Tests of hypotheses for the evolution of thermal physiology often rely on mean temperatures, but mounting evidence suggests geographic variation in temperature extremes is also an important predictor of species’ thermal tolerances. Although the tropics are less thermally variable than higher latitude regions, rain shadows on the leeward sides of mountains can experience greater diel and seasonal variation in temperature than windward sites. Rain shadows provide opportunities to test predictions about the relationships of extreme temperatures with thermal physiology while controlling for latitude. We tested the hypothesis that populations of leaf-cutting ants (Atta cephalotes) in leeward, montane, and windward sites in Costa Rica would differ in upper thermal tolerances (CT_max) of workers. As predicted from rain shadow effects via extreme high temperatures, the leeward rain-shadow site yielded the highest mean CT_max (rain shadow site 42.1±0.3 °C, Montane site 38.2±0.5 °C, windward site 38.2±0.3 °C). This suggests that high-temperature extremes in tropical rain shadow forests can select for higher thermal tolerances. CT_max increased with worker body size within sites, but CT_max increased with body size more gradually at the two lowland sites, as predicted if local high temperatures selected more strongly on the most thermally vulnerable society members (small workers). This suggests that warmer lowland climates selected for colonies with less variation in heat tolerance than cooler high elevation climates.

Ant collections

We studied thermal tolerance of the leaf cutting ant Atta cephalotes across its elevational range extremes in Costa Rica, sampling high elevations near the continental divide and lowlands on both the Pacific rain shadow (leeward) and Caribbean less variable (windward) slopes of the Tilarán mountain range. All field work was performed during the rainy seasons of 2014 and 2016 in premontane wet forests of Monteverde (N10.31°, W84.31° 1296-1311 m asl), lowland rainforests of La Selva Biological Station (N10.42°, W84.02° 50-58 m asl), and seasonal dry forests of Santa Sosa National Park (N10.85°, W85.64° 297-300 m asl). Although we searched the continental divide near Monteverde up to over 1800 m above sea level (masl), A. cephalotes was only common at and below circa 1300 masl, suggesting that our sampling was representative of the elevational extremes for this species in the region. Within each site, active and mature nests of Atta cephalotes at least 500 m linear distance apart from one another were selected. Twenty worker ants across the range of body sizes we encountered were collected from the tops of the nest mounds and transported (within 2 hours of collection) to the lab to use as subjects in thermal tolerance assays. Three colonies were sampled in Santa Rosa and La Selva, and two colonies were sampled in Monteverde. Voucher specimens from each colony were deposited in the entomology collection of the Academy of Natural Sciences of Drexel University, Philadelphia, PA, U.S.A.

Ambient temperature measurements

Daily maximum, minimum, and mean temperatures were calculated from temperatures recorded round-the-clock in each site using iButton thermochron data loggers (Maxim Integrated, San Jose, CA). Temperature sampling was concurrent with CT_max sampling of the ants (rainy seasons of 2014 and 2016). Each temperature logger was placed in a plastic housing and beneath a plastic sheet (to shield the probe from direct insolation) and inserted beneath the leaf litter on the forest floor. Loggers recorded surface soil temperature once every 5 minutes for three continuous days. Loggers were arranged in transects of three, spaced 5 m apart from one another and 3 m into the forest alongside dirt trails at each site. A total of 8 such transects were deployed in Santa Rosa, 9 in Monteverde, and 2 in La Selva. Both raw ibutton temperature data files and summaries of these temperature data showing daily temperature range, means and standard devition have been uploaded to this digital repository. We also obtained bioclim extrapolations of local temperatures at each site. These extrapolated climate data are reported in the master file alonside CT_max and head width. 

Thermal tolerance assays

We ran standard dynamic heat tolerance assays using digital dry heat blocks (USA Scientific). Ants were placed in 1.5 ml microcentrifuge tubes stoppered with cotton at the top to prevent access to thermal refuge in the tubes during the assay. Tubes were placed in the dry heat block and started at 30ºC, increasing in temperature at a rate of 1ºC every 10 minutes. Every 10 minutes we checked for loss of mobility response to light tapping on the tube. For each ant lack of motor response to tapping was interpreted as having surpassed the critical thermal maximum (CT_max). For each colony, five ants across the range of caste sizes were kept in tubes but not put into heat blocks for the duration of the assays, to ensure that factors other than increasing temperature did not account for loss of mobility. None of the control ants in this study lost mobility, suggesting that the cause of loss of mobility was due to temperatures experienced within the heat blocks rather than handling and confinement.

Body size

After performing CT_max assays, we collected all ant workers and measured body sizes from head capsules. Each ant was photographed at 15X to 40X magnification (depending on size of ant) using a scope-mounted digital camera. ImageJ software was used to measure the width of heads at eye-height from photographs. A micrometer accurate to 0.1 mm was used to convert head width measurements from pixels to millimeters. Head width was used as a proxy for body size across all analyses. Both CT_max and head width data are reported for each assayed ant in these uploaded files.

Most headers have been named intuitively, with abbreviations such as "Avg." meaning "average" and "SD" indicating "standard deviation".  All temperatures and CT_max values are reported in degrees celcius. Below is a list of less intuitive headers within these files and their exact meaning.

Downloaded bioclim data based on nest coordinates:

BC_MAT - mean annual temperature (BIO₁)

BC_MDRngT - Mean Diurnal Range (Mean of monthly (max temp - min temp)) (BIO₂)

BC_TARng - temperature annual range (BIO₇)

BC_Aprecip - Annual precipitation (BIO₁₂)

BC_PrecipSnlty - Precipitation seasonality (Coefficient of Variation) (BIO₁₅)

BC_MaxTWM - Maximum temperature of warmest month (BIO₅)

BC_MinTCM - Minimum temperature of coldest month (BIO₆)

Ibutton-based data in master CTmax file:

ibutton_Min - average minimum daily temperature for this site

ibutton_Max - average maximum daily temperature for this site

ibutton_Avg - average daily temperature for this site

Warming tolerances (discussed only in supplementary material):

Night_Warming_Tolerance - individual ant CTmax minus average daily minimum temperature (ibutton data)

Day_Warming_Tolerance - individual ant CTmax minus average daily maximum temperature (ibutton data)

Mean_Warming_Tolerance - individual ant CTmax minus average mean daily temperature (ibutton data)