Introduction

The desert pocket gopher (Geomys arenarius) is an herbivorous rodent of the family Geomyidae (^{Williams and Baker 1974}). Its distribution is restricted to the states of New Mexico and Texas in the United States and northern Chihuahua in México (^{Anderson 1972}; ^{Chambers et al. 2009}). This species usually builds its burrows in sandy soils ( ˃ 40 %) and avoids clay, gravel, or stone (^{Mauk et al. 1999}). Gophers live near water bodies (rivers, ponds, irrigation canals) and agricultural areas (^{Lacher et al. 2019}). These organisms feed mainly on leaves, roots, tubers, wood, bark, seeds, grains, nuts, fruits, and flowers of different plant species (^{Templeton 2006}). Gophers cause significant damage to crops within their range (^{Monge 1999}; ^{Witmer et al. 1999}; ^{Engeman and Witmer 2000}; ^{Monge 2013}; ^{Baldwin et al. 2013}), although they have been reported to feed mainly on the herbaceous plants and grass encountered while tunneling (^{Myers and Vaughan 1965}; ^{Foster and Stubbendieck 1980}; ^{Luce et al. 1980}; ^{Williams and Cameron 1986}).

The Médanos de Samalayuca Flora and Fauna Protection Area (MSFFPA) is a Natural Protected Area (NPA) located in northern Chihuahua, México. It has a program that includes the management and sustainable use of wildlife, including the implementation of population ecology studies (^{CONANP 2013}). Different economic activities are carried out within this NPA, most notably irrigated crops of pecans (Carya illinoinensis), where a population of G. arenarius is established successfully. There is a conflict between this species and local farmers, who consider it a pest. Damages to the irrigation system and crops are attributed to gophers, as they are suspected of feeding on the roots of walnut trees, thus affecting the establishment, health, and production of these plants. Consequently, pest control is conducted using traps to reduce the abundance of gophers within crop areas. It is hypothesized that gophers are generalists that feed mainly on the vegetation associated with crops and that changes in plant cover affect the physical condition of this species. Therefore, the objective of this study was to define the extent of the trophic niche, changes in the physical condition of G. arenarius throughout the year, and whether they are actively feeding on walnut tree roots.

Materials and methods

Study area. The Arantxa Ranch comprises 1000 ha and is located within the Médanos de Samalayuca Flora and Fauna Protection Area at coordinates 31° 12' 2.13” N, -106° 28' 11.36” W (Figure 1). The prevailing climate is very dry, with warm summer and cold winter, mean annual temperature of 15 ºC to 25 °C, and mean annual precipitation of 212 mm (^{Enríquez-Anchondo 2003}).

The soil is sandy, originally covered by microphyllous desert shrubland (^{CONANP 2013}). This vegetation has been replaced by irrigated crops of vines (Vitis vinifera), pistachios (Pistacia vera), and pecans (Carya illinoinensis); the latter is the most important crop, covering 400 ha. The vegetation associated with crops is mainly composed of species of the families Poaceae, Asteraceae, Solanaceae, and Boraginaceae.

Figure 1 Location of the study area in the Arantxa Ranch within the Médanos de Samalayuca Flora and Fauna Protected Area. 

Capture of Geomys arenarius specimens. Eight commercial gopher traps (Sweeney's) were installed along a 1,400 m transect within the crop area in the following seasons: dry (21 to 22 June 2020), wet (26 to 27 September 2020), and post-wet (14 to 15 November 2020). Traps were placed following the methodology by ^{Chávez-León (2017}), monitoring and relocating them every 30 minutes for six hours. Simultaneously, we collected the corpses of G. arenarius from the trapping conducted by the ranch pest control department.

Fitness Index. This index evaluates the differences between weights according to breed, sex, age, sexual and social status, season, climate, disease, and food, reflecting genotypic, phenotypic, and environmental interactions (^{Bailey, 1968}). It is based on the body weight: length ratio, calculated with the formula: IK = (W/LC³) x 10^⁵, where W = weight (kg) and LC = body length (cm) to the base of the tail, with values ranging from 1 to 10 (^{Corriale et al. 2013}). To note, the values of the morphometric and IK measurements correspond to 23 specimens (5 males and 18 females). A total of 17 specimens were excluded, 11 lacking the tail and 6 pregnant or lactating females (4 in the dry season and 2 in the wet season).

The individuals captured were euthanized by cervical dislocation following the Guidelines of the American Society of Mammalogists for the use of wild mammals in research (^{Sikes and Gannon 2011}). Conventional morphometric measurements (total length, tail length, right leg length, and ear length) were recorded using a ruler; the weight was measured with a Pesola® scale, and sex was determined based on the presence/absence of the baculum (^{Lorenzo et al. 2006}). The digestive tract was removed, placed in a flask containing 10 % formol, and transported to the Animal Ecology and Biodiversity Laboratory (LEBA, for its acronym is Spanish) at Universidad Autónoma de Ciudad Juárez (UACJ). The skulls of the collected specimens were deposited in the Scientific Collection of Vertebrates at UACJ (CHI-VER 189-08-06) registered with SEMARNAT.

Vegetation sampling. Seven 25 m² quadrants separated by 200 m were established along the trapping transect. The plant species comprising the vegetation cover were recorded in each season following the methodology by ^{Mostacedo and Fredericksen (2000}). As reference material, we used plant samples from the PJ018 CONABIO Project deposited in the UACJ Herbarium (HERB-UACJ) and samples available in the Animal Ecology and Biodiversity Laboratory (LEBA, for its acronym in Spanish) at UACJ. Additionally, some plant samples were collected in the field following the methodology by ^{Ricker (2019}) and deposited in the LEBA reference collection as a botanical catalog (Appendix 1).

Plant material processing. Pressed plant specimens were dried at room temperature for two weeks. Once dry, the family, genus, and species were determined using the ^{SEINet database (2020)} with the assistance of the UACJ Herbarium staff. Subsequently, we performed the histological technique described by ^{Gallina-Tessaro (2011}).

We built a catalog of reference plant material with diagnostic characteristics that comprised a total of 32 species, 21 collected in the dry season, 29 in the wet season, and 9 in the post-wet season.

Processing of the digestive tracts of Geomys arenarius. The stomach contents of 40 desert pocket gophers were recovered. This material was dehydrated at 80 °C for 4 hours, grounded with a porcelain mortar with pestle, and sieved through a 1 mm-diameter mesh. Histological slides were prepared from the sieved material following the methodology by ^{Castellaro et al. (2004}). Five slides were prepared from each sample and examined under a light microscope at 40' including 10 fields of view. The frequency was calculated using the formula: Fr = (ai/A)*100, where ai = number of observations of a particular food element and A = number of total observations.

Levin's Niche Breadth Index. This index represents the specialization of an organism by measuring the elements that make up its diet. The standardized form has values ranging from 0 to 1, where 0 corresponds to a specialist and 1 to a generalist (^{Alarcón-Nieto and Palacios 2009}). The Levin's niche index is expressed as B = 1/ΣPj², where Σ = sum, Pj= ratio of individuals using resource j, and the standardized index as Bₐ = B - 1/n+1, where B = Levin's index and n = number of resources used by organisms, where values lower than 0.6 correspond to a specialist diet and those higher than 0.6, to a generalist diet (^{Krebs 1989}).

Statistical analysis. The mean and standard deviation of morphometric measurements, weight, and IK of 23 specimens were obtained and sorted by season and sex. Due to the number of samples for each season, nonparametric statistics were used to determine significant differences in male IK by season using the Mood test. The Kruskal-Wallis test was used for females, and significant differences were analyzed with the Conover test. The degree of association between plant cover and IK was estimated using Spearman’s correlation coefficient. Statistical analyses were performed in Excel using the SPSS Statistics Base 22.0 statistical package.

Results

A total of 40 adult specimens of Geomys arenarius were collected as follows: 11 females and 3 males in the dry season, 13 females and 8 males in the wet season, and 5 females in the post-wet season.

Males (n = 5) had the following mean measurements: total length (TL) 267.40 ± 9.60 mm, tail length (T) 77.6 ± 3.78 mm, hind foot (HF) 33.60 ± 3.71 mm, ear length (E) 5.0 ± 1.0 mm, and weight (w) 194.80 ± 43.75 g. Females (n = 18) had the following mean measurements: TL = 234.94 ± 19.24 mm, T = 68.05 ± 11.04 mm, HF = 32.05 ± 3.05, E = 4.05 ± 0.87, and w = 120.83 ± 26.03 g in all three seasons. The measurements and weights by season corresponded to 10 specimens (3 males and 7 females) in the dry season, 8 (2 males and 6 females) in the wet season, and 5 females in the post-wet season (See Table 1).

Fitness Index (IK). The average IK of males was 2.82 ± 0.47, with 3.12 ± 0.30 in the dry season and 2.38 ± 0.22 in the wet season. For females, the average IK was 2.64 ± 0.61, with 3.28 ± 0.12 in the dry season, 2.29 ± 0.54 in the wet season, and 2.15 ± 0.13 in the post-wet season.

The Mood median test showed no statistically significant differences (α = 0.05; d.f = 1; p = 0.13) in the fitness index (IK) between dry-season and wet-season males.

The Kruskal-Wallis test for the seasonal effect on the IK of females was statistically significant (α = 0.05, d.f. = 2; p = 0.005). The Conover test (α = 0.05) showed statistically significant differences between the dry season versus the wet (p= 0.012) and post-wet (p = 0.0005) seasons. On the other hand, there were no significant differences in the IK between the wet and post-wet seasons (p = 0.57).

A reduction in plant cover associated with the crop was observed throughout the study, mainly due to trimming and plot cleaning. A 72.88 % cover was reported for the dry season, 16.12 % for the wet season, and 4.14 % for the post-wet season. Spearman's correlation coefficient for males (R= 0.86) and females (R = 0.74) showed a strong correlation between the fitness index (IK) and plant cover (Figure 2).

Figure 2 Correlation between plant cover and physical condition index (IK) by season and sex (males marked in black, females marked in blue). 

Diet of Geomys arenarius. In all seasons, the diet of the 40 desert pocket gophers consisted mainly of the ivyleaf groundcherry (Physalis hederifolia) with 29.26 % of the consumed items, followed by the spectacle pod (Dimorphocarpa wislizeni; 16.29%) and the common sandbur (Cenchrus incertus; 14.40 %; Table 2).

In the dry season, the diet consisted of 11 species, 11 genera, and 9 families. The common sandbur (Cenchrus incertus) was the most abundant food item (27.97 %), followed by the spectacle pod (Dimorphocarpa wislizeni; 21.14 %) and the ivyleaf groundcherry (Physalis hederifolia; 15.78 %; Table 2).

In the wet season, 11 species, 10 genera, and 7 families were recorded in the diet. The ivyleaf groundcherry (Physalis hederifolia) was the dominant food item (31.24 %), followed by the spectacle pod (Dimorphocarpa wislizeni; 17.88 %), and Hall's panicgrass (Panicum hallii; 15.29 %; Table 2).

Finally, six species, six genera, and five families were recorded in the post-wet season. The ivyleaf groundcherry (Physalis hederifolia) was the most abundant food item (58.05 %), followed by the plains bristle grass (Setaria macrostachya; 25.84 %), and the salt heliotrope (Heliotropium curassavicum; 13.20 %; Table 2).

Levin's Niche Breadth Index: Levin's index for G. arenarius was B = 5.27, with a standardized value of Bₐ = 0.37, indicating that the desert pocket gopher is a specialist herbivorous. Levin's values by season suggest that the gopher has a narrow trophic niche (Table 3).