Habitat preference and species interactions of the desert woodrat (Neotoma lepida) in the Mojave Desert

Becca Cosmero1, Emily Fieberling2, Katherine Marlin3, Ricardo Ruiz4, Rizzie Vermont4, and Lara Volski5

1University of California, Merced; 2University of California, Los Angeles; 3University of California, Santa Barbara; 4University of California, Santa Cruz; 5University of California, Berkeley

ABSTRACT

Ecosystem engineers often have cascading impacts on the ecosystems in which they occur. Understanding the factors that contribute to their habitat selection can provide valuable information on how ecosystems function. Our study investigates the habitat distribution of desert woodrats (Neotoma lepida) in the Eastern Mojave Desert. We show that there is a difference in desert woodrat abundance among habitats, and found that specific aspects of these environments are associated with woodrats. These aspects include grass cover, shrub richness, and especially substrate architecture. Our findings indicate that woodrats may facilitate plant productivity within this arid environment.

INTRODUCTION and in turn support biodiversity. For example, American pikas (Ochotona Preferential habitat selection is one of princeps) are ecosystem engineers in alpine the driving factors that allow species to environments because their burrows coexist (Rosenzweig 1981). Habitat increase the levels of nutrients available quality, determined by resource to plants (Aho et al. 1998). Thus, the availability, is a primary driver of habitat places that pikas prefer to live would selection (Orians and Wittenberger therefore be associated with the 1991). A greater understanding of what distribution of alpine vegetation. In leads an organism to choose a specific other words, understanding what habitat can uncover ecological dynamics determines the habitat preference of a and interactions between species. This single organism can reveal what drives is especially the case for ecosystem habitat selection of other species. This engineers that modify their environment

CEC Research | https://doi.org/10.21973/N3766T Fall 2017 1/7

in turn can lend greater understanding In order to examine the reciprocal of the spatial distribution of species and relationship between woodrat habitat resource availability within that habitat. preference and the associated Abiotically extreme environments that surrounding vegetation, we investigated contain limited resources, such as whether (1) woodrats prefer specific deserts, can provide insight into the habitats and microhabitats, (2) midden greater ecological impacts associated substrate vegetation differed from non- with habitat selection of key organisms. midden vegetation, and (3) woodrat Ecosystem engineers play major roles in presence drives changes in surrounding the distribution of patchy and limited vegetation. Together, our study helps resources. Rodents that create soil contribute to a better understanding of disturbances in arid regions, such as the the underlying mechanisms driving desert woodrat (Neotoma lepida), are habitat preference of desert woodrats, often classified as ecosystem engineers and in turn, the potentially large impact because collected organic materials and of desert woodrats on desert fecal detritus surrounding desert vegetation. Furthering our knowledge of woodrat middens promote nitrogen this relationship will allow us to better mineralization that in turn promotes understand the dynamics of desert local biodiversity (Whitford and ecosystems at large, and what enables Steinberger 2010). Reciprocally, desert desert natives to persist in this harsh, woodrats are also heavily reliant on resource limited environments. their habitat. For example, desert woodrat survivorship was found to be METHODS heavily correlated with the success or Study site failure of desert annual and perennial plant production (Smith 1995). When We conducted our study in three Mojave yucca (Yucca schidigera) failed separate sites in the University of to set seed during drought years, desert California Sweeney Granite Mountains woodrats also failed to reproduce Desert Research Center (UCGMR) in the (Smith 1995). Thus, vegetation can Eastern Mojave Desert (24°47' N, influence woodrat survivorship, and the 115°43' W). This area is a transition fact that woodrats have been cited as zone and contains flora and fauna ecosystem engineers indicates that characteristic of the Great Basin, woodrats may also be responsible for Sonoran, and Mojave Deserts. There are the composition of their surrounding 14 species of rodents in the area and vegetation. Nevertheless, the specific 502 species of plants. At roughly 1200 m composition of these surrounding plant elevation, our study sites did not breach communities has yet to be documented. dusky-footed woodrat (Neotoma In this study, we investigated how fuscipes) territory and instead featured woodrat habitat preference correlates only desert woodrat middens. Among with differences in their communities. three sites, we characterized each according to their predominant

CEC Research | https://doi.org/10.21973/B8R93X 2/7

vegetation. “Mixed Cholla/Yucca Scrub” associated with shrub richness, we was abundant in Mojave yucca (Yucca measured the following variables for schidigera) and buckhorn cholla every substrate: surrounding plant (Cylindropuntia acanthocarpa). “Mixed species richness, percent grass cover Yucca Scrub,” was abundant in yucca under shrubs, and plant architecture (as but had fewer buckhorn cholla. “Mixed number of clones). We defined an Acacia Scrub” had few to no yucca and individual yucca clone as a new cholla, but was abundant in cat-claw rhizomatous growth and an individual acacia (Senegalia greggii). All three sites cholla clone as a main branch at the had a varying number of creosote bush base. Then, we chose the next closest (Larrea tridentata). non-midden substrate of the same Shrub richness species of similar height, and measured the same variables. To test whether desert woodrat Statistical analyses densities correlated with richness of surrounding shrubs we used five 50 m2 We used a one-way ANOVA to test macro-plots within each site. To reduce differences in midden abundance bias, we randomly determined the among sites, followed by Tukey’s HSD distance between macro-plots. Each post-hoc test to distinguish specific macro-plot was divided into four differences among sites. To determine transects to give five 10 m x 50 m bands what habitat characteristics were and within three bands we counted the associated with woodrat abundance, we total number of yucca, cholla, creosote, used linear regressions to examine and acacia individuals. We used this relationships between midden data to estimate the abundance of each abundance and shrub species richness, plant species. Additionally, a belt shrub density, proportion of cholla, and transect was performed along the x-axis proportion of creosote. of each macro-plot in order to measure We conducted two way ANOVAs to test species richness of all plants in addition the relationships of midden presence to the previously surveyed yucca, cholla, and substrate species on surrounding creosote, and acacia individuals. shrub richness, percent grass cover, and Midden data shrub architecture. To determine specific difference among the effects of To assess woodrat habitat preference, substrate species on surrounding shrub we surveyed every woodrat midden richness, understory percent grass within each macro-plot. For each cover, and substrate architecture we midden we encountered within macro- used Tukey HSD post-hoc tests. Acacia plots, we recorded its position, the plant was excluded from the ANOVA analysis species that hosted the midden of number of clones because acacia (hereafter referred to as substrate), and does not exhibit the same growth the midden construction material. In patterns as yucca and cholla. order to test if woodrat presence was

CEC Research | https://doi.org/10.21973/B8R93X 3/7

Lastly we used ANCOVA to compare creosote increased (p < 0.001, R2 = 0.99; number of yucca clones and Figure 2D). Also, 76% of middens surrounding species richness by midden contained cholla as midden material. presence. All statistical analyses were Midden presence was positively conducted using JMP 13.0.0. correlated with shrub richness of surrounding vegetation (p < 0.002; RESULTS Figure 3A). Shrub richness was Mixed cholla scrub predominantly significantly higher in yucca than in consisted of Mojave yucca (30%), cholla, but acacia shrub richness did not buckhorn cholla (60%), and desert significantly differ from either yucca or creosote bush (10%). Mixed yucca scrub cholla (p < 0.04; Figure 3B). Midden was composed of Mojave yucca (41%), substrates also had greater percent buckhorn cholla (6%), and desert grass cover than non-midden substrates creosote bush (53%). Mixed acacia scrub (p < 0.03; Figure 3C). Grass percent consisted of desert creosote bush (57%) cover was highest under acacia, and catclaw acacia (43%). followed by cholla, then yucca. Acacia There were significantly more was significantly different from cholla middens in the mixed cholla scrub than and yucca, and cholla and yucca were any other habitat (p < 0.001; Figure 1). not significantly different from each Mixed cholla scrub also had higher other (p < 0.0001; Figure 3D). Number shrub richness and total shrub density of clones had the strongest correlation than the other two habitats. Midden with midden presence (p < 0.0001; density was positively correlated with Figure 3E). The number of clones was shrub richness (p < 0.01; R2 = 0.8; Figure significantly higher in yucca than in 2A) and shrub density (p < 0.001; R2 = cholla (p < 0.0001; Figure 3F). 0.85; Figure 2B). Midden abundance Woodrat presence is more strongly increased as proportion of cholla associated with a high number of clones increased (p < 0.0001; R2 = 0.8; Figure in yucca than in cholla (p < .001; Figure 2C) but decreased as proportion of 4). Yuccas with woodrat middens had almost double the number of clones A" P"<"0.0001" than yuccas without woodrat middens. In yucca with middens, as number of clones increased, shrub richness slightly increased. In yucca without middens, as B" number of clones increased, shrub richness slightly decreased. However, C" these results were not significant (P(midden presence) = 0.09, P(midden Figure 1. Mean number of middens ± 1 S.E. per presence*no. of clones) = 0.15, P(no. of habitat. Difference Tukey HSD: Mixed Cholla clones) = 0.76. R2 = 0.045. Linear Scrub (A), Mixed Cholla/Yucca Scrub (B), Mixed regression R2 pres = 0.02, R2 abs = 0.02; Acacia Scrub (B). Figure 5.)

CEC Research | https://doi.org/10.21973/B8R93X 4/7

Figure 2. The relationship between habitat characteristics and midden density. Graphs demonstrate response of midden density to shrub richness (A), shrub density (B), proportion of cholla (C), and proportion of creosote (D).

Figure 4. Relationship of midden presence and substrate species with substrate architecture. Dark and light bars represent midden presence and absence, respectively. Bars represent

means ± 1 S.E. Bars not connected by the same Figure 3. Mean vegetation characteristics ±1 S.E letters were significantly different after Tukey’s by midden presence (A, C, E) and midden HSD post-hoc tests. substrate species (B, D, F). Bars not connected by the same letters were significantly different after Tukey’s HSD post-hoc tests.

CEC Research | https://doi.org/10.21973/B8R93X 5/7

the desert woodrat’s position as an ecosystem engineer. There were also habitats that desert woodrats selected against. For example, midden density was negatively correlated with proportion of creosote. Habitats with a greater proportion of creosote were also less dense and had lower diversity. This relationship may be Figure 5. Relationships between yucca due to the fact that creosote has architecture and shrub richness for middens allelopathic effects on surrounding and non-middens. Yucca architecture was vegetation and avoids growing around measured by number of clones. Midden competing root systems (Brisson and presence is represented by blue and midden Reynolds 1994). Since woodrats prefer absence is represented by red. dense areas that provide ample cover, it is appropriate that a woodrat would DISCUSSION avoid this habitat (Bonaccorso and Our research suggests that woodrat Brown 1972). habitat preference influences the There were strong differences in the distribution of desert vegetation by biotic communities around middens promoting growth around middens. We when compared to non-midden found higher woodrat midden densities substrates. Midden presence was in areas with the highest proportion of strongly linked with greater diversity of cholla, demonstrating that woodrats surrounding shrubs, grass cover, and exhibit habitat preference. This finding architectural complexity, most notably complements the fact that woodrat in yucca. Since woodrats promote survivorship is greater in cholla middens nitrogen mineralization, it is quite because it confers more protection possible that midden presence is against predators (Smith 1995). This responsible for the correspondingly cholla-woodrat relationship is further higher diversity and density of underscored by the high proportion of surrounding vegetation. Nevertheless, cholla used as midden construction the association of woodrat middens material. The fact that the species with greater shrub diversity and percent richness and the percent cover of grass grass cover could be confounded by around middens were higher than in inherent differences in the three areas that lacked middens indicates that habitats we surveyed. It is also unclear there is clear relationship between whether woodrat presence allows yucca desert woodrats and desert vegetation. to live longer and thus generate more We suggest a manipulation study to clones, or whether woodrats simply confirm a cause-and-effect relationship, prefer to build their middens in yuccas but these findings nevertheless endorse that are more architecturally complex. Yet the relationship between woodrats

CEC Research | https://doi.org/10.21973/B8R93X 6/7

and yucca shows promise because REFERENCES woodrats pass down their middens for Aho, K., N. Huntly, J. Moen, and T. Oksanen. several generations (Dial 1988) and the 1998 Pikas (Ochotona princeps: number of clones in yucca increases Lagomorpha) as allogenic engineers in an with age (LaPre 1979). alpine ecosystem. Oecologia, 114:405–409. Desert woodrats exhibited clear habitat preference and there were Bonaccorso, F.J. and J.H. Brown. 1972. House differences between the vegetation construction of the desert woodrat, around middens versus non-middens. Neotoma lepida lepida. Journal of This demonstrates that desert woodrat Mammalogy 53(2):283–288. habitat preference is biotically driven and there are strong interspecific Brisson, J. and J.F. Reynolds. 1994. The effect of relationships between woodrats and neighbors on root distribution in a creosote desert vegetation. Desert woodrat bush (Larrea tridentata) population. habitat preference was associated with Ecology 75:1693–1702. shrub density and diversity, but this Dial, K.P. 1988. Three sympatric species of preference was driven by the Neotoma: dietary specialization and proportion cholla. Our findings further coexistence. Oecologia 76(4):531–537. illustrate the complexity of desert woodrat habitat selection in a resource- LaPre, L.F. 1979. Physiological ecology of Yucca deficient environment. There are likely schidigera. Riverside, CA: University of other species interactions that we have California. 117 p. Dissertation. not accounted for, but this study portrays a baseline understanding of the Orians, G.H. and J.F. Wittenberger. 1991. Spatial dynamic relationship woodrats have and temporal scales in habitat selection. with desert vegetation. Our results The American Naturalist 137:29–49. indicate that shrub diversity and Rosenzweig, M.L. 1981. A theory of habitat abundance may drive midden presence selection. Ecology 62(2):327–335. which in turn can have ecosystem level consequences. These findings further Smith, F.A. 1995. Den characteristics and clarify what drives habitat selection, and survivorship of woodrats (Neotoma lepida) what enables these organisms to survive in the Eastern Mojave Desert. The in arid environments. Southwestern Naturalist 40(4):366–372.

ACKNOWLEDGEMENTS Whitford, W.G. and Y. Steinberger. 2010. Pack This work was performed at the rats (Neotoma spp.): Keystone ecological engineers? Journal of Arid Environments University of California’s Sweeney 74:1450–1455. Granite Mountains Desert Research Center, doi:10.21973/N3S942.

CEC Research | https://doi.org/10.21973/B8R93X 7/7