Introduction

The resources that animals rely on are rarely homogeneously distributed in space. The resulting patches typically vary with respect to properties that govern the quantity and quality of resources available in any given area. These properties can determine the presence and abundance of the individuals that depend on these patches (Karczmarski et al. 2000; Mazía et al. 2006; Shima et al. 2008; Rukke 2013). As patches age or change through time, dynamic patterns of use emerge within a habitat, such as those seen in ecological succession (Wiens 1976). Patches with more high-quality resources generally attract higher densities of consumers (Pettorelli et al. 2001; Shima et al. 2008; Mortelliti et al. 2014), and, as a result, generate conditions that foster more conspecific interactions (Freeberg et al. 2012; Webster et al. 2013; Leu et al. 2016).

Individual and sex differences in motivational, social, and reproductive factors drive differential use of habitat patches among individuals within a population (Emlen and Oring 1977; Main and Coblentz 1996; Pröhl and Berke 2001). Sex-specific trade-offs in fitness components can be reflected in the strength and type of habitat preference (Fuchs 1980; Stokke 1999; Loe et al. 2006). In some species, males use habitat patches whose qualities incur greater predation risk but provide higher resource levels, while females prefer the opposite (Rowe 1994; Merilaita and Jormalainen 1997, 2000). One sex may be driven to avoid some patch types due to their attraction of high densities of the other sex (Eldakar et al. 2009; Turlure and Dyck 2009). Conspecific interactions incur variable costs depending upon the social composition of those interactions. In water striders, for example, male-male competition and male–female harassment behavior drives individuals to avoid certain patches and prefer others (Krupa and Sih 1993; Eldakar et al. 2009). In other cases, female preference for certain habitat patches may draw males to them, and depending on the patchiness may result in male-male competition over the high-value patches (Shuster and Wade 2003). The consequences of conspecific interactions and the sex-specific preferences for different habitat characteristics could generate differences in the kinds of individuals and the behaviors that occur in different patches.

As the environment alters where and how individuals interact with conspecifics, it directly and indirectly influences the social structure of a population. Ecological constraints are thought to lead the evolution of social systems (Faulkes et al. 1997; Shuster and Wade 2003; Lövy et al. 2012). Characteristics of habitat patches can impact variation and polymorphism in social behavior within species (Corley and Fjerdingstad 2011; Gall and Manser 2018). Interactions driven by variation in resource properties can themselves mediate how organisms exploit and compete for resources present in higher value patches (Goubault et al. 2005). Alternatively, individuals may choose to occupy patches based on the presence of conspecifics, either as an indirect indicator of habitat quality or as a direct effect of preference for higher density (Donahue 2006). Resource quantity, quality, and distribution can impact multiple aspects of social interactions, so it is critical to consider these habitat conditions when evaluating the patterns and evolution of social behavior in natural populations (as reviewed in (He et al. 2019)). As habitat patches age, a cross-sectional approach can be used to view the effects of different stages of succession or different patch quality on the amount and type of social interactions.

To investigate how variation among aging resource patches may shape social interactions, we examined how forked fungus beetles (Bolitotherus cornutus) used host fungal resources for a variety of social and reproductive behaviors. B. cornutus live in temperate forests in eastern North America and are found on dead logs that are colonized by wood decaying fungi (Fig. 1) (Liles 1956). These beetles use the fruiting bodies of host fungi (brackets) as discrete microhabitat patches on the surface of a log in all stages of their life cycle and conduct most of their social interactions on brackets. Previous work has shown that variation in both reproductive and non-reproductive social behavior has substantial fitness consequences within natural populations of forked fungus beetles (Conner 1988; Formica et al. 2012, 2016b).

Fig. 1
figure 1

Living Ganoderma applanatum brackets on a fallen log with white hymenium visible on the underside. Inset: Partially decayed dead bracket with many holes visible

Full size image

Across taxa, females have been found to select habitat characteristics that fulfill their specific reproductive needs (Orians and Wittenberger 1991; Stokke 1999; Pröhl and Berke 2001). Males congregate in these preferred patches to maximize mating opportunities, and as a result male spatial distribution can also be governed by female reproductive needs (Shuster and Wade 2003; Przybylski et al. 2007). To best characterize the drivers of habitat preferences in both sexes, we must examine how females use their habitat for reproduction. Female B. cornutus in captivity exhibit preferences for different fungal species for oviposition, though the specific properties of the fungi that correlated with these choices were not examined (Wood et al. 2018). Larval cannibalism is frequently observed in B. cornutus and could influence female preference for brackets that minimize cannibalism of her offspring (Wood et al. 2014). However, other than preference for certain fungus species, the factors involved in female oviposition decisions in B. cornutus remains largely unknown. Female preference for oviposition sites may then drive greater female and male presence at brackets, resulting in greater social interactions between and within-sexes. Variation in the kinds of social interactions on different patch types that occur as a result of variation in bracket preference is largely unexplored (but see Costello et al. 2022b). Two aspects of fungus fruiting bodies that are commonly thought to impact the presence and abundance of insects are bracket size and presence of living growth. Generally, adult B. cornutus preferred less decayed fungi under experimental conditions (Heatwole and Heatwole 1968). In other insect-host fungus relationships, mycophagous beetles also were more commonly encountered on less decayed brackets (Jonsell and Nordlander 2004). However, fungus brackets with live growth may be more chemically defended, deterring consumers and producing different preferences for beetles in their natural habitat.

Here we test whether properties of individual fungus brackets play a significant role in structuring the type and frequency of social interactions in natural subpopulations. We predicted that female beetles would oviposit more often on live brackets, and that beetles of both sexes would visit live brackets more often following the pattern observed by Heatwole and Heatwole (1968) because of the value of living tissue as an oviposition site. Female abundance is expected to be more strongly predicted by patch properties than male abundance, due to the importance of bracket quality on offspring success. By scoring the social interactions observed on fungal brackets, we further examined how the habitat may shape the social environment. We predicted that between-sex interactions would primarily occur at brackets preferred by females for oviposition, as males are more likely to encounter a female beetle there and are known to court laying females in an attempt to mate. Male-male interactions, which can result from competition over females, are expected to occur at patches frequented by females and with higher numbers of between-sex interactions.

Methods

Study system

The brackets (fruiting bodies) of several species of polypore fungi (Ganoderma applanatum, Ganoderma tsugae, and Fomes fomentarius) are central to all aspects of the B. cornutus life cycle (Liles 1956). Adults and larvae consume the fungus, females oviposit on bracket surfaces, laying one egg at a time over the course of 30 min to 1 h; larvae develop for up to 48 months within brackets (Pace 1967) until they eclose as adults (Wood et al. 2018). Most adults spend an entire breeding season on a single log, leading to repeated social interactions on the surface of the fungal brackets (Formica et al. 2011, 2012, 2016a, b). Competition for access to mates, extended courtship behavior, copulation, and mate guarding also take place on, or adjacent to, the bracket surfaces. Males have two sets of horns that they use in combat with other males (Conner 1988). Male body and horn size experience strong sexual and social selection in low density populations (Conner 1989; Formica et al. 2021). Furthermore, non-mating social interactions, as determined by close physical proximity, have fitness consequences for both male social partners (Formica et al. 2011, 2012).

We focused our study on one species of host fungus (G. applanatum) with perennial brackets that most frequently hosted the largest subpopulations of beetles in our study area. The brackets of this species are woody and can remain on a log for several years. A single colonized log, which may include one or more genotypes of fungus, can produce anywhere from a few to over one hundred brackets during the lifespan of the fungus. Each spring and summer, new brackets emerge, and some existing fruiting bodies add new hymenium tissue through the growing season, expanding the surface area of the bracket. Eventually individual brackets stop getting replenished with new hymenium and start to decay, even while new brackets are being produced on the same log. The variable rates of generation, maintenance, and decay of brackets creates a patchy and dynamic landscape of resource quality for fungivorous B. cornutus (Paviour-Smith 1960).

Field methods

We conducted observational scan sampling on ten unique logs colonized by the fungus G. applanatum during the summers of 2015 and 2016 (six in 2015 and ten in 2016) within the Pond Drain metapopulation at Mountain Lake Biological Station in Giles County, VA (37.376°N, 80.522°W) (Table 1). Each sampled log is a subpopulation selected because it was one of the largest subpopulations of B. cornutus in the forest and made up of mostly G. applanatum brackets. Each bracket was tagged with a permanent aluminum marker (National Band & Tag Company, Style 50, ½ inch round). We measured the height, width, and depth of each bracket at its widest point. We approximated the surface area from these measurements using the formula for a half cylinder. Using surface area allowed us to reflect the available social space for beetles on a bracket, both on top of and under the brackets. We categorized the condition of the bracket as “live” (a bracket with any white hymenium on the underside) or “dead” (a bracket with no white hymenium on the underside). Dead brackets subsumed considerable variation in the level of decay, including brackets that ranged from fully intact but without live growth to those that had significant rotting and structural damage. Bracket senescence is highly variable. In yearly surveys of G. applanatum conducted by our lab from 2015–2020, new live brackets in one survey were generally observed to be dead in the next year, though some brackets continued have live growth for up to four years (unpublished data). Brackets were observed to persist without live growth for anywhere from 1 to 5 years (unpublished data). Brackets on each subpopulation were surveyed three times a day (0600, 1400 and 2100) from June through August in 2015 and 2016. During sampling, we observed each bracket and recorded the sex and social interactions of all B. cornutus present on it. It was not possible to record data blind because our study involved focal animals in the field. These analyses do not contain information about individual beetles. Communication between insects occurs at close distances allowing chemical, visual, and acoustic information to be transferred between pairs (d’Ettorre and Moore 2008), we followed previous work that defined social interactions as instances when two or more beetles were observed within 5 cm (~ 2 body lengths) of one another (Formica et al. 2011, 2012). Specific behaviors of each beetle were also recorded. Courtship and post-insemination mate-guarding each last for several hours and are recognizable from distinct behavioral postures of males and females (Brown et al. 1985; Brown and Bartalon 1986).

Table 1 Subpopulation demographics
Full size table

Statistical analysis

To evaluate the effect of bracket properties on the presence of, and use by forked fungus beetles, we developed four generalized linear mixed models using the glmmTMB package in R (Brooks et al. 2017; R Core Team 2022; RStudio version 2022.12.0; Table 2). In each model, the unit of analysis is a bracket, centering our analysis on the qualities of the environment as opposed to the choices of an individual beetle. For Model 1, we asked how the condition and size of the bracket influenced the total number of visits by male and female beetles. The dependent variable was the total number of recorded beetle observations on each bracket over the entire breeding season and included all G. applanatum brackets on all subpopulations (n = 432 brackets). Each bracket was included in the analysis twice, once for a count for males and once for a count for females and included zero-counts; the bracket ID was fitted as a random effect to adjust for repeated measures (see Table 2 for details on all fixed and random effects in each model). The condition of the bracket, size of the bracket, and sex of the visiting beetles were included as fixed effects, with condition by sex and size by sex included as interactions. Subpopulation ID was also included as a random effect to adjust for possible variation across subpopulations, allowing us to control for variation in density, sex ratio, and bracket condition.

Table 2 Descriptions of the four generalized linear mixed models used in this study
Full size table

In Model 2 and Model 3, we examined how bracket properties influenced the number and type of social interactions observed. For these analyses, we included only brackets where at least one beetle had been observed (n = 336 brackets). The dependent variable for Model 2 was the number of beetle interactions that were observed on the bracket as a fraction of the total number of beetle observations on that bracket over the entire breeding season. We used per observation count because the results from Model 1 indicated that both size and condition predicted beetle abundance of individual B. cornutus on brackets. Including the dependent variable as a fraction of the total number of beetle observations allowed us to determine if the properties of the brackets contribute to social interactions above and beyond the properties that might drive general beetle abundance on each bracket. We separated these social interactions into three categories: male-male, male–female, and female-female, which resulted in each bracket being included in the model three times with the Bracket ID as random effect to control for these repeated measures (similar to model 2). We conducted a follow up analysis (Model 3) examining just male–female interactions in which we separated them into reproductive (courtship, copulation attempts, and guarding) and non-reproductive (proximity and touching) behaviors. For both models, bracket condition and size and social interaction type were included as fixed effects, with condition by social interaction type and size by social interaction type included as interactions. Bracket ID and subpopulation ID were included as random effects.

For Model 4, we examined how bracket properties influenced female egg laying. For this analysis, we included only brackets where at least one female beetle had been observed (n = 336). For our dependent variable we used the number of observations of egg laying per total female beetle visits on each bracket. Condition and size of the bracket were included as fixed effects, while bracket ID and subpopulation ID were included as random effects.

Each of the models passed overdispersion tests and zero inflation tests. The data were slightly underdispersed and so we used a tweedie distribution for all models. Size was standardized using the base R scale function, with centering done by subtracting column means and scaling by dividing the centered columns by their standard deviations. Model effect significance was calculated using type III Wald chi-square tests from the ‘car’ package and the packages ‘ggeffects’ and ‘ggemmeans’ to conduct post-hoc analyses of the interactions (Ludecke 2018; Fox and Weisberg 2019). These tests produced estimates and significance tests for each slope as well as pairwise comparisons among categories. Figures 2, 3, 4 show marginal effects of predicted values of the independent variables extracted directly from the glmmTMB models using the packages ggeffects and ggemmeans (Lüdecke 2018).

Fig. 2
figure 2

Model 1 indicates that properties of fungal brackets, (A) bracket surface area and (B) bracket condition, predict use similarly for males and females. Both male and female use is predicted by both patch properties. Female (yellow) use is more strongly predicted by bracket size as compared to male (blue) use. Predicted values in (B) are calculated at the mean bracket size of 203 mm2. Raw data and 95% confidence interval are plotted for (B). Error bars for (A) represent 95% confidence interval. Though the error bars overlap for large bracket sizes in (B), the sex and sex by surface area remain significant predictors of the mean number of beetle observations

Full size image
Fig. 3
figure 3

The proportion of each type of interaction (yellow = Female-Female interactions, green = Male-Female, and blue = Male-Male interactions) observed per individual visit to a bracket, given (A) bracket size and (B) bracket condition. Generally, more social interactions are observed on larger brackets and between-sex interactions occur more often on live growth, though the condition of the bracket did not significantly predict the total number of social interactions. Raw data and 95% confidence interval are plotted for (A). Error bars for (B) represent 95% confidence interval. Predicted values in (B) are calculated at the mean bracket size of 203 mm2

Full size image
Fig. 4
figure 4

The proportion of each type of between-sex interaction observed per individual visit to a bracket, given bracket age. Both non-reproductive and reproductive interactions occur more often on live brackets, though the difference is greater for reproductive behaviors. Error bars for represent 95% confidence intervals. Predicted values are set at the mean bracket size

Full size image

Results

Abundance of beetles

We found a greater abundance of beetles on larger, living brackets (Fig. 2, Table 3). Surface area had a steeper relationship with female abundance than male abundance as indicated by the significant sex by surface area interaction. In other words, more females are found on larger brackets than males, while at smaller bracket sizes the difference between sexes is not as notable. Live growth on the bracket had no significant interaction with sex.

Table 3 Results from Model 1, predicting B. cornutus observations on fungal brackets given the size and condition of the brackets and the sex of the beetle
Full size table

Social interactions

More social interactions were observed on larger brackets, above and beyond the differences in beetle abundance among the brackets, (Fig. 3, Table 4). More between-sex interactions were observed on living brackets, though presence of live growth did not significantly predict the number of social interactions overall. There were more between-sex interactions than same sex interactions, and more female-female than male-male interactions.

Table 4 Results from model 2, predicting the number of social interactions (male-male, female-female, or male–female) given the size and condition of the brackets and the sex of the beetle
Full size table

Between-sex interactions

The effects from Model 3 are driven by both reproductive and non-reproductive interactions, though more non-reproductive behaviors were observed than reproductive behaviors. More between-sex interactions were observed on living brackets and larger brackets (Fig. 4, Table 5). Additionally living growth on brackets had a stronger effect on reproductive behaviors than non-reproductive behaviors, though the effect is small (Table 5).

Table 5 Results from Model 3 predicting the number of between-sex social interactions given the size and condition of the brackets and the sex of the beetle
Full size table

Oviposition

Contrary to our expectations, more observations of egg laying were observed on brackets with live growth (Table 6) and the size of the bracket did not predict female egg-laying behavior.

Table 6 Results from model 4 predicting laying events given the size and condition of the bracket
Full size table

Discussion

Our data suggest that patch use is influenced by the sex of individuals, the nature of social interactions, and the characteristics of resource patches themselves. Larger, living brackets hosted more beetles, with more social interactions observed on larger brackets (Fig. 2, Fig. 3). Between-sex interactions were observed more on living brackets (Fig. 3). We further found that the presence of live growth on a bracket more strongly predicted reproductive behaviors than non-reproductive behavior between the sexes (Fig. 4). The number of egg-laying events on a single bracket was also predicted by the presence of live growth, with more egg laying occurring on live brackets. The strength of bracket size as a predictor of behavior differed between males and females, which could indicate that brackets are used by the sexes for different purposes or that sexes maintain different drivers for such a preference (Fig. 2). Unlike between-sex interactions, same-sex interactions were not predicted by habitat patch characteristics (Fig. 3).

Our results could be explained by two, non-mutually exclusive alternative hypotheses − beetles may be choosing brackets based on their quality or responding to the presence of conspecifics on the brackets. Our findings may be a result of microhabitat selection by beetles for higher quality brackets. Greater attendance on living brackets has been corroborated by similar studies examining mycophagous beetle preference and properties that have demonstrated fungal quality, including B. cornutus (Heatwole and Heatwole 1968; Jonsell and Nordlander 2004). As brackets decay, they undoubtedly change in their nutritional value as they have been eaten and decrease in their amount of live growth (Lambert et al. 1980). Ciid beetles have been shown to detect changes in the chemical compounds of an aging bracket, allowing them to distinguish the amount of live growth of a bracket (Guevara et al. 2000). The greater magnitude of between-sex social interactions on living brackets may be a result of female preference for living brackets. However, living brackets do not host more interactions overall, only more between-sex contacts, suggesting that between-sex interactions may be influenced by female oviposition preference (Fig. 3).

Resource quality could drive female preference for laying on brackets with living growth. Female forked fungus beetles lay and encapsulate a single egg atop the bracket’s surface. Once hatched, larvae burrow into the host bracket and feed on the inner, hymenial layers for up a year or more (Liles 1956; VF and EDBIII, unpubl. data). The “preference-performance” hypothesis argues that selective pressure will drive females to lay offspring on hosts that increase offspring fitness (Jaenike 1978; Gripenberg et al. 2010). Living brackets represent a resource that generally is likely to exist for a greater duration of time than decaying brackets and may be of greater nutritional value to larva (Heatwole and Heatwole 1968). Female preference for living brackets as oviposition sites thereby may drive the presence of more between-sex and reproductive interactions on living brackets. Males may be under selection to frequent female-preferred brackets in order to acquire mating opportunities as they are frequently observed courting females during egg-laying. This pattern would lead to males more frequently visiting live brackets, and lead to between-sex interactions taking place on those same brackets. The fact that live growth predicts only the frequency of between-sex interactions further supports this interpretation (Fig. 3).

Alternatively, the differences we observe in the frequencies of between-sex (reproductive and nonreproductive) and same-sex social interactions could be related to the consequences of the social interactions themselves. Differences in the distribution of individuals and their social interactions under varying ecological conditions are a response to competition or between-sex conflict in a variety of taxa (Conradt et al. 1999; Stokke 1999; Weckerly et al. 2001; Breed et al. 2006; Eldakar et al. 2009). In B. cornutus, two types of social interaction could explain the pattern of bracket attendance—male-male competition and male harassment of females.

Intrasexual competition between males could play a role in the lack of relationship of male-male interactions to patch properties. When competing for females, male B. cornutus often engage in physical combat (Conner 1988). We might expect that because males are competing for females, more male-male interactions would be observed on patches preferred by females and where more reproductive behaviors occur. However, although males are most likely to interact with females on living brackets, intrasexual interactions between males do not follow the same pattern. Since male-male interactions are not predicted by either condition or size of the fungus bracket, males are not interacting more with one another on brackets that host reproductive behaviors (Fig. 3). However, even if males are competing over female-preferred brackets, the duration of their interaction will be short as one excludes the other from the resource, and their interaction would be less likely to be detected during our observations. Males compete even without fungal resources and in the absence of females (Mitchem et al. 2019), possibly explaining why resource patch properties do not drive male-male interaction patterns, even if they influence female use patterns, between-sex interactions, and reproductive behavior. Females do not compete physically over fungal resources, so their interactions are likely not driven by fungal preference (Mitchem et al. 2022).

Another explanation for why male-male interactions do not solely occur on brackets with high reproductive interactions is that brackets with fewer females could produce a higher local operational sex ratio thereby creating spatial variance in the competitive environment experienced by males (Emlen and Oring 1977). Therefore, male competition may be just as prevalent on brackets used by few females as brackets used by many females. Additionally, intrasexual association can be costly for males depending upon the size of the interacting males (Formica et al. 2011). Although males may still be motivated to use the brackets of females and compete for mating opportunities, male-male interactions may be further driven by avoidance or preference of certain male partners and consequently fail to exhibit patch characteristic preferences.

Environmental properties represent the stage on which social and reproductive behaviors play out. As the properties determine the frequency and distribution of such behaviors in space and time, they are integral to understanding the evolution of social behaviors. Dynamic habitat characteristics, such as host bracket size and condition, influence the prevalence and distribution of between-sex social interactions and oviposition in B. cornutus, and could have dramatic effects on the evolution of social behaviors, especially given the fitness consequences of social interactions and social networks in this species (Formica et al. 2011, 2012; Costello et al. 2022a, b). While our study focused on the use of and social behavior within a patch given varied resource quality, further research is necessary to determine how preferences at the patch scale affect emergent spatial patterns of social structure at the subpopulation and metapopulation level. To connect these levels of scale, further research should assess both variability in resource quality among patches and how the distribution of that variation in space affects the spatial structure of social behavior. The changing nature of ephemeral resources like fungus brackets could form a critical link between the ecological time span of a habitat patch and the evolutionary consequences of social interactions through these effects on frequency and distribution of behaviors. While the cross-sectional analysis of habitat patch use presented here gives us insight into the ways resource characteristics might determine interactions, longitudinal studies that track aging resources and successional environments could elucidate how dynamic ecological landscapes drive the distribution of social behavior and ultimately determine the landscape of selection.