Our study is the first of its kind within the Black Belt prairie, which is an understudied yet critically endangered ecoregion (Schotz & Barbour 2009; Noss 2013). This study provides insight into how ecologically influential beetles respond to local landcover and management practices and can inform grassland management and restoration. The species we modeled were relatively common across the study area; few rare species were modelled. This is consistent with needing at least 20 observations to fit models.
Calculating the proportion of coefficient posterior distribution above 0 helped determine whether a covariate had a positive, negative, or no relationship with occupancy probability because the proportion of the posterior distribution above 0 equals the probability a covariate had a positive relationship with occupancy probability. In our study, when a covariate had a consistent relationship with occupancy across morphospecies, 0.96 − 0.65 of the β posterior distribution was greater than 0 for covariates with β mean. We feel confident concluding these covariates were positively associated with occupancy. Additionally, 0.14–0.37 of the posterior distribution was above 0 for covariates with a negative β mean. Consequently, we conclude the covariates were negatively associated with occupancy. Moreover, agreement between SSVS and LASSO provides confidence in the direction of the relationship between a covariate and occupancy probability.
We expect occupancy had a positive relationship with agricultural area, disturbed area, vegetation diversity, and burn frequency because these covariates mediate habitat heterogeneity, which is a driver of insect diversity and abundance (da Silva et al. 2008; Joern & Laws 2013; Seibold et al. 2016). Diverse plant communities are structurally heterogeneous, which increases microhabitats and niches and provisions heterogeneous resources for phytophagous insects (Tilman 1986; Siemann 1998; Duquette et al. 2022). Habitat heterogeneity due to plant diversity in grasslands has been shown to positively influence diversity and abundance of arthropod communities across feeding guilds, including carabids and scarabs (Murdoch et al. 1972; Joern & Laws 2013; Song et al. 2023).
Low-intensity agricultural practices, such as low-intensity cattle grazing, can positively affect insect populations by increasing both biotic and abiotic heterogeneity (van Klink et al. 2015; WallisDeVries et al. 2016; Silva-Monteiro et al. 2022). Moreover, intermediate levels of disturbance, can benefit species by producing heterogeneous physical habitat and preventing dominance of one or a few species, which opens more niches and enables co-occurrence of many species (Sousa 1984; Fuhlendorf & Engle 2001; Mittlebach & McGill 2019). Disturbance and low-intensity agriculture produce habitat heterogeneity that can support diverse and abundant arthropod communities, including diverse and abundant carabid and scarab assemblages (Verdu et al. 2008; Tocco et al. 2013; Joern & Laws 2013). Around our sites, disturbance was low intensity, largely from historical agriculture. Further, no sites had greater than 30% disturbed area within a 200m radius, and only 4 sites had greater than 15% disturbed area in 200m radius. Although our landcover data did not specify types of agriculture, agriculture in the Black Belt is largely pastoral (Blevins 1998).
Fire maintains heterogeneity in grasslands (Fuhlendorf & Engle 2001; Van Auken 2009; Wilsey 2018). Prescribed fire has been investigated frequently as a driver of carabid richness, community composition, and abundance (Mason et al. 2023), but less research has investigated scarab responses to fire in grasslands, especially in North America (Smith et al. 2019). For carabids, there is an ambiguous relationship between fire and taxonomic and functional changes in community composition. Where some studies found taxonomic or functional diversity increased in response to fire, other studies found no response (Valko et al. 2016; Nelson et al. 2021; Mason et al. 2023). Mason et al. (2023) argues these inconsistencies are due to not characterizing fire severity. For scarabs, the few studies available have found species diversity of coprophagous scarabs increased in response to prescribed fire because herbivorous mammals preferred recently burned patches, thus producing more dung resources (Smith et al. 2019; Hosler et al. 2021). In our study, frequent prescribed fire was associated with increased occupancy for carabid and scarab species. We expect frequent fire contributed to habitat heterogeneity and prevented establishment of woody plant species (Van Auken 2009; Fill et al. 2015; Mason et al. 2023).
We expect occupancy had a negative relationship with urban landcover and number of RIFA mounds due to the multiple stressors associated with urban areas and predation and competition from RIFA, respectively. Urban areas are associated with increased artificial light at night, impervious surfaces, exotic plant species, and pollution, all of which are implicated as drivers of insect declines (Niemela et al. 2002; Vaz et al. 2023). Both carabid and scarab diversity and abundance have been negatively affected by urbanization (Niemela et al. 2002; Hartley et al. 2007; Correa et al. 2021).
Literature on how RIFA influences insect communities is equivocal (Porter & Savignano 1990; Calixto et al. 2007; King & Tschinkel 2013). Some studies found community-wide decreases in diversity and abundance of native ants where RIFA invaded (Lu et al. 2012), direct predation on other arthropods by RIFA (Forys et al. 2001) and shifts in insect community composition (Porter & Saviganano 1990). Conversely, other studies concluded RIFA invasion did not competitively suppress native ants or other insects and suggested declines in native insects were a result of a disturbance event that enabled establishment of RIFA (King & Tschinkel 2006; King & Tschinkel 2013). Our results support the notion that RIFA negatively affects other insects. For example, when there were no RIFA mounds, mean occupancy probability for Dyschirius sp. (Carabidae) was greater than 0.80, and when there were more than 10 RIFA mounds per 450m2, occupancy probability dropped to under 0.25. RIFA may compete with predaceous beetles such as Dyschirius sp. for prey resources, and RIFA may prey on larvae or adults of predaceous and herbivorous beetles (Forys et al. 2001). Nearly all sampled sites had RIFA mounds, and we observed variation in the number of RIFA mounds per site (0–13). Therefore, we expect our results reflect a relationship with RIFA rather than a relationship with an indirect factor, such as a disturbance event.
Basal area, years since last burn, and restoration type were positively associated with occupancy for some taxa but negatively associated with other taxa due to variations in morphospecies’ habitat requirements. We found seeded sites generally had higher scarab and carabid occupancy. However, occupancy estimates at unseeded and seeded sites often overlapped, and there was a negative response to seeded restoration for three taxa. This ambiguity is consistent with the literature; there is conflicting evidence on the success of seeded restoration of grasslands (Hutchings & Booth 1996; Fagan et al. 2008; Paraskevopoulos et al. 2023). Seeding restoration sites can increase plant diversity, especially in sites without a nearby source of native plants for colonization (Guo et al. 2023) but introducing propagules may create non-target habitat due to improper ratios of forbs to grasses or the presence of non-native genetic variants or species (Fagan et al. 2008; Simmers & Galatowitsch 2010; Fraser et al. 2015). We expect that where we observed increased occupancy probability with seeded restoration, species were responding to increased plant diversity, and where we observed a negative relationship between occupancy and seeded restoration, we expect species were negatively influenced by soil disturbance that is required to prepare a site for seeding (Woodcock et al. 2008). We note this differs from the disturbed landcover covariate because preparing a site for seeding involves killing existing vegetation via suppression or herbicides and plowing the top layer of soil (Applestein et al. 2018). Additionally, age of seeded restoration site may be important, and depending on the species, occupancy probability may increase or decrease as time since seeding increases. For instance, some species with a positive response to seeded restoration may be early colonizers, and as time since seeding increases, occupancy probability may decrease. Conversely, species with a negative response to seeding may have been extirpated by the initial site preparation and may display increased occupancy probability as time since seeding progresses.
The negative relationship between basal area and family-level scarab abundance was likely due to more scarab species in our dataset responding negatively to basal area than positively. Our abundance model included data for scarab species that were analyzed with occupancy models as well as scarab species that did not have enough detections for occupancy modeling. The species that were not included in the occupancy modelling could be influencing the negative association between basal area and scarab abundance.
We expect few covariates were selected in the abundance models due to the diversity within families. For example, if abundances of some carabid species were negatively affected by a covariate, other species were positively affected by that covariate, and other species had no association with that covariate. The variation in species’ responses could confound detecting a pattern for the family. A future avenue of research could be to model species-specific abundance and compare covariates associated with abundance to the covariates associated with occupancy presented here.
The low detection probabilities in our study were not surprising given our sampling methods. Of the three sampling methods, no scarabs or carabids were collected in malaise traps, and a relatively small proportion of the collected beetles came from pitfall traps. The black light traps collected the bulk of specimens, and although they were effective, they experienced technical issues on several surveys, resulting in low detection probabilities at those sites and surveys. Further, our attempt to broadly sample carabid and scarab species resulted in surveys that were spread across several months, thus it is probable our surveys missed peak activity of many of the taxa, which have heterogenous emergence and peak activity times. The ability to explicitly estimate detection probability in our modelling approach enabled unbiased occupancy estimates despite low detection probabilities (MacKenzie et al. 2001; Kery & Schaub 2012).
Grasslands are imperiled and valuable for their biodiversity and ecosystem services, so conservation and restoration of grasslands is imperative. Considering how animal taxa, especially taxa that mediate many ecosystem services, are associated with landscape conditions and management practices could benefit restoration efforts. Scarabaeidae and Carabidae are suitable taxa to study because of the important role insects play in grassland ecology, the diversity of species within these families, and the contribution to knowledge beyond highly studied taxa such as pollinators. Further, there is value in studying understudied grasslands to address geographic variation in grassland ecosystems. Little remains of the historic range of Black Belt prairie, and previous studies have primarily inventoried species in prairie remnants. Our study contributes to understanding and restoration of Black Belt prairies and grasslands generally by confirming positive relationships between insect occupancy and low-intensity agricultural area, disturbed area, and vegetation diversity and a negative relationship between insect occupancy and urban landcover as detected in previous studies. Our results also elucidated important variations in beetle habitat requirements, showing positive relationships with basal area, years since last burn, and restoration type for some morphospecies and negative relationships with these covariates for other morphospecies. Our results and the literature indicate variable responses to seeded versus unseeded restoration, highlighting a subject where increased understanding could advance restoration efforts. Additionally, our results support the idea that RIFA has a direct negative relationship with insect communities when there has been debate in the literature about whether the negative effect associated with RIFA is due to RIFA or a correlated disturbance event. Our study contributed new knowledge about the relationship of scarabs with burn frequency, which has been understudied previously, and confirmed the need to study characteristics of fire rather than simply presence of fire (Mason et al. 2023). We confirm the importance of fire to grassland management with results that showed fire should be frequent (for vegetation diversity and insect species that prefer recently burned habitat) but temporally heterogeneous (for habitat heterogeneity and reestablishment of insects extirpated by fire). In general, our results highlight the important role habitat heterogeneity plays in beetle occupancy in grasslands.