Sex-dependent resource defense in a nectar-feeding bat

Aggressive resource defense spans from the transient monopolization of a resource up to the long-term maintenance of a territory. While such interference competition is common in nectar-feeding birds, reports in nectar-feeding bats are rare. Glossophaga bats have been observed to temporarily defend flowers but the extent of this monopolization, its effects on nectar intake, and underlying sexual differences have remained unknown. We investigated resource defense behavior of Glossophaga mutica in the laboratory. We presented bats with two patches of computer-controlled artificial flowers and tracked individual nectar intake. Furthermore, we established an automated method for detecting aggressive interactions at the artificial flowers. Theoretical models of interference competition predict more aggressive interactions when resources are spatially more clumped. To test this, we varied resource distribution across two patches from clumped to dispersed and monitored bats’ interactions in one male, one female, and four mixed-sex groups. Males engaged in aggressive interactions more often than females and in each group some individuals defended clumped artificial flowers against others. Subordinate males experienced a substantial decrease in nectar intake, while females were only marginally affected by male aggression. These results suggest that aggressive interactions and their effect on nectar intake are sex-dependent in G. mutica. Furthermore, aggressive interactions were more frequent and resource defense was only successful when resources were clumped. Our experimental set-up allowed us to perform an automated test of models of interference competition with a mammal under controlled laboratory conditions. This approach may pave the way for similar studies with other animals. Lay summary Males bully other males to get more food, but only when food is easy to defend. When flowers are spread out nectar-feeding bats rarely engage in fights. However, when there are rich flowers in one spot and no flowers elsewhere, some males start attacking others, denying them access to the nectar. Females do not seem bothered by such male bullies, but when there are no males around, some females become bullies themselves.

the whole night; only the spatial distribution of food changed from the clumped resource treatment with one 142 patch rewarding (five flowers) during the first phase of the night to the dispersed resource treatment with 143 two patches rewarding (ten flowers) during the second phase of the night. With this experimental schedule, 144 the maximal amount of nectar the bats could collect was 108mL, which corresponds to 18mL nectar per 145 individual per night, roughly 150% of their daily requirement (Winter and von Helversen 2001). The side of 146 the rewarding patch during the first phase of the night was chosen pseudo-randomly and the same patch 147 was never chosen in more than two consecutive nights. For the mixed groups, the duration of the clumped 148 resource treatment was six hours and the experiment lasted nine nights (seven nights for the first mixed 149 group). For the same-sex groups, the duration of the first part of the night was variable (range = 4-8h, mean 150 = 6h) and the experiment lasted eight nights for the male group and seven nights for the female groups. 152 We took the frequency of individuals chasing each other at the artificial flowers as an indicator of the intensity 153 of aggressive interactions between group members. We developed a method to automatically detect and score 154 chasing events using the computer-collected animal identification data from the RFID sensors and flower 155 sensors. In a previous pilot study (Wintergerst 2018), three mixed groups of bats were video recorded for 24h 156 over 14 nights, and the video data were synchronized to the computer-collected data. During this pilot study and only 4.1 ± 2.8 during the training nights, mean ± SD) for the 36 participating bats, we considered the automated approach adequate for quantifying within-group dominance relationships. The total number of 175 individual detections per night constrains the number of chasing opportunities detectable with our method. 176 Therefore, we calculated a chase score and a chased score by dividing the number of observed chases (directed 177 to others or received from others, respectively) for each bat by the total number of detections for that bat on 178 each night.

(d) Chasing behavior
(e) Statistical analysis 181 To investigate the difference in chasing behavior between males and females and between the resource 182 treatments (one versus two rewarding patches) a Bayesian generalized linear mixed model (MCMCglmm,183 Hadfield 2010) with a binomial error structure and a parameter expanded prior was used. Body weight as an 184 approximation of size and the full interaction between resource treatment and sex were included as fixed 185 effects and the influence of these fixed effects on the proportion of chasing events was assessed. Experimental 186 group and individual were included as random effects. The same model structure was used to address the 187 question whether the proportion of visits on which the visitor was chased was influenced by these independent 188 variables. If one or more individuals start to defend flowers and thus exclude others from drinking, nectar 189 consumption should increasingly differ between individuals since the successful chaser should gain a higher 190 nectar intake thus reducing the intake of the chased individuals. Therefore, the between-individual difference 191 in nectar consumption over the course of the experiment was compared between experimental groups and 192 resource treatments (clumped vs. dispersed). First, each individual's total nectar consumption standardized 193 by the number of hours of foraging during the clumped (one rewarding patch) and dispersed (two rewarding 194 patches) resource treatment was determined for each experimental night. Then these data were used to 195 calculate group standard deviations, separately for the males and females of each group. In order to assess 196 the influence of resource defense on the individual differences in nectar consumption (standard deviation 197 of nectar intake) we fit a MCMCglmm model with a Gaussian error structure and the following fixed effects: 198 sex, experimental night (centered), and resource treatment (clumped or dispersed), as well as all two-way 199 interactions. Again, we included group and individual as random effects.

215
(a) Behavioral observations 216 The goal of our experiment was to investigate the sex-specific effects of resource defense in Glossophaga 217 mutica, in addition to the potential influence of interference competition on individual nectar intake. Quali-218 tative behavioral observations of four hours of video recordings revealed several behaviors that seem to be 219 characteristic for some males, which according to further analyses (see below) we designated as dominant 220 males. Instead of just visiting the flowers and leaving the patch as the other individuals did, dominant males 221 remained hanging between the flowers within the patch for a significant amount of time (Fig. S1). When 222 other individuals came close due to visits of directly adjacent flowers, dominant males often spread one wing in the direction of the other individual which could be interpreted as a threatening posture. Some individuals 224 were attacked and chased away by dominant males while visiting artificial flowers. In this case, dominant 225 males mostly attacked from above with their mouth wide open, and followed the intruder for a short distance.

226
Sometimes the chasing escalated into fighting with both bats tumbling towards the ground and resuming 227 their flight only shortly above the floor. In rare cases, these fights might have led to small injuries. One 228 subordinate male had several fresh scratches on its wing that were not present before the experiment and 229 that were possibly caused by bites (Fig. S2). After a successful flower defense, the dominant male normally 230 visited most of the five flowers within the patch before returning to its hanging position between the flowers.

231
(b) Example of nectar intake in one experimental group 232 One of the first striking observations we made was the uneven distribution of nectar consumed between the 233 sexes and individuals. For example, in the first mixed group of bats tested, after only two nights the nectar 234 consumption of two males was nearly reduced to zero, whereas the third male increased its consumption 235 substantially ( Fig. 2A). However, this pattern occurred mostly for males during the clumped resource   Table 1). Notably, the frequency of females as active chasers in female-only groups was higher 244 than chasing by females in the mixed groups (Fig. 3A). Although the rate of nectar availability remained 245 constant throughout the night and only the spatial distribution of the resources changed, chase frequencies 246 were significantly lower during the dispersed resource treatment when rewards were available at both patches 247 (Table 1). There was no significant difference between the sexes in how often a bat was chased by another 248 individual ( Fig. 3B) but individuals were chased less during the dispersed resource treatment (Table 1).

249
Weight as an indicator of size had no significant effect on either the chase score or the chased score (Table 1). . During the clumped resource treatment (first part of the experimental night) rewards were only available at one patch. The nectar consumption of two subordinate males approached zero after only two nights, whereas the third, dominant, male greatly increased nectar intake during the experiment (males filled symbols). Females (open symbols) on the other hand maintained a stable level of nectar intake. (B) During the dispersed resource treatment (second part of the experimental night) rewards were available at both patches. Under this treatment, individuals nearly equalized their level of nectar intake over the course of the experiment. The second part of night 4 was excluded due to technical problems.   Figure 3: Sexes differed in the frequency of chasing or being chased during the clumped resource treatment.
(A) Males (dark symbols) chased others significantly more than females did (light symbols, Table 1). Shown are the individual proportions of chasing events (chase scores) over the whole experiment. Notably, in the females-only groups some females chased more than any female in the mixed groups. (B) Being chased by other bats did not differ significantly between sexes (Table 1). Resource defense should lead to a larger between-individual difference in nectar consumption (Brown 1964).

252
Differences in nectar consumption were quantified as the standard deviation of nectar intake in each group, 253 separately for males and females. During the clumped resource treatment, the standard deviation was higher 254 for males than for females (     To our knowledge, this study is the first report of sex-dependent differences in resource defense behavior of 308 neotropical nectar-feeding bats. In mixed sex groups, females seemed to be much less affected by the behavior 309 of dominant males whereas subordinate males were excluded at least partially from the defended flower patch.   (Table S1). In single-sex groups, but not in mixed groups there was a 347 higher frequency of chases in the clumped than in the dispersed resource treatment (Table S1). The chased 348 score was only affected by treatment but not by group type and was higher in the clumped than in the 349 dispersed resource treatment (Table S1). Over the course of the experiment, the standard deviations in 350 nectar intake increased for females in the single-sex, but not in the mixed groups (Table S2). This increase 351 was only significant in the clumped, but not in the dispersed resource treatment (Table S2). The standard 352 deviations were higher in the single-sex groups than in the mixed groups, both in the clumped (group type 353 effect in Table S2) and in the dispersed resource treatments (estimate = -0.06, 95% CI = -0.12, -0.01). Thus,

381
The ability of an individual to successfully defend and monopolize resources is often correlated with distinct 382 physical characteristics such as body size (Searcy 1979). However, in our results weight as an approximation 383 of size did not correlate significantly with the chase score of individuals (Table 1) and therefore did not 384 predict which male succeeded in defending a flower patch. Another factor that could influence the success 385 in defending flowers is age and therefore experience (Yasukawa 1979;Arcese 1987). Since we could only 386 discriminate between young and adult animals, we cannot dismiss age and experience as a predictor of 387 successful flower defense.

388
In this study, subordinate males received considerably less nectar than dominant males and females (Fig. 5).

389
However, except in mixed group 1, subordinate males were rarely completely excluded from the flower patch  There were 89 chase occurrences observed (f->f 4 times, f->m 2 times, m->f 59 times, m->m 24 times).

421
Every time the algorithm marked an event as a chase event, there were two individuals following each other.

422
Some chase sequences did not get detected. The individual that chased never drank immediately after the 423 chase at the same flower location where the chase occurred. There were 16 incidences that were difficult to 424 classify by observation or did not appear to be aggressive interactions.         Figure S25: Glicko ratings within the six experimental groups. Over the last two experimental nights, the males (closed symbols) with the highest chase scores were also the individuals with the highest Glicko rating in each group (panels) during the clumped resource treatment. In female-only groups this correspondence was found only in group 2. Numbers at symbols give the Glicko rating in thousands.    Figure S28: Distribution of rewarded visits across flowers for the six bats in the males-only group. The colored bars give the number of rewarded visits of each individual at the ten flowers during the clumped (top) and dispersed (bottom) resource treatments for each experimental night (columns). The dominant males are shown with stripes and the subordinate males are shown with dots. This was the only group with two males behaving as dominant. On the last night, rather than sharing all flowers within the defended patch, the dominant males partitioned the patch into two subpatches, with each bat defending its own partition.  Figure S31: Distribution of rewarded visits across flowers for the six bats in mixed group 3. Same notation as in Fig. S29, but the colors correspond to different individuals. Due to a technical malfunction on night 9, there were no rewards delivered in the dispersed resource treatment and the data were excluded from analysis.  Competing interests 439 We declare we have no competing interests.