The salt marsh harvest mouse (Reithrodontomys raviventris; RERA) is an endangered species endemic to the coastal wetlands of the San Francisco Estuary, California. RERA are specialized to saline coastal wetlands, and their historical range has been severely impacted by landscape conversion and the introduction of non-native plant and rodent species. A better understanding of their diet is needed to assess habitat quality, particularly in relation to potential competitors. We investigated three questions using DNA metabarcoding with ITS2 and trnL markers: (1) Do RERA specialize on the native plant, pickleweed (Salicornia pacifica), (2) Do RERA consume non-native plants, and (3) What is the dietary niche breadth and overlap with three sympatric native and non-native rodents? RERA diet was dominated by two plants, native Salicornia and non-native salt bush (Atriplex spp.), but included 48 plant genera. RERA diet breadth was narrowest in fall, when they consumed the highest frequencies of Salicornia and Atriplex, and broadest in spring, when the frequencies of these two plants were lowest. Diet breadth was slightly lower for RERA than for co-occurring species in pairwise comparisons. All four species consumed similarly high frequencies of wetland plants, but RERA consumed fewer grasses and upland plants, suggesting that it may be less suited to fragmented habitat than sympatric rodents. Diet overlap was lowest between RERA and the native California vole (Microtis californicus). In contrast, RERA diet overlapped substantially with the native western harvest mouse (R. megalotis) and non-native house mouse (Mus musculus), suggesting potential for competition if these species become sufficiently abundant.

The critically endangered Amargosa vole (Microtus californicus scirpensis) is found only in rare marsh habitat near Tecopa, California in a plant community dominated by three-square bulrush (Schoenoplectus americanus). Since the earliest research on the Amargosa vole, the existing paradigm has been that these voles are obligatorily dependent on bulrush as their only food source and for the three-dimensional canopy and litter structure it provides for predator avoidance. However, no prior research has confirmed the diet of the Amargosa vole. In this study we characterized the Amargosa vole' nutritional needs, analyzed the quality of bulrush by forage analysis, and performed microhistological and metabarcoding analyses of vole feces to determine what foods were consumed in the wild. All bulrush plant tissues analyzed were low in fat (from 0.9% of dry matter in roots to 3.6% in seeds), high in neutral detergent fiber (from 5.9% in rhizomes to 33.6% in seeds), and low in protein (7.3-8.4%). These findings support the conclusion that bulrush alone is unlikely to support vole survival and reproduction. Fecal microhistology and DNA metabarcoding revealed relatively diverse diets including plants in 14 families, with rushes (Juncaceae), bulrushes (Cyperaceae), and grasses (Poaceae) being the most common diet items. On microhistology, all analyzed samples contained bulrush, sedges (Carex sp.), rushes (Juncus sp.), and beaked spikerush (Eleocharis rostrellata) even from marshes where non-bulrush plants were uncommon. There was evidence of insects at <1% in two marshes but none in the remaining marshes. Metabarcoding detected ten genera of plants. When considering non-Schoenoplectus targets, for which metabarcoding had poor sensitivity, saltgrass (Distichlis spicata) was the most commonly detected species, with prominent contributions from seaside arrowgrass (Triglochin concinna) and yerba mansa (Anemopsis californica) as well. Diversity of vole diets generally increased with increasing site plant diversity, but differences were not statistically significant. Confirming details about dietary behaviors is critical for informing appropriate conservation planning including habitat management and reintroduction of voles into new sites.