Foraging valor linked with aggression: selection against completely abandoning aggression in the high‐elevation ant Tetramorium alpestre?

Aggression has multiple benefits and is often coupled with other behaviors (“behavioral syndromes”). The level of aggressiveness is influenced by an adaptive benefit–cost ratio suggesting that benefits should outweigh the costs of aggression. Here, we assess if several behaviors are coupled in two behaviorally different populations (aggressive, peaceful) of the high‐elevation ant Tetramorium alpestre. For three weeks, we collected colony fragments and analyzed boldness, exploring, foraging, and risk‐taking behaviors. We hypothesized that the aggressive population is bolder, more explorative and risk‐prone, and forages more food than the peaceful population. To test whether (a) the combination of experiments and parameters used yields a good setup, (b) populations differ behaviorally, and (c) populations display behavioral syndromes, we assessed (a) the frequency of repeatable behaviors of each experiment, (b) the behavioral means among populations, and (c) the behavioral repeatability, respectively. We found that (a) boldness and exploring were most repeatable and represent a good experimental setup, (b) the aggressive population was bolder and more explorative and risk‐prone than the peaceful population, (c) boldness and exploring behaviors were highly repeatable in both populations, thus corroborating our hypothesis. The results suggest that boldness, exploring, and risk‐taking but not foraging are presumably coupled with aggression and indicate the presence of behavioral syndromes in this ant. Under specific ecological conditions, aggression may be coupled with other behaviors and important for finding food. Aggression is probably adaptive in T. alpestre, possibly indicating that selection favors aggression at least partially, which may counteract the complete loss of intraspecific aggression.


Introduction
Aggression is ubiquitously present in animals and occurs, for example, when the interests of individuals conflict (Huntingford & Turner, 1987) or to attain fitness benefits (Bubak et al., 2014).Aggressive animals often gain benefits such as high-quality resources and/or more mates.Such benefits likely promote animals' fitness, directly or indirectly.Natural selection thus may influence the formation and maintenance of aggression.
In ants, as in other social insects, workers often interact aggressively with each other when encountering nonnestmates, and aggression frequently correlates with the social structure of the colony (Martin et al., 2009).For example, workers of single-queened colonies (monogyny) are usually more aggressive toward non-nestmates while workers of multiple-queened colonies (polygyny) interact more peacefully with non-nestmates.However, in several species, workers of monogynous colonies also interact peacefully with non-nestmates (Steiner et al., 2007;Krapf et al., 2018Krapf et al., , 2023)).Therefore, the intraspecific behavior likely differs within and across colonies as well as populations (Pinter-Wollman, 2012;Lai et al., 2015;Krapf et al., 2018).
Behavioral syndromes are adaptive and can change and evolve (Sih et al., 2004;Jandt et al., 2014).This means that the structure of behavioral syndromes may differ among populations that live in different environments.Consequently, behavioral syndromes are often associated with trade-offs where beneficial responses to one stimulus (e.g., physiological, environmental, social, or individual factors) might be coupled with sub-optimal behaviors in response to other stimuli.For instance, bold and aggressive individuals may allocate more food but are possibly more prone to end up as prey during foraging when predators are present.Hence, behavior is unlikely to remain consistent over time and across contexts due to the adaptation of the benefit-cost ratio at different organizational levels.The benefit-cost ratio influences individual differences in aggression where the benefits of aggression (e.g., food or territory acquired) should exceed the costs of aggression (e.g., physical injury or exposure to depletion) (Georgiev et al., 2013).Behavioral syn-dromes have only been studied in a few ant species so far (Chapman et al., 2011;Pinter-Wollman, 2012;Bengston & Dornhaus, 2015;Blight et al., 2016;Segev et al., 2017;Maák et al., 2021), but detailed knowledge of behavioral syndromes in ants and their implications for the behavior of workers is largely lacking.
Here, we used the high-elevation ant Tetramorium alpestre to investigate if specific behaviors are coupled in an aggressive and/or a peaceful population.We collected workers from two populations (Austria and Italy) which are both known to be monogynous (Krapf et al., 2018;Krapf et al., unpubl. data).Workers from one population (Kue, Austria) were peaceful in earlier research compared with those of the other population (Pen, Italy), which were aggressive (Krapf et al., 2018(Krapf et al., , 2019;;unpublished pilot data).We hypothesized that T. alpestre workers from the aggressive population demonstrate repeatable beneficial behavior in the behavioral experiments compared with workers from the peaceful population.Specifically, we hypothesized that workers from the aggressive population are more abundant (i.e., the number of workers that choose to enter the experimental setup is higher) and active in the various experiments and need less time to arrive on the experimental setup (see Table S1 for detailed hypotheses).We (a) assessed the suitability of four different parameters tested in four behavioral experiments (boldness, exploration, foraging, and risk taking, adapted from Blight et al., 2016) as good experimental setup by assessing the number of repeatable behavior values of the experiments, (b) compared the behavioral means of both populations, and (c) analyzed the behavioral repeatability at two levels (replicates within the colony; colony).Addressing (a) allows to identify the best possible experimental setup among the options explored to describe behavioral variation and behavioral repeatability in T. alpestre.The latter two allow to evaluate the behavior of T. alpestre including (b) behavioral differences in the mean and (c) the potential presence of behavioral syndromes.

Study species
The high-elevation ant species Tetramorium alpestre Steiner et al., 2010 was used for all behavioral experiments.This species belongs to the Tetramorium caespitum complex and is located in European montane to subalpine habitats between 1300 and 2300 m above sea level (Schlick-Steiner et al., 2006;Wagner et al., 2017).Tetramorium alpestre displays a social polymorphism (monogyny, polygyny), but the populations used for the experiments here are known to be monogynous (Krapf et al., 2018;Krapf et al., unpublished data).Moreover, this species exhibits broad behavioral plasticity, including peaceful and aggressive behavior among colonies (Krapf et al., 2019(Krapf et al., , 2023)).

Worker collection and laboratory maintenance
In September 2019, workers were collected from six nests in two populations: three nests at Kühtai ("Kue"; Tyrol, Austria, Fig. 1) and three nests at Penser Joch ("Pen"; South Tyrol, Italy).Due to the difficulty in morphologically determining this species at the species level, the COI gene was partially sequenced to ensure that the collected individuals were not from another resident Tetramorium species (Wagner et al., 2017).Nests were found under rocks on south-facing nonforested mountainous-alpine meadows in climatically similar environments (Table 1).The exact collection locations were determined with NAVSTAR-GPS system (Garmin eTrex ® Legend HCx, Olathe, USA; Table 1).Standardized surface air temperature (TAS) was calculated for each colony following the rationale from Seifert and Pannier (2007): TAS was calculated as the mean air temperature of the period from May 1st to August 31st averaged over the years 1961 to 1990 of the nearest three meteorological stations (data provided by Klimaabteilung der Zentralanstalt für Meteorologie und Geodynamik (1996), Vienna, Austria).The data were corrected for an altitudinal decrease in temperature of 0.661 °C per 100 m according to the equation of Seifert and Pannier (2007): TAS = −0.694× LAT + 0.078 × LON − 0.00661 × ALT + 52.20,where TAS is the predicted standardized air temperature in °C, LAT and LON denote the geographical latitude and longitude in decimal format, respectively, and ALT is the altitude in meters above sea level.The selected nests of both populations were at least 10 m apart from each other to ensure that they represented independent colonies (Fig. 1; Krapf et al., 2018Krapf et al., , 2019)).Colony fragments were collected once a week for 3 weeks (4, 13, and 20 September 2019) using a polypropylene aspirator.Approximately 280 workers were collected from the entire range of nest compartments that were accessible.Thus, workers from a range of different behavior tasks (foraging for food, tending brood, or exiting the nest to attack intruders) except callow workers were collected each week whenever possible, totaling approximately 840 workers per colony.Collecting workers from different nest compartments was achieved by turning around stones.This procedure enables capturing the behavior of the colony as the colony behavior is determined by the behavior of workers (Pinter-Wollman, 2012).Additionally, collecting these numbers of workers allowed us to make experimental replicates without drastically decreasing worker numbers per colony or causing colony imbalance through biased collection from workers with specific behavior tasks.As T. alpestre colony size was estimated to be similar to T. caespitum, a congener which has approximately 15 000 to 75 000 workers per colony (Sugawara & Nikaido, 2014;Seifert, 2017), a drastic decrease in worker numbers causing behavioral changes due to our sampling is unlikely.Additionally, the collection period of three weeks was chosen not to cause too much disturbance inside the colony but also to validate the behavioral repeatability of the different colonies and populations.
Each colony fragment of each sampling week was transported to a laboratory at the University of Innsbruck and transferred to a polypropylene box ("nest"; 12 cm × 12 cm × 6 cm; Fig. 2A1) shortly after their collection.Nests were stored in a climate chamber (MIR-254, Panasonic, Etten Leur, Netherlands) at a constant temperature of 18 °C with approximately 70% humidity and in darkness based on the characteristics of their subterranean lifestyle (Seifert, 2018;Cicconardi et al., 2020).The walls of the nests were Fluon-coated (GP1, De Monchy International BV, Rotterdam, Netherlands) to prevent the workers from escaping.Small holes were drilled in the lid to ensure sufficient ventilation.Boxes were equipped with paper towels as well as cotton to create retreats and provide material for different constructions.The diet consisted of honey water (1 : 10 dilution) and water, provided ad libitum inside the nest (Fig. 2A1).Based on preliminary food preference tests, honey water was the food most likely to be accepted by T. alpestre workers.

Experimental setup
Collected workers were acclimatized in the nest for at least 4 d before being used in the behavioral experiments.Workers were fed on the day of collection and subsequently starved for 5 d before the behavioral experiments.Water was always provided ad libitum.Whenever possible, each colony fragment was split into two replicates (Blight et al., 2016).Each replicate comprised 100 workers that were transferred to an experimental nest (screwcap tube; Fig. 2A2) where workers acclimatized for 30 min before the start of each experiment (Blight et al., 2016).The experimental nests contained cotton soaked with water, plus small holes drilled into the lid to provide oxygen.
Four experiments were subsequently conducted (boldness, exploration, foraging, and risk taking; adapted from Blight et al., 2016).Experiments 1 and 2 were conducted on Day 5 of the diet because no food was included in these experiments.Experiments 3 and 4 were performed on the following day (Day 6 of the diet), that is, after the workers had fasted for a total of 5 days.The number of starvation days was determined in preliminary experiments.To ensure that the same workers were tested in the experiments, the workers were kept in the experimental nests until the last experiment was conducted.This means that after conducting Experiments 1 and 2, the experimental nests were transferred to the climate chamber overnight until used for Experiments 3 and 4 on the following day.The order in which experimental nests were tested in each experiment was randomized.The experiments started with connecting the arena (coated with paraffin oil) to the experimental nest by a connection tube (Fig. 2B).The connection tube consisted of two parts with a fixed point at a junction between the two parts (triangle in Fig. 2B).Inside the arena, a foam ramp was placed between the arena floor and the connection tube, which facilitated workers to enter the arena floor.In all experiments, workers had the choice to enter the arena or to remain in the experimental nest.All experiments were recorded by a high-definition camera (Handycam HDR-PJ810E, HDR-CX625, or HDR-XR 155, all Sony, Tokyo, Japan).For accessibility, experiments are from now on written in lower-case letters (e.g., boldness) and behaviors in capital letters (e.g., BOLDNESS).By measuring behaviors in the experiments, it can be assessed if the behaviors are coupled in the aggressive and/or a peaceful population.Four specific behaviors were chosen to be tested that represent behaviors according to different ecological situations (Sih et al., 2004;Réale et al., 2007).BOLDNESS and EXPLORING represent a (potentially new) situation that could be dangerous in terms of predators, FORAGING represents a known situation, and RISK TAKING represents the willingness

Experiment 1: Boldness
The aim of this experiment was to investigate the individual worker's BOLDNESS, which represents a behavior in a new situation and may relate to aggressive behavior.This experiment was assessed together with Experiment 2 (see below).The experiment lasted for 30 min and started with connecting the experimental nest with the arena.The first 10 workers of each replicate that crossed the junction were observed (Fig. 2C).To evaluate the boldness of these 10 workers, four parameters were recorded or calculated over the first 10 workers: (A) the mean number of workers crossing the junction, (B) the mean number of workers depositing pheromones while leaving the arena, (C) the time in seconds that the first 10 workers needed to cross the junction, and (D) the maximum time needed until 10 workers had crossed the junction (Table S2).The pheromone deposition was observed during the video analysis.In detail, workers that were on the way back from the arena to the experimental nest stopped briefly and lowered the tip of their gaster to the floor to deposit pheromones.These workers were noted as depositing pheromones, as similarly described for Tetramorium impurum by Verhaeghe (1982).No food was used as an attractant during this experiment.

Experiment 2: Exploring
The aim of this experiment was to investigate the EX-PLORING behavior, which represents a behavior in a new situation and may be important for a colony in identifying a well-suited part of the habitat to colonize or to find food.This experiment was run together with Experiment 1, that is, it lasted 30 min and started with connecting the experimental nest with the arena.During the analysis, four parameters were recorded every 10 s and calculated for 30 min observation time: (E) the average and (F) the maximum number of workers in the arena, (G) the time in seconds until the first, and (H) the maximum number of workers were present in the arena (Table S2).A worker was defined as inside the arena when at least 2/3 of its body was inside the arena at the second observed.Such a worker was almost always inside the arena with its whole body in the following second and practically never retreated from the arena without having entered it completely.Workers that had, for example, only the head in the arena (1/3 of their body) were defined as workers outside the arena because they often retreated to the connection tube or the experimental nest.Additionally, this definition allowed us to determine the exact second at which the worker entered the arena.In the analysis of Experiment 2 (exploring), the specific number of workers observed was not limited compared with Experiment 1 (boldness).The values recorded for this experiment were retrieved from the same video as used in the boldness experiment but were different in that parameters were recorded for workers that entered the arena (Fig. 2C): Not all workers crossing the junction entered the arena, which is why both experiments were conducted and evaluated.After this experiment, workers were returned to their experimental nest and stored in the climate chamber for the experiments conducted the next day.In doing so, workers of the replicates were kept separate.

Experiment 3: Foraging
The aim of this experiment was to determine the foraging activity within both populations by analyzing two behaviors, FORAGING-AT FOOD and FORAGING-IN ARENA (Table S2).The former allows assessing the foraging activity when food is found and the latter assessing the activity when a food source was found but workers still explored the surroundings.The experiment lasted 15 min and started at the time when at least one worker was present at the honey water.During the analysis, four parameters were recorded every 10 s and calculated for 15 min observation time: (I) the average and (J) the maximum number of workers at the honey water, (K) the time in seconds until the first, and (L) the maximum number of workers were feeding on honey water (FORAGING-AT FOOD).The same parameters as presented above were recorded for workers that were inside the arena but not feeding on honey water (FORAGING-IN ARENA; M-P; Table S2).As a precondition, 50 µL of honey water (1 : 10 dilution) was placed centrally in the arena so that workers could access it from all directions (Fig. 2D).After this experiment, all workers were returned to their experimental nest to perform the next experiment with an experimental setup cleaned with ethanol (except the experimental nest) to avoid any pheromone residuals.Between Experiments 3 and 4, workers were allowed to acclimatize for 30 min.

Experiment 4: Risk taking
The aim of this experiment was to investigate two risktaking behaviors, RISK TAKING-AT FOOD and RISK TAKING-IN ARENA, to study how workers behave in exceptional and risky circumstances when temperatures are quickly increasing.The former allows assessing for the proneness to risk when food is available, and the latter allows assessing the general proneness to risk.This experiment thus represents a dangerous situation when workers are foraging.During the analysis, three parameters were recorded every 10 s and calculated for 13 min observation time: (Q) the average and (R) the maximum number of workers in the arena and (S) the time in seconds until the maximum number of workers feeding on honey water (RISK TAKING-AT FOOD).The same parameters as described above were also recorded for workers that were inside the arena but not feeding on honey water (RISK TAKING-IN ARENA; T−V; Table S2).As a precondition, 50 µL of honey water (1 : 10 dilution) was placed centrally in the arena so that workers could access it from all directions.The experimental setup was placed on a heating plate (TP, VWR International, Radnor, PA, USA; as described in Tratter Kinzner et al. (2019) and here Fig. 2D) to simulate a rapid temperature increase in the experimental arena.Commercial olive oil was placed onto the underside of the arena to diminish a loss of temperature.A foam was placed underneath the experimental nest to avoid heat exposure so that workers could escape the rising temperature by entering the experimental nest.During the experiment, the plate was heated from 25 °C (daily average temperature during summer at 2000 m above sea level; unpubl.data) to 49 °C in 13 min (max.>50 °C measured during summer; unpubl.data).Every 30 s, the temperature increased by one degree and stagnated at 49 °C during the last minute.This temperature range was determined by measuring soil temperature in the field as well as preliminary experiments using a heating plate.At higher altitudes, ants of this species can be exposed to high temperatures, especially in their specific stony habitat.The selected temperature range, especially above 40 °C, thus represents a temperature at which workers' locomotion ability is reduced.This means that they are not able to escape anymore, and long exposure eventually leads to the death of the individuals.The temperature increase tests the proneness to risk of the workers (Cerdá et al., 1998a,b).The experiment, and thus the heating, was started when at least five workers were present in the arena or feeding on honey water.This condition was necessary to examine their risk-taking behavior, and honey water was used here as an attractant.After completion of this experiment, both replicates were returned to the nest boxes and kept in the climate chamber.

Statistical analyses
Repeatability To address research questions (a) assessing the suitability of four different parameters tested in four behavioral experiments (boldness, exploration, foraging, and risk taking) and identifying the best possible experimental setup among the various options explored and (c) analyzing the behavioral repeatability at two levels (replicates within the colony; colony), we assessed (a) the frequency of repeatable behaviors of the parameters and experiments and (c) the behavioral repeatability values of the replicates and colonies.
To analyze the data, all videos were analyzed individually and manually.For each video, data for three to four parameters (Q−V in Experiment 4; A−P in Experiments 1−3) were obtained for all experiments (Table S2).Data were recorded for the number of workers as well as for individual workers at a certain time (in seconds).Workers were observed individually to avoid double counting.The data were manually curated in an Excel file and analyzed using R (version 4.0.5;R Core Team, 2018).
To assess whether the colonies and replicates were behaviorally consistent and whether the behavior changed over the 3 weeks, linear mixed effect models (LMEMs) were conducted for each population, experiment, and parameter separately.LMEMs were calculated using the function lmer applying a restricted maximum likelihood (lme4 package; Bates et al., 2015).In the LMEMs, the parameters (e.g., mean number of workers in the arena) were set as dependent factors, the days of the year when the experiments were conducted were set as a fixed factor, and "replicate" and the "identity of colony fragments" were set as random factors.The model fit was checked visually using the package "DHARMa" (Hartig, 2021).Coefficients of determination were calculated using the r.squaredGLMM function (MuMin package; Barton, 2020).Repeatability and credible intervals to infer the statistical significance of the repeatability values were calculated by simulating the model 1000 times using the sim function (arm package; Gelman & Su, 2020) and applying the quantile function (Hertel et al., 2020).Repeatability is defined as the closeness of agreement between test results (here experiments, parameters, etc.) obtained with the same experimental procedure (Jones & Marengo, 2016).As a prerequisite for interpreting the repeatability value, the calculated mean had to reach at least 0.10 and the minimum value had to be larger than 0.001 at the lower confidence interval.Model significance was assessed using the "lmerTest" package (Kuznetsova et al., 2017).All graphs were created using the packages "gg-plot2" (Wickham, 2009), "ggpubr" (Kassambara, 2021), "ggdist" (Kay, 2021), as well as Biorender ® (2021).

Differences in the behavioral means of populations Kue and Pen
To address research question (b) assessing if the behaviors differed between the two populations regardless of the sampling date, the data from all 3 weeks from each colony within populations were analyzed together.One-tailed tests were applied following the hypothesis that the aggressive population is bolder and more explorative, forages more food and is more risk-prone compared with the peaceful population.The normality of data was checked using a Shapiro-Wilk test (shapiro.testfunction) for each experiment and parameter.If data were deviating from normality, a Mann-Whitney U test was conducted using the function wilcox.test(R Core Team, 2018).If no deviation was detected, a Welch's t-test was conducted using the function t.test (base R package).For the risk-taking experiment, a linear regression was calculated between the temperature and the number of workers feeding on honey water for both populations separately (but again combining data from all colonies within populations).The linear regression was conducted using the function lm (base R package).Model fit was checked visually using DHARMa.Additionally, a LMEM was calculated between the two populations and the number of workers feeding on honey water to assess a potential difference in the number of workers among the two populations.The LMEM (restricted maximum likelihood) was calculated using the function lmer (Bates et al., 2015) with the workers feeding on honey water as a dependent factor, the populations as a fixed (explanatory) factor, and temperature as a random factor.The model fit was checked visually using DHARMa and was slightly deviating from a normal distribution and within-group deviations from uniformity (Hartig, 2021).Thus, the number of workers leaving the arena was log n -transformed, which improved model fit.Model significance was assessed using the "lmerTest" package.

Repeatability
Significant repeatable behavior was unequally distributed across parameters, experiments, and populations (Fig. 3, Table S3).Except for FORAGING-IN ARENA, all behaviors were repeatable at least once at the colony or replicate level.At the replicate level, patterns of repeatable behavior were more often observed in replicates from the aggressive population Pen (3 out of 22 assays) compared with replicates from the peaceful population Kue (1 out of 22 assays).The repeatability analysis of replicates resulted in repeatability values for population Kue in EXPLORING (Fig. 3H), while for population Pen, the random factor replicate was repeatable in FORAGING-AT FOOD, RISK TAKING-AT FOOD, and RISK TAKING-IN ARENA (Fig. 3L, Q, V).For EXPLORING, the repeatability of replicates in population Kue was significant with a value of 0.34 (Fig. 3H), and replicates in population Pen had a significant repeatability value of 0.31 in RISK TAKING-AT FOOD (Fig. 3Q).Within the latter behaviors, the highest repeatability value was found for replicates in population Pen in RISK TAKING-IN ARENA (0.67; Fig. 3V) followed by the FORAGING-AT FOOD behavior of replicates in population Pen (0.51; Fig. 3L).
At the colony level, patterns of repeatable behavior were more often observed in colonies from the aggressive population Pen (10 out of 22 assays) compared with colonies from the peaceful population Kue (6 out of 22 assays).In BOLDNESS and EXPLORING behaviors, the colonies of population Pen behaved frequently more consistently compared with the colonies of population Kue.Within these two behaviors, only the parameter "S until max workers in arena" (Fig. 3H) was not repeatable for Pen.The colony repeatability of these behaviors ranged between 0.19 (Fig. 3G; population Pen) and 0.63 (Fig. 3A; population Pen).Additionally, for workers from population Kue, significant changes over time were identified in the behavior EXPLORING (Fig. 3F, G): The "Max n of workers in arena" significantly increased over three weeks (LMER, t 11 = 2.278, R 2 = 0.30, P = 0.0436; Fig. 3F, Table S3), while the "S until the first workers in arena" significantly decreased over three weeks (LMER, t 9.305 = −2.808,R 2 = 0.59, P = 0.0198; Fig. 3G).For FORAGING-AT FOOD, repeatable behavior for each population was found for a single parameter (Fig. 3L): Population Kue revealed a highly repeatable behavior with a value of 0.77 at the colony level, while the value of population Pen was 0.17.For FORAGING-IN ARENA, the colony repeatability for Kue was 0.12 ("S until max workers in arena"; Fig. 3P).The colony repeatability value for RISK TAKING-AT FOOD for Kue was 0.25 (Fig. 3S), while the values for RISK TAKING-IN ARENA were 0.32 and 0.33 for the mean and maximum number of workers in the arena in the population Pen, respectively (Fig. 3T−U).

Differences in the behavioral means of populations Kue and Pen
The mean behavioral values were calculated for each of the two populations Kue and Pen and differed in two out of six behaviors scored, namely, BOLDNESS, RISK TAKING-AT FOOD, and RISK TAKING-IN ARENA (Fig. 4A, Q, R, V, Table S4).The "Mean n of workers crossing junction" was significantly different between the populations (BOLDNESS; mean Kue = 6.08; mean Pen = 8.93; Mann-Whitney U, W = 48.5,P = 0.040; Fig. 4A), with slightly more workers from population Pen crossing the junction.For RISK TAKING-AT FOOD, two parameters were significantly different: For the "Mean n of workers at honey" and "Max n of workers at honey," significantly more workers from population Pen were feeding on honey water (Welch's t-test, mean Kue = 3.54, mean Pen = 8.69, t 13.631 = −3.29,P = 0.006, Fig. 4Q; mean Kue = 1.22,mean Pen = 4.63, t 14.337 = −2.75,P = 0.016, Fig. 4R, respectively).Population Kue needed significantly less time to reach the maximum number of workers in the arena than population Pen (RISK TAKING-IN ARENA "S until max workers in arena"; Welch's t-test, mean Kue = 242.31;mean Pen = 397.69;t 21.696 = −1.95,P = 0.032; Fig. 4V).
In addition to the significant differences in certain behaviors and parameters between the populations, further trends were observed.For the behaviors BOLDNESS and EXPLORING, workers from population Pen were frequently present in greater numbers and faster to cross the junction or enter the arena (Fig. 4B−H).In contrast, workers from population Pen needed more time to arrive at the honey than workers from Kue for FORAGING-AT FOOD (Fig. 4L).For FORAGING-IN ARENA, more workers from population Kue needed more time to enter the arena compared with workers from population Pen (Fig. 4P).For RISK TAKING-AT FOOD and RISK TAKING-IN ARENA, behavioral trends emerged, in that more workers from population Kue were present in the arena compared with population Pen (Fig. 4T−U).Furthermore, the time workers needed to enter the arena or arrive at the honey while the temperature was rising differed slightly for each population but not significantly (Fig. 4S, V).To determine differences between the populations for RISK TAKING-AT FOOD, a linear regression was additionally performed between the temperature and the number of workers feeding at honey at a specific time point.Negative correlations (fewer workers at higher temperatures) were found for both populations (Fig. 5), but only in the population Pen, this effect was significant (LM; Kue, R 2 = 0.09, P = 0.066; Pen, R 2 = 0.33, P < 0.001).A LMEM conducted between workers feeding at honey and the populations by controlling for the rising temperature also revealed a significant difference between populations Kue and Pen in the number of workers present at honey (LMEM, t 101.185 = 5.19, P < 0.001).

Discussion
In this study, we (a) analyzed the repeatability of four different parameters tested in four behavioral experiments (boldness, exploration, foraging, and risk taking, adapted from (Blight et al., 2016)) to assess their suitability for such experimental tests; (b) compared the behavioral means of both populations to detect population differences; and (c) analyzed the behavioral repeatability at two levels (replicates within the colonies and colony) to determine the presence of behavioral syndromes.We tested three to four parameters for each behavior.(a) We found that BOLDNESS and EXPLORING were the most repeatable behaviors, and FORAGING and RISK TAK-ING were less repeatable.We thus infer that the former two behaviors yield the best possible experimental setup and are thus favorable to analyze the repeatability and variation of behaviors, at least in this species.Two behavioral changes over time were found for EXPLOR-ING in population Kue: Over three weeks, the number of workers in the experiment "Maximum number of workers in the arena" increased from Week 1 Week 3, and the "Time workers needed to enter the arena" decreased over the same period.(b) The two populations differed mainly in the behavioral means of BOLDNESS, RISK TAKING-AT FOOD and RISK TAKING-IN ARENA behaviors.In a risky situation (e.g., increasing temperatures), workers from the aggressive population Pen remained longer and with more workers at the food source compared with the peaceful population.(c) When analyzing the repeatability of replicates, population Pen was more consistent in its behaviors compared with population Kue.This was similar at the colony level where the behaviors BOLDNESS and EXPLORING included the most significant repeatability values compared with the other behaviors.For FORAGING-IN ARENA, one repeatability was found, while the other three behaviors (FORAGING-AT FOOD, RISK TAKING-AT FOOD, and RISK TAKING-IN ARENA) varied in repeatability levels (replicate and colony) between the populations and parameters.
Behavioral repeatability over time and/or across situations is a prerequisite for making statements about behavioral syndromes in animals.In this study, we assessed several experiments not only across four behaviors (boldness, exploring, foraging, and risk taking) but also at the parameters as well as colony organization level within and between populations.This allows us to evaluate the experimental setup (i.e., the combination of experiments and parameters used) and gives us information on whether continuing to use a setup is warranted or whether a setup should rather be adjusted or discontinued to be used.For both populations, four out of 22 repeatability values were significant at the replicate level (range 0.31−0.67,Fig. 3, Table S3) giving us insights into the repeatability or variability of within-colony behavior.The highest levels were found in RISK TAKING-IN ARENA (0.67) and FORAGING-AT FOOD (0.51).This implies that heat acting as a kind of threat as well as honey water as an attractant have similar effects on the behavior which may be based on survival instinct (Blumberg, 2017).In contrast, the two other behaviors (EXPLORING and RISK TAKING-AT FOOD) were relatively low in repeatability, indicating a higher behavioral variation at the replicate level.This is interesting as we included workers putatively from a range of different behavior tasks except for callow workers (see Materials and methods, section "Worker collection and laboratory maintenance," for details).Because each colony acts as a superorganism (Wilson & Hölldobler, 2009), individuals and groups of colonies (here, for example, replicates with 100 workers) can behave differently in specific situations compared with replicates of the same or other colonies, as was found here and in previous (Cronin, 2015;Jandt & Gordon, 2016;Bockoven et al., 2017;Carere et al., 2018;Krapf et al., 2019).Increasing the number of replicates and weeks (e.g., increase to n = 500 workers per colony per week yielding five replicates and collecting them over several weeks) would allow gaining additional information for each colony and assessing the uniformity of the behavior at the replicate and colony level.On the downside, such additional sampling would also create a stronger impact on the natural colony system, because more workers would need to be collected over a longer time.In the case of large colonies, such an ef-fect would be minor but would be more severe for smaller colonies.Another option, in our opinion, would be to test the individual behavioral level to assess FORAGING and RISK TAKING.This would likely reduce the variation in the experimental group and also address the "replicationcrisis" problem known in behavioral studies (Tincani & Travers, 2019;Locey, 2020).
At the colony level, more significant repeatability values were found for both populations (16 out of 22 values) compared with the replicate level (4 out of 22 values).Especially for population Pen, significantly repeatable values were returned in the experiments BOLDNESS and EXPLORING.The highest repeatability values for BOLDNESS behavior were found for two parameters (Fig. 3A: 0.63 and 3B: 0.47, both Pen).Also, the behavior of population Kue was significantly repeatable for these two parameters at the colony level, even if values were lower (Fig. 3A: 0.20 and 3B: 0.24, both Kue).This indicates that the behavior BOLDNESS and its parameters analyzed at the colony level represent a sound experimental setup to analyze the number of workers and the time needed to enter the arena similar to the results by Blight et al. (2016).Additionally, we suggest including other parameters that focus on time because it could be compared among assays increasing the comparability and allowing for generalizations.For example, "S until first workers crossed junction" and "S until max workers crossed junction" were both consistent in population Pen, but not in population Kue.This also applies to EXPLORING (Fig. 3G): For this behavior, the colonies of population Pen were consistent for the first two parameters ("Mean n of workers in arena" and "Max n of workers in arena"), which corroborates the sound design of the experimental setup.The two FORAGING behaviors and their corresponding parameters are, in our estimation, less meaningful to describe consistent behavior at the colony and population level, although behaviors were to some degree repeatable between the two populations (Fig. 3L).Among the parameters assessed, the repeatability of "S until the max workers at honey" differed strongly between the populations (FORAGING-AT FOOD, Kue: 0.77, Pen: 0.17).Overall repeatable behavior in this experiment is not surprising, because we had an attractant (honey water), and the ants were hungry due to the previous fasting.
The behavior of the population Kue changed significantly over time .This change over time cannot coincide with a learning effect as new colony fragments were collected and tested over the three weeks.The change over time in EXPLORING, comprising more and faster workers, may rather be associated with the season when ants were collected.Due to the beginning of autumn at the time of fieldwork, ants were likely already foraging for food to store for the winter (Reyes-López & Fernández-Haeger, 2002;Judd, 2011;Khalife & Peeters, 2020), which would also explain the higher mean and maximum number of workers for FORAGING compared with the equivalent parameters of EXPLORING.However, it remains inexplicable why this change only exists in workers from population Kue because the two population sites are very similar in altitude and TAS (standardized surface air temperature), which would exclude a temporal and seasonal difference in food storage.Further studies are needed to assess these behaviors in spring and summer to compare the results here and elucidate this observation.
Besides high behavioral repeatability values within populations, significant behavioral were found between the means of the two populations in BOLDNESS, RISK TAKING-AT FOOD, and RISK TAKING-IN ARENA (Fig. 4, Table S4).All other behaviors and respective parameters revealed similar trends for both populations: Workers from the aggressive population Pen were overall bolder, more explorative, and more abundant in FORAGING compared with those of the peaceful population Kue, which needed more time to reach the maximum number of workers on honey (Fig. 4L).In RISK TAKING, the workers of population Pen were more abundant and faster, especially when feeding on honey.Additionally, significantly more workers of the population Pen were feeding for a longer time at honey compared with Kue when the temperature was rising (Fig. 5), which represents a risky situation.
Possible explanations for the behavioral differences between these two populations could be based on colony size, temporal polyethism, environmental conditions of the habitat of origin, or the aggression level (Suarez et al., 2002;Larsen et al., 2016;Segev & Foitzik, 2019;Maák et al., 2021).The colony size T. alpestre is similar to the colony size of Tetramorium caespitum, a congener, which has between 15 000 and 75 000 workers per colony (Sugawara & Nikaido, 2014;Seifert, 2017).Larger colony sizes in population Pen might be one explanation for the bold and explorative behavior observed.However, both populations are known since 2015, and we assume that all colonies were mature when tested and that other factors may explain the differences observed.
Another explanation for the behavioral differences between Kue and Pen could be the temporal polyethism of workers (i.e., workers conducting different tasks depending on their age).Li et al. (2014) found that older workers have a higher foraging efficiency, will likely die sooner, and are more willing to take risks than younger workers (Moroń et al., 2012;Hartmann et al., 2020).However, we do not expect an uneven distribution of polyethism stages to have an impact on behavioral differences because we likely collected and included workers with a large range of different behavior tasks (except for callow workers) in all behavioral experiments performed.
Environmental differences in the habitats may also explain behavioral differences.However, populations Kue and Pen exhibit relatively similar environmental conditions (similar flora and fauna, both south slopes).Moreover, no correlation was found between risk taking and standardized surface air temperature (TAS; Table 1).This suggests that the slightly longer foraging duration of population Pen in higher temperatures (RISK TAKING) cannot be due to living in a warmer habitat.One theoretical possibility is the fact that the colonies in population Pen are located at a slightly higher elevation (Pen = approximately 1950 m a.s.l.; Kue = approximately 1790 m a.s.l.; 160 m difference) and that these ants are adapted to greater environmental differences.However, we think that the behavior traits observed are unlikely influenced due to such a small elevation difference, although behavioral modifications due to environmental divergence are known (Segev & Foitzik, 2019).
In our opinion, the most plausible explanation for the behavioral difference between Kue and Pen is the presence of behavioral syndromes in this species.Behavioral syndromes are behaviors consistently coupled to each other over time and/or across contexts (Réale et al., 2007).In other animals, it was shown that boldness, exploration, foraging, and risk behavior are often correlated in some way (Chapman et al., 2011;Monceau et al., 2015;Maák et al., 2021).When aggressive behavior is included in such assays, it is also frequently positively correlated with other behaviors (Bengston & Dornhaus, 2014;Jandt et al., 2014;Blight et al., 2016).As hypothesized, we expected colonies from the aggressive population Pen to be bolder, display higher exploratory behavior, be faster in foraging, and be more willing to take risks compared with colonies from the peaceful population Kue (Table S1).Our results corroborate this hypothesis.Workers from Pen were bolder (BOLDNESS) and more risk-prone in RISK TAKING than workers from Kue.Additionally, BOLDNESS and EXPLORING behaviors were more often repeatable in population Pen than in population Kue.These findings combined with previous results that demonstrated a higher aggressiveness in the population Pen (Krapf et al., 2019;Krapf et al., unpubl. data) strongly indicate the presence of behavioral syndromes.We speculate that the two populations may be differently positioned in a trade-off continuum: Workers from population Pen are aggressive (based on earlier studies) (Krapf et al., 2019;Krapf et al., unpubl. data), bold, and forage more food (this study) but possibly have a lower life expectancy due to taking more risks.This, in turn, may negatively impact colony growth and reproduction in the long run (Sih et al., 2004;Modlmeier & Foitzik, 2011;Moroń et al., 2012;Jandt et al., 2014;Sih et al., 2015;Hartmann et al., 2020).In contrast, the opposites may apply to population Kue: Workers are peaceful (earlier studies) (Krapf et al., 2019;Krapf et al., unpubl. data) and shy and forage less (this study) and possibly display a longer life expectancy, which may promote colony growth and reproduction.These opposite positions in such a trade-off are the results of evolutionary processes that may have been driven by the adaptive value of predominantly either aggressive or peaceful behavior.
Within the same ant species, workers of different colonies are frequently aggressive toward each other (Julian & Fewell, 2004;D'Ettorre & Lenoir, 2009;Waddington et al., 2010).This, however, is not always the case: In T. alpestre, peaceful intercolonial behavior has been observed as has aggressive behavior (Krapf et al., 2018(Krapf et al., , 2019)).This behavioral variation was also seen in other multicolonial species such as Lasius flavus (Steinmeyer et al., 2012) and Monomorium pharaonis (Schmidt et al., 2010).In some situations, peaceful behavior between workers or colonies may be beneficial for the colonies and their stability over time and may increase their fitness.For example, more workers can each devote more energy and time to tasks like foraging, brood care, and expanding the territory, while aggressive behavior and fights entail trade-offs such as injuries and deaths (Steiner et al., 2007).However, aggressive behavior also provides advantages (Bubak et al., 2014).For example, aggressive colonies can be more successful in monopolizing a food source or a territory compared with peaceful colonies (Adams, 2016;Lessard et al., 2020).Interestingly, T. alpestre is predominantly peaceful but has not completely abandoned aggression, possibly highlighting an adaptive value of aggressive behavior for this species: Under specific (ecological) conditions, aggression may be important for finding food or expanding the territory.This could indicate that selection favors aggression and coupled behaviors and that selection may counteract the complete loss of aggressive behavior in T. alpestre.Nevertheless, aggression and its trade-offs are unlikely the baseline in this species as recently shown in a large-scale study across the European Alps, where colonies behaved predominantly peacefully (Krapf et al., 2023).Quantifying the specific costs of aggressive behavior and measuring their actual fitness consequences on different organizational levels will be one of the next steps herealso in light of recent findings that across the Alps, the level of aggression correlates with environmental conditions (Krapf et al., 2023).
Addressing and quantifying intra-and interspecific aggression and behavioral syndromes will also become more important for other ant species, especially invasive species, which cause large economic damage globally (Bertelsmeier, 2021).The few studies available addressing specific behaviors and invasive species such as Solenopsis invicta (Heller, 2004) and Linepithema humile (Blight et al., 2017) revealed that colonies in invasive ranges are more aggressive toward con-and allospecifics as well as quicker to explore novel habitats and detect food than colonies in the native ranges.These findings suggest that some behaviors that are important for the success of an invasive species might be coupled (e.g., aggression, foraging), at least in S. invicta and L. humile.Such coupled behaviors may thus allow colonies to establish and spread quickly in invasive ranges (Bockoven et al., 2015).However, more studies addressing behavioral syndromes in invasive ants are needed to assess if these observations can be generalized to other invasive species as well (Horna-Lowell et al., 2021).
So far, only a few studies have been addressing behavioral syndromes in ants (Chapman et al., 2011;Pinter-Wollman, 2012;Bengston & Dornhaus, 2015;Blight et al., 2016;Segev et al., 2017) and tested behaviors such as boldness, exploring, and foraging.However, studies addressing the underlying physiological pathways of behavioral syndromes are currently lacking as studies either investigate behavioral syndromes or physiological pathways.The studies addressing the latter found that biogenic amines such as octopamine, dopamine, and serotonin are linked with aggression (Bubak et al., 2016;Kamhi et al., 2017;Felden et al., 2019).Interestingly, gene expression of these amines is suggested to be linked to behaviors such as activity and aggression (Kamhi et al., 2017;Friedman et al., 2018), and a dopamine receptor gene has been associated with behavioral syndromes in birds (Fidler et al., 2007;Korsten et al., 2010).Whether the same or a similar link also exists between behavioral syndromes and these genes in ants needs to be addressed in future work.Additionally, future studies should also test more diverse behaviors that capture a more complete range of behavioral tasks of ants such as risk-taking, brood care, aggression, and foraging.
To summarize, in our study using the high-elevation ant T. alpestre, we (a) evaluated the experimental setup (i.e., the combination of experiments and parameters used) and detected that BOLDNESS and EXPLORING are two favorable behaviors for quantitative analyses.(b) In addition, we compared two behaviorally different populations (one aggressive and one peaceful) and detected behavioral differences between both populations with the aggressive population being generally bolder and more risk-prone.(c) Further, we evaluated the repeatability of the behaviors tested and observed repeatability in several behaviors.The results indicate that boldness, exploring, and risk-taking behaviors are coupled in the aggressive population possibly indicating that they are also coupled with aggression thus suggesting the presence of behavioral syndromes.This hypothesis is in line with findings from other ant species in which behavioral syndromes consist of coupled behaviors such as aggression, boldness, and exploratory (Bengston & Dornhaus, 2014;Blight et al., 2016).Interestingly, T. alpestre is mainly peaceful in intercolonial interactions (Krapf et al., 2018(Krapf et al., , 2019)), with some exceptions at the colony and/or population level (Krapf et al., 2023).We speculate that aggression can have an adaptive value (benefit-cost ratio): Workers behaving more aggressively may also be bolder and risk-prone compared with more peaceful workers.Such workers may forage more food thus promoting colony growth and reproduction.In turn, those colonies may have a selective advantage compared with less aggressive colonies.We further speculate that under specific ecological conditions such coupled behaviors may persist over time and thus counteract the complete loss of aggression in this species.

Fig. 2
Fig. 2 Setup of experimental nests used for laboratory maintenance (A1, A2) and detailed experimental setup (B).Experimental set up for Experiments 1 and 2 (red arrow marks the junction) (C) and for Experiments 3 and 4. Please note that Experiment 4 includes a heating plate (D).Created with BioRender.com

Fig. 3
Fig. 3 Repeatability and change over time of the six behaviors and their corresponding parameters.Significant repeatability results of the different levels (replicates and colony) are marked with green dots in the header of each plot.Arrows within squares and blue trend lines indicate the directions of significant changes over time.S = seconds, Max = maximum, n = number, Kue = peaceful population Kühtai, Pen = aggressive population Penser Joch.

Fig. 5
Fig.5Linear regression between workers feeding on honey and the rising temperature displayed for both populations.The regression was calculated using a linear model and includes a significant level, correlation equations, and the coefficients of determination R 2 .n = number.Kue = peaceful population Kühtai, Pen = aggressive population Penser Joch.

Table 1
Colony characteristics and numbers of replicates used for the experiments.