Study species

Spiny-footed lizards are medium-sized lacertids living in sandy, open habitats of Northern Africa and the Iberian Peninsula (Pérez-Mellado, 1998). The phenology of the reproductive cycle varies among populations, but in southern Spain, most spiny-footed lizards hibernate during November-February, emerge from hibernation in March, and begin to mate in April, females lay eggs during May-July, and most hatchings take place in August and September (Busack and Jaksić, 1982; Seva, 1982; Busack and Klosterman, 1987). There is no evidence of multiple clutches in populations in southern Spain (Busack and Klosterman, 1987), but a second clutch might be possible in other populations (Castilla, Barbadillo and Bauwens, 1992). Average clutch size is 4.4 eggs (range 1-8; Belliure, 2009). The species only lives for about two years (Busack and Jaksić, 1982) and reaches sexual maturity at the age of one year and a half (Belliure, 2009). This means that spiny-footed lizards have a single breeding season, and females probably produce only one clutch, in their lifetime. The mating system of the species is unknown, but most lacertids studied so far are promiscuous and mate multiply (Olsson and Madsen, 2001; Laloi et al., 2004; Salvador et al., 2008).

Field procedures

The study was carried out during February-May 2009 in Almería, south-eastern Spain. All individuals were captured in Cabo de Gata-Níjar Natural Park (36°49′08″-36°50′13″N, 2°16′59″-2°18′36″W), in open coastal shrubland with sandy soils and Mediterranean semi-arid climate conditions (average annual rainfall less than 200 mm). Lizards were captured with a noose at the end of a 2-m-long fishing pole and placed in individual cloth bags inside a cooler to avoid overheating. All animals were taken to the lab, where they were measured (snout-vent length (SVL) and total length with a ruler to the nearest 0.1 cm) and weighed (with an electronic balance to the nearest 0.1 g). For field observations, 37 females were caught during April and May (fig. 1), tail and hind-leg colour was quantified (see Colour Measurement below) immediately after the other measurements were taken, and the individuals were released less than 24 h after capture in exactly the same places where they had been captured. Lizards were not individually marked, so to decrease the risk of catching the same individual twice, during April and May we always looked for lizards in areas where none had previously been released. These areas were at least 200 m away from previous release points. Recaptures were highly improbable because home ranges of adult females are usually 500-600 m² (Seva, 1982). For the experimental study, individuals (31 males and 47 females) were caught during February and March, and were kept in captivity for at least two months (see Experiment in Captivity below). All females were more than 60 mm SVL (range 61-71 mm) and all males more than 70 mm SVL (range 71-77 mm), so they were considered to be adults. These thresholds strongly reduce the chance of considering immature individuals as mature, at least in spiny-footed lizard populations in southern Spain, where females reach sexual maturity at around 57-58 mm and males at around 60-61 mm (Busack and Jaksić, 1982; Seva, 1982). Males were easily distinguishable from females because the base of the tail is much wider in the male (Blasco, 1975).

Experiment in captivity

Individuals involved in the experiment were mostly captured during March (from the 6^th to the 26^th (fig. 1), although one male was captured on February 26^th), because at this time of the year spiny-footed lizards in this area are already active after hibernation but have not begun mating (Seva, 1982; Busack and Klosterman, 1987). Females were randomly assigned to one of three experimental groups. Sixteen females were assigned to a group in which one female and one male shared a terrarium where they could freely interact. Fifteen females were assigned to another group consisting of one female and one male in the same terrarium, but separated by a transparent methacrylate wall with holes, so that females could see and smell the male but physical contact, and thus copulation, was impossible. Holes in the methacrylate wall were made because chemical intraspecific communication might be important in this species (López and Martín, 2005). Finally, 16 females were assigned to a group consisting of two females sharing a terrarium in which they could freely interact. Females in the eight two-female pairs were individually identified by measuring their total length, which differed, on average, by 2.1 mm within pairs. Long and short females within pairs, however, did not differ significantly in SVL (mean (SE) SVL of short females = 66.3 (0.7) mm; mean (SE) SVL of long females = 65.9 (0.9) mm; paired Student’s t-test; $t_7=0.89$, $P=0.40$). Females in different experimental groups did not differ significantly in SVL, total length, body weight, weight relative to body size (an estimate of underlying energy reserves expressed as the residuals of weight regressed on SVL; hereafter condition) or capture date (ANOVAs; $F_{2,44}\le 2.29$, $P\ge 0.11$ in the five tests).

The experiment took place in outdoor terraria located at the Finca Experimental La Hoya, a facility of the Estación Experimental de Zonas Áridas (CSIC) in Almería city (36°50′38″N, 2°28′22″W), about 20 km from the area where lizards were captured. Weather conditions in both areas are very similar. All terraria were made of opaque white high-density polyethylene, were the same size (80 cm long × 60 cm wide × 42 cm high) and were placed in an open area, thus receiving direct sunlight most of the day. Although the area appeared to have identical conditions, terraria containing females from different experimental groups were intermixed to avoid any microclimatic differences among groups. The bottoms of the terraria were covered with a 3-5-cm-thick layer of sand, and every terrarium also contained hollow bricks and pieces of wood for refuge and thermoregulation. Lizards were fed live crickets dusted with calcium and vitamins three times per week and water was provided ad libitum. Colouration of all experimental lizards (males and females) was quantified (see Colour Measurement below) on March 21^st-27^th (fig. 1). In addition, female colour was measured about every two weeks on April 9^th, April 22^nd-23^rd, May 9^th, and May 23^rd (i.e., five measurement batches in total; fig. 1). Experimental lizards were released May 24^th-28^th in exactly the same places where they had been captured. Females at the time of release had already laid eggs or were close to lay. Fecundity (number of eggs), egg characteristics (colour: white or yellowish; turgidity: turgid or flaccid), and laying date were determined by checking females every second day during the egg-laying period. In the case of females that had not laid eggs at the time of release ($n=15$; female and male together: $n=8$; female and male separated: $n=4$; two females together: $n=3$), the number of eggs was estimated by gently palpating the abdomen, and laying date was considered the day after the release date. When the eggs did not seem to have been laid the same day they were found because they looked dry ($n=7$ clutches; female and male together: $n=2$; female and male separated: $n=3$; two females together: $n=2$), laying date was considered the day before the finding date. Number of eggs could not be determined for one female because eggs were not found. A sample of 38 eggs from 16 females from the 3 experimental groups (female and male together: 4 females, 14 eggs; female and male separated: 6 females, 12 eggs; two females together: 6 females, 12 eggs) was carefully examined by candling to look for the blastodisc, a circular spot with a reddish border (due to vascularisation) typically visible through the shell in fertilized eggs (Olsson and Shine, 1997).

Colour measurement

Colour was quantified in terms of reflectance, i.e., the ratio of the total amount of radiation reflected by a surface to the total amount of radiation incident on the surface. Reflectance spectra (325-700 nm) were recorded with an Ocean Optics USB2000 spectrophotometer (Ocean Optics, Inc., Dunedin, FL, USA) connected to a deuterium tungsten halogen light source (DT-MINI-2-GS) by a coaxial reflection probe (QR400-7-UV/BX), using SpectraSuite (Ocean Optics, Inc., Dunedin, FL, USA) software. Reflectance was always measured with the probe placed at a constant distance from the object at a 45° angle. The measurements were referenced to standard white (WS-1) and dark, which were calibrated before measuring each lizard. Reflectance was measured 12 times for each individual and batch of measurements, three on the rear part of the right hind leg, three on the rear part of the left hind leg, three on the ventral part of the tail around 1 cm from the cloaca, and three around 2.5 cm from the cloaca. The three measurements of each body part were taken in a row, and the probe was repositioned after each measurement. To our knowledge, this is the first time spiny-footed lizard colouration is studied using spectrophotometry.

Colour parameters were determined using a principal component analysis (PCA) on raw spectral values (Cuthill, 1999; Cuervo and Shine, 2007). This method is not free from criticism (e.g., Grill and Rush, 2000; Endler and Mielke, 2005), but was chosen because it makes no a priori assumptions about which aspects of the spectrum might be important. Only one PCA was done for both captive and wild females. The first two factors from the PCA explained 69.1% and 27.2% of the variance (96.3% in total) and were retained for subsequent analyses. The first factor (PC1) represented variation in brightness and the second factor (PC2) represented mainly variation in orange-red colouration (fig. 2). PC values were multiplied by −1 to render positive relationships between PC1 and brightness, and between PC2 and redness (Cuthill et al., 1999). Measurement error of the two PCs was small for the four body parts (repeatability according to Lessells and Boag, 1987; PC1: $0.595\le r\le 0.786$, $P<0.0001$; PC2: $0.769\le r\le 0.825$, $P<0.0001$), thus justifying the use of means for each body part and measurement batch. We admit, however, that repeatability estimates might have been lower if measurements had not been taken in a row but at longer intervals (e.g., hours). Colour estimates of different body parts were strongly and significantly correlated (Pearson correlations; when all measurement batches for the 84 females are included: PC1, $0.560\le r\le 0.715$, PC2, $0.643\le r\le 0.784$, $n=272$, $P<0.0001$ in the six tests; when only one measurement is included for each female: PC1, $0.502\le r\le 0.702$, PC2, $0.576\le r\le 0.728$, $n=84$, $P<0.0001$ in the six tests), so we calculated a single PC1 value and a single PC2 value for each individual and measurement batch (the mean of the four body parts) and used these in later analyses.

View Full Size

Loadings of the first two factors from a principal component analysis on reflectances measured at 1-nm intervals between 325 and 700 nm on the tail and hind-legs of spiny-footed lizards (Acanthodactylus erythrurus). The first factor (broken line) represents mainly variation in brightness and the second factor (solid line) in red colouration. Both loading and factor values were multiplied by −1 to render positive relationships between factors and colour traits (see the text).

Citation: Amphibia-Reptilia 34, 4 (2013) ; 10.1163/15685381-00002912

• Download Figure
• Download figure as PowerPoint slide

Statistical analyses

Most variables were approximately normally distributed (Kolmogorov-Smirnov tests; $P\ge 0.10$) and, consequently, parametric tests were used in these cases. The only exception was number of eggs, which we considered a discrete variable and was analyzed using non-parametric statistics (Kruskal-Wallis tests and Spearman correlations). Differences in PC1 and PC2 among measurement batches (i.e., temporal variation) and among experimental groups, and the relationships between colour parameters (PC1 and PC2) and body size and weight were tested using two-way ANCOVAs with repeated measures in one factor (measurement batch) followed by Tukey post-hoc tests, ANCOVAs or Pearson correlations. All variables and the interactions time × experiment, time × SVL, and time × weight were included in initial models, but terms that did not explain a significant amount of the variance ($P\ge 0.10$) were removed from final models following a backward stepwise procedure. Females captured in the wild during April and May were compared with females kept in captivity using ANOVAs and Tukey post-hoc tests as follows: colouration of captive females in the second measurement batch (April 9^th) was compared to wild females captured on April 3^rd-14^th ($n=12$), colouration in the third measurement batch (April 22^nd-23^rd) was compared to females captured on April 20^th-29^th ($n=11$), colouration in the fourth measurement batch (May 9^th) was compared to females captured on May 8^th-13^th ($n=7$) and colouration in the fifth measurement batch (May 23^rd) was compared to females captured on May 19^th-22^nd ($n=7$). In the experimental treatment consisting of two females sharing a terrarium, the change in colour of the two females of the same pair might not be completely independent if the terrarium per se had any influence on colour change. Consequently, to avoid any problem of pseudo-replication, one female of each pair was chosen randomly and only these eight females were included in subsequent analyses for this treatment. All statistical analyses were two-tailed with a significance level of 0.05, and performed with the Statistica (StatSoft, 2011) programme.

View Full Size

Average (SD) reflectance spectra of ventral tail and hind legs of spiny-footed lizards (Acanthodactylus erythrurus): (a) adult males ($n=31$) and (b) adult females ($n=47$) before the onset of reproduction (March 21^st-27^th), (c) adult females around egg-laying time ($n=47$, May 23^rd). Reflectances for each individual are means of tail and hind leg measurements. This figure is published in colour in the online version.

Citation: Amphibia-Reptilia 34, 4 (2013) ; 10.1163/15685381-00002912

• Download Figure
• Download figure as PowerPoint slide

Female colour change in captivity

Sexual dichromatism before the onset of reproduction was obvious to a human observer, as adult females had red tails and hind legs but adult males did not (fig. 3a and b). Females kept in captivity underwent a noticeable colour change during the breeding season, and in all cases, by the time of egg-laying, red had become light buff-grey (fig. 3c). Redness differed significantly among measurement batches (table 1), and was particularly reduced between late April and early May (fig. 4b), possibly coinciding with the beginning of egg formation. In fact, redness differed significantly among all measurement batches (Tukey post-hoc tests; $P\le 0.035$ in the nine tests) with the exception of the comparison between early and late May ($P=0.35$). This also means that redness temporarily increased between March and early April (fig. 4b). However, redness did not differ significantly among experimental groups, and all groups showed similar temporal variation in this colour (table 1, fig. 4b).

Table 1.

Two-way ANCOVAs with repeated measures in one factor (measurement batch) testing for temporal effects (differences among five measurement batches, the first one at the end of March, the last one at the end of May) and experimental treatment effects (differences among three experimental groups: female with a male, fertilization possible; female with a male, fertilization impossible; and two females) on PC1 (representing brightness) and PC2 (representing redness) in tail and hind legs of adult female spiny-footed lizards (Acanthodactylus erythrurus) while simultaneously controlling for body weight at the time of capture (March).

View Table

Brightness differed significantly among measurement batches and among experimental treatments, but measurement batch effect depended on the treatment (table 1). Specifically, brightness increased along the breeding season to reach a peak in early May (Tukey post-hoc tests; $P<0.001$ in the three treatments when comparing March and early May), but the subsequent decrease between early and late May was only statistically significant in females sharing the terrarium with a male and with no restriction for copulation ($P=0.0033$; in the other two treatments $P\ge 0.70$; fig. 4a). Differences among experimental groups were significant only in May, but exclusively between the two groups in which females shared the terrarium with a male (Tukey post-hoc tests; $P=0.0054$ in early May, $P<0.001$ in late May, $P\ge 0.52$ in all other comparisons). Females sharing the terrarium with a male and interacting freely showed lower brightness than females separated from males by a methacrylate wall (fig. 4a).

All females in the three experimental groups laid eggs or were unequivocally gravid at the end of May. The first eggs in captivity were laid on May 22^nd, and laying date did not differ significantly among experimental groups, either considering the 39 females (ANOVA; $F_{2,36}=0.56$, $P=0.57$) or only females that laid eggs before the time of release ($F_{2,24}=0.31$, $P=0.74$). All eggs ($n=25$) laid by females that had shared the terrarium with a male and interacted freely were white and turgid. Blastodiscs were found in 13 of 14 eggs (from 4 females) examined (93%). On the contrary, all eggs ($n=89$) laid by females in the other two groups were yellowish and flaccid, indicating they were unfertilized. No blastodiscs were found in any of the 24 eggs (from 12 females) examined. The number of eggs did not differ significantly among experimental groups, either considering all females for which clutch size was estimated (mean (SE) clutch size; female and male together: 3.87 (0.19), $n=15$; female and male separated: 4.07 (0.30), $n=15$; two females together: 3.50 (0.42), $n=8$; Kruskal-Wallis test; $H_2=2.02$, $n=38$, $P=0.36$) or only females that laid eggs before the time of release (female and male together: 3.57 (0.20), $n=7$; female and male separated: 4.09 (0.34), $n=11$; two females together: 3.50 (0.42), $n=8$; $H_2=2.41$, $n=26$, $P=0.30$).

Colour parameters measured in five batches from late March to late May were not significantly related to body size (SVL) at the time of capture (March), but we found significant interactions between measurement batch and weight at the time of capture for both brightness and redness (table 1). Regarding brightness, although the interaction between measurement batch and weight was statistically significant (table 1), analyses within each measurement batch showed non-significant relationships between brightness and weight in all five batches (ANCOVAs with experimental treatment as a factor; $F_{1,35}\le 3.74$, $P\ge 0.061$). Redness, however, was negatively related to weight in early and late April (Pearson correlations; $r\le -0.367$, $n=39$, $P\le 0.022$ in the two tests), but the relationship was not statistically significant in the other three batches ($-0.064\le r\le 0.010$, $n=39$, $P\ge 0.70$). These results suggest that female body weight before reproduction (March) predicted the timing of change in red colouration, even though weight was not significantly related to initial (late March) or final (late May) redness. Specifically, red colouration faded earlier in heavy females. Female body size, weight and condition before reproduction were negatively correlated with laying date when all 39 females were considered (Pearson correlations; $r\le -0.325$, $n=39$, $P\le 0.043$ in the three cases), but female body size was no longer significantly correlated to laying date when only females that laid eggs before the time of release were included in the analyses ($r=-0.234$, $n=27$, $P=0.24$; for the other two variables $r\le -0.422$, $n=27$, $P\le 0.028$). Female body size, weight and condition before reproduction were not significantly correlated with clutch size, either considering all females for which clutch size was estimated (Spearman correlations; $0.044\le r\le 0.201$, $n=38$, $P\ge 0.23$ in the three tests) or only females that laid eggs before the time of release ($-0.081\le r\le -0.036$, $n=26$, $P\ge 0.69$ in the three tests).

View Full Size

Mean (SE) (a) PC1 (representing brightness) and (b) PC2 (representing redness) in ventral tail and hind legs of female spiny-footed lizards (Acanthodactylus erythrurus) in five measurements from late March to late May. Solid symbols: females kept in captivity in the experiment (circles ●: with a male and fertilization possible ($n=16$); triangles ▲: with a male but fertilization impossible ($n=15$); squares ■: with a female ($n=8$)). Open circles ○: free-ranging females (these individuals were measured only once). Colouration was measured on March 21^st-27^th, April 9^th, April 22^nd-23^rd, May 9^th, and May 23^rd in captive females and on April 3^rd-14^th ($n=12$), April 20^th-29^th ($n=11$), May 8^th-13^th ($n=7$) and May 19^th-22^nd ($n=7$) in free-ranging females. The first copulation was observed on April 3^rd. Females began to lay eggs around May 8^th in the wild and May 22^nd in captivity (see the text for more details).

Citation: Amphibia-Reptilia 34, 4 (2013) ; 10.1163/15685381-00002912

• Download Figure
• Download figure as PowerPoint slide

The time (number of days) elapsed between colour measurement and laying date was correlated with colour parameters in late April, either considering the 39 females (Pearson correlations; brightness: $r=-0.357$, $n=39$, $P=0.026$; redness: $r=0.563$, $n=39$, $P<0.001$) or only females that laid eggs before the time of release (brightness: $r=-0.439$, $n=27$, $P=0.022$; redness: $r=0.686$, $n=27$, $P<0.001$). In contrast, these correlations were not statistically significant in the other measurement batches (all females: $-0.223\le r\le 0.227$, $n=39$, $P\ge 0.16$ in the eight tests; females that laid eggs before release: $-0.278\le r\le 0.379$, $n=27$, $P\ge 0.051$ in the eight tests). Specifically, at the time of red colour fading (late April), brighter/less red females were closer to egg-laying (i.e., more advanced in the reproductive cycle) than less bright/redder females. This suggests that colour changes occurred at certain phases of the female reproductive cycle.

Colour parameters before the onset of reproduction (March) were related to colour parameters at the time of egg-laying (late May) (Pearson correlations; brightness: $r=0.403$, $n=39$, $P=0.011$; redness: $r=0.363$, $n=39$, $P=0.023$), suggesting high intra-individual consistency of female colouration through time.

Female colour change in the wild

Intensive field observations begun at the beginning of March, but the first copulation in the wild was not observed until April 3^rd. The first female with evident signs of having laid eggs (loose folds of lateral and ventral skin) was observed on May 8^th. Brightness of adult females captured in the wild during April and May differed significantly in the four periods considered (ANOVA; $F_{3,33}=3.72$, $P=0.021$), although differences were statistically significant only between early and late April (Tukey post-hoc test; $P=0.045$) and between early April and late May ($P=0.034$; in the other four tests $P\ge 0.43$). This means that brightness increased during April and remained unchanged during the rest of the breeding season (fig. 4a). Redness also differed significantly in the four periods (ANOVA; $F_{3,33}=5.90$, $P=0.0024$), but differences were statistically significant only between early April and later periods (Tukey post-hoc tests; $P\le 0.045$ in the three tests; in the other three tests $P\ge 0.60$). In other words, redness decreased during April and remained unchanged until the end of May (fig. 4b).

Colour change during the breeding season in wild and captive female lizards showed similarities because in general both became lighter and less red (see fig. 4). When brightness was compared among the four groups of females (three experimental treatments in captivity and wild females), differences between wild females and experimental groups were significant only in late April (ANOVA; $F_{3,46}=8.95$, $P<0.001$; Tukey post-hoc tests; $P\le 0.016$ in the three comparisons) and late May (ANOVA; $F_{3,42}=25.69$, $P<0.001$; Tukey post-hoc tests; $P=0.16$ when wild females were compared to captive females sharing the terrarium with a male but separated by a methacrylate wall, $P<0.001$ in the other two comparisons). In these two periods, captive females generally showed lower brightness, indicating that colour lightening was more pronounced and happened earlier in wild females. Redness did not differ significantly among the three groups in captivity (table 1), so they were pooled together and compared to wild females. Captive females showed lower redness in early and late May (ANOVA; $F_{1,44}\ge 9.81$, $P<0.001$ in the two tests), but not in early or late April ($F_{1,44}\le 0.33$, $P\ge 0.57$ in the two tests), implying that reduction of redness was less pronounced in wild females.