Physiological thermal responses of three Mexican snakes with distinct lifestyles

Study species and sites

Crotalus polystictus is an endemic rattlesnake from central Mexico (Mociño-Deloya et al., 2009). It is a diurnal, medium-sized viper with an average snout-vent length (SVL) of about 60 to 70 cm (Campbell & Lamar, 2004). It inhabits pine-oak forests, scrublands, and both dry and humid grasslands between 1,450 and 2,739 m above sea level (Meik et al., 2012). This terrestrial species spends considerable time under rocks, and it alternates between shelters and the open for thermoregulation. It feeds mostly on small rodents as a sit-and-wait predator (Mackessy et al., 2018). The study site is located in San Bartolo Morelos, on the northwest side of Estado de México. The site is composed of a series of croplands surrounded by grasslands and oak forest patches, and it is located at an elevation of 2,660 m. Climate is subhumid temperate with summer rains. Rocky areas between croplands provide suitable shelters for the rattlesnakes.

Conopsis lineata is a small-sized fossorial snake with a mean SVL of around 18 cm. This diurnal species spends most of its time within burrows or under rocks and logs, where it feeds on larvae, insects, and other arthropods (Ramírez-Bautista et al., 2009; García-Balderas et al., 2014). It inhabits pine-oak forests, montane cloud forests, xerophytic shrubs, and fir forests in an altitude range from 1,700 to 3,100 m above sea level (Goyenechea & Flores-Villela, 2006; Ramírez-Bautista et al., 2009). The study site is located in Los Dinamos National Park, which is part of the forested and montane area in the western part of Mexico City. Climate is subhumid semicold with annual rainfall of 900 to 1,300 mm and a mean annual temperature range of 9 °C to 15 °C (Kovács et al., 2021). The study site is an open area used for shepherding and plum tree cultivation, located at an elevation of 2,720 m.

Thamnophis melanogaster is a medium-sized, semiaquatic colubrid with an average SVL of about 40 cm (Manjarrez, Macías Garcia & Drummond, 2017). It is distributed widely throughout Mexico, and it inhabits riverbanks and lake shores (Ramírez-Bautista et al., 2009). It uses shelters on land, near aquatic bodies, and forages actively underwater, feeding on aquatic prey such as tadpoles, fish, and invertebrates (Manjarrez, García & Drummond, 2013). It is mostly a diurnal species but can also be active during particularly warm nights (Rossman, Ford & Seigel, 1996). The study site is the ecotourism park Arcos del Sitio, which is located within the Tepotzotlán mountain range in Estado de México, at an elevation of 2,280 m. Climate is subhumid temperate with summer rains, with an annual precipitation of 700–800 mm (Espinosa-Graciano & García-Collazo, 2017). The dominant vegetation in the area is crasicaule shrubland, heavily fragmented by shepherding areas. Gallery forest is present along the Arcos del Sitio River, where individuals of T. melanogaster are mostly found in areas with reduced vegetation cover.

Field work and snake captivity

We visited each site every 2 months from April 2021 to July 2023, excluding the winter period (December to February) when snakes are rarely found active. Furthermore, many squamate species present a decrease in metabolic rates during winter as part of the brumation process (Dubiner et al., 2023), which would have biased our measurements. For C. polystictus and T. melanogaster, each visit lasted 3 days, during which a team of 2–4 people searched for the snakes in their preferred microhabitats from 10:00 to 18:00 h. As the study site of C. lineata is considerably smaller, sampling occasions for this species were shorter, lasting 1 or 2 days, with a sampling schedule from 10:00 to 14:00 h. For each snake encountered, we registered the time of capture, body mass (g), SVL (mm), sex, and geographic coordinates. We determined sex by cloacal probing; however, to avoid hurting them, we did not use the technique on the small juveniles, whose size prevented the probe from entering easily. We identified pregnant females by palpation, which were not considered for the experiments. After capturing the snakes, we kept them inside cloth sacks and transported them to the laboratory. We planned our field trips and experiments so that all individuals spent a similar amount of time in captivity (between 5 and 8 days) before the onset of the experiments, which lasted between 7 and 10 days. On average, the time between capture and the end of the experiments was just over 2 weeks.

While in captivity, we maintained the snakes in either 40 × 30 × 30 cm terrariums (C. polystictus and C. lineata) or in 50 × 40 × 30 cm fish tanks conditioned with a dry area (T. melanogaster). We placed the enclosures in a room at ambient temperature with natural light, under a natural photoperiod. We conditioned all enclosures with cardboard boxes to serve as refuges and water bowls that we filled constantly. We placed heating mats under one extreme of each enclosure so that snakes could have the opportunity to thermoregulate. Using a digital timer, we set the heating mats to be on at a constant schedule of 09:00 to 17:00 h. We did not feed the snakes prior to the experiments so that digestion did not interfere with the tests, which is a safe procedure since snakes can spend extended periods of time without eating (McCue, Lillywhite & Beaupre, 2012). After the experiments were finished, we maintained the snakes in their enclosures and fed them once every 2 weeks until the subsequent field trip, during which we liberated the organisms at their respective study site. We fed C. lineata with crickets (Acheta domesticus), C. polystictus with mice (Mus musculus), and T. melanogaster with small fish (Chirostoma sp.).

Permission to collect the animals was given by the Secretary of Environment and Natural Resources (SEMARNAT) through the permits SGPA/DGVS/06935/21 and SPARN/DGVS/00316/23. Experimental procedures were made under the approval of the ethics committee of the Faculty of Higher Studies (FES) Iztacala of the National Autonomous University of Mexico (UNAM) through the permit CE/FESI/022023/1578.

Thermal tolerance

To avoid subjecting a large number of animals to stressful experimentation, we decided to perform the CT_min and CT_max trials with a subset of individuals. We utilized the snakes collected in 2023: 10 individuals of C. polystictus (six females and four males), eight of C. lineata (three females, two males, and three juveniles), and seven of T. melanogaster (three females and four males). We followed the most commonly used procedure to determine thermal tolerance in reptiles, which is to heat and cool the animals to the point of losing the righting response (Taylor et al., 2021). To avoid confounding negative effects, we made these experiments last after performance and RMR tests. To determine CT_min, we placed the snakes in a 50 × 40 × 50 cm plastic container, of which the bottom was covered with frozen refrigerant gel packs. With this procedure, snakes were cooled at a rate of 1 °C/min. After 15 min, using a pair of 30 cm metal tweezers, we started turning snakes on their backs every minute. At the moment in which the snakes were unable to right themselves, we terminated the experiment and recorded their body temperature with a contact thermometer (Fluke model 52-II, ± 0.1 °C), whose sensor was inserted 0.5–1 cm into the cloaca, and recorded it as the CT_min. To measure CT_max, we placed a 200 W heat lamp 50 cm above a plastic bucket (30 cm in diameter and 40 cm high), in which we placed the snakes. With these conditions, we were able to heat the snakes steadily at a rate of 1.5 °C/min. When the snakes recurred to panting as a cooling strategy, we started turning them on their backs every 30 s. When they were unable to right themselves up, we recorded that temperature as CT_max, in the same way as we did with CT_min. Right after each CT_min or CT_max test ended, we placed the snakes in direct contact with water at ambient temperature (1 cm deep) inside a 60 × 40 × 50 cm plastic container and allowed them to recover. To avoid subjecting the snakes to consecutive stressful experiments, we first performed the CT_min tests and let the snakes recover for 24 h before the CT_max tests. With these data, we determined TTB as the distance in °C between CT_max and CT_min (Stellatelli et al., 2022; Valdez Ovallez et al., 2022).

Thermal treatments preparation

We ran both the performance and RMR trials on a set of thermal treatments that, without exceeding the TTB limits, encompasses temperatures below, within, and over their set point temperature range (T_set). This range represents the target body temperatures that organisms attempt to achieve (Hertz, Huey & Stevenson, 1993), and we calculated it for the three species in another study focused on their thermal ecology (R. Figueroa-Huitron et al., 2023, unpublished data). We found that, while their overall thermoregulation strategies are different, their T_set and field body temperatures are similar. This allowed us to make a generalization to determine five ecologically relevant temperatures for the three species: 15 °C, 25 °C, 30 °C, 33 °C, and 36 °C. Approximately, 15 °C is the lowest body temperature we registered on the field, 25 °C is near the mean T_set, 30 °C is the upper limit of T_set, and 33 °C is the maximum body temperature registered in the field. Though it was not considered at first, we included the 36 °C treatment subsequently to observe the response of the snakes above voluntarily maintained body temperatures in natural conditions and to better adjust the thermal performance curves over a wide temperature range (Taylor et al., 2021).

Before the onset of each test, we set the snakes to acclimatize to the target temperature for at least 30 min. To achieve this, we put the individuals in hermetically closed chambers placed in a darkened incubator. Depending on the size of each individual, we used 50 ml centrifuge tubes or plastic containers of 300 or 900 ml, which allowed the snakes to adopt a coiled resting position but not move freely. In both experimental procedures, we randomized the sequence of thermal treatments for each individual so that none passed through the treatments in direct ascending or descending order. Since metabolic rates change during rest hours (Dubiner et al., 2023), all tests were performed between 9:00 and 18:00 h, within the observed activity period of the snakes in the field, so that our data accurately represent RMR. Individuals performed only one experimental trial per day to reduce possible confounding effects derived from acute stress.

Locomotor performance

We evaluated locomotor performance through swimming speed. We decided to use swimming speed over crawling speed because in terrestrial conditions snakes frequently adopt defensive behaviors, such as coiling and biting, which would interfere with the locomotion trials. Swimming forward is a reflex behavior of snakes in water; thereby, swimming speed is an adequate measurement to assess the locomotor capacity of both semiaquatic and terrestrial snakes (Aïdam, Michel & Bonnet, 2013). To carry out the swimming speed trials, we designed a rectangular lane made from wood, covered with impermeable resin. The lane was 350 cm long, 30 cm wide, and 40 cm high. Along the lane, we placed marks every 30 cm to serve as a reference for the determination of the distance covered by the snakes. For each thermal treatment, we filled the lane with water up to a level of 15 cm and either heated or cooled the water up to the targeted trial temperature using refrigerated gel packs and electric water heaters.

On each trial, we placed the snake at one end of the lane and followed it as it went down the lane. If a snake stopped, we stimulated it to continue swimming forward by gently touching it with a herpetological hook. We made sure that snakes advanced at least through five marks (150 cm) throughout the trials, which lasted a maximum of 2 min. We filmed trials from an overhead view with a digital video camera (Akaso EK7000) at the highest frame rate: 60 frames per second (fps). We analyzed the resulting videos in the video editing program Shotcut (Version 21.05.18), in which we determined the number of frames it took for the snakes to move through each 30 cm segment. We then divided each value by 60 to get time in seconds, and with that information, we calculated swimming speed for each segment. We chose the highest value as the maximum swimming speed at the given temperature (Table S1).

Resting metabolic rate

To measure RMR, we used the rate of CO₂, since it is a reliable measure of basic maintenance in a steady state (Holden et al., 2022). We used an open flow respirometry system, following the protocol of Plasman et al. (2020). Incurrent air was scrubbed with Drierite (calcium sulfate for water vapor) and Ascarite (sodium hydroxide for CO₂) so that dry, CO₂-free air was pumped into the chambers at a rate of 100 ml·min⁻¹. Water vapor was also removed from the excurrent air sample, and the proportion of CO₂ in the excurrent gas was analyzed using a Foxbox O₂/CO₂ analyzer (Sable Systems International). After the acclimation period, the trials ran over 30 min, unless more time was required for the CO₂ levels to stabilize. Due to logistical difficulties, we did not perform trials for the 33 °C treatment with the animals collected in 2023 (10 C. polystictus, eight C. lineata, and seven T. melanogaster). We utilized the Expedata software (Sable Systems International, North Las Vegas, NV, USA) to visualize the CO₂ curves and to make the calculation of the lowest average levels of CO₂ during 100 continuous seconds. From the data obtained, we calculated RMR in terms of rates of carbon dioxide production (VCO₂; ml CO₂/min) using the following equation (Lighton, 2008),

$$VC{O}_2=F{R}_i\left(F_{e}C{O}_2-F_{i}C{O}_2\right)$$where FR_i is the incurrent air flow rate in ml/min, and F_eCO₂ and F_iCO₂ are the fractional concentrations of CO₂ in the excurrent and incurrent gas samples, respectively. Finally, we calculated the thermal sensitivity through the Q₁₀ index. We calculated Q₁₀ both over the entire thermal interval of the experiments and between subsequent thermal treatments using the following equation:

$$Q_{10}={\left(RM{R}_2/RM{R}_1\right)}^{\left[10/\left(T_2-T_1\right)\right]}$$

Statistical analyses

To assess the thermal sensitivity of locomotor performance, we constructed TPCs. For each species, we utilized the average swimming speed of all individuals at each of the five thermal trials as the base data on which the curves were fitted. The boundaries of the curves, where the performance value is zero, were set at the mean CT_min and CT_max. Following common practice in the field (see Angilletta, 2006; Taylor et al., 2021), we fitted a set of curve equations and obtained their R² values and the Akaike information criterion adjusted for small samples (AICc, our sample size was 33 for C. polystictus, 31 for C. lineata and 38 for T. melanogaster). To fit the curves, we utilized the software TableCurve 2D (Systat Software Inc., version 5.01). From the pool of fitted curves, we filtered the ones with the highest R² (over 0.9) and selected the best curve according to the lowest AICc (Table S2) as their TPC. We then calculated maximum swimming speed (V_max), T_o, TSMs, and B₈₀ from the TPC of each species.

We analyzed the data on resting metabolic rates using generalized linear mixed models (GLMMs), which allow for study designs with repeated measures. We log₁₀ transformed RMR and body mass to normalize the data and account for the non-linearity of their relationship (Glazier, 2021). We set thermal treatment, body mass, and sex as fixed predictors and snake identification as the random factor. We tested seven models with combinations of these predictors, including the interaction of temperature and body mass (Table S3). The best model was selected based on the lowest AICc.

To account for body mass while testing for differences in RMR between species, we ran a GLM with species set as a fixed factor and tested the effect of the species × body mass interaction. We then calculated the estimated marginal means and performed marginal contrasts analysis. To test for differences in swimming speeds, direct RMR among species, and overall Q₁₀, we used the Kruskal-Wallis and Dunn tests. We tested normality with the Shapiro-Wilk test and homoscedasticity with the Bartlett test. Analyses were made in R version 4.3.2 (R Core Team, 2023) using the following packages: AICcmodavg (Mazerolle, 2023) for computing AICc values; FSA (Ogle et al., 2023) for performing the Kruskal-Wallis and Dunn tests; lme4 (Bates et al., 2015) for fitting GLMMs; and modelbased (Makowski et al., 2020) for estimating and comparing marginal means. We used ggplot2 (Wickham, 2016) to generate the figures.

Locomotor performance and thermal tolerance

We examined the effect of temperature on performance on 33 C. polystictus (12 males, 14 females, and seven unsexed juveniles), 31 C. lineata (16 males, 15 females), and 38 T. melanogaster (12 males, 15 females, and 11 juveniles). The curves that best fit the performance data were an exponentially modified Gaussian for C. polystictus (R² = 0.92, Table S2), an asymmetric logistic for C. lineata (R² = 0.94, Table S2), and an extreme value distribution for T. melanogaster (R² = 0.93, Table S2). The TPCs of the three species are represented in Fig. 1. Overall, T. melanogaster reached the highest swimming speeds, whereas C. lineata presented the lowest (X² = 176.02, P < 0.001). C. polystictus presents the highest T_o, followed by T. melanogaster and then C. lineata (Table 1). The TTB of C. polystictus is 37.2 °C, ranging from 5.9 °C to 43.1 °C (Table 1). The T_o (36.7 °C) and the upper limit of the B₈₀ (39.2 °C) of this species are close to the CT_max, which resulted in the lowest TSM among the three species (Fig. 1, Table 1). The TTB of C. lineata is 31.9 °C, ranging from 9.3 to 41.2 °C, whereas that of T. melanogaster is 33.1 °C, ranging from 8.5 °C to 41.6 °C (Table 1). These two species present similar values of T_o (27.8 °C, 28.6 °C) and B₈₀ (20.1–34.3 °C, 20.7–34.9 °C, Table 1).

Resting metabolic rate

We calculated the RMR of 41 C. polystictus (15 males, 19 females, and seven unsexed juveniles), 26 C. lineata (11 males, 13 females, and two juveniles), and 36 T. melanogaster (13 males, 15 females, and eight juveniles). The bigger species had higher absolute values of RMR, with C. polystictus presenting the highest values of RMR (X² = 149.18, P < 0.001; Figs. 2, 3; Table S4). However, after variable transformation and taking body mass into consideration, T. melanogaster had a higher estimated marginal mean (t = −2.56, P < 0.05). No significant differences were found between C. polystictus and C. lineata.

For all three species, the GLMM that best explained RMR included the additive effect of temperature and body mass (RMR~Temperature + Body mass. Conditional R² = 0.96 for C. polystictus, 0.85 for C. lineata, and 0.92 for T. melanogaster). The coefficients of the best models for the three species are presented in Table 2, whereas all tested models with their AICc values are presented in Table S3. In all three species, the thermal treatments affected RMR positively, and this effect increased along with temperature (Table 2). Body mass also had a positive, significant effect on RMR in the three species.

Conopsis lineata and T. melanogaster had very similar values of overall Q₁₀ (2.20, 2.11, Table 3), but C. polystictus had a significantly lower Q₁₀ than the other two species (1.62, X² = 28.447, P < 0.001; post-hoc Dunn test: P < 0.001 in both cases). The variation of Q₁₀ across the thermal treatments was also similar in C. lineata and T. melanogaster. In both species, Q₁₀ progressively decreased from 15 °C to 33 °C but increased abruptly in the 33 °C to 36 °C interval (Table 3). On the other hand, Q₁₀ of C. polystictus increased slightly from 15 °C to 33 °C but also experienced a noticeable increment between 33 °C and 36 °C (Table 3).