The fragmentation and transformation of land cover modify the microclimate of ecosystems. These changes have the potential to modify the foraging activity of animals, but few studies have examined this topic. In this paper, we investigated whether and how the foraging activity of the leafcutter ant Atta cephalotes is modified by microclimatic variations due to land cover change from forest to pasture. We characterized the microclimate of each habitat and identified alterations in foraging behavior in response to relative humidity (RH), air temperature and surface temperature along ant foraging trails by synchronously assessing foraging activity (number of ants per 5 mins including incoming laden and unladen and outgoing ants) and microclimatic variables (air temperature, RH, and maximum and minimum surface temperature along the foraging trail). There were climatic differences between habitats during the day but not throughout the night, and A. cephalotes was found to have a high tolerance for foraging under severe microclimatic changes. This species can forage at surface temperatures between 17 and 45 °C, air temperatures between 20 and 36 °C, and an RH between 40 and 100%. We found a positive effect of temperature on the foraging activity of A. cephalotes in the pasture, where the species displayed thermophilic behavior and the ability to forage across a wide range of temperatures and RH. These results provide a mechanism to partially explain why A. cephalotes becomes highly prolific as anthropogenic disturbances increase and why it has turned into a key player of human-modified neotropical landscapes.

Study site

This study was carried out in a 13.4-ha remnant of tropical dry forest (Holdridge et al. 1971, GEMA 1998) and its surrounding matrix of managed cattle pasture at the Alejandría Farm (4°51´ North, 75°52´ West), a private property located in the Cauca Valley (Risaralda, Colombia) between the Central and Oriental Andean Cordilleras at 900-1000 m asl. The site receives ~1700 mm of rainfall annually (8-year mean, 1993-2000) concentrated in two rainy seasons (March-May and September-November). Sunrise is at 5:00 hours and sunset at 17:00 hours throughout the year, and the average temperature is approximately 24 °C. The forest vegetation forms a closed, diverse canopy, the height of which varies from 20 to 40 m, and the dominant tree species include Oxandra panamensis, Trophis caucana Berg., Chamaedorea linearis Mart., Clarisia biflora Ruiz., and Gustavia speciosa Kunth. This forest patch is one of the last remnants of the original continuous dry forest along the central Cauca River Valley (Ramírez et al. 2002), which has undergone large-scale deforestation to less than 1% of its original area due to intensive agriculture (sorghum, soybean, sugarcane and cotton) and cattle-ranching since 1920 (GEMA 1998).

Study species

Atta cephalotes (Hymenoptera: Formicidae) is considered a “woodland species” and is commonly found in mature, well-conserved forest (Rockwood 1973, Correa et al. 2005). Surprisingly, the occurrence of A. cephalotes in Colombia is not restricted to closed forests, and the species exhibits more disturbance tolerance than indicated by published observations. In fact, it is common throughout almost the entire country from sea level to 2500 m asl (Fernández et al. 2015). Although the Cauca Valley has been extensively converted from forest to pasture and sugarcane crops, A. cephalotes is broadly distributed and easily found, even in parks within cities such as Cali and Pereira (Montoya-Lerma et al. 2012) and in the grasslands, croplands and forests of the region (Armbrecht et al. 2001, Chacon et al. 2012).

Study colonies

We systematically surveyed A. cephalotes colonies in 13 ha of forest and in a similar area of the surrounding pasture. After a preliminary 15-day monitoring period of their foraging activity and general health status, we choose 12 active mature colonies as study colonies, six within the forest and six within the pasture. The minimum distance of each colony from the edge of its habitat was 50 m, a distance that should have excluded edge effects (Meyer et al. 2009). Because nest size can affect foraging activity (Lewis et al. 1974a), the size of the colonies was calculated as the minimum elliptical area circumscribing the central nest mound (Hernández et al. 1999), and although mature (e.g., presence of soldiers and well-established foraging trails), the colonies were consistently smaller in the pasture than in the forest, with mean nest sizes of 27.39 m² ± 22.19 and 109.3 m² ± 65.35, respectively (Welch two-sample t-test: t = -2.90, df = 6.14, N = 12, P = 0.026). Measurements were taken on the main foraging trail of each nest, which was identified as the trail with the highest number of ants at the time of the measurement. The studied foraging trails represented cleared, persistent trunk trails as typically described for Atta ants (Kost et al. 2005).

Microclimate Measurements

To characterize the effects of microclimatic conditions on the main foraging trails of A. cephalotes, we assessed the following parameters for each nest: (1) air temperature, (2) RH, (3) maximum surface temperature, and (4) minimum surface temperature. The air temperature and RH were measured using a HOBO Pro Temperature/Relative Humidity sensor (Onset Computer Corp., Bourne, MA, USA ®) equipped with internal data loggers. The sensor was located 5 cm above the soil surface on the main foraging trail near the entrance hole of each colony nest, and it recorded measurements at hourly intervals from the 18^th of May to the 12^th of December 2013.  The maximum and minimum surface temperature were recorded for all nests across the two habitats during twelve field work periods of eight hours: four from 6:00 to 14:00, four from 14:00 to 20:00, and four from 20:00 to 6:00 (totaling 96 hours of data at each nest). The surface temperatures were measured by walking slowly from the nest entrance to the end of the foraging trail while registering the maximum and minimum temperatures observed using an infrared thermometer (Fluke 62 MAX).

Foraging Activity

To describe and compare the foraging activity of A. cephalotes between the two habitats and to examine relationships with microclimatic factors, the activity rates of all above ground workers involved in foraging-related tasks were assessed on the foraging trails. Particularly, we counted the number of outgoing and incoming unladen and laden ants at each nest. Outgoing ants are those that leave the colony to search and harvest resources for fungal substrate, to maintain the trail, e.g., by reinforcing the pheromone trail (Moser 1967), or to protect the foragers, e.g., by clearing foraged leaves to control parasites such as phorid flies (Yackulic & Lewis 2007). Incoming unladen ants are foragers that have lost their loads in the route or workers returning to the nest after being involved in other above ground tasks. Laden ants directly reflect the efficiency of foraging as they consist of ants that are transporting harvested leaves, flowers, seeds or other materials to feed the fungus garden.

Foraging rates were calculated as the number of ants passing by a fixed point on the main foraging trail near the main hole of each nest in 5 minutes (Wirth et al. 2003) and were measured at the same time as the maximum and minimum surface temperatures. We used a video camera (Canon PowerShot A490) fixed to a tripod to count the ants, and the total number of ants, both incoming and outgoing, were determined using AntCounter V.1.0 software, which counts the number of ants that pass through a given point in two directions with respect to the nest entrance (Bustamante & Amarillo-Suárez 2016). Because this software does not distinguish between laden and unladen ants, the number of laden ants was counted manually by playing the recorded videos in slow motion. To control for differences in nest size when comparing the foraging rates between forest and pasture, we used the relative foraging activity (i.e., the hourly foraging activity divided by the maximum foraging activity observed of a given nest).

Data analysis

Differences in microclimatic conditions between habitats and hours were determined for each microclimate parameter by a two-factorial ANOVA with repeated measures on one factor, where habitat and time were fixed factors. “Habitat” had two levels, pasture and forest, and “hour” had 24 levels representing each hour of the day. “Nest” corresponded to each of the subjects with repeated measures each hour and with each nest located either in the forest or the pasture.

Additionally, to identify differences in hourly foraging between habitats, we performed a nested ANOVA with colony as a random factor nested within habitat (fixed factor) for each microclimatic condition and hour. Because there were 24 ANOVAs, one for each hour, we used an alpha of 0.01 instead of 0.05 to reduce the probability of a type II statistical error (Zar 2010).

Differences in foraging activity between habitats and hours were analyzed for each foraging variable (unladen and laden incoming and outgoing ants) in the same way as for the microclimate variables. To identify differences in foraging activity between habitats, we performed a nested ANOVA with the same parameters as with the microclimatic variables for each foraging activity variable and for each hour.

The assumptions of normality and homoscedasticity were met in many, though not all, cases, even after the data were log or arcsine transformed. However, we decided to use nested ANOVA because there is no equivalent non-parametric approach (Dytham 2011) and it is robust against deviations from normality and homoscedasticity (Sokal & Rohlf 1995, Zar 2010). Due to unequal sample sizes, type III sums of squares are reported.

Because the numbers of outgoing and unladen and laden incoming ants were expressed as percentages of the maximum flow, they were arcsine-transformed prior to ANOVAs (Doncaster & Davey 2007). To determine the relationships between foraging activity and microclimatic variables, we performed simple linear regressions between each microclimatic variable (air temperature, RH, and maximum and minimum temperature along the foraging trail) and the response variables (number of unladen or laden incoming or outgoing ants). All analyses were carried out using R v. 3.1.0 (R Core Team 2015).