The population of Rocky Mountain water shrew Sorex navigator occurring in the White Mountains, Arizona, is isolated, genetically divergent, and of conservation concern. However, little is known about its distribution and habitat use due to difficulty capturing animals during previous surveys. The objectives of this study were to report captures of S. navigator that occurred during a survey for the New Mexico jumping mouse Zapus luteus luteus that expand its known distribution, evaluate habitat of capture sites, report natural history observations, and describe methods for capturing S. navigator. We captured 17 S. navigator at six sites, making this the most successful survey for this population. The records included two new sites and confirmation of the persistence of two historical populations considered extirpated. At the landscape scale, elevation was the best predictor of sites for capture of S. navigator. We captured Rocky Mountain water shrews on small cold-water streams and seeps on saturated soil with high vertical cover of herbaceous plants primarily consisting of sedges Carex. Sherman live traps were ideal for capturing S. navigator, if set appropriately. Given the population's restricted distribution, fragile habitat, and ongoing threats, conservation measures may be warranted.
Traditionally, the water shrew Sorex palustris was considered to occur throughout much of the boreal zone of Alaska and Canada, with southern extensions of range in the Appalachian Mountains in the east and the Sierra Nevada, Great Basin, and Rocky Mountain ranges in the west (Hall 1981). However, based on molecular data, Hope et al. (2014) demonstrated that S. palustris was actually composed of three distinct clades, with eastern populations referred to Sorex albibarbis, boreal populations referred to S. palustris, and western “cordilleran” populations referred to S. navigator, the Rocky Mountain water shrew (Woodman 2018). Subsequently, Nagorsen et al. (2017) demonstrated the morphological distinctiveness of S. palustris and S. navigator. Members of the S. palustris species group are uniquely adapted to freshwater environments; for example, modified hairs provide insulation by repelling water and trapping air, and a fringe of stiff hairs on the feet and toes allows them to swim underwater and to walk on the surface of water (Beneski and Stinson 1987). S. navigator occurs along the Rocky Mountain cordillera from Alaska south to Arizona and New Mexico (Hope et al. 2014). Within S. navigator, there is considerable phylogeographic structuring, with the two most divergent lineages occurring in the American Southwest (southern Utah, Arizona, and New Mexico; Hope et al. 2014). The lineage from the White Mountains, Arizona, had the oldest split from other lineages of S. navigator and was most divergent (e.g., 2.15% sequence divergence compared with other American Southwest populations; Hope et al. 2014). The White Mountains region is located in east central Arizona and centered on Mount Baldy, an extinct volcano that reaches a 3,476-m elevation. No other populations of S. navigator are known from Arizona. The next nearest population of S. navigator occurs in the Jemez Mountains of north central New Mexico; the intervening approximately 350-km gap separating these two populations is a mostly low-elevation arid desert and grassland environment largely devoid of perennial water. Given the degree of differentiation and isolated nature of the White Mountains population, Hope et al. (2014) concluded that it represents a distinct evolutionarily significant unit that may warrant taxonomic recognition and that due to threats of climate warming, additional studies are necessary to better understand its distribution and population status. As a consequence of its restricted distribution, rarity, and association with sensitive riparian habitats, S. navigator is a Species of Greatest Conservation Need (Tier 1b) in Arizona's State Wildlife Action Plan, thus identifying it for immediate conservation action (Arizona Game and Fish Department 2012).
General accounts of S. navigator usually report it as occurring along swift streams that individuals enter to elude predators and to hunt aquatic insects (e.g., Ingles 1965; Beneski and Stinson 1987). Much information formerly ascribed to S. navigator actually refers to S. palustris or S. albibarbis (e.g., Wrigley et al. 1979). Information about habitat associations of S. navigator is meager and mainly limited to anecdotal descriptions of habitat or classifications into broad vegetation types, primarily in mammal inventories (e.g., Negus and Findley 1959; Spencer and Pettus 1966; Brown 1967; Rickart and Heaney 2001). In general, those studies describe S. navigator as occurring in wet places in proximity to flowing water in a variety of vegetation types. The only such study of S. navigator based on large sample sizes was Conway (1952), who provided a general description of habitat features of locations where 101 Rocky Mountain water shrews (hereafter water shrews) were taken in western Montana. Those water shrews were taken primarily along swift streams under overhanging banks or where rocks or logs provided cover (Conway 1952). Several studies have evaluated how different vegetation types or management treatments influence small-mammal communities that include S. navigator, but low capture rates have usually prevented conclusions about response of S. navigator (Anthony et al. 1987; Carey and Johnson 1995; Rowe 2007; Storm and Choate 2012). One exception is Wilk et al. (2010), who concluded that S. navigator responded negatively to riparian logging treatments, although based on small samples. Reichel (1986) is the only study that measured quantitative aspects of habitat at capture sites of S. navigator. Based on the capture of seven S. navigator in the alpine zone of Oregon and Washington, Reichel (1986) found that use of wet meadow and willow Salix vegetation types was higher than expected but that none of the quantitative habitat variables that he measured were significant.
In the White Mountains, S. navigator has been considered rare and difficult to survey, with few known occurrence records and meager information on habitat. In the most recent published account, Hoffmeister (1986; also reported by Lange  and Cockrum ) reported three occurrence records, all museum specimens collected decades earlier, from three locations in the White Mountains region: south end of Blue Range, Prieto Plateau; Horseshoe Cienega, North Fork White River; and Sheep's Crossing, West Fork Little Colorado River. Hoffmeister (1986) noted that recent attempts to capture the species were unsuccessful and speculated that livestock grazing could have resulted in loss of the species' habitat. No further scientific information about S. navigator in Arizona has been published. In the intervening years, six studies aimed to better understand distribution, status, and habitat of S. navigator in the White Mountains (Table 1). Those studies made only modest advances in providing new information, due primarily to difficulty capturing animals (Text S1, Supplemental Material). Those surveys resulted in documentation of S. navigator at five new locations: Fish Creek, Lee Valley Reservoir (Lee Valley Creek), Phelps Cabin (East Fork Little Colorado River), West Fork Black River, and Burro Creek (Smith 1993; Hanna 1994; Markow and Hocutt 1998; see Text S1 for additional details). In sum, the White Mountains population of S. navigator is currently known from eight locations on the basis of 35 captures, and no data exist on habitat use.
During 2008 and 2009, the senior author conducted surveys for the endangered New Mexico jumping mouse Zapus luteus luteus (formerly referred to as Zapus hudsonius luteus [Malaney et al. 2017]) in the White Mountains (Frey 2011; Frey 2017a, archived data). Z. l. luteus is a riparian obligate occurring along perennial streams, and S. navigator was captured as a nontarget species during surveys for Z. l. luteus. The new records of S. navigator provide important distributional information. In addition, given the paucity of information about habitat relations, we describe habitat used by S. navigator at the landscape and microhabitat scales based on quantitative data. Objectives of this study were to 1) refine the known distribution of S. navigator in the White Mountains, 2) evaluate habitat used by S. navigator in the White Mountains at the landscape and microhabitat scales and compare the findings with those of Z. l. luteus, 3) report natural history observations, 4) describe methods for capturing S. navigator, and 5) consider conservation implications of the results.
The White Mountains are located on the southern edge of the Colorado Plateau in Apache, Navaho, and Greenlee counties in east central Arizona. The Prieto Plateau is a southern high-elevation extension of the White Mountains region indicated on historical maps (e.g., Rand McNally and Company 1897), which is located southeast of the Black River, west of the San Francisco River, and north of the Mogollon Rim. The headwaters of two major watersheds of the Colorado River converge in the White Mountains: the Little Colorado River on the north side of the mountains and the Gila River on the east (San Francisco River), south (Blue River), and west (White River) sides of the mountains. Elevations range from 1,008 to 3,476 m, with lower elevations dominated by desertscrub and grassland biotic communities, mid elevations dominated by conifer woodlands, and high elevations dominated by several montane and subalpine grassland and conifer forest biotic communities (Frey 2010).
We conducted trapping surveys at 33 sites in the White Mountains, Apache and Greenlee counties, by using 10,706 trap-nights from 18 July to 13 September 2008 (n = 20) and from 25 June to 6 September 2009 (n = 13; Table S1, Supplemental Material). We selected survey sites to better describe the current distribution and habitat associations of Z. l. luteus; therefore, sites included historical locations for Z. l. luteus (many of which are also historical locations for S. navigator) and other areas with well-developed riparian vegetation. Together, these sites represented a wide array of stream and riparian vegetation types, although no effort was made to systematically sample all types. At each site, we set large (7.62 × 8.89 × 22.86 cm) Sherman live traps (model LFATDG; H.B. Sherman, Tallahassee, FL) baited with commercial horse sweet feed spaced approximately 3–5 m apart, especially targeting patches of the best developed foraging habitat for Z. l. luteus that was present. Foraging habitat for Z. l. luteus had the following characteristics: near water; saturated soil; vegetation composed of tall, dense herbaceous vegetation; and lacking trees, rocks, coarse woody debris, or bare ground. Transects were typically set via wading through the stream and setting traps on adjacent banks. Because foraging habitat usually occurred in patches, the trapping transects also traversed areas with other conditions where traps were also set. Thus, the traps sampled a wide range of streamside environments. Importantly, because Z. l. luteus associates with wet areas and saturated soils, we frequently set traps on shallow water (<2 cm) by placing pebbles, sticks, or crushed vegetation under the trap to ensure that the trap opening was at the water level, but water was not flowing into the trap, which could cause hydrostatic pressure to interfere with treadle mechanism or increase risk of trap mortalities.
Trap-night is a measure of survey effort wherein 1 trap-night is equivalent to one trap set for 1 night (Wilson et al. 1996). We aimed to achieve a sampling effort of 400 trap-nights at each site, but we usually discontinued trapping when we captured Z. l. luteus to prevent trap mortalities, resulting in smaller effort at some sites. Some sites had fewer trap-nights for logistical reasons. We calculated capture rate as the number of S. navigator captured per 100 trap-nights. We evaluated some sites for habitat but did not trap due to human activity. Because of a confidentiality agreement, we do not present detailed survey location data but they are available from theArizona Game and Fish. Because detailed location data are not presented, we use the same survey site names as in Frey (2017a, archived data) to prevent confusion (Table S1). S. navigator is a distinctive shrew and unlikely to be misidentified as any of the other sympatric shrew species (Hoffmeister 1986). Trap mortalities were collected and prepared as voucher specimens and deposited in the Museum of Southwestern Biology, University of New Mexico (Table S2, Supplemental Material).
Habitat data collection
We collected quantitative habitat data at two scales, landscape (i.e., along an ∼1-km reach of stream) and microhabitat (i.e., on 4-m radius plot), and they are the same as detailed in Frey (2017a, archived data). At the landscape scale, we compared habitat attributes at the site level (i.e., comparison of sites where we captured S. navigator vs. sites where we did not capture S. navigator). Because this was a mensurative study design that did not control for confounding variables, use replication, or manipulate the system, the scope of inference is limited to the study sites (Morrison et al. 2008). At the microhabitat scale, we used the data to describe conditions at traps that captured S. navigator. Consequently, the scope of inference is limited to those capture locations. We only collected microhabitat data at capture locations for Z. l. luteus, capture locations for S. navigator, or a representative location of well-developed herbaceous vegetation at sites where we captured neither species. Consequently, it was not possible to compare microhabitat at locations where we captured S. navigator vs. locations where we did not capture S. navigator.
For the landscape scale (Frey 2017a, archived data), at each survey site we established paired transects paralleling the stream and located 0.5 m (i.e., “stream-edge transect”) and 4.5 m (i.e., “inland transect”) from the edge of the green-line (i.e., vegetation closest to the water). We located the paired transects on the same side of the stream that trapping occurred, or we randomly determined the side if trapping occurred on both sides of the stream. We established sample stations each 20 m along transects, except at two sites where the survey area was small and the interval reduced to 10 m to increase sample size. The goal for transect length was 1 km, but site characteristics or lack of access reduced the length at some sites. We measured vertical cover with a Robel pole (Robel et al. 1970). At each station, we read a Robel pole from the opposing transect (i.e., 4 m away) at 1-m eye level, and we recorded the lowest 1-in. (25.4-mm) segment that was not obstructed by cover. In addition, we recorded the dominant plant or other structure that provided the cover: conifer, rush (Juncaceae), sedge Carex, grass (Poaceae), forb, willow, alder Alnus, shrub cinquefoil Dasiphora fruticosa, Wood's rose Rosa woodsii, other shrub, other plant, dead standing plant, coarse woody debris, rock, and bank. We did not use this method at sites with complex wetlands spread across broad valleys because there was no defined stream edge.
We collected microhabitat data at traps where we captured S. navigator (Table S3, Supplemental Material). At the trap, we visually estimated slope and aspect. We measured canopy cover with a densitometer in the four cardinal directions. We obtained an index of soil moisture ranging from 1 (dry) to 10 (saturated) by using a soil moisture probe (Lincoln Irrigation, Lincoln, NE) inserted into the ground approximately 4 cm. We assessed vertical cover with a Robel pole (read in inches) from a 4-m distance at a 1-m eye level. We read the Robel pole at the trap site from three random azimuths as well as away from the trap along three random azimuths. We established four 4-m perpendicular transects at a random azimuth from the trap. At each 1-m interval along a transect, a Daubenmire frame was used to assess the percent cover of sedges, rushes, field horsetail Equisetum arvense, forbs, grass, willow, alder, redosier dogwood Cornus sericea, shrub cinquefoil, Wood's rose, moss, other plants, coarse woody debris, litter, rocks, gravel, bare ground, and open water. Cover classes were 1 for 0–5% cover, 2 for 5–25% cover, 3 for 25–50% cover, 4 for 50–75% cover, 5 for 75–95% cover, and 6 for 95–100% cover. In addition, we recorded soil moisture, litter depth, and stubble height in each frame. We measured stubble height with a ruler and recorded height as both the laid-over stubble height and vertical stubble height (in millimeters). We measured laid-over stubble height as the representative height of the vegetation as it naturally lay. We obtained vertical stubble height by measuring the height of a representative blade of the dominant herbaceous vegetation that was fully extended vertically from the ground. Grazing by livestock was not common on most of the survey sites; thus, the stubble height often reflected the natural growth of the plants. Finally, we recorded the number and identity of each tree and shrub within 1 m of the transect. For each trap location, we averaged measurements of canopy cover, soil moisture, vertical cover, stubble height, and ground cover class estimates. We determined water temperature at capture locations by using a thermometer. We did not collect some measurements that might be meaningful for S. navigator (e.g., stream velocity) because we aimed the survey methods at Z. l. luteus.
We calculated statistics using SPSS 10.0 for Windows (SPSS 1999). We tested variables for normality by using one-sample Kolmogorov–Smirnov tests and used Pearson and Spearman correlations to assess relationships among variables for normal and nonnormal variables, respectively. For microhabitat variables, we report descriptive statistics for capture locations. For the stream reach variables, we tested for differences between survey sites where S. navigator was captured or not captured by using two-tailed t-tests (corrected using Levene's test for equality of variances) and two-sample Kolmogorov–Smirnov tests for parametric and nonparametric data, respectively. We used discriminant function analysis (DFA) to evaluate multivariate differences in stream reach cover at sites where we captured S. navigator vs. sites where we did not capture S. navigator. To reduce the chance for multicollinearity problems in the data sets, we excluded variables from the analysis that exhibited high (i.e., r > 0.6) correlations with other variables (McGarigal et al. 2000). We used a stepwise selection procedure and used Wilks' lambda to rank the variables in ability to discriminate by passing the tolerance tests (0.05 to enter; 0.10 to remove). We used a chi-square transformation of the overall Wilks' lambda to test for multivariate differences in habitat between locations where we captured or did not capture S. navigator. We used a classification routine in the DFA to predict whether S. navigator would be captured or not captured at sites that we did not trap and to predict presence of S. navigator at sites where we did not capture S. navigator.
We had 17 captures of S. navigator at six sites, all in 2009 (Figures 1 and 2; Table 2). Two of these sites, Lee Valley Creek Lower and Rudd Creek Upper, represent new distributional records. The 33 survey sites included five of the eight historical locations for S. navigator in the White Mountains. Historical locations not surveyed were Horseshoe Cienega, White River (due to lack of permission), West Fork Little Colorado River (due to human activity; only habitat data were collected), and Burro Creek (due to lack of information about the specific capture location). We captured S. navigator at four of the five historical sites surveyed: KP Creek (=Prieto Plateau), Fish Creek, Lee Valley Creek Upper, and East Fork Little Colorado River Upper. West Fork Black River Upper was the only historical location where we failed to capture water shrews. Capture rates of S. navigator were low (0.23–1.25 captures/100 trap-nights; Table S1). At sites where we captured S. navigator, it required 1–3 nights of trapping by using a mean effort of 243 trap-nights (range = 139–440; SD = 145; Table S1). However, at four of the six sites it required only 1 night of trapping by using an effort of 139–160 trap-nights to capture the first S. navigator (Table S1).
At the stream reach scale, sites where we captured S. navigator were dominated by grass and sedge cover and had significantly less forb and alder cover than noncapture sites (Figures S1–S5, Supplemental Material; Table 3). The stepwise DFA of stream reach cover data revealed elevation as the only significant predictor of presence of S. navigator (Wilks' lambda = 0.649; χ2 = 10.595; df = 1; P = 0.001). We classified sites with elevations of 2,695 m or higher (i.e., elevation of West Fork Black River Upper) as S. navigator present, whereas we classified sites less than or equal to 2,596 m in elevation (i.e., elevation of Hannagan Creek Upper) as S. navigator absent. Consequently, the DFA misclassified Rudd Creek Upper (elevation 2,491 m) as absent. Sites predicted to have S. navigator, but no captures, were West Fork Black River Upper, Fish Creek, Spruce Creek, Buckshot Creek, and Corduroy Creek. We classified the sites on the West Fork of the Little Colorado River that we did not trap as present for the Upper site and absent for the Lower site. We noted that elevation correlated with capture rate of water shrews (rs = 0.530; P = 0.002).
Microhabitat at capture locations was typified by flat slopes with uniformly saturated soil (Table 4). Canopy cover tended to be low (mean = 39%) but was variable (SD = 35.2; range = 0–98%). Vertical cover (mean = 45.7 cm) and vertical stubble height (mean = 37.2 cm) were high. Litter depth tended to be low (mean = 1.8 cm) but was variable (SD = 20.3) and sometimes fairly deep (maximum = 8.6 cm). Sedges were the dominant ground cover, although grasses, forbs, and open water also had high coverage. Shrubs and trees were rare components of S. navigator microhabitat. Mean water temperature at capture locations averaged 11°C (range = 7.5–17°C).
Natural history observations
The mean (SD and range in parentheses) for standard measurements (in millimeters) and mass (in grams) of 15 specimens were as follows: total length = 145.7 mm (SD = 9.8; range = 115–156 g), tail length = 66.3 mm (SD = 6.5; range = 44–71 g), hindfoot = 19.0 mm (SD = 1.9; range = 13–20.5 g), ear = 4.4 mm (SD = 1.2; range = 3–7), and mass = 11.2 g (SD = 2.3; range = 7.3–14.5 g; Table S2). Observed pelage variation within our sample of S. navigator included individuals with buffy or white venters and individuals with white-tipped tails; in one individual, the white pelage covered as much as half of the terminal portion of the tail. Only 2 of the 15 specimens were sexually active and both were from Rudd Creek Upper: one specimen captured on 23 July was a 10.5-g female with an open, swollen vagina and the other specimen captured on 24 July was an exceptionally large male (17.5 g) that had enlarged testes (8 × 5 mm) and a large, easily protrudable penis (Table S3). This male also had unusual brown pelage compared with the usual charcoal gray pelage. Sexually active male S. navigator develop skin glands along the flanks between the fore- and hind limbs that are recognizable by an oval patch of white hair (Conaway 1952); no such glands were observed in our specimens. A female captured 12 July at West Fork Fish Creek appeared to be a juvenile based on small body size (mass = 7.3 g, total length = 115 mm, tail = 44 mm, hindfoot = 13 mm).
Prior studies speculated that streams on and near the base of Mount Baldy were the last refugium for S. navigator in the White Mountains (Smith 1993; Hanna 1994; Markow and Hocutt 1998; Bogan and Ramotnik 2001). We confirmed persistence of S. navigator at two historical sites (East Fork Little Colorado River Upper and Lee Valley Creek Upper) and one new site (Lee Valley Creek Lower) in the Mount Baldy area. We also verified S. navigator at other sites. Our capture of S. navigator at West Fork Fish Creek (Apache County) confirms the persistence of the Fish Creek population, which was last documented in 1987 and thought extirpated (Bogan and Ramotnik 2001). A specimen of S. navigator collected in 1914 by Edgar A. Goldman from “Prieto Plateau, S End Blue Range, 9,000 ft” has remained enigmatic because of the obscure location, changes in place names, and because all other historical records were from streams draining the vicinity of Mount Baldy. Field notes written by Goldman verify that the collection location was on KP Creek at KP Cienega, Greenlee County, located approximately 40 km southeast of Mount Baldy (Figures S6 and S7, Supplemental Material). Our capture of S. navigator at KP Creek confirms that a population persists on the Prieto Plateau. The KP Creek records also are the only from the Blue River watershed. Thus, S. navigator may occur along other nearby streams draining the Prieto Plateau and other areas of the Blue River watershed, including in adjacent New Mexico.
Our capture of S. navigator at Rudd Creek Upper represents an important new distributional record. Rudd Creek is a tributary to Nutrioso Creek, which drains a portion of the eastern edge of a large plateau composed of Holocene-to-middle Pliocene basaltic rock situated east of the Mount Baldy volcano and bounded by the drainages of the Little Colorado River to the north, Nutrioso Creek to the east, and the Black River to the south. Prominent features of the plateau include Pool Knoll, Rudd Knoll, and Big Lake. This plateau (hereafter called the Pool-Rudd Knolls Plateau) averages 2,800 m in elevation and is dominated by expansive montane grassland with coniferous forests primarily restricted to isolated knolls and the rougher terrain of the rugged edges of the plateau. Fish Creek (Apache County) originates on a similar but smaller plateau of the same geologic origins located north of Mount Baldy and includes Sunrise Lake. The canyon of the West Fork of the Little Colorado River isolates the two plateaus. Thus, the records from Fish Creek and Rudd Creek Upper suggest S. navigator may occur on other nearby streams that drain these plateaus. Similarly, the Nutrioso Creek watershed also drains portions of Escudilla Mountain. Consequently, we recommend that future survey efforts concentrate on areas away from Mount Baldy, particularly streams draining the adjacent basaltic plateaus, Escudilla Mountain, and the Prieto Plateau.
Because Z. l. luteus was the target of the surveys, we did not set all traps in situations ideal for capturing S. navigator. For example, at West Fork Black River Upper, a historical location for S. navigator, we set most traps in vegetation on the bank of the stream above the level of the water, rather than at the edge of the stream channel where S. navigator is typically active. Consequently, our failure to capture S. navigator at this site may be due to trap placement and hence represent sampling error rather than extirpation. Results of this study almost certainly underestimate the distribution of S. navigator in the White Mountains, and our failure to capture S. navigator at a site does not necessarily imply that S. navigator was absent at the site.
Generic accounts of S. navigator typically describe the species as primarily occurring along rocky, swift-flowing mountain streams (e.g., Beneski and Stinson 1987). By contrast, although our results cannot refute that habitat description (because we did not sample a large number of such places), the locations where we caught S. navigator in the White Mountains did not match that description. Places where we captured S. navigator in the White Mountains were along small (<2 m wide, often <1 m wide) streams in mostly low gradient (e.g., <2%) reaches with slow-moving water that ran through meadows with few rocks. At some sites (e.g., KP Creek and Rudd Creek Upper), the streambed had a rocky substrate due to steeper gradient (e.g., <4%), but the banks were mostly soil and well vegetated. We captured several water shrews on tiny seeps (∼0.3 m wide, 2 cm deep) that ran into the small creeks. Kinsella (1967) described capturing S. navigator in similar situations in Montana. We almost always caught our specimens in traps set on shallow (∼1-cm-deep) slack or slow-moving water at the edge of the stream where vegetation cover existed. It is likely that S. navigator uses these areas for movements and foraging on aquatic and terrestrial insects, its primary prey (Conaway 1952). Although S. navigator can dive and swim underwater, it has limited ability for sustained underwater swimming (<50 s; Calder 1969). Its pelage traps air for insulation, but this trapped air makes the animal buoyant, which could result in it being swept downstream cork-like in swift water. In addition, the pelage becomes wetted after a few minutes at which point the animals must exit the water to dry its pelage (Conaway 1952). Consequently, in terms of understanding habitat use of S. navigator, we consider it important to view them as terrestrial animals with adaptations for foraging in and near water, rather than as aquatic animals that occasionally take to land as apparently was the view during some prior surveys and that probably influenced low trapping success (e.g., Markow and Hocutt 1998).
The finding that elevation was the best predictor of sites where we captured S. navigator was not surprising given that S. navigator is primarily restricted to Canadian and Hudsonian life zones and that S. navigator has various morphological and physiological adaptations for living in cold environments (e.g., Beneski and Stinson 1987). Elevation can be considered a “master” variable that serves as a proxy for other more proximate influences on distribution such as climate, vegetation, and biotic interactions. However, because of the overall poor sampling of S. navigator in the White Mountains, it is also likely that elevation limits are not completely understood. For example, given that Rudd Creek is located on the northeastern face of the Pool-Rudd Knoll Plateau, it is possible that local cooler edaphic conditions allow for a lower elevation limit than would be permitted on a stream with a southern exposure. Of the six sites where S. navigator was predicted to occur (i.e., sites ≥2,695 m), but where we did not capture any (West Fork Little Colorado River Upper, Fish Creek, Buckshot Creek, Spruce Creek, Corduroy Creek, and West Fork Black River Upper), S. navigator had previously been documented at three of these sites (West Fork Little Colorado River Upper, Fish Creek, and West Fork Black River Upper), lending support to the DFA analysis. The DFA does not accurately identify the lower elevation limits of S. navigator. The lowest elevation site where we captured S. navigator, Rudd Creek Upper, was a misclassification in the DFA (i.e., S. navigator was predicted to be absent based on elevation), and the lowest elevation trap at that site that captured S. navigator was 2,406 m. Thus, it remains a possibility that other sites above 2,406 m could potentially harbor S. navigator.
Because the main goal of our surveys was to document distribution and habitat of Z. l. luteus, our results may not reflect the full range of habitat conditions used by S. navigator in the White Mountains. However, it is clear that our methods were able to identify different habitats selected by the two species within context of the sites sampled. For example, at the stream reach scale S. navigator occurred at higher elevation sites dominated by grasses and sedges and containing significantly less alder and forbs than sites where they were not captured. By contrast, based on the same suite of sites surveyed, Z. l. luteus also used sites with high cover of grasses and sedges but that contained significantly more forbs and alder than where they were not captured (Frey 2017a, archived data). Similarly, at the microhabitat scale, S. navigator capture locations had substantially less forb and alder cover and higher sedge and grass cover than represented in Z. l. luteus microhabitat (Frey 2017a, archived data). In addition, S. navigator microhabitat contained more bare ground, coarse woody debris, litter, rocks, moss, and trees than did microhabitat for Z. l. luteus, which may reflect broader habitat tolerances for S. navigator. As a result of differences in microhabitat, vertical cover and stubble heights were lower for S. navigator than for Z. l. luteus.
Natural history observations
Our specimens of S. navigator from the White Mountains were small in external measurements, particularly tail length, in comparison with specimens from Colorado and Montana (Conaway 1952; Armstrong 1972). Jackson (1928) reported that the specimen from Horseshoe Cienega had a white-tipped tail. That some of our specimens also had white-tipped tails suggests that this trait is not uncommon in the White Mountains. The senior author has also observed white-tipped tails on S. navigator captured in northern New Mexico and southern Colorado. We are not aware of any other reports of specimens of S. navigator having white-tipped tails. Thus, this trait could be confined to the Southwestern clade identified by Hope et al. (2014). Additional research is necessary to evaluate the phylogenetic relationships and taxonomy of water shrews in this region.
Conaway (1952) concluded that male and female S. navigator generally do not sexually mature and breed until their second year (i.e., after their first winter). Because the average life span of S. navigator is approximately 18 mo, most individuals in a population are therefore likely immature first-year animals (Conaway 1952). This is consistent with our observation that all but two of our specimens were sexually immature. S. navigator is active year-round, and in Montana females with embryos were found from February to July (Conaway 1952). Our capture of a juvenile on 12 July is consistent with this reproductive time frame. By contrast, our only captures of sexually mature individuals was in late July. This suggests that at least some pregnancies would be expected later in the summer (e.g., August) in the White Mountains. Thus, the population of S. navigator in the White Mountains may have a different reproductive phenology compared with that of S. navigator in Montana. The large size of the sexually active male compared with the sexually active female at Rudd Creek Upper is consistent with other studies (e.g., Conaway 1952).
Recommendations for surveys
Although we did not design this study to detect S. navigator, it represents the most successful survey for S. navigator in the White Mountains, both in terms of numbers of individuals captured and numbers of sites where the species was documented. Low capture success during prior studies was likely due to choice of capture device and trapping methods. Most prior surveys relied primarily on pitfall traps. Pitfall traps are often considered a superior method for capturing shrews because their small body size can render other trap types ineffective, particularly those that depend on an animal of particular weight or size to trigger a mechanism. However, the size of S. navigator captured during this study (n = 16; mean = 11.2 g; range = 7.3–17.5 g) was as large, or larger, than many rodents (e.g., pocket mice Perognathus, harvest mice Reithrodontomys) that are routinely captured with Sherman traps. Therefore, there is no reason to use pitfall traps on basis of body size of S. navigator. In addition, pitfall traps are labor intensive and difficult to effectively deploy in montane streamside habitat due to rocky soils or hydrostatic pressure that pops buckets out of the ground. Strategies some researchers have used to overcome these problems, such as partially burying the bucket and building a cone of earth around the rim, placing large rocks on top of the bucket, or setting traps on floating rafts (e.g., Markow and Hocutt 1998), likely interfere with the trap's effectiveness. A major advantage of Sherman traps is that they can be efficiently set in large numbers and in a diversity of situations, including on saturated soils and on shallow water, which we consider essential for capturing water shrews. Small snap traps are also effective for capturing S. navigator (Conaway 1952), although use of kill traps may not be appropriate for populations of conservation concern, such as the White Mountain population of S. navigator. Pitfall traps that collect water are usually similarly lethal for small mammals, including for those with semiaquatic adaptations (e.g., Frey 2017b). See Text S2, Supplemental Material for additional recommendations for trapping S. navigator.
The White Mountains population of S. navigator is genetically distinct and restricted to a small, isolated region of high elevation. Based on occurrences in several watersheds that have headwaters in isolated high-elevation areas of the White Mountains region (Mount Baldy, Pool-Rudd Knolls Plateau, Prieto Plateau), it is possible that the population is composed of several isolated or semi-isolated subpopulations due to natural fragmentation caused by topography. In addition, S. navigator is a habitat specialist. We found that S. navigator in the White Mountains was associated with low gradient reaches of small, high-elevation streams in places with dense herbaceous streamside vegetation on saturated soil, especially composed of sedges. However, it is possible that these results were influenced by the selection of survey sites that targeted Z. l. luteus, which depends upon similar habitats (Frey 2017a, archived data). Consequently, we recommend that our results be confirmed by future studies aimed specifically at establishing distribution and habitat use by S. navigator in the White Mountains. Z. l. luteus is listed as endangered (as Z. h. luteus) pursuant to the U.S. Endangered Species Act (ESA 1973, as amended) due to habitat loss as a result of livestock grazing, loss of beaver Castor canadensis, wildfire, drought, and hydrological alterations (U.S. Fish and Wildlife Service 2014). Given the similar habitat associations of S. navigator and Z. l. luteus, it seems possible that S. navigator has also experienced population losses, which is of significant conservation concern given that the White Mountains population of S. navigator is genetically distinct and uses a narrower range of elevations than does Z. l. luteus. The specialized habitat required by S. navigator is sensitive to disturbance and prone to habitat loss. However, the hypothesis that habitat loss has caused population losses requires further investigation.
It is possible that S. navigator is experiencing ongoing threats to its survival in the White Mountains. Human-caused habitat alteration, such as reservoirs, development, heavy livestock grazing, wildfire, and other factors, have reduced availability and fragmented its habitat. Given the adaptations S. navigator has to cold conditions (e.g., relatively high basal metabolic rate [Gusztak et al. 2005]), climate change that increases temperature of streams and terrestrial environment may be threats. In the Ruby Mountains, Nevada, the range of S. navigator has shifted up-slope, consistent with predictions of a warming climate (Rowe et al. 2010). Besides changing temperatures, climate change may also result in other threats to this species including loss of perennial flows and increased catastrophic wildfire. Habitat restoration and management used to benefit Z. l. luteus are likely to also benefit S. navigator, given their similarities in habitat. However, most of the sites occupied by Z. l. luteus in the White Mountains are below the known elevation limits for S. navigator (Frey 2017a, archived data), which could leave populations of S. navigator exposed to threats. Consequently, additional research is necessary to better understand the taxonomic status, genetic variation, distribution, habitats, and threats to S. navigator in the White Mountains. Improved sampling methods are likely to reveal new locations for S. navigator in the White Mountains. However, such discoveries, per se, do not negate potential conservation concern. Rather, we recommend that future studies focus on understanding prospects for long-term persistence of this evolutionarily distinct population.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Text S1. Synopsis of surveys for the Rocky Mountain water shrew Sorex navigator in the White Mountains, Arizona, conducted before 2008.
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S1 (20 KB DOCX).
Text S2. Recommended methods for capturing the Rocky Mountain water shrew Sorex navigator using Sherman live traps during future studies in the White Mountains, Arizona, or elsewhere in the species' range.
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S2 (22 KB DOCX).
Table S1. Sites surveyed for the New Mexico jumping mouse Zapus luteus luteus, survey dates, trapping effort (trap-nights), and captures rate of the Rocky Mountain water shrew Sorex navigator during 2008–2009 in the White Mountains, Apache and Greenlee counties, Arizona. Specific locality information is protected by a confidentiality agreement and is available from the Arizona Game and Fish Department. Site designations and names are as reported in Frey (2017a, archived data).
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S3 (20 KB XLSX).
Table S2. Museum of Southwestern Biology catalog number, J. K. Frey field number, capture location, date, reproductive condition, external measurements, mass, and disposition for specimens of the Rocky Mountain water shrew Sorex navigator captured in the White Mountains, Apache and Greenlee counties, Arizona, 2009. Site names follow Frey (2017). Specific locality information is protected by a confidentiality agreement and is available from the Arizona Game and Fish Department. Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S4 (20 KB XLSX).
Table S3. Archive of microhabitat data collected at capture locations for the Rocky Mountain water shrew Sorex navigator in the White Mountains, Apache and Greenlee counties, Arizona, 2009.
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S5 (19 KB XLSX).
Figure S1. Photographs of a representative capture locations for the Rocky Mountain water shrew Sorex navigator at East Fork Little Colorado River Upper, in the White Mountains, Arizona, 2009. The Robel pole is at the trap location.
Figure S2. Photographs of representative capture locations for the Rocky Mountain water shrew Sorex navigator at Lee Valley Creek Lower (top row, bottom left) and Lee Valley Creek Upper (bottom right) in the White Mountains, Arizona, 2009. The Robel pole is at the trap location.
Figure S3. Photograph of the capture location for the Rocky Mountain water shrew Sorex navigator at West Fork Fish Creek, in the White Mountains, Arizona, 2009. The red flagging is at the trap location.
Figure S4. Photograph of the capture location for the Rocky Mountain water shrew Sorex navigator at KP Creek, in the White Mountains, Arizona, 2009. The Robel pole is at the trap location.
Figure S5. Photographs of representative capture locations for the Rocky Mountain water shrew Sorex navigator at Rudd Creek Upper, in the White Mountains, Arizona, 2009. The Robel pole is at the trap location.
Figure S6. Reproduction of the 20 August to 15 September 1914 itinerary and description of physical features by Edgar A. Goldman describing the capture location of a specimen (USNM 22510) of the Rocky Mountain water shrew Sorex navigator captured from KP Cienega, on the Prieto Plateau, Arizona (Smithsonian Archives, Record Unit 7176, U.S. Fish and Wildlife Service, Field Reports, Box 23, Folder 17).
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S7 (5.75 MB PDF).
Figure S7. Reproduction of a portion of the 20 August to 19 September 1914 mammal specimen catalog of Edgar A. Goldman describing the capture of a specimen (USNM 22510) of the Rocky Mountain water shrew Sorex navigator at KP Cienega, on the Prieto Plateau, Arizona (Smithsonian Institution Archives, Record Unit 7176, U.S. Fish and Wildlife Service, Field Reports, Box 23, Folder 16).
Found at DOI: https://doi.org/10.3996/102019-JFWM-092.S8 (7.18 MB PDF).
We thank J. Underwood, M. Bogan, and C. Ramatnik for providing information about water shrews captured during their surveys and J. Malaney, M. Moses, J. Redman, and G. Wright for assistance in the field. We thank D. Shapiro and H. Stover for assistance providing E. A. Goldman's field notes in the Smithsonian Institution Archives. We thank the Apache-Sitgreaves National Forest and Arizona Game and Fish Department for permits, access to survey sites, and other logistical help. We thank R. Goljani for making the map. We thank the reviewers and editors for helpful comments that improved this article. This study was made possible by funding provided by Arizona Game and Fish Heritage Fund, Grant I09004.
Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Citation: Frey JK, Calkins MT. 2020. Habitat use of the Rocky Mountain water shrew in the White Mountains, Arizona. Journal of Fish and Wildlife Management 11(1):196–209; e1944-687X. https://doi.org/10.3996/102019-JFWM-092
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.