Altitude and ground brightness explain interpopulation variation in dorsal coloration in a lizard

Abstract

Non-signaling functions of coloration include thermoregulation (thermal melanism hypothesis), protection against ultraviolet radiation (photoprotection hypothesis), and concealment from predators (crypsis hypothesis). We investigated whether dorsal coloration in 19 populations of spiny-footed lizards, Acanthodactylus erythrurus, across the Iberian Peninsula varies according to these functions. We captured adult males and females in each population and calculated standardized dorsum brightness estimates from photographs. We also calculated standardized ground luminosity estimates and gathered information on latitude, altitude, mean annual temperature, and mean annual solar radiation for each location. Males showed a higher percentage of black coloration and a more contrasted dorsum than females, suggesting different selection pressures on dorsal coloration in both sexes. Furthermore, males showed a darker dorsum and a higher percentage of black coloration at higher altitudes and when the ground was darker. In contrast, females exhibited a darker dorsum only when the ground was darker and a higher percentage of black coloration only at higher altitudes. We also observed that the variation of dorsum luminosity within males and the variation of dorsum luminosity among females within populations were both positively related to the variation of ground luminosity among different points within locations. Latitude, temperature, and solar radiation were not significantly related to dorsal coloration in either sex. Our results support the photoprotection and crypsis hypotheses in males and, to some extent, in females, whereas the thermal melanism hypothesis is weakly supported in both sexes. These findings suggest that there is local adaptation in the dorsal coloration of the spiny-footed lizard.

Coloration in animals fulfills many different functions, broadly categorized as signaling and non-signaling (reviews in Caro 2005; Protas and Patel 2008; Cuthill et al. 2017). Signaling functions include intraspecific communication in the contexts of social selection, including sexual selection and parent–offspring communication (Lyon et al. 1994; Hill 2006; Senar 2006). Coloration also plays a role in interspecific communication in some species, with signals often directed to potential predators (Caro and Allen 2017). Among the non-signaling functions of coloration, the most important are thermoregulation, protection against ultraviolet (UV) radiation, and camouflage (Porter and Norris 1969; Merilaita et al. 2017; Stuart-Fox et al. 2017). This study specifically focuses on these 3 non-signaling functions.

The thermal melanism hypothesis suggests that dark skin coloration provides an advantage for thermoregulation under conditions of low temperature, particularly in ectotherms (Clusella-Trullas et al. 2007, 2008). Dark surfaces, in general, absorb more solar radiation than light surfaces, leading to faster heating. Animals with darker skin or integumentary tissues would reach active body temperature more rapidly, potentially reducing basking time (and decreasing predation risk) while increasing time available for fitness-related activities such as foraging or mate searching. Additionally, dark-skinned animals might achieve higher body temperatures, leading to enhanced performance. The thermal melanism hypothesis has found support in studies of invertebrates (Forsman 1997; Pereboom and Biesmeijer 2003; Ellers and Boggs 2004) and vertebrates (Bittner et al. 2002; Vences et al. 2002; Geen and Johnston 2014). However, studies not supporting the hypothesis or yielding mixed results are also common, possibly because other factors such as body size, convective heat loss, or thermoregulatory behaviors can potentially mask the effect of body coloration (Turner and Lombard 1990; Forsman 1995; Moreno Azócar et al. 2016). According to the thermal melanism hypothesis, species or populations in colder regions have evolved to exhibit darker coloration than those in warmer regions. Consequently, the hypothesis predicts that body brightness (reflectance) will be positively related to environmental temperature or solar radiation and negatively related to latitude or altitude. These relationships have been observed in some studies (Harris et al. 2013; Reguera et al. 2014; Martínez-Freiría et al. 2020). An alternative, nonexclusive interpretation of the thermal melanism hypothesis is that bright skin coloration may provide a thermoregulatory advantage under high temperatures, preventing overheating and increasing the time for fitness-related activities (Norris 1967; Ottenheim et al. 1999; Tattersall et al. 2006). This interpretation would lead to the same predictions, with species or populations in warmer regions evolving to exhibit brighter coloration than those in colder regions.

The photoprotection hypothesis (Law et al. 2020) proposes that certain pigments, such as melanin, offer protection against the harmful effects of UV radiation, a component of the solar electromagnetic spectrum. High levels of UV radiation have been shown to have detrimental effects on fitness components (Linaza and Pedraza 1998; Wang et al. 2008), and pigments would prevent these harmful effects by absorbing a portion of the radiation (Mosse and Lyakh 1994; Delmore et al. 2018). According to the photoprotection hypothesis, species or populations exposed to higher levels of UV radiation have evolved to show darker coloration than those experiencing lower levels. This hypothesis has been invoked to explain color variation in many different taxa (e.g., Cooper 2010; Jablonski and Chaplin 2010; Bastide et al. 2014; Katoh et al. 2018). Given that UV radiation increases with altitude (Blumthaler et al. 1997) and total solar radiation (Gandía et al. 2015; Aculinin et al. 2016), but decreases with latitude (Diffey 1991), the photoprotection hypothesis predicts that body brightness will be negatively related to altitude or solar radiation, but positively related to latitude. Previous studies have found a negative relationship between body brightness and altitude in both invertebrates (Ellers and Boggs 2002; de Souza et al. 2020; Stanbrook et al. 2021) and vertebrates (Reguera et al. 2014; Martínez-Freiría et al. 2020; González-Morales et al. 2021). However, it is important to note that a negative relationship between body brightness and altitude is also predicted by the thermal melanism hypothesis (see above), as environmental temperature generally decreases with altitude (Körner 2007).

The crypsis hypothesis, also known as the background-matching hypothesis, posits that animals exhibit colors or patterns that resemble those of the surrounding environment (Merilaita and Stevens 2011), making detection by other animals (e.g., visual predators) more challenging and improving survival (Stuart-Fox et al. 2003; Walton and Stevens 2018). Supporting this hypothesis, when the environmental background undergoes long-term changes, becoming darker or lighter, body coloration in some species has been observed to become also darker or lighter to match the environment (Cook 2003; Nachman et al. 2003; Hoekstra et al. 2004; Rosemblum 2006; Linnen et al. 2009). Therefore, a prediction of the hypothesis is that body brightness will be positively related to background brightness, for example, the brightness of bare soil in the case of terrestrial animals. Moreover, most habitats are heterogeneous, with patches of different colors, patterns, and textures, and the spatial scale of patches may vary, sometimes being larger and sometimes smaller than the size of the animals. Under some conditions, animal coloration needs to be patchy (variable within individuals) or polymorphic (variable among individuals) to be cryptic (Bond and Kamil 2006; Chiao et al. 2009; Akkaynak et al. 2017; Duarte et al. 2018). Generally, the diversity of color patterns in animals is expected to be proportional to the complexity of their background (Endler 1978). Quantifying this complexity can be challenging, but simple estimates of body color variation within and among individuals (e.g., brightness variation) can be assessed and related to background color variability at different spatial scales. Consequently, another prediction of the crypsis hypothesis is that body brightness variation will be positively related to background brightness variation.

We examined whether predictions from these 3 nonexclusive hypotheses (thermal melanism, photoprotection, and crypsis) were consistent with the interpopulation variation in the dorsal coloration of an ectothermic reptile species, the spiny-footed lizard, Acanthodactylus erythrurus. This species is widely distributed across western North Africa and the Iberian Peninsula (Salvador 2014), primarily inhabiting open habitats with scattered vegetation and loose soils (Belliure 2015). Their activities, including basking and foraging for prey (mainly insects), are almost entirely terrestrial, rarely climbing plants (Seva 1989). Common predators include snakes and birds (Salvador 2014). The dorsal pattern changes from strongly marked dark and light bands in hatchlings to a reticulated or spotted pattern of brown, black, and light gray in adults (Seva 1982; Salvador 2014). Adult dorsal coloration does not appear to change seasonally, but varies greatly among individuals (Figure 1).

Figure 1.

Adult male (upper row) and female (lower row) spiny-footed lizards displaying diverse dorsal color patterns. (A) Male from Granada (GR1), (B) male from Zaragoza, (C) male from La Rioja, (D) female from Navarra, (E) female from Sevilla, (F) female from Jaén.

According to the thermal melanism hypothesis, mean dorsal brightness of spiny-footed lizards would be positively related to mean annual temperature or mean annual solar radiation and negatively related to latitude or altitude (Table 1). In contrast, the photoprotection hypothesis predicts that mean dorsal brightness would be negatively related to altitude or mean annual solar radiation, positively related to latitude, and unrelated to mean annual temperature (Table 1). Finally, the crypsis hypothesis suggests that mean dorsal brightness would be positively related to mean ground brightness, whereas dorsal brightness variation within and among lizards would be positively related to ground brightness variation at different spatial scales (Table 1).

Table 1.

Predicted relationships (positive: +; negative: −; or unrelated: 0) between lizard color traits (mean dorsal brightness and dorsal brightness variation) and environmental or geographical variables (mean temperature, mean solar radiation, latitude, altitude, mean ground brightness, and ground brightness variation) according to the thermal melanism, photoprotection, and crypsis hypotheses explaining interpopulation color variation

Lizard color trait:	Mean dorsal brightness					Dorsal brightness variation
Environmental variable:	Temperature	Solar radiation	Latitude	Altitude	Ground brightness	Ground brightness variation
Thermal melanism hypothesis	+	+	−	−	0	0
Photoprotection hypothesis	0	−	+	−	0	0
Crypsis hypothesis	0	0	0	0	+	+

Materials and Methods

Study species

Spiny-footed lizards, Acanthodactylus erythrurus, are medium-sized lacertids, with a snout–vent length of up to approximately 80 mm and a total length of up to around 230 mm (Salvador 2014). In central Spain, spiny-footed lizards hibernate from November to March, emerge from hibernation in April, and start to mate in May (Pollo and Pérez-Mellado 1990; Castilla et al. 1992). Females lay eggs during June and July, with most eggs hatching in late August and September (Pollo and Pérez-Mellado 1990; Castilla et al. 1992). In southern Spain, reproductive events occur approximately 1 month earlier (Busack and Jaksić 1982; Seva 1982; Cuervo and Belliure 2013). This species reaches sexual maturity during their second spring and, at least in southern Spain, lives for about 2 years (Busack and Jaksić 1982).

Field procedures

We captured 350 adult spiny-footed lizards (193 males and 157 females) during the spring and early summer of 2010, 2011, and 2013 in 19 locations across Spain (Table 2 and Figure 2). These locations were selected to cover most of the species distribution range in Spain and to maximize variation in latitude, longitude, and altitude. We considered that each location corresponded to a different lizard population. As phenology varies with latitude, lizards were captured from April to June in southern Spain (locations with codes ALC, ALM, GR1, GR2, HUE, JAE, MAL, and SEV in Figure 2) and from May to July in central and northern Spain. Restricting captures to the first months of their annual activity period allowed us an easy distinction between juveniles (born the previous year) and adults because juveniles are much smaller than adults during this period. The minimum snout–vent length for adults was considered to be 6.0 cm. We were uncertain regarding the adulthood of 20 individuals (from 9 populations) and excluded these individuals from the study. Adult lizards were sexed based on the width of their tail base, which is much wider in males than in females (Blasco 1975). Our target was to capture 20 lizards (10 males and 10 females) in each location, but this goal could not be achieved in some cases (Table 2). However, certain locations were visited in 2 different years, enabling us to capture more than 20 lizards for some populations (range 14–30; Table 2). We were unable to capture any adult females in one of the locations (Zaragoza).

Table 2.

Information on the locations where spiny-footed lizards were captured (including code, municipality, province, longitude, and latitude), the number of male and female lizards captured in each location, disaggregated by year for locations with captures spanning 2 years (second figure for 2013), and the year (or years) of capture

Code	Municipality	Province	Longitude	Latitude	Males	Females	Year
ALB	Chinchilla de Montearagón	Albacete	01°28.9ʹW	38°53.7ʹN	9	9	2010
ALC	Santa Pola	Alicante	00°35.6ʹW	38°11.4ʹN	8	9	2010
ALM	Almería	Almería	02°18.4ʹW	36°50.0ʹN	15 (10, 5)	15 (10, 5)	2010, 2013
BAD	Albuquerque	Badajoz	07°02.4ʹW	39°15.9ʹN	8	9	2011
CUE	Sotorribas	Cuenca	02°11.1ʹW	40°11.6ʹN	6	9	2011
GR1	Monachil	Granada	03°31.4ʹW	37°05.1ʹN	13 (10, 3)	6 (6, 0)	2011, 2013
GR2	La Calahorra	Granada	03°01.6ʹW	37°15.7ʹN	10	10	2010
HUE	Almonte	Huelva	06°31.5ʹW	36°59.4ʹN	16 (10, 6)	13 (10, 3)	2010, 2013
JAE	Linares	Jaén	03°40.1ʹW	38°09.5ʹN	10	10	2011
LRJ	Alfaro	La Rioja	01°51.5ʹ W	42°12.4ʹN	10	5	2011
MAD	Chapinería	Madrid	04°13.6ʹW	40°22.7ʹN	10	10	2010
MAL	Marbella	Málaga	04°44.9ʹW	36 29.2ʹN	10	10	2010
NAV	Ablitas	Navarra	01°36.2ʹW	41°57.3ʹN	11 (7, 4)	8 (4, 4)	2011, 2013
SEV	San Nicolás del Puerto	Sevilla	05°37.0ʹW	37°57.8ʹN	10	5	2011
TAR	Torredembarra	Tarragona	01°26.1ʹE	41°09.3ʹN	9	5	2011
TER	Alcañiz	Teruel	00°14.4ʹW	41°03.8ʹN	10	4	2011
VLC	Valencia	Valencia	00°18.7ʹW	39°20.6ʹN	9	10	2010
VLD	Castronuño	Valladolid	05°15.8ʹW	41°24.4ʹN	5	10	2011
ZAR	Leciñena	Zaragoza	00°39.2ʹW	41°49.4ʹN	14 (10, 4)	0	2011, 2013

Figure 2.

Map of the Iberian Peninsula indicating the locations where spiny-footed lizards were captured. Codes correspond to province name: ALB (Albacete), ALC (Alicante), ALM (Almería), BAD (Badajoz), CUE (Cuenca), GR1 (Granada 1), GR2 (Granada 2), HUE (Huelva), JAE (Jaén), LRJ (La Rioja), MAD (Madrid), MAL (Málaga), NAV (Navarra), SEV (Sevilla), TAR (Tarragona), TER (Teruel), VLC (Valencia), VLD (Valladolid), ZAR (Zaragoza). See Table 2 for more information. The base map, which has been modified, is freely available from Trabajos de Prehistoria journal, Spanish National Research Council (http://tp.revistas.csic.es/public/journals/1/tp_mapa2010.jpg).

Lizards were captured by noosing and immediately placed in individual cloth bags in the shade to prevent overheating. The longitude and latitude of the exact point where each lizard was captured were recorded with a GPS. Subsequently, we took a picture of the ground at each of these points using a compact digital camera (Canon IXUS 70 by JB and Canon PowerShot A620 by J.J.C.), positioned perpendicular to the ground at a distance of around 30 cm (Supplementary Figure S1). These pictures included a reference chart (FT8 scanner reference chart 6611, FOTOWAND-Technic, Sudwalde, Germany) to allow subsequent standardization of light conditions (see Image analysis). In every location, we took 20 pictures of the ground at the points where the lizards were captured, generally where the lizards were first spotted or nearby. When we were not able to capture 20 lizards, the remaining pictures were taken at points where an adult lizard had been observed but not captured. When more than 20 lizards were captured, we photographed only the first 20 points to maintain a consistent representation of ground coloration across locations. Pictures of the ground were generally taken in the shade, with the observer providing shade if necessary, although in 3 locations they were taken in the sun (see Supplementary Table S1). We preferred taking ground pictures in the shade to ensure uniform luminosity and minimize variations in brightness caused by direct sunlight and shadows. Although the pictures would be standardized later, starting with similar pre-standardization conditions made the process more efficient. Direct sunlight introduces bright spots, reflections, and shadows, which can create unrealistic contrasts and affect the accuracy of ground brightness measurements. Shade reduces these issues, resulting in more homogeneous and comparable images across locations.

On the day of capture, the snout–vent length of each lizard was measured with a ruler to the nearest 0.1 cm. Two pictures of the dorsal part of each lizard were taken with a digital camera positioned perpendicular to the lizard at a distance of around 20 cm (Figure 1). These pictures included the same reference chart used for the ground pictures. All lizard pictures were taken in the shade with natural light. To prevent unnoticed recapture of the same individuals, we clipped the most distal phalanx of one toe, usually toe 20 (Tinkle 1967), using surgical scissors. Toe clipping was preferred over alternative marking methods because it is permanent and reliable, not particularly stressful (Langkilde and Shine 2006), and adverse effects on performance, at least for terrestrial lizards, do not seem to be important (Borges-Landáez and Shine 2003). All lizards were released less than 24 h after capture in exactly the same places where they had been captured. On release, all lizards were in good condition and exhibited normal behavior.

Image analysis

The light conditions of all ground and lizard pictures were standardized using Adobe Photoshop CS5 version 12.0 (Adobe Systems Inc., San Jose, CA). Standardization involved selecting and setting black, white, and gray points in the picture with the aid of the reference chart and adjusting the light parameters of the entire picture based on this setting. For a comprehensive and detailed description of the standardization process, see Supplementary Appendix S1 and Thomas (2016). All pictures were standardized by the same person (M.C.D.-G.) to prevent interobserver variability.

Ground brightness was estimated on standardized pictures using Adobe Photoshop. We selected the ground area (excluding the reference chart) with the rectangular marquee tool and checked the luminosity channel in the histogram. Luminosity, ranging from 0 (black) to 255 (white) with no units, was our brightness estimate, representing the mean luminosity of all pixels within the selected area. The program also provided the standard deviation (SD), allowing calculation of the coefficient of variation (CV) for ground luminosity within pictures (CV = SD / mean). The CV for ground luminosity among pictures within locations was also computed (see Supplementary Table S1). Similarly, lizard dorsum brightness was estimated by selecting the area from the neck to the beginning of the tail using the lasso tool, which allows freehand selection. Both the mean and SD of dorsum luminosity were provided, and the CV of dorsum luminosity within individuals was calculated. The CV of dorsum luminosity among individuals within populations was also calculated, but as sample size varied among populations, this CV was corrected for sample size (corrected CV = (1 + (1 / 4n)) CV; where n = sample size; Sokal and Rohlf 1995). We also estimated the percentage of black coloration in the lizard dorsum, a parameter that is less influenced by light conditions. After trimming the dorsum area, we determined the total number of pixels in the histogram window. Subsequently, using the magic wand tool, we selected all black areas and calculated the percentage of black coloration by dividing the number of black pixels by the total number of pixels in the dorsum, multiplying by 100.

As 2 pictures were taken for each lizard, we were able to calculate the repeatability of color parameters for a total of 186 males and 148 females. The figures are slightly smaller than the total number of lizards captured because one of the 2 pictures was unfocused and excluded from the study for 7 males and 9 females. Repeatability (Lessells and Boag 1987) was very high for dorsum luminosity (males: r = 0.923; females: r = 0.940) and percentage of black coloration (males: r = 0.956; females: r = 0.947), and high for the CV of dorsum luminosity within individuals (males: r = 0.796; females: r = 0.805; F ≥ 8.80, P < 0.0001 in all 6 tests). The average values of the 2 pictures from each lizard (or the single value for individuals with only one picture) were used in further analyses.

Parameters of interest and confounding variables

Mean longitude and latitude for all points where lizards had been captured in each location were considered the geographic coordinates of the location (Table 2). These geographic coordinates were used to estimate altitude above sea level using Google Earth (Google LLC, Mountain View, CA), as well as mean annual temperature and mean annual solar radiation using the Digital Climatic Atlas of the Iberian Peninsula (Ninyerola et al. 2005). Temperature values were derived from interpolated algorithms and temporal series from the nearest weather stations, whereas solar radiation referred to potential values estimated according to a digital elevation model and the position of the sun, without correcting the cloud effect (Ninyerola et al. 2007). We used mean annual temperature because the precise activity periods for most populations included in this study are unknown. However, mean annual temperature includes temperatures during hibernation, which are not relevant for lizard coloration. Therefore, we repeated all the analyses using mean spring–summer (April–September) temperature instead of mean annual temperature, assuming this is the period when lizards are active in most, if not all, studied populations. It is important to note that even mean spring–summer temperature would only provide a rough estimate of the air temperature when lizards are actually active, because this is a strictly diurnal species that seeks refuge in underground shelters not only when temperatures are too low but also too high (e.g., Busack 1976). In analyses including mean spring–summer temperature, we also included mean spring–summer (April–September) solar radiation. Longitude might also play a role in dorsal coloration in this species, because some geographic variation has been described in the Iberian Peninsula, with eastern populations showing paler and less marked dorsal patterns than western ones (Salvador 1982). Information on longitude, latitude, altitude, mean annual temperature, mean spring–summer temperature, mean annual solar radiation, and mean spring–summer solar radiation for each location can be found in Supplementary Table S1.

Statistical analysis

Possible relationships between the 2 estimates of lizard brightness (dorsum luminosity and percentage of black coloration) and potential predictors, such as latitude, altitude, mean annual temperature (or mean April–September temperature), mean annual solar radiation (or mean April–September solar radiation), and ground brightness, were tested using phylogenetic generalized least squares regression models (Martins and Hansen 1997; Pagel 1997, 1999) implemented in R statistical environment (R Core Team 2014). The function pglm3.3, developed by R. P. Freckleton (University of Sheffield, UK), and libraries “ape,” “MASS,” and “mvtnorm” were used for these analyses. The method employs a maximum likelihood modeling approach to estimate the phylogenetically corrected partial correlation between the variables of interest (Freckleton et al. 2002). Confounding variables such as longitude and the observer who took the pictures (J.B. or J.J.C.) were also included in the models. The observer variable was included because the 2 observers used different cameras and may have also selected different lighting conditions, given that shaded areas can vary considerably in brightness. Whether ground pictures were taken in the sun or in the shade was included only in models that also incorporated ground luminosity. To account for phylogenetic relationships among the 19 lizard populations, we built a phylogenetic tree based on genetic distances among 138 lizards from the same 19 populations (Harris et al. 2019). The branch lengths in the phylogenetic tree presented in Figure 2 of Harris et al. (2019) were measured with the program ImageJ (https://imagej.net/ij/), and these lengths were converted to genetic distances using the scale bar provided in the figure. Mean interpopulation genetic distance was then calculated for every pair of populations. Using the software PhyD* v1.1 (Criscuolo and Gascuel 2008) with algorithm BioNJ*, we converted mean genetic distances among the 19 populations into a binary phylogenetic tree (Supplementary Figure S2), which was included in the models as a design matrix. The optimum degree of phylogenetic dependence was identified for each model, and the corresponding lambda parameter (λ) was included in subsequent analyses. Our estimates of lizard brightness were derived from a different number of individuals for each lizard population (Table 2), so we performed weighted analyses using the number of individuals as a weight. Specifically, a matrix of 1/weight was added as an error term, and this term was multiplied by different values until the value providing the highest maximum likelihood was found. This method has been used and described in detail in previous studies (e.g., Garamszegi and Møller 2007). Akaike’s (1974) information criterion corrected for sample size (AICc) was used to select the best models explaining lizard brightness. The model with the lowest AICc was considered the most parsimonious and plausible (Burnham and Anderson 2002). However, models were not considered to differ in plausibility if they did not differ by at least 2 AICc units (Burnham and Anderson 2002). Therefore, among models differing by less than 2 AICc units from the one with the lowest AICc, the model with the fewest variables was chosen as the most parsimonious. Some geographic and environmental variables in the study were strongly correlated, particularly altitude with mean annual and mean April–September temperatures (Supplementary Table S2). Therefore, altitude and temperature were never included in the same model.

Possible relationships between the 2 estimates of lizard brightness variation (CV of dorsum luminosity within individuals and among individuals within populations) and potential predictors (CV of ground luminosity within pictures and among pictures within locations) were also tested with phylogenetic generalized least squares regression models. Confounding variables such as dorsum luminosity and the observer who took the pictures (J.B. or J.J.C.) were included in the models. Dorsum luminosity was included because of the inherent relationship between the CV and the mean. When analyzing the CV of dorsum luminosity among individuals within populations, we included the CV of dorsum luminosity within individuals as a confounding variable. Whether ground pictures were taken in the sun or in the shade was included only in models that also incorporated the CV of ground luminosity within or among pictures. Phylogenetic relationships among lizard populations and the number of individuals for each lizard population were also taken into account in these analyses. Similar to previous analyses, Akaike’s information criterion was used to select the best models explaining lizard brightness variation within and among individuals.

Possible sexual differences in dorsum luminosity, percentage of black coloration, and CV of dorsum luminosity within individuals were examined using phylogenetic generalized least squares regression models, but including in the analyses 336 individuals rather than population means. Male lizards from Zaragoza were excluded due to the unavailability of females from this population. For these analyses, we built a phylogenetic tree that included every individual lizard as a terminal branch and had equal branch length polytomies for individuals within populations. The genetic relationships and distances among the 18 lizard populations were identical to those used in the previously mentioned phylogenetic tree (Supplementary Figure S2), and the genetic distance among individuals within populations was arbitrarily set to 0.00001 (Supplementary Appendix S2).

Spatial autocorrelation is probably not an issue in this study because Moran’s I values for dorsum luminosity, percentage of black coloration, and CV of dorsum luminosity within and among individuals in both males and females, calculated with the software SAM v4.0 (Spatial Analysis in Macroecology; Rangel et al. 2010), were not statistically significant after sequential Bonferroni correction (Rice 1989; Chandler 1995) (Supplementary Table S3). Mean annual solar radiation was x²/10³-transformed and mean April–September solar radiation was x⁵/10¹²-transformed before further analyses to approach a normal distribution (Kolmogorov–Smirnov test; P > 0.05). All other continuous variables followed an approximately normal distribution (Kolmogorov–Smirnov test; P > 0.20 in all cases) and were analyzed untransformed.

Results

Male and female lizards did not differ significantly in dorsum luminosity (males mean (standard error (SE)) = 102.26 (1.22), females mean (SE) = 104.51 (1.48); estimate (SE) = −1.830 (1.226), t = −1.493, P = 0.136), but males showed a higher percentage of black coloration (males mean (SE) = 28.60 (0.88), females mean (SE) = 22.12 (0.82); estimate (SE) = 6.186 (1.027), t = 6.023, P < 0.0001) and a more contrasted dorsum (CV of dorsum luminosity within individuals; males mean (SE) = 53.74 (0.59), females mean (SE) = 50.89 (0.64); estimate (SE) = 2.861 (0.686), t = 4.173, P < 0.0001) than females (n = 179 males and 157 females in all cases).

Best models explaining dorsum luminosity included ground luminosity, the observer who took the pictures, and whether ground pictures were taken in the sun or in the shade for both sexes, and also altitude in the case of males (Supplementary Table S4). Regarding the percentage of black coloration, the best models included altitude for both sexes and also ground luminosity in the case of males (Supplementary Table S4). Males showed a darker dorsum and a higher percentage of black coloration at higher altitudes and when the ground was darker (Table 3, Figure 3). In contrast, females showed a darker dorsum only when the ground was darker (not at higher altitudes) and a higher percentage of black coloration only at higher altitudes (not when the ground was darker) (Table 3, Figure 4). Results were qualitatively identical, with the same best models obtained, when mean April–September temperature and mean April–September solar radiation were used instead of mean annual values (Supplementary Table S5).

Table 3.

Phylogenetic generalized least squares regression models with variables from the best models (determined using AICc, see main text and Supplementary Table S4) as independent variables and dorsum luminosity, percentage of black coloration in the dorsum, CV of dorsum luminosity within (intra) individuals, and CV of dorsum luminosity among (inter) individuals within populations in male and female spiny-footed lizards as dependent variables. The sample size (number of populations) is 19 for males and 18 for females

Sex	Dependent variable	Independent variables	Estimate (SE)	t	P
Males	Dorsum luminosity	Observer (J.J.C.)	−19.966 (3.795)	−5.261	0.0001
		Sun/shade (shade)	20.287 (5.005)	4.053	0.0012
		Altitude	−0.010 (0.004)	−2.644	0.0192
		Ground luminosity	0.255 (0.099)	2.576	0.0220
	Dorsum % black	Altitude	0.011 (0.003)	3.911	0.0012
		Ground luminosity	−0.158 (0.062)	−2.545	0.0216
	CV dorsum luminosity intra	CV ground luminosity inter	0.858 (0.141)	6.083	<0.0001
		Dorsum luminosity	−0.287 (0.044)	−6.474	<0.0001
	CV dorsum luminosity inter	Observer (J.J.C.)	−3.289 (1.669)	−1.970	0.0654
Females	Dorsum luminosity	Observer (J.J.C.)	−25.376 (5.301)	−4.787	0.0003
		Sun/shade (shade)	25.980 (6.935)	3.746	0.0022
		Ground luminosity	0.316 (0.147)	2.148	0.0497
	Dorsum % black	Altitude	0.008 (0.003)	2.729	0.0149
	CV dorsum luminosity intra	Dorsum luminosity	−0.227 (0.062)	−3.662	0.0021
	CV dorsum luminosity inter	CV ground luminosity inter	0.448 (0.168)	2.674	0.0166

The models had the statistics: dorsum luminosity in males, F = 15.748, adj-r² = 0.766, P < 0.0001, λ = 6.61 × 10⁻⁵; dorsum % black in males, F = 15.100, adj-r² = 0.610, P = 0.0002, λ = 6.61 × 10⁻⁵; CV dorsum luminosity (intra) in males, F = 42.351, adj-r² = 0.821, P < 0.0001, λ = 6.61 × 10⁻⁵; CV dorsum luminosity (inter) in males, F = 3.880, adj-r² = 0.138, P = 0.0654, λ = 6.61 × 10⁻⁵; dorsum luminosity in females, F = 11.466, adj-r² = 0.649, P = 0.0005, λ = 6.61 × 10⁻⁵; dorsum % black in females, F = 7.446, adj-r² = 0.275, P = 0.0149, λ = 6.61 × 10⁻⁵; CV dorsum luminosity (intra) in females, F = 13.407, adj-r² = 0.422, P = 0.0021, λ = 0.436; CV dorsum luminosity (inter) in females, F = 7.150, adj-r² = 0.266, P = 0.0166, λ = 6.61 × 10⁻⁵. The lambda parameter (λ) significantly differed from 1 (χ² ≥ 5.367, P ≤ 0.0205) but showed nonsignificant differences from 0 (χ² ≤ 0.572, P ≥ 0.4496) in all 8 models.

Figure 3.

Relationships between relative dorsum luminosity and (A) relative altitude or (B) relative ground luminosity, and between relative percentage of black coloration and (C) relative altitude or (D) relative ground luminosity in male spiny-footed lizards from 19 populations. Information on regressions and how relative estimates were calculated can be found in Supplementary Appendix S3. All models and regressions took into account similarities among populations due to common phylogenetic descent (see Statistical analysis section for details).

Figure 4.

Relationships between (A) relative dorsum luminosity and relative ground luminosity, and between (B) percentage of black coloration and altitude in female spiny-footed lizards from 18 populations. Information on regressions and how relative estimates were calculated can be found in Supplementary Appendix S3. All models and regressions took into account similarities among populations due to common phylogenetic descent (see Statistical analysis section for details).

The best models explaining the CV of dorsum luminosity within individuals included dorsum luminosity for both sexes and, in the case of males, also the CV of ground luminosity among pictures within locations (Supplementary Table S4). Regarding the CV of dorsum luminosity among individuals within populations, the best models included the observer who took the pictures for males and the CV of ground luminosity among pictures within locations for females (Supplementary Table S4). In locations with higher variation of ground luminosity among pictures (i.e., among different points within locations), males showed higher variation of dorsum luminosity within individuals (Table 3, Figure 5A) and females showed higher variation of dorsum luminosity among individuals (Table 3, Figure 5B).

Figure 5.

Relationships between (A) relative CV of dorsum luminosity within (intra) individuals and relative CV of ground luminosity among (inter) pictures within locations in male spiny-footed lizards from 19 populations, and between (B) the CV of dorsum luminosity among (inter) individuals within populations and the CV of ground luminosity among (inter) pictures within locations in female spiny-footed lizards from 18 populations. Information on regressions and how relative estimates were calculated can be found in Supplementary Appendix S3. All models and regressions took into account similarities among populations due to common phylogenetic descent (see Statistical analysis section for details).

It is worth noting that the degree of phylogenetic dependence (λ) was significantly different from one and not significantly different from zero in all the best models explaining dorsum brightness, percentage of black coloration, and dorsum brightness variation (Table 3). This suggests that controlling for the phylogenetic relationships among populations had very little, if any, influence on the results.

Discussion

Thermal melanism hypothesis

The relationships identified between dorsum luminosity or percentage of black coloration and altitude were the only results consistent with the thermal melanism hypothesis. These relationships were more evident in males, because in females the percentage of black coloration was related to altitude, but dorsum luminosity was not. Some studies that found a negative relationship between body brightness and altitude interpreted this as support for the thermal melanism hypothesis (e.g., de Souza et al., 2020; González-Morales et al., 2021). However, without additional evidence (e.g., a positive relationship with solar radiation or a negative relationship with latitude), the negative association between body luminosity and altitude could also support the photoprotection hypothesis. In our study, the observation that dorsum luminosity was negatively related to altitude, but nonsignificantly related to environmental temperature, is more consistent with the photoprotection than with the thermal melanism hypothesis. Therefore, our results provide limited support for the thermal melanism hypothesis in both sexes. However, some influence of thermoregulation on the evolution of dorsal coloration in this species cannot be entirely ruled out, as temperature may still play a minor role in the observed association between coloration and altitude.

An alternative explanation for our results could be that the study design or model was not optimally suited to detect a thermoregulation effect on dorsal coloration. The spiny-footed lizard is predominantly heliothermic (gaining heat by basking; Belliure et al. 1996; Belliure and Carrascal 2002), but environmental temperature might not be the main driver of evolutionary changes in the coloration of heliothermic reptiles. Previous studies found solar radiation and/or latitude to be better predictors of dorsal coloration than temperature in heliothermic reptiles (Clusella-Trullas et al. 2008; Martínez-Freiría et al. 2020). Moreover, the limited variation in latitude and solar radiation within our sample may not contribute substantially to color variation. Notably, studies showing a significant relationship between latitude or solar radiation and reptile coloration were conducted on a continental or global scale (Clusella-Trullas et al. 2008; Martínez-Freiría et al. 2020). The 5.7-degree difference in latitude (about 630 km) between the northernmost and southernmost populations in our sample (Supplementary Table S1) may limit the efficacy of testing the thermal melanism hypothesis, especially if the impact of thermoregulation on coloration is small. Expanding the latitudinal range by incorporating Moroccan populations could address this limitation, but these populations belong to different subspecies (Harris et al. 2004), introducing potential genetic heterogeneity.

We must also note that the mean population values used in our analyses were based on a small number of lizards in some cases (Table 2) and might not fully represent the variation within populations, given that intrapopulation variability is expected. Consequently, the lack of significant relationships between some parameters should be interpreted with caution. A larger sample size within each population would likely provide more robust and reliable results.

Photoprotection hypothesis

As explained above, the finding that dorsum luminosity in male spiny-footed lizards was negatively related (and the percentage of black coloration positively related) to altitude, but nonsignificantly related to environmental temperature, is more consistent with the photoprotection hypothesis than with the thermal melanism hypothesis. These results suggest that dorsal coloration in males has evolved as an adaptive response to local UV radiation conditions. Similar conclusions were drawn from studies on another lizard species in the Iberian Peninsula, where altitude effects on dorsal coloration also supported the photoprotection hypothesis over the thermal melanism hypothesis (Reguera et al. 2014). In contrast, our findings for females were less conclusive, providing only partial support for the photoprotection hypothesis, as the percentage of black coloration was significantly related to altitude, but dorsum luminosity was not. The reason for this inconsistency is unclear, but it is plausible that dorsum luminosity estimates were more influenced by light conditions or other confounding factors than the percentage of black coloration. This was in fact the initial reason for considering both color parameters. Supporting this idea, we found in both sexes that dorsum luminosity, but not percentage of black coloration, was associated with the observer who took the photographs and with whether they were taken in the sun or in the shade (Table 3).

The photoprotection hypothesis assumes that the relationship between dorsal coloration and altitude is mediated by variation in UV radiation. Although we did not measure UV radiation, the substantial altitudinal range across the studied localities (approximately 1,300 m; Supplementary Table S1) likely provided considerable variation in UV radiation (Sola et al. 2008; Reguera et al. 2014). Why then did we not observe a significant relationship between dorsal coloration and mean solar radiation? One explanation could be that we relied on potential values from a digital elevation model and sun position (Ninyerola et al. 2007) rather than actual solar radiation measurements. These potential values, while useful for incorporating topographic information, assume constant atmospheric conditions across the study area and do not account for cloud effects. Clouds can substantially alter the amount of UV and total radiation reaching the ground, with their impact being less pronounced on UV radiation (Calbó et al. 2005; Paulescu et al. 2010; Aculinin et al. 2016). Interestingly, water vapor and aerosols might even enhance UV radiation (Calbó et al. 2005). When the association between UV and total solar radiation is not strong, different results for these parameters may not indicate inconsistency.

Crypsis hypothesis

The primary prediction of the crypsis hypothesis was that both estimates of dorsum brightness would be related to ground luminosity. Our results fully support this prediction for males, but only partially for females, where dorsum luminosity was related to ground luminosity, but the percentage of black coloration was not. Overall, dorsum brightness appears to be associated with ground brightness across the studied lizard populations, probably enhancing lizard crypsis, and potentially reducing predation risk and improving fitness. This suggest that dorsal coloration in spiny-footed lizards has evolved, at least in part, as an adaptive response to predation pressure. Similar results supporting the crypsis hypothesis have been found in other reptiles (e.g., Rosemblum 2006; Sun et al. 2024) and other vertebrates (e.g., Nachman et al. 2003; Linnen et al. 2009).

Additionally, the crypsis hypothesis predicted a positive association between dorsum brightness variation and ground brightness variation. Our dataset enabled us to test this prediction at 2 spatial scales, corresponding to within-individual and among-individual brightness variation or, in other words, to patchiness (e.g., Chiao et al. 2009; Akkaynak et al. 2017) and polymorphism (e.g., Bond and Kamil 2006; Duarte et al. 2018). Although it was unclear at which levels camouflage might operate, an association within each spatial scale seemed more probable. We found that the variation of dorsum luminosity within males was positively related to the variation of ground luminosity among pictures (i.e., among different points within locations) but not within pictures. This result is consistent with the general prediction of higher dorsum luminosity variation where ground luminosity varies more, but does not fit well with the expected association within spatial scales. In contrast, the variation of dorsum luminosity among females was positively related to the variation of ground luminosity among pictures, supporting both the crypsis hypothesis and the expected within-scale association. Overall, these results support the crypsis hypothesis, suggesting that ground brightness variation influences male patchiness and female polymorphism. Moreover, they imply that dorsum luminosity variation is more strongly affected by ground brightness variation at larger spatial scales (meters to hectometers) than at smaller scales (millimeters to centimeters). For a more comprehensive understanding of the selection pressures related to camouflage in this species, additional estimates of dorsal color and pattern may be necessary (e.g., Stuart-Fox et al. 2004; Allen et al. 2020).

Sexual dichromatism

Male spiny-footed lizards exhibited a higher percentage of black coloration and a more contrasted dorsum than females, indicating that dorsum coloration is subject to different selection pressures in each sex. Although sexual dimorphism in behavior or activity patterns has not been extensively studied in this species, males appear more active and have larger home ranges than females during the reproductive season (Seva 1982). In other lizard species, males are also often more active than females (e.g., Cooper 2011; Childers and Eifler 2015). As a result, males might need to be more cryptic to reduce predation risk associated with increased activity (Merilaita and Stevens 2011). Additionally, if males are more active and spend less time in refuges, their dorsal coloration may need to better counteract potential UV damage, possibly through more melanin-based black coloration (Porter and Norris 1969). Our findings, specifically the stronger association between dorsum and ground luminosity and the higher percentage of black coloration in males compared with females, are consistent with expectations based on sexual dimorphism in activity patterns. Another plausible, non-mutually exclusive explanation for dorsal sexual dichromatism is that signaling to conspecifics, particularly in the context of sexual selection, may be more crucial for males than for females (Andersson 1994), which could require males to be more conspicuous. Future studies could investigate this hypothesis by focusing on the role of male dorsal conspicuousness in male–male competition for access to females or in female choice. Some predictions of the sexual selection hypothesis, such as weaker evidence for local adaptations aimed at improving camouflage in males compared with females, have already been corroborated by studies on other lizard species (Stuart-Fox et al. 2004).

Supplementary Material

Supplementary material can be found at https://academic.oup.com/cz.

Funding

This study was funded by the European Regional Development Fund (ERDF A way of making Europe) and the Spanish Ministry of Education and Science (grant CGL2008-00137/BOS).

Conflict of Interest

The authors declare that they have no conflict of interest.

Ethics Statement

This study was conducted following the ASAB/ABS (2006) and ASIH (2004) guidelines for the treatment of animals in behavioral research, and it complied with Spanish laws and the regulations of the 10 Spanish Autonomous Regions where lizards were captured.