Covariations between personality behaviors and metabolic/performance traits in an Asian agamid lizard (Phrynocephalus vlangalii)

Animal sampling

All lizards were adult males (average snout-vent length (SVL) = 66.58 ± 5.68 mm, N = 47) from the Zoige Wetland Nature Reserve, in southwestern China (Qi et al., 2012). The subjects were caught by hand or in pitfall traps in early July 2015. A total of 13 individuals were captured from site A (latitude, 33°42′57.7″N; longitude, 102°29′23.4″E; altitude, 3,471 m a.s.l.), and another 34 individuals were obtained from site B (latitude, 33°44′31.5″N; longitude, 102°30′15.9″E; altitude, 3,454 m a.s.l.). At both sites, the sandy substrate is similar but the vegetation density varies across sites (it was one-fold higher at site B). The vegetation is predominantly composed of Dracocephalum heterophyllum and Carex aridula, occasionally accompanied with Astagalus sutchenensis, Anaphalis lacteal, Vicia cracca, Morina kokonorica, Oxytropsis glabra, Linum stelleroides, and Clematis tangutica. After capture, all lizards were transferred to the Xiaman Conservation Station. Captured lizards were individually housed in card-boxes (40L × 30W × 20H cm) with sand substrate from the field. Water and mealworms were provided ad libitum. To approximate conditions in the field, we moved the boxed outside and basked the lizards when the weather allowed. All lizards were collected in 4 days. To minimizing the stress response of capturing, we housed all individuals for 5 days before we began the assessments. At the Xiaman Conservation Station in the Zoige Wetland Nature Reserve, we completed the performance measures in 2 days at first, then did the behavioral tests in another 2 days. Finally, all individuals were sent to the Chengdu Institute of Biology for 3-days metabolic tests. All animal procedures in this study were carried out in accordance with and approved by the Animal Care and Use Committee at the Chengdu Institute of Biology, Chinese Academy of Sciences (20151220) and the field experiments were approved by the Management Office of the Zoige Nature Reserve (20150701).

Morphology and performance measurement

As we wished to evaluate how body size covaries with behavioral, metabolic and performance traits, we measured the SVL to the nearest mm and body mass, to the nearest mg, of each lizard. Each lizard was marked with acrylic paint on the dorsum to facilitate identification. Due to the close link between performance and temperature in ectothermic species (Qi et al., 2014), we standardized temperature during physical testing. The subjects’ bite force was measured at room temperature (15 °C) by inducing each lizard to bite the free end of a steel bite-plate connected to an isometric force transducer (S1-100NHL, 0–100N; Nanjing Bio-inspired Intelligent Technology Co. Ltd., Nanjing, China). Bite forces were measured three times per individual, and the maximum value was used in further analysis (Lailvaux, Gilbert & Edwards, 2012). Second, the subjects were warmed with an infrared heater before measuring endurance and sprint speed in order to ensure the lizards were active and to standardize body temperature (22–35 °C). Detailed protocols for measuring sprint speed and endurance have been described previously (Qi et al., 2014). Briefly, lizards were stimulated to run by tapping the base of the tail with a paintbrush (Noble, Fanson & Whiting, 2014). All runs were recorded using a camera (HDC-HS60; Panasonic, Kadoma, Japan) and sprint speed was calculated as the fastest speed along a 1.5 m racetrack. To insure that the lizards experienced environmental conditions similar to those in the wild, the racetrack was covered with sand from the field. Endurance was measured immediately following after sprint speed tests. Lizards were placed in a circular racetrack and stimulated to run by tapping the base of the tail using a paintbrush. We considered the subject as exhausted if the lizard did not run after tapping the tail 10 times. Total movement time before exhaustion was regarded as a proxy of endurance. Sprint speed and endurance were measured once per day over 2 consecutive days. We used the maximum value for subsequent analyses, because lizard performance can be highly variable due to differences in motivation and other factors (Losos, Creer & Schulte, 2002).

Personality measurement

Two behavioral traits, exploration and risk-taking, were measured and used as proxies of the lizards’ personalities. Similar traits have been measured in many lizard species and have been shown to be consistent across different contexts (Cote & Clobert, 2007; Lopez et al., 2005; Stapley & Keogh, 2004). All lizards received risk-taking measurement after exploration tests.

Exploration

Lizards were individually introduced into a novel enclosure (60L × 44W × 37H cm) and were initially placed in a opaque plastic box in the center of the enclosure (Fig. 1A). Two polyvinyl chloride (PVC) refuge burrows (10L × 4R cm) were set at each end of the enclosure; these served as the goal of lizard exploration. After a 2 min acclimation period, the plastic box was lifted, exposing the lizard to the novel enclosure. Over 20 min we scored the following aspects of the subject’s behavior: (i) time in locomotion and (ii) latency to enter each of the two PVC refuges in the enclosure (time to enter refuge). The time to enter refuge was taken as measurement of quickness to explore a novel environment, and time in locomotion was taken as a measurement of exploration intensity.

Metabolic capacity measurement

Minimum metabolic rate represent the maintenance costs from metabolically active organs such as gut, intestines, and liver (Biro & Stamps, 2010). The minimum metabolic rate at a specified temperature is usually called standard metabolic rate (SMR) for ectotherm, and basal metabolic rate for endotherm is named when it is measured in thermal neutral zone. These maintenance costs may profoundly affect individuals’ behaviors. Individuals with a higher metabolic rate could potentially afford a faster pace of life, because a higher metabolic rate allows them to mobilize the energy needed to express a high level of activity, a fast exploration, or high levels of aggressiveness. Minimum metabolic rate can be also considered as “competitor” because it monopolizes energy that would be otherwise available for other energy-demanding activities (Biro & Stamps, 2010; Careau et al., 2008). For these reasons, we used SMR as a proxy for energy metabolism to test the coadaptation hypothesis.

The oxygen volume consumed by animals provides a convenient and readily measured estimate of their metabolic rates. We measured lizard oxygen consumption using an open air flow respirometer from TSE Systems (Bad Homburg, Germany) comprised of a source of pressurized gas (14% O₂, 86% N₂, this oxygen pressure is equal to the oxygen concentration at their habitats), Low Pressure Regulator G261, eight Channel Gas Switcher G244, Sample Chambers, eight Channel Gas Flow Controller and Monitor G246, Differential Oxygen Analyzer S104 and CO₂ analyzer S157 with Gas Pump P651. After 2 days’ fasting, the lizard was placed individually in a 190 ml darkened chamber. To standardize the measuring condition, we controlled the airflow at 110 ml/min and set the temperature at 25 ± 0.5 °C. Oxygen concentration was measured for 300 s every hour and the measurement for each individual last 6 h, and the animals were weighed once at the end of the 6 h’s measurement. We calculated the oxygen consumption as the difference between the oxygen concentration in the ambient air and at the chamber exit. The equation was follows: $${\text{VO}}_{\text{2}}=\frac{\text{FR*}\left({\text{F}}_{\text{i}}{\text{O}}_{\text{2}}-{\text{F}}_{\text{e}}{\text{O}}_{\text{2}}\right)-\left({\text{F}}_{\text{e}}{\text{O}}_{\text{2}}{\text{*VCO}}_{\text{2}}\right)}{\text{1}-{\text{F}}_{\text{e}}{\text{O}}_{\text{2}}}$$

Where FR is the flow rate, F_iO₂ and F_eO₂ are the fractional concentration of O₂ entering and exiting the respirometry chamber, respectively, and VCO₂ is the rate of CO₂ production. The minimum consumption based on six time points was used to calculate the SMR (ml O₂/h).

Statistical analyses

All data were analyzed using R software version 3.2.2 (R Core Team, 2013) and SPSS 19.0 (IBM Corporation, New York, NY, USA). Five individuals of the 47 lizards were excluded from further analyses, because they were highly active and did not meet the SMR measurement assumption (Kristín & Gvoždík, 2012). A high proportion (33/42) of the remaining lizards did never entered either of the two refuges during the exploration test. For this reason, we regarded the time to enter refuge as a binomial variable, with lizards that entered the refuge assigned as “1,” and lizards that did not entered assigned as “0.” Similarly, a high proportion (25/42) of the lizards did not return to the basking site during the risk-taking test after the simulated predator attack. For this reason, we regarded risk-taking intensity as a binomial variable, with lizards that returned to the basking site assigned as “1,” and lizards that did not return to the basking site assigned as “0.”

All continuous variables were normalized (SMR and endurance were log-transformed and time in locomotion was square root-transformed) and standardized (mean = 0, SD = 1) to facilitate the following analyses.

We used the residual of the regression of performance against body temperature if we detected any significant correlation between them. We examined the correlations among behavioral traits (exploration and risk-taking), body size (body mass and SVL), metabolic rate and performances (sprint speed, endurance, and bite force) to verify the relationship between behavior, physiology, and performance predicted by the POLS. For the correlation analyses between time to enter refuge, risk-taking intensity, and the other variables, we used Spearman correlations since these variables are binominal. For the correlation tests between continuous variables, Pearson’s product-moment correlations were applied. If performance traits were significantly affected by body size (e.g., bite force), we used the residuals of the observed variable against SVL for the correlation test. All means are reported as mean ± 1 SE.