Behavioral dominance interactions between Nicrophorus orbicollis and N. tomentosus burying beetles (Coleoptera: Silphidae)

Species collection and identification

We collected live N. orbicollis and N. tomentosus during July and August 2018 using non-lethal pitfall traps set on the properties of the Queen’s University Biological Station (44.5653°N, 76.3242°W, 115–170 m elevation) near Elgin, Ontario. We identified all Nicrophorus beetles to species and sex following Anderson & Peck (1985) and placed each beetle individually into 4oz clear plastic containers (Reditainer, model no. RTSC100400) containing moist soil and ventilated with small airholes, with their species and sex marked on the lid. All beetles were housed outdoors in shade with natural light and temperature conditions until used in our trials; individuals were held in captivity 4–14 days before trials began. Beetles held for more than 1 week were fed ground beef every 7 days. All beetles were given water every 5 days, at which time the soil was also moistened.

Captive experiments

Trials involved single N. orbicollis and N. tomentosus of the same sex competing for one mouse (Mus musculus) carcass. Mouse carcasses were donated by the Queen’s University Health Sciences breeding facilities. The mice were culled using carbon dioxide gas and immediately frozen at −20 °C. For our trials, we used mice weighing 25–30 g, and thawed carcasses for 48 h at room temperature before experiments. We randomly selected burying beetles of the same sex representing our two focal species based on the lid labels, without regard to size, but within the same capture cohort, such that two individuals within a trial were originally captured within a day of each other. Nonetheless, we did not know the nutritional condition of our beetles or their feeding history prior to capture, and differences among individuals in condition could have affected our results.

Each pair of beetles was placed into a clear plastic container (31.4 × 20.8 × 11.9 cm) with 2 cm of fresh top soil and a mouse carcass placed in the center. Inside the container (fastened to either end) were two red, battery-powered, 5-LED bicycle lights for illumination; these red lights appeared to have little effect on beetle activity. We then covered the container with a glass pane taken from a picture frame. Trials ran from 18:00 until 07:00 the following day (13 h total), and were subject to natural light at a window, but controlled temperatures (21–24 °C). Sunset times ranged from 20:18 to 19:33; sunrise ranged from 06:04 to 06:35. We ran trials 10 August–6 September 2018; trials attempted after 6 September were abandoned because N. orbicollis ceased preparing carcasses for breeding. We recorded behaviors and interactions through the night using a Sony Handycam digital video recorder (DCR-SR100; Sony Corporation, Tokyo, Japan), with the “nightshot” option that allowed us to record video at night using the red lights. We covered the “nightshot” red light source of the video camera because it reflected off the glass, obscuring the view of the beetles, and instead relied on the red bicycle lights installed under the glass. After each trial, we disposed of the soil and carcass, washed the container with warm water, and introduced fresh soil. In total, we conducted 14 trials (7 male-male, 7 female-female).

Body size

After each trial, the individual beetles were frozen within sealed plastic bags for later measurement. Within a few weeks of completing the trials, we weighed each beetle to an accuracy of 0.001 grams using a GEM20 Smart Weigh jewelry scale (Smart Weigh Packaging Machinery Co., Ltd., Zhongshan City, Guangdong Province, China) and measured the width of the pronotum and length of the elytra to the nearest 0.01 mm using digital calipers (Table 1). To control for variation in the scale measurements, we also weighed a known 10 g standard and calculated beetle mass as: (beetle mass (g)) = (uncorrected weight of beetle (g)) × {(10g)/(weight of the 10g standard (g))}. We did not weigh beetles prior to the experiments because we wanted to minimize handling effects on their behaviors at the beginning of our experiments.

We preferentially used mass as an index of body size in our statistical analyses because elytra length and pronotum width may vary across species independent of size (e.g., as a result of adaptation to different ecological strategies), mass is directly important in some aspects of aggressive contests (e.g., inertia), and mass has been a standard estimate of size in previous work (Peters, 1983). Previous work also suggests that mass and elytra length are highly correlated within our two focal species at our study site (Collard et al., 2020). Nonetheless, we repeated our analyses with elytra length and pronotum width as indices of body size to ensure that results were similar (see “Statistical Analyses” below).

Categorization of interactions

After conducting our experiments, we reviewed the video and categorized all interactions between our two focal species over the 13 h of each experiment. We first defined an interaction between beetles as any occasion when two individuals came within 1 cm of each other. We used 1 cm as a cutoff point for interactions because our previous observations suggested that two individual beetles will be aware of each other’s presence at this distance. Beetles may also be aware of each other at distances greater than 1 cm (e.g., through olfaction or stridulations), and thus our definition of an interaction may underestimate the overall number of interactions among beetles. Nonetheless, we had difficulty determining if an interaction was occurring at distances greater than 1 cm, so we excluded these from this study. For each interaction, we defined the initiator of the interaction as the individual that moved towards the other, such that the two individuals came within 1 cm of each other. We did not designate an initiator in cases where both individuals moved towards each other, bringing them <1 cm apart.

We separated interactions between beetles into four categories: (1) physical interactions and chases, (2) avoidance interactions, (3) neutral interactions, and (4) other interactions. Physical interactions and chases were defined as interactions that involved grappling, biting, chasing, and digging after a buried beetle. We further noted whether the outcome of these interactions was symmetric (no winner or loser in the interaction) or asymmetric (one beetle won and the other lost the interaction). A beetle that won an interaction retained its position, whereas a beetle that lost an interaction retreated (moved away) from the winner or carcass (no beetles were physically injured in any of our trials, thus all beetles that lost an interaction were able to retreat).

Avoidance interactions included interactions where we observed evidence for a dominant and subordinate, despite there being no physical contact or overt chasing. Examples include cases where one individual actively avoided the other individual (e.g., changing direction or approach) after coming within 1 cm of each other. Within avoidance interactions, we defined the subordinate as the individual moving away from the resource (the mouse).

Neutral interactions included all instances where an individual moved to <1 cm of the other without any sign of aggression or negative influence on each other’s behaviors (e.g., no alteration of their path of movement or avoidance of the other beetle). Examples of neutral interactions include: the two beetles passed by each other without changing their direction, both beetles moved away from the mouse without avoidance or aggression, both beetles fed on the mouse without aggression, and one beetle passed over the other beetle (buried under the soil) without digging after or attacking it.

Some interactions were difficult for us to categorize. In some cases, interactions began as one category (e.g., neutral interactions) but changed over time into another category (e.g., physical interactions and chases). In these cases, we considered the sequential aggressive and non-aggressive interactions as distinct, and considered the initiator of the second interaction as the individual that closed the remaining distance between the two beetles. In other cases, both individuals were under the mouse simultaneously, and thus, we could not observe the occurrence or details of any interactions; in these cases, we recorded no interactions, or “na” for any component of an interaction that could not be observed (e.g., the initiating species). We designated “other interactions” as interactions where we could not view the details of the interaction (e.g., physical contact vs avoidance), but nonetheless could tell that an interaction took place, and in some cases, the outcome of the interaction (e.g., one beetle retreated from the other).

Assessment of behavioral dominance

Within each trial (n = 14), we designated the behaviorally dominant individual as the individual who won the most interactions with asymmetric outcomes (i.e., a winner and a loser), and the subordinate as the other individual (following Drews (1993)). Dominance designations based only on physical interactions and chases (i.e., avoidance and other interactions were excluded) were identical to dominance designations based on all interactions in 12/14 trials.

First species to arrive at the carcass

For each of our 14 trials, we recorded the first beetle to physically touch the mouse as the first species to arrive at the carcass. This measure allowed us to test for priority effects on the outcomes of aggressive contests for the carcass on a time scale of minutes to hours, but did not address the potential for priority effects occurring on longer time scales (such as over days) that could also be important in nature.

Behavioral dominance and distance to carcass

Competition experiments cannot always record the outcomes of individual fights, and thus we measured the distance of each beetle to the mouse carcass (cm) at the end of each trial (07:00) for future use as a proxy for behavioral dominance. Previous observations suggest that behaviorally dominant beetles will usually stay in close proximity to the carcass, while subordinate species are displaced further away (Wilson, Knollenburg & Fudge, 1984). We tested this hypothesis across the 14 trials in our experiment, predicting that the behaviorally dominant beetle (i.e., the beetle that had won most interactions) would be closer in proximity to the carcass than the subordinate, and thus the proximity to the carcass at the end of the trials would provide a useful estimate of behavioral dominance in competitive experiments.

Statistical analyses

We performed all of our plotting and analyses in R (version 3.5.2; R Development Core Team, 2018).

We first tested among our alternative hypotheses using a binomial generalized linear model, with the proportion of asymmetric dominance interactions (i.e., physical interactions and chases, avoidance interactions, and other interactions with an asymmetric outcome; Table 2) won by each species as the response variable. Specifically, we combined the number of interactions won by N. orbicollis, and the number of interactions won by N. tomentosus, into one response variable (using the cbind command in R; Crawley, 2013), with the relative mass of the two species (ln (mass N. orbicollis in g/mass N. tomentosus in g)) and the first species to the carcass as the predictor variables in a saturated model (i.e., all interactions included). This model was overdispersed, so we corrected the standard errors using a quasibinomial generalized linear model. We tested the hypothesis that species identity consistently predicted interaction outcomes using the intercept of the model, which provides a test of asymmetry in the interaction outcomes when the difference in mass between individuals is zero. We removed one trial that had no interactions between the two species during the 13 h of observation.

We ran models with all possible combinations of predictor variables and compared their performances using quasi Akaike information criterion values (QAIC) and the dredge command in the R package MuMIn (Bartoń, 2019). We calculated QAIC values because AIC values are not available for quasibinomial generalized linear models; we calculated QAIC values following Bolker (2014). We identified the best-performing model as the model with the lowest QAIC value. We summarize the results of the best-performing model in the Results section, and present model performance results in Table 3. In addition to running models with mass as our proxy for body size, we repeated our analyses with both elytra length and pronotum width as estimates of body size.

Interactions among the species

The total number of interactions, and their distribution among the different categories, varied between trials, with most interactions being neutral (Table 2). Most individuals engaged in both physical interactions and chases and avoidance interactions, with no clear patterns between trials (Table 2). The majority of physical interactions and chases were asymmetric, with a clear loser retreating from its position. These interactions usually consisted of grappling or chasing, with some other less common interactions, including digging after the other beetle. No beetles were visibly injured due to an aggressive interaction. Interaction time varied across experiments with an average of 11.85 +/− 40.65s (SD; n = 6), with neutral interactions lasting longer than other interactions. Both species initiated interactions with asymmetric outcomes (i.e., a winner and loser), despite N. orbicollis usually winning (Fig. 1; Table 2). N. orbicollis initiated more interactions with an asymmetric outcome than N. tomentosus in 9 of the 14 trials (Table 1); N. orbicollis initiated an average of 14.50 +/− 11.64 (SD) interactions, while N. tomentosus initiated an average of 10.43 +/− 12.13 (SD) interactions (Table 2).

Factors predicting behavioral dominance

Larger relative body size was the best predictor of behavioral dominance between N. orbicollis and N. tomentosus. Our best-performing model (Table 3) included only relative body size as a predictor variable (quasibinomial linear model, estimate = 1.60 +/− 0.69SE, t = 2.3, p = 0.04, n = 13; Fig. 1). The intercept of the model—our test of a species effect—was not significantly different from zero (estimate = −0.20 +/− 0.46SE, t = −0.43, p = 0.68), while the first species to arrive at the carcass (i.e., priority effect) was not retained in any top model (Table 3). The first beetle to the carcass arrived between zero (i.e., the beetle went almost immediately to the carcass, and was on the carcass by the time we started recording) and 159.0 min (mean = 32.1 min) after video recording began; the second beetle arrived 1.2–193.2 min (mean = 45.7 min) after the first (Table 2).

Generalized linear model results were similar if we used elytra length or pronotum width as indices of body size (elytra length: best-performing model included only elytra length as a predictor; elytra length estimate = 5.83 +/− 2.16SE, t = 2.7, p = 0.02, intercept = −1.10 +/− 0.70SE, t = −1.6, p = 0.14; pronotum width: best-performing model included only pronotum width as a predictor; pronotum width estimate = 3.73 +/− 1.78SE, t = 2.1, p = 0.06, intercept = −0.19 +/− 0.49SE, t = −0.39, p = 0.70).

Behavioral dominance and distance to carcass

When the two species differed in their distance to the carcass (measured at 07:00, just after the competition experiment), the closest individual to the carcass also won the most interactions in 10 of 12 trials (83%; binomial test on accuracy, p = 0.039). In only one trial was the subordinate individual (i.e., the individual that lost most interactions) closer to the carcass than the dominant individual (in the other trial, the two beetles won the same number of interactions). In the remaining two of the 14 trials, the two beetles were equidistant from the carcass at the end of the experiment; in one of these trials, the two beetles also won equal numbers of interactions, and thus the distance to the carcass accurately described the lack of a dominance relationship. In one of the 14 trials, the two beetles were not observed to interact during the 13-h experiment (Table 2); however, the larger N. orbicollis was closer to the carcass at the end of the experiment.