to the study.
The Soutpansberg Mountains are extremely
heterogeneous, with diverse habitat types (Hahn
2006), complex rocky areas and three main ridges
(Hahn 2011), all creating many unique microhabi-
tats. As heterogeneous landscapes create many
microhabitats, this allows reptiles with different phys-
iological and morphological requirements to occur in
an area, facilitating high diversity in reptile communi-
ties (Pianka 1967; Tews et al. 2004; Alexander 2009;
Price et al. 2010). This in turn also facilitates niche
separation. The heterogeneous nature of the Sout-
pansberg appears to be allowing the two Lygodactylus
species to coexist, despite their seemingly different
requirements.
The findings of this study suggest that anthro-
pogenic activities may affect the conservation status
of L. incognitus and L. soutpansbergensis in the future.
Neither species were recorded in plantations, sug-
gesting that these areas are unsuitable. Therefore,
expansion of silviculture would have a negative
impact on both species. As L. soutpansbergensis is reli-
ant on specific rock structures, the removal of rocks
for agricultural purposes and building could be a
threat to this species (Shine et al. 1998; Bates et al.
2014). Although there is an abundance of loose rocks
in mountainous environments, sustained and com-
plete removal of these rocks in an area could lead to
local extirpation.
doi:10.1111/aec.12857
CONCLUSIONS
This study suggests that interspecific aggression and
competition is not driving the differences in spatial
separation between L. incognitus and L. soutpansber-
gensis. Instead, it is hypothesised that their morpho-
logical and physiological tolerances are influencing
their microhabitat selection. The data gathered in
this study indicate that niche separation of organisms
is facilitated by complex interactions with the envi-
ronment. Ultimately, it is likely the heterogeneous
nature of the Soutpansberg that allows these two spe-
cies to coexist in syntopy. Identifying ecological
requirements for endemic and restricted animals is
extremely important to ensure that conservation
strategies in the area are effective for these species.
ACKNOWLEDGEMENTS
We thank Jabu Linden from the Lajuma Research
Centre for his valuable suggestions on initial project
design and feedback on the first draft. We also would
like to thank Ryan van Huyssteen for his expertise
and assistance with fieldwork and critical comments
on the manuscript. We are grateful to all land owners
across the Soutpansberg who allowed access onto
their property and for their hospitality, particularly to
Ian Gaigher, Marietjie Underhay and Hannes Under-
hay. Finally, we thank the anonymous reviewers for
their constructive and helpful comments, all of which
greatly improved this manuscript.
AUTHOR CONTRIBUTIONS
Conceptualization: MAP, GJA, Data curation, MAP,
Formal analysis, MAP, GJA, Investigation, MAP,
Methodology, MAP, GJA; Supervision, GJA, Project
administration, MAP, Validation, MAP, GJA, Visual-
ization, MAP, GJA, Writing-original draft, MAP,
GJA, Writing-review & editing, MAP, GJA.
REFERENCES
Adolph S. C. (1990) Influence of behavioural thermoregulation
on microhabitat use by two Sceloporus lizards. Ecology 71,
315–27.
Akani G. C., Capizzi D. & Luiselli L. (2002) Community
ecology of scincid lizards in a swamp rainforest of south-
eastern Nigeria. Russ. J. Herpetol. 9, 125–34.
Alexander G. J. (2009) An Evaluation of Expected Impacts of the
Proposed Chapudi Coal Mine on the Herpetofauna. School of
Animal, Plant and Environmental Sciences University of
the Witwatersrand, South Africa.
Bates M. F., Branch W. R., Bauer A. M. et al. (eds). (2014)
(CD set). Atlas and Red List of the Reptiles of South Africa,
Lesotho and Swaziland. Suricata 1. South African National
Biodiversity Institute, Pretoria, South Africa.
© 2020 Ecological Society of Australia
 NICHE PARTITIONING IN TWO SYNTOPIC GECKOS
Bay L. K., Jones G. P. & McCormick M. I. (2001) Habitat
selection and aggression as determinants of spatial
segregation among damselfish on a coral reef. Coral Reefs
20, 289–98.
Chen J., Saunders S. C., Crow T. R. et al. (1999)
Microclimate in forest ecosystem and landscape ecology:
variations in local climate can be used to monitor and
compare the effects of different management regimes.
Bioscience 49, 288–97.
Crandell K. E., Herrel A., Sasa M., Losos J. B. & Autumn K.
(2014) Stick or grip? Co-evolution of adhesive toepads and
claws in Anolis lizards. Zoology 117, 363–9.
Elstrott J. & Irschick D. J. (2003) Evolutionary correlations
among morphology, habitat use and clinging performance
in Caribbean Anolis lizards. Biol. J. Linn. Soc. 83,
389–98.


Ferreira D., Zagar A. & Santos X. (2017) Uncovering the rules
of (reptile) species coexistence in transition zones between
bioregions. Salamandra 53, 193–200.
Filippi E. & Luiselli L. (2006) Changes in community
composition, habitats and abundance of snakes over 10 +
years in a protected area in Italy: conservation
implications. J. Herpetol. 16, 29–36.
Hahn N. (2006) Floristic diversity of the Soutpansberg, Limpopo
Province, South Africa. Unpublished Doctoral Thesis.
University of Pretoria.
Hahn N. (2011) Refinement of the Soutpansberg geomorphic
province, Limpopo, South Africa. T. Roy. Soc. S. Afr. 66,
32–40.
Hamilton T. H. (1962) Species relationships and adaptations
for sympatry in the avian genus Vireo. Condor 64, 40–68.
Heller H. C. (1971) Altitudinal Zonation of Chipmunks
(Eutamias): interspecific aggression. Ecology 51, 312–9.
Hess N. E. & Losos J. B. (1991) Interspecific aggression
between Anolis cristatellus and A. gundlachi: comparison of
sympatric and allopatric populations. J. Herpetol. 25,
256–9.
Huang S. P., Porter W. P., Tu M. C. & Chiou C. R. (2014)
Forest cover reduces thermally suitable habitats and affects
responses to a warmer climate predicted in a high-
elevation lizard. Oecologia 175, 25–35.
Huey R. B. (1991) Physiological consequences of habitat
selection. Am. Nat. 137, 91–115.
Huey R. B. & Pianka E. R. (1977) Patterns of niche overlap
among broadly sympatric versus narrowly sympatric
Kalahari lizards (Scincidae: Mabuya). Ecology 58, 119–128.
Huey R. B., Pianka E. R., Egan M. E. & Coons L. W. (1974)
Ecological shifts in sympatry; Kalahari fossorial lizards
(Typhlosurus). Ecology 55, 304–16.
Huey R. B., Peterson C. R., Arnold S. J. & Porter W. P.
(1989) Hot rocks and not-so-hot rocks: retreat-site
selection by garter snakes and its thermal consequences.
Ecology 70, 931–44.
Jacobsen N. H. G. (1992) New Lygodactylus taxa (Reptillia:
Gekkonidae) from the Transvaal. Bonn. Zool. Beitr. 43,
527–42.
Jacobsen N. H. G. (1994) A new subspecies of Lygodactylus
ocellatus (Roux) (Lacertilia: Gekkonidae) from the
Soutpansberg, South Africa. Afr. Zool. 108, 231–6.
Kabanda T. A. (2003) Climate. In: A first synthesis of the
Environmental, Biological & Cultural assets of the
Soutpansberg. Workshop held at Lajuma Mountain Retreat 9–
10 May 2003 (eds K. Berger, J. E. Crafford, I. Gaigher,
M. J. Gaigher, N. Hahn & I.A.W. MacDonald). pp. 9–12.
Leach Printers & Signs, Louis Trichardt.
© 2020 Ecological Society of Australia
9
Kirchhoff S., Kr€amer M., Linden J. & Richter K. (2010) The
reptile species assemblage of the Soutpansberg (Limpopo
Province, South Africa) and its characteristics. Salamandra
46, 147–66.
Kitchen A. M., Gese E. M. & Schauster E. R. (1999) Resource
partitioning between coyotes and swift foxes: space, time,
and diet. Can. J. Zool. 77, 1645–56.
Langkilde T. & Shine R. (2004) Competing for crevices:
interspecific conflict influences retreat-site selection in
montane lizards. Oecologia 140, 684–91.
Langkilde T., Lance V. A. & Shine R. (2005) Ecological
consequences of agonistic interactions in lizards. Ecology
86, 1650–9.
Lisi
cic D., Drakulic S., Herrel A., Dikic D., Benkovic V. &
Tadic Z. (2012) Effect of competition on habitat utilisation
in two temperate climate gecko species. Ecol. Res. 27,
551–60.
Losos J. B. (1990) The evolution of form and function:
morphology and locomotor performance in West Indian
Anolis lizards. Evolution 44, 1189–203.
Luiselli L. (2008) Community ecology of African reptiles:
historical perspective and a meta-analysis using null
models. Afr. J. Ecol. 46, 384–94.
Macrini T. E., Irschick D. J. & Losos J. B. (2003)
Ecomorphological differences in toepad characteristics
between mainland and island Anoles. J. Herpetol. 37,
52–58.
Melville J. (2002) Competition and character displacement in
two species of scincid lizards. Ecol. Lett. 5, 386–93.
Mostert T. H. C. (2006) Vegetation ecology of the Soutpansberg
and Blouberg area in the Limpopo Province. Unpublished
Doctoral Thesis, University of Pretoria.
Mostert T. H., Bredenkamp G. J., Klopper H. L., Verwey C.,
Mostert R. E. & Hahn N. (2008) Major vegetation types
of the Soutpansberg conservancy and the Blouberg nature
reserve, South Africa. Koedoe 50, 32–48.
Mucina L. & Rutherford M. C. (2006) The vegetation of
southern Africa, Lesotho and Swaziland. Strelitzia 19.
South African National Biodiversity Institute, Pretoria.
Myrberg A. A. & Thresher R. E. (1974) Interspecific
aggression and its relevance to the concept of territoriality
in reef fishes. Am. Zool. 14, 81–96.
Neilson K. A. (2002) Evaporative water loss as a restriction on
habitat use in endangered New Zealand endemic skinks. J.
Herpetol. 36, 342–9.
Noble T., Bunbury N., Kaiser-Bunbury C. N. & Bell D. J.
(2010) Ecology and coexistence of two endemic day gecko
(Phelsuma) species in Seychelles native palm forest. J. Zool.
283, 73–80.
Nordberg E. J. & Schwarzkopf L. (2019) Reduced competition
may allow generalist species to benefit from habitat
homogenization. J. Appl. Ecol. 56, 305–18.
Pianka E. R. (1967) On lizard species diversity: North
American flatland deserts. Ecology 48, 333–51.
Pianka E. R. (1973) The structure of lizard communities.
Annu. Rev. Ecol. Evol. S. 4, 53–74.
Pianka E. R. (1986) Ecology and Natural History of Desert
Lizards: Analyses of the Ecological Niche and Community
Structure. Princeton University Press, Princeton, New Jersey.
Piantoni C., Curcio F. F., Ibarg€ uengoytıa N. R. & Navas C. A.
(2019) Implications of climate change on the habitat shifts
of tropical lizards. Austral Ecol. 44, 1174–86.
Price B., Kutt A. S. & McAlpine C. A. (2010) The importance
of fine-scale savanna heterogeneity for reptiles and small
mammals. Biol. Conserv. 143, 2504–13.
doi:10.1111/aec.12857
10 M . A . P E T F O R D ET AL.
Pringle R. M., Webb J. K. & Shine R. (2003) Canopy
structure, microclimate, and habitat selection by a
nocturnal snake, Hoplocephalus bungaroides. Ecology 84,
2668–79.
Quirt K. C., Blouin-Demers G., Howes B. J. & Lougheed S.
C. (2006) Microhabitat selection of Five-Lined Skinks in
northern peripheral populations. J. Herpetol. 40,
335–42.
Rivas L. R. (1964) A reinterpretation of the concepts
“sympatric” and “allopatric” with proposal of the
additional terms “syntopic” and “allotopic”. Syst. Zool. 13,
42–43.
Robinson S. K. & Terborgh J. (1995) Animal interspecific
aggression and habitat selection by Amazonian birds.
Ecology 64, 1–11.
Ruiz-Gutierrez V. & Zipkin E. F. (2011) Detection biases yield
misleading patterns of species persistence and colonization
in fragmented landscapes. Ecosphere 2, 1–14.
Russell A. P. (1979) Parallelism and integrated design in the
foot structure of Gekkonine and Diplodactyline Geckos.
Copeia 1979, 1–21.
Russell A. P. & Bauer A. M. (1989) The morphology of the
digits of the golden gecko, Calodactylodes aureus and its
implications for the occupation of rupicolous habitats.
Amphibia-Reptilia 10, 125–140.
Russell A. P. & Delaugerre M. J. (2017) Left in the dust:
differential effectiveness of the two alternative adhesive pad
configurations in geckos (Reptilia: Gekkota). J. Zool. 301,
61–68.
Sanford T., Frumhoff P. C., Luers A. & Gulledge J. (2014)
The climate policy narrative for a dangerously warming
world. Nat. Clim. Change 4, 164–66.
Shine R., Webb J. K., Fitzgerald M. & Sumner J. (1998) The
impact of bush-rock removal on an endangered snake
doi:10.1111/aec.12857
species, Hoplocephalus bungaroides (Serpentes: Elapidae).
Wildlife Res. 25, 285–95.
Snyder J. K. (1979) Respiratory metabolism and evaporative
water loss in a small tropical lizard. J. Comp. Physiol. 104,
13–18.
Starkweather J. (2010) Principal components analysis vs. factor
analysis. . . and appropriate alternatives. Available in
original form at Benchmarks: https://it.unt.edu/sites/defa
ult/files/pcavsfavsaa_jds_july2010.pdf
Steinberg D. S., Powell S. D., Powell R., Parmerlee J. S. Jr &
Henderson R. W. (2007) Population densities, water-loss
rates, and diets of Sphaerodactylus vincenti on St. Vincent,
West Indies. J. Herpetol. 41, 330–6.
Still C. J., Foster P. N. & Schneider S. H. (1999) Simulating
the effects of climate change on tropical montane cloud
forests. Nature 398, 608–10.
Tews J., Brose U., Grimm V. et al. (2004) Animal species
diversity driven by habitat heterogeneity/diversity: the
importance of keystone structures. J. Biogeogr. 31, 79–92.
Toft C. A. (1985) Resource partitioning in amphibians and
reptiles. Copeia 1, 1–21.
Triska M. D., Craig M. D., Stokes V. L., Pech R. P. & Hobbs
R. J. (2017) Conserving reptiles within a multiple-use
landscape: determining habitat affiliations of reptile
communities in the northern jarrah forest of south-western
Australia. Aust. J. Zool. 65, 21–32.
Tulli M. J., Cruz F. B., Herrel A., Vanhooydonck B. & Abdala
V. (2009) The interplay between claw morphology and
microhabitat use in neotropical iguanian lizards. Zoology
112, 379–92.


Zagar A., Sim
ci
c T., Carretero M. A. & Vrezec A. (2015) The
role of metabolism in understanding the altitudinal
segregation pattern of two potentially interacting lizards.
Comp. Biochem. Physiol. A 179, 1–6.
© 2020 Ecological Society of Australia