Invertebrate community response to fire and rodent activity in the Mojave and Great Basin Deserts

Abstract Recent increases in the frequency and size of desert wildfires bring into question the impacts of fire on desert invertebrate communities. Furthermore, consumer communities can strongly impact invertebrates through predation and top‐down effects on plant community assembly. We experimentally applied burn and rodent exclusion treatments in a full factorial design at sites in both the Mojave and Great Basin deserts to examine the impact that fire and rodent consumers have on invertebrate communities. Pitfall traps were used to survey invertebrates from April through September 2016 to determine changes in abundance, richness, and diversity of invertebrate communities in response to fire and rodent treatments. Generally speaking, rodent exclusion had very little effect on invertebrate abundance or ant abundance, richness or diversity. The one exception was ant abundance, which was higher in rodent access plots than in rodent exclusion plots in June 2016, but only at the Great Basin site. Fire had little effect on the abundances of invertebrate groups at either desert site, with the exception of a negative effect on flying‐forager abundance at our Great Basin site. However, fire reduced ant species richness and Shannon's diversity at both desert sites. Fire did appear to indirectly affect ant community composition by altering plant community composition. Structural equation models suggest that fire increased invasive plant cover, which negatively impacted ant species richness and Shannon's diversity, a pattern that was consistent at both desert sites. These results suggest that invertebrate communities demonstrate some resilience to fire and invasions but increasing fire and spread of invasive due to invasive grass fire cycles may put increasing pressure on the stability of invertebrate communities.

grasses have altered fire regimes by increasing the size, frequency, and severity of fires (Brooks et al., 2004;Brooks & Matchett, 2006), which could have long-term effects on the stability and biodiversity of these systems.
Invertebrates make up a large proportion of ecosystem diversity (May, 1988) and provide a wide range of ecosystem functions.
Invertebrates often have specialized relationships with plants, vertebrates, and microbes. Invertebrates are critical to food webs in serving as prey for many vertebrate species and have important interactions with plants through herbivory, seed dispersal, and pollination. Many invertebrates have small home ranges, making them less able to escape unfavorable changes in their environment. These qualities make invertebrates good indicators of ecosystem function and resilience (Andersen, 1990;Lavelle et al., 2006;Majer, 1983).
Ants (Hymenoptera: Formicidae) are particularly good indicators of ecosystem stability (Andersen, 1997) because they are among the most abundant and diverse group of invertebrates, and occupy a variety of specialized niches across multiple trophic levels (Majer, 1983).
The recent increase in the frequency and size of desert wildfires (Brooks et al., 2004) brings into question the direct and indirect impacts of fire on desert insect communities. Direct fire mortality is influenced by the degree of exposure and the mobility of the species or life stage (Swengel, 2001). Rice (1932) and Morris (1971) show that mortality can often continue to occur postfire from starvation and exposure while others report shifts in insect abundance and diversity after repeat burns (Wright & Samways, 1998. Flying insects and other highly mobile insects are often the first to recolonize into burned landscapes (Swengel, 2001). Grasshoppers have been shown to increase in abundance in burned areas (Lamotte, 1975); however, grasshopper richness is usually lower in frequently burned areas (Evans, 1984(Evans, , 1988. Evans (1984Evans ( , 1988 found that forb-feeding grasshopper richness declined in more frequently burned areas because of fewer forbs in those areas, and grass-feeding grasshoppers increased because of relatively higher grass cover in burned areas. Insect species that require a specific plant community structure that does not reoccur in the first few years after fire can lose resource availability for generations, and, if fires are too frequent, this can dramatically reduce their population size (Wright & Samways, 1998. Fire tends to favor some ant species (Holbrook et al., 2016), while reducing overall ant species richness (Ostoja, Schupp, & Sivy, 2009). In many cases, fire decreases the diversity of the entire insect community (Swengel, 2001).
The ability of rodents to modify plant community structure, and their sensitivity to fire, could result in rodent communities having important effects on invertebrate communities in postfire environments. Many rodent species include insects as part of their diet, and small mammal insectivory has been shown to have strong effects on grassland invertebrate communities (Churchfield, Hollier, & Brown, 1991). The effects of rodents on plant community structure via granivory and folivory (Sharp-Bowman, McMillan, & St. Clair, 2017a, 2017b are also likely to have indirect effects on the abundance and diversity of insect communities. A previous study at our Great Basin site determined that rodents can suppress cheatgrass invasion (St. Clair, O'Connor, Gill, & McMillan, 2016); rodent exclusion produced a plant community dominated by invasive grasses, and where rodents had access, the plant community was a much more diverse annual forb community. In both burned treatments (with and without rodents), plant diversity was reduced compared to unburned plots, but burned plots with rodent access had higher plant diversity than burned plots without rodents; these changes in plant habitat could have bottom-up influences on invertebrate diversity. Banner-tailed kangaroo (Dipodomys spectabilis) rats have been known to alter ant community composition (Schooley, Bestelmeyer, & Kelly, 2000) via changes in plant community structure through mound building (Moroka, Beck, & Pieper, 1982). The indirect effects of rodents on insect communities through the modification of the plant community are not well characterized. Our study was designed to increase the characterization of the indirect effects of rodents on insect communities.
There have been many studies in the deserts of North America documenting plant-insect interactions (Ostoja et al., 2009) and rodent-insect interactions (Brown & Davidson, 1977;Brown, Davidson, & Reichman, 1979), but there are far fewer studies that compared these relationships across different desert ecosystems. The deserts of western North America vary in climate and have unique biotic communities. The Great Basin and the Mojave Desert share a border but are very different from one another, one is semi-arid while the other is hyperarid. Despite these differences, both are facing a similar threat of changing fire regimes caused by invasive annual grasses.
Because of their inherent differences, the biological communities in each desert may respond differently to these changes. The purpose of our study was to characterize the influence of fire and rodent exclusion on invertebrate community abundance and diversity in the Great Basin and Mojave Deserts and whether they were related to changes in the plant community. This study addressed the following

| Mojave Desert
Our Mojave Desert site is located at the Lytle Ranch Preserve, which is a 680-acre nature preserve owned and managed by Brigham Young University. Lytle Ranch is located in the northern Mojave Desert, in western Washington Co., Utah (37°08′54″N 114°00′50″W).
Elevation is 915 m, mean annual temperature is 16.3°C, average mean January temperature is 6.2°C, and average mean July temperature is 28.1°C (Lytle Ranch GHCN, Utah Climate Center).

| Experimental design
The experimental design at both sites was the same and consisted of 60 × 60 m experimental blocks replicated five times. Each block was split into four equal (30 × 30 m) subplots, which were assigned to one of four factorial treatment combinations: burned or unburned, and rodent access or rodent exclusion (St. Clair et al., 2016  fire between shrubs at the Great Basin site, we placed 300 g/m −2 of wheat straw in the shrub interspaces in our burn plots (St. Clair et al., 2016). Fire spread naturally without straw at the Mojave site.

| Invertebrate trapping
There were 4 pitfall traps placed in each experimental subplot (Andersen, 1991) (Triplehorn & Johnson, 2005).We used the most abundant taxa from each group for our analysis. We selected taxa that were represented at both sites and had ≥four individuals. We excluded rare invertebrate families because it was impossible for us to determine whether they were simply rare in our system or rare because of our trapping method. The ground-dwelling group across both sites comprised invertebrates from the taxa Acari, Entomobryidae,  (Majer, 1983). Pitfall trapping is a well-established method for capturing ants (Andersen, 1991). Ants are also relatively easy to identify and are common in most terrestrial ecosystems throughout the world. Therefore, we were able to identify ants to species and determine changes in richness and diversity of ant species in response to treatment conditions. Our results can also thus be readily compared to a wider range of studies.

| Vegetation surveys
Vegetation cover and density were measured at both sites. Vegetation surveys were conducted in April 2016 at the Mojave site and in June 2016 at the Great Basin site. Cover was measured using the step point intercept method (Bonham, 1989

| Statistical analysis
We

| RE SULTS
We identified 101 families or orders from the Great Basin site. We also identified 10 ant species representing nine genera in the Great

F I G U R E 1 Ant forager abundance (a and b), species richness (c and d), and Shannon's diversity (e and f) responses to burn treatment separated by month and site. Error bars represent standard error for each value
Basin site. We identified 108 families or orders from the Mojave Desert site. We also identified twelve ant species representing nine genera in the Mojave Desert (Tables 3 and 4).

| Fire effects
Fire affected ant species richness and diversity at both sites (Tables   5 and 6), with higher species richness and diversity in unburned plots than in burned plots (Figure 1). In the Great Basin, the effect of fire on ant diversity was only significant in May and June (Table 6; Figure 1).
Fire did not significantly affect ant abundance at either site (Tables 5   and 6). Fire also had little effect on the abundances of ground dwellers or ground foragers at either site (Tables 5 and 6). Fire played a significant role in determining flying-forager abundance at the Great Basin site (Table 6), with higher flying-forager abundance in unburned areas than in burned areas (Figure 2), but fire had little effect on flying-forager abundance at the Mojave site (Table 5). Structural equation models suggest that the effects of fire on ant species richness and Shannon's diversity at both sites are mediated through changes in the plant communities (Tables 7 and 8; Figure 3). Specifically, fire had a positive influence on invasive plant cover, which then negatively influenced ant species richness and diversity (Tables 7 and 8; Figure 3).

F I G U R E 2 Ground-dweller abundance (a and b), flying-forager abundance (c and d), and ground-forager abundance (e and f) responses to burn treatments separated by month and site. Error bars represent standard error for each value
Structural equation models did not show any effects of plant cover on ground-dweller, flying-forager, or ground-forager abundance at either site (Tables 7 and 8; Figure 4).

| Rodent effects
Rodent treatments had little to no effect on ant abundance, richness, or diversity at either location when averaged across months (Tables 4 and 5). At our Great Basin site, rodents had a significant effect on ant abundance depending on the month, with abundance being higher in rodent access plots than in rodent exclusion plots in May and June but being lower in rodent exclusion plots in August and September (Table 6; Figure 5). Rodent treatments had little to no effect on flying-forager abundance or ground-forager abundance at either site (Tables 5 and 6).

| Fire and rodent interactions
Fire and rodent interaction terms in our models were generally not significant (Tables 5 and 6). The only exception to this was the abundance of ground-dweller invertebrates at our Great Basin site (Table 6), where abundance was higher in rodent exclusion plots, particularly in unburned conditions ( Figure 5).

| Time effects
Ant abundance, species richness, and Shannon's diversity at both locations changed significantly across months (Tables 5   and 6), with abundance and species richness peaking in June at both locations (Figure 1). Ant diversity was highest in June in the Mojave and highest in May in the Great Basin (Figure 1). At the Mojave site, ground-dweller abundance and flying-forager abundance changed significantly across time (Table 5), with grounddweller abundance peaking in June and flying-forager abundance being highest in April (Figure 2). At the Great Basin site, grounddweller abundance, flying-forager abundance, and groundforager abundance were all significantly affected by month (Table 6), with ground-dweller and flying-forager abundances peaking in June, and ground-forager abundance peaking in May ( Figure 2).

| Invertebrate responses to fire
Fire can have both positive and negative effects on invertebrate abundance and diversity (Swengel, 2001), with the effects varying depending on the taxa measured and other environmental conditions (Warren, Scifres, & Teel, 1987). In our study, fire had very little effect on ant abundance; however, fire reduced ant species diversity at both sites (Tables 5 and 6; Figures 1 and 2). Ostoja et al. (2009) reported lower ant diversity in cheatgrass-dominated plots compared with sagebrush intact plots in the Great Basin which is a typical vegetation conversion after fire as seen in our plots. These results are consistent with our previous research at the Great Basin site, where ant species diversity was reduced in burned areas but ant abundance was unaffected (Day, Bishop, & St. Clair, 2018). In that study, abundance of most species was reduced in burned plots, but the abundances of some dominant ant species increased, which kept overall ant abundance in burned areas similar to those in unburned areas. This same pattern occurred at our Mojave site, where reduction in abundance for some ant species was balanced by the increase in abundance of others (Table 3). Among the dominant ant species that responded positively to fire were harvester ants in the genus Pogonomyrmex (Tables 3 and 4). Holbrook et al. (2016) reported increased P. occidentalis nest density in burned areas over unburned areas in the Great Basin. Our surveys show that P. occidentalis forager abundance increased in burned plots at our Great Basin site (Table 4) while P. rugosus forager abundance at our Mojave site nearly tripled in burned plots compared to unburned plots (Table 3).
This increase in Pogonomyrmex abundance may be the result of shrub removal, allowing for increased colony densities (Day et al., 2018;Sneva, 1979), or it may also be caused by increased seed resources from increases in annual plant cover.
The abundances of most of the invertebrate groups in our study were unaffected by fire (Tables 5 and 6; Figures 1 and 2). We did observe that the flying-forager group saw reductions in abundance in burned plots compared to unburned plots in the Great Basin ( Pitfall trapping is not a reliable method for sampling flying insects, so we were surprised to have collected so many. Our findings in this study suggest further research is needed on the effects of fire on flying insects in the Great Basin.

| Indirect effects of fire mediated through changes in plant communities
Vegetation structure and plant community composition are important determiners in invertebrate community composition (Bromham, Cardillo, Bennett, & Elgar, 1999;Denno et al., 2002;Herrera & Dudley, 2003;Pearson, 2009). Lower ant diversity in burned plots may be a response to reduced resource availability or unfavorable abiotic conditions as a result of an altered plant community. Ant species richness and diversity were negatively influenced by invasive plant cover at both sites (Tables 7 and 8; Figure 3). This is consistent with findings of Ostoja et al. (2009) who found that ant diversity decreased in B. tectorum-dominated sites compared to sagebrush intact sites in the Great Basin. Invasive plants were also reported to reduce ant species richness in a grassland (Lenda, Witek, Skorka, Moron, & Woyciechowski, 2013). Ants are generally thermophilic, but have varying levels of temperature tolerance (Hölldobler & Wilson, 1990). In sagebrush systems, shrub removal increases soil surface temperature (Chambers & Linnerooth, 2001) and reduces soil moisture in surface soils (Inouye, 2006). This change in abiotic conditions may favor some ant species, such as Pogonomyrmex (Bucy & Breed, 2006), but may restrict foraging time for other ant species. The abundances of arboreal ant species were reduced in burned plots at both sites; Crematogaster depilis in the Mojave was not found at all in burned areas during our study (Table 3), and Camponotus vicinus in the Great Basin was eight times more abundant in unburned plots than in burned plots during our study (Table 4). The life histories of more arboreal ants such as C. depilis and C. vicinus are closely tied to woody plants (Hölldobler & Wilson, 1990), which are greatly reduced in burned plots. The carpenter ant, C. vicinus, was reported to stop foraging when temperatures reach 23°C (Bernstein, 1979), so more shaded unburned areas may allow longer foraging times in summer. Nocturnal nectivorous ants, which may rely more on perennial plants for nectar resources, were also reduced in burned areas compared to unburned areas, Myrmecocystus mexicanus in the Mojave (Table 3) and M. testaceus in the Great Basin (Table 4).

Rodents can have strong top-down effects on Great Basin and
Mojave plant communities (Sharp-Bowman et al.,2017a, 2017b. Previous research in our Great Basin plots shows that rodent exclusion in burned areas dramatically increased the cover of cheatgrass leading to loss of plant biodiversity (St. Clair et al., 2016). We therefore expected to see top-down effects of rodents on plant communities translate to shifts in invertebrate community composition and structure. However, we observed no significant main effects of rodent exclusion on ant community richness and diversity or invertebrate community abundance (Tables 5 and 6). Our results suggest that invasive plant cover strongly affects ant diversity (Tables 7 and   8; Figure 3), and while rodents may alter which types of invasive plants dominate in burned areas (St. Clair et al., 2016), they seem to have less effect on the percent cover of invasive plants in burned plots (Tables 7 and 8; Figure 3). For example, invasive plant cover in burned-rodent access (BRA) and burned-rodent exclusion (BRX) plots were nearly identical between sites (68% and 67% in BRA plots in the Mojave and Great Basin, respectively, and 72% in BRX plots at both sites). This suggests that the loss of native shrubs and their replacement by invasive annuals following fire has a larger impact on invertebrate communities than differences in the composition of invasive annual communities (annual grasses vs. annual forbs) created by rodents (St. Clair et al., 2016). Abiotic changes associated with shifts from native perennial shrublands to invasive annual plant communities, to which invertebrates are sensitive, are likely much greater than differences between invasive annual grass and annual forb communities.

| Invertebrate responses over time
Seasonality played a significant role on invertebrate abundance and diversity in the Mojave (Table 5) and Great Basin (  (Stinson & Brown, 1983). After a burn, annual grasses can quickly fill open space and make the system more flammable as they quickly dry out toward the end of spring (Knapp, 1996). Changes in the seasonality of the vegetation likely alter the site selection of flying foragers, such as leafhoppers, that utilize the area for forage or laying eggs (Stinson & Brown, 1983). by seed production, as most annual plants have gone to seed by June (Gordon, Holmes, & Nacu, 2008). The peaks in ant species richness and Shannon's diversity in May and June in unburned plots in the Great Basin ( Figure 1) may be related to flowering events, seed production (Pol, Casenave, & Pirk, 2011), nectar production (Dáttilo et al., 2015), and increasing summer temperatures (Crist & MacMahon, 1991). The

Ant
Mojave site saw higher Shannon's diversity in unburned plots every month, but total abundance was significantly higher in burned plots in May, June, and July ( Figure 1). The Mojave site has much higher shrub diversity than the Great Basin site, so burned plots may be seeing more losses in ant-shrub mutualisms (like that of C. depilis and cacti F I G U R E 5 Responses of ant forager abundance to rodent treatment (a) and ground-dweller abundance to fire and rodent treatments (b) separated by month in the Great Basin. Error bars represent standard error for each value (Chamberlain & Holland, 2008)), lowering diversity throughout the study.

| Desert comparison
Ants in general responded similarly to treatments at both sites when averaged over the whole collecting season; however, the way those responses played out over time differed between sites. In the Mojave, ant richness and diversity were higher in unburned than in burned plots throughout the entire collecting season (Figure 1). In the Great Basin, however, differences in ant richness and diversity between burn treatments were more variable (Figure 1). The Mojave site has higher mean temperatures than our Great Basin site through most of the year, and plant diversity is much higher at our Mojave site than at our Great Basin site. Ant foraging rates are dependent on both temperature (Crist & MacMahon, 1991;MacKay & MacKay, 1989) and food availability (Gordon et al., 2008). The higher plant diversity and higher mean temperatures in the Mojave may allow for the sustained difference in ant richness and diversity through the season.

| IMPLIC ATIONS
Exotic invasive plants are changing desert fire regimes (D'Antonio & Vitousek, 1992), and their downstream impacts on invertebrate communities can have important ecological consequences. Fire facilitates invasion (Brooks et al., 2004) and invasion in turn facilitates fire (Balch, Bradley, D'Antonio, & Gomez-Dans, 2013), creating a positive feedback loop and threshold resulting in potential state changes. Our data suggest that invertebrate community abundance is generally stable in response to desert fires but that species and taxonomic groups can vary dramatically. Invasive grass fire cycles pose a serious threat to arid systems where we may see significant modification to ecosystem function (Hooper et al., 2005), and perhaps local extinctions of some species.
Biodiversity is already lower in arid systems than in more mesic systems because of abiotic limitations, and many of the species are operating near tolerance limits. This makes functional redundancy less likely in arid systems, which increases the importance of each invertebrate species in the system (Whitford, 1996). Our results show reductions in ant diversity in response to fire, suggesting potential losses of key ecosystem services tied to invertebrate diversity due to increases in fire frequency and extent. Even within a genus, there are important physiological, behavioral, and life-history differences between species that minimize competition and affect the role that the individual species play in the ecosystem (Whitford, 1976(Whitford, , 1978. The replacement of less common species with more common ones may not adequately replace the services provided by the less common species. Fires and exotic annual grass invasion are combining to change the world for invertebrates in desert ecosystems.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
JDD collected samples, identified ants, analyzed data, and wrote the majority of the manuscript. JHB and TJT identified invertebrates, assisted with data analysis, and assisted in writing and editing the manuscript. SBB conceived and financially supported the study, directed analysis, and provided input and revisions during the writing process.
PA contributed to several discussions related to ants and reviewed multiple drafts of the manuscript. AC collected and analyzed insect samples and data.

DATA ACCE SS I B I LIT Y
Data available through the Dryad data repository: https://doi.