Population Demographics and Dynamics of Colonizing Elk in a Desert Grassland–Scrubland Free

Desert grassland and desert-scrub (hereafter, grassland–scrubland) habitats are anecdotally considered marginal habitat for elk Cervus elaphus (Skovlin et al. 2002). Presumed detriments of arid habitats include scarcity of water, low forage quantity and quality, lack of vegetative cover, and thermal stress; this potentially results in nutritive deficiencies, dehydration, high thermoregulatory costs, and increased risk of predation and harvest (Young 1988; Jiang 1993; Sowell 2001; Skovlin et al. 2002). These limitations would result in lower individual condition and consequently reduced demographic vigor and population performance (Hanks 1981; Gaillard et al. 2000; Piasecke 2006; Bender et al. 2008). Despite this, elk have colonized or been introduced into arid grassland or scrubland habitats in several areas, and many of these populations subsequently grew at rates comparable to elk populations in montane habitats (McCorquodale et al. 1988; Carpenter and Silvy 1991; Strohmeyer and Peek 1996).

Around January 2000, about 20 elk colonized the Chaco Culture National Historical Park (CCNHP) area, an arid grassland–scrubland in northwestern New Mexico. These elk have subsequently established a productive population in and around CCNHP, increasing to ≥ 53 individuals by 2006 (see below). Because Southwestern desert grassland–scrublands have generally been considered unsuitable for elk, no current information was available on actual suitability of these habitats for elk or the potential for elk populations to increase in these habitats. Consequently, our goal was to determine population demographics and growth potential of elk in the Southwestern desert habitats in and around CCNHP. Specifically, our objectives were to: 1) determine nutritional condition, 2) determine population size, composition, and trend, 3) determine survival and productivity and the environmental factors influencing them, and 4) model population growth for elk in CCNHP.

Our study area covered about 308 km² on and adjacent to CCNHP in northwestern New Mexico, centered at about 36°00′N, 108°00′W (Figure 1). Elevations ranged from 1,670 to 2,079 m, and topography was characterized by rolling plains and mesas interspersed with steep ravines. Average high temperature in July was 32°C and average low temperature in January and December was −11°C. Average annual precipitation was 23 cm, with 52% falling July–October, and average annual snowfall was 37 cm.

The dominant plant community on mesas consisted of fourwing saltbush Atriplex canescens, rubber rabbitbrush Chrysothamnus nauseosus, mountain mahogany Cercocarpus intricatus, winterfat Krascheninnikovia lanata, galleta Pleuraphis jamesii, blue grama Bouteloua gracilis, Indian ricegrass Achnatherum hymenoides, and bigelow sage Artemisia bigelovii, with scattered pinyon pine Pinus edulis and oneseed juniper Juniperus monosperma. Characteristic vegetation of ravines descending from mesas included mound saltbush Atriplex obovata, galleta, blue grama, alkali sacaton Sporobolus airoides, bigelow sage, winterfat, Mormon tea Ephedra spp., and oneseed juniper. Vegetation in large washes included fourwing saltbush, galleta, giant dropseed S. giganteus, and sand dropseed S. cryptandrus. Only about 15% of the type of ecosystem represented by CCNHP remains intact in North America, with most perturbation due to grazing (Ricketts et al. 1999).

Ungulates other than elk and mule deer Odocoileus hemionus were rare in CCNHP. Potential predators of elk present on CCNHP include coyotes Canis latrans, pumas Puma concolor, and bobcats Lynx rufus. Domestic livestock including cattle and horses were common on adjacent Tribal, State of New Mexico, and Bureau of Land Management holdings.

We captured (and recaptured previously radiocollared) cow elk by darting from a Bell 206B Jet Ranger helicopter using carfentanil citrate (3.0 mg/elk) and xylazine hydrochloride (100 mg/elk) each April and November, 2003–2006. We similarly captured a limited number of bulls to aid in population estimation (see below). We captured calves by hand if found less than about 5 d after birth, 2003–2006. We captured calves > 5 d old by net-gunning or aerial darting from a Bell 206B Jet Ranger helicopter, 2004–2006. We blindfolded all captured elk to minimize stress during handling and administered penicillin, vitamin B, vitamin E–selenium (MUSE), and an eight-way Clostridium bacterin to alleviate capture stress. We fit elk with radiocollars (adults, 1.5 y and older) or ear-tag transmitters (calves; Advanced Telemetry Solutions, Asanti, MN).

We collected blood by venal puncture, checked lactation status, and measured chest girth at each capture. We tested serum for pregnancy-specific placental protein B to determine pregnancy status (Piasecke 2006). We checked lactation status of each cow to assess calf survival up to the point of capture (November). Lactation indicated successful nursing by a calf within 3–11 d (Bender et al. 2002). We used spring scales or girth to estimate body mass (kg; Cook et al. 2003), where calves  =  0.0003901 × G^2.6430078; yearlings  =  2.038 × G − 97.533; and adults  =  2.549 × G − 156.760, where G  =  chest girth (cm).

We estimated a rump body condition score (rBCS) by palpation of the soft tissue of the rump near the base of the tail and a withers body condition score (wBCS) by palpation of the top of the withers just posterior to the shoulder hump (Cook 2000). We scored results of rBCS and wBCS measures from standards that ranged from one (emaciated) to five (obese) in intervals of 0.25 (Cook 2000). We used a SonoVet 2000 ultrasound (Medison, Seoul, South Korea) with a 5-mHz probe to measure subcutaneous fat thickness along a straight line midway between the spine, at its closest point to the coxal tuber (hip bone), and the ischial tuber (pin bone; MAXFAT; Cook 2000). We estimated percent body fat (BF) by combining rBCS and MAXFAT into an index (rLIVINDEX) using: rLIVINDEX  =  rBCS when MAXFAT < 0.3 cm; and rLIVINDEX  =  (MAXFAT − 0.3) + rBCS when MAXFAT ≥ 0.3 cm (Cook et al. 2001). We then calculated BF from rLIVINDEX using BF  =  −7.1527185 + 7.323081 × L − 0.98980456 × L² + 0.057445567 × L³, where L  =  rLIVINDEX (Cook et al. 2001). If MAXFAT was not measured, we calculated BF directly from rBCS using BF  =  4.478 × rBCS − 4.618. We also used an ultrasound to measure the thickness of the loin (longissimus dorsi) muscle between the 12th and 13th ribs near the spine to index lean tissue (muscle) catabolism (Cook et al. 2001).

We visually located all radiocollared individuals a minimum of once per week. When located, we recorded survival status (live or dead) of each elk and determined causes of death for all mortalities following Bender et al. (2004). We considered deaths that occurred within 30 d of capture to be capture-related and censored these from survival analyses (Berringer et al. 1996).

We calculated annual survival rates of cows and calves (for pooled radiotagged calves and calves not radiotagged but whose dam was radiocollared [i.e., calves-at-heel]) using monthly intervals with the staggered-entry Kaplan–Meier estimator and used Z-tests to compare survival among years (Pollock et al. 1989). We also used pregnancy status of individual cows and their lactation status the following autumn to determine calf survival (Bender et al. 2002; i.e., Ŝ_calf  =  Lactating_y / Pregnant_y−1). This estimate of calf survival incorporates calves that may have died shortly after parturition but before we were able to either capture them or observe them with their dam, and thus would not have been included in the radiotagged or calf-at-heel samples. However, this estimate is potentially confounded by fetuses that are lost in utero. Because observed in utero losses are < 3% for elk (Piasecke 2006), this potential positive bias is low.

We used logistic regression (Hosmer and Lemeshow 1989) to model the dichotomous outcome of calf survival (i.e., live or die) as a function of its mother's age, girth, mass, MAXFAT, rBCS, wBCS, loin, and BF the previous autumn and late winter, which represent the approximate seasonal peak (autumn) and low (late winter) in condition, respectively. We also modeled calf survival as a function of precipitation during several biologically relevant periods, including: winter prior to birth (December–March); late gestation period (April–June); lactation period (June–October); annually based on calendar (January–December) and biological (June–May) years; and cumulative annual precipitation through parturition (i.e., January–June) and lactation (January–September). We used odds ratios to interpret results of logistic analyses.

We determined proportions pregnant and proportions lactating by age class (yearlings, ≥ 2.5-y-olds) by dividing the number of pregnant cows by the total numbers tested, determined standard errors of estimates using the normal approximation (Zar 1996), and used Fisher's exact tests (Zar 1996) to compare proportions among years. We used logistic regression (Hosmer and Lemeshow 1989) to model variables related to the probability of an individual elk being pregnant. Our dichotomous outcome variable was pregnant or not pregnant and our predictors were age, mass, girth, MAXFAT, rBCS, wBCS, loin, and BF, with all variables collected from cows during the autumn when pregnancy status was assessed.

We estimated minimum population size from herd composition surveys during November 2004–2006 capture events. Surveys covered the entire study area and were flown two to three times during each capture event. We recorded numbers of bulls, cows, and calves during surveys, and determined variance around ratios (bulls/100 cows, calves/100 cows) following Czaplewski et al. (1983). We compared ratios among years using parametric bootstrapping (Bender et al. 1996).

We estimated the minimum number of cows in the population by summing the numbers of radiocollared cows and the highest number of uncollared cows seen during surveys. Because each individual survey was accomplished in < 1 d, two or three complete surveys occurred during each capture event, groups of elk were usually separated by > 5 km, uncollared elk were generally associated with collared elk, and all collared elk were observed multiple times, we were confident that there was little or no intermixing during surveys and, hence, little potential for duplication of counts. In addition, we also tallied the number of uncollared cows seen throughout the summer to corroborate their minimum numbers. Minimum calf and bull numbers were determined identically. We also determined calving success for each radiocollared cow to corroborate minimum calf numbers. Last, we correlated minimum population estimates of all individuals and only cows and calves with pregnancy rate, lactation rates, calf survival rates, and calf ∶ cow ratios to detect any trends in population demographics as population size or density increased.

We estimated maximum potential finite rate of population increase (λ) annually as λ  =  Ŝ_F + 0.5 × (C/C), where Ŝ_F  =  annual survival rate of adult cows and C/C  =  observed calf ∶ cow ratio (White and Bartmann 1997). This was a maximum estimate for λ, because our observed calf ∶ cow ratios (collected in November) did not include calf mortality from December through the following September, when calves would be recruited into the adult population. We also estimated mean annual λ since colonization using λ  =  (N_y / N_initial)^1/y, where y  =  number of years since colonization.

We captured and processed 11 elk (4 adult cows, 7 adult bulls), 14 adult cows (7 new captures, 7 recaptures), 21 adult cows (3 new captures, 18 recaptures), and 23 adult cows (all recaptures), April 2003–2006, respectively. In addition, we captured and processed 10 (8 new captures, 2 recaptures), 18 (4 new captures, 14 recaptures), 20 (two new captures, 18 recaptures), and 24 adult cows (all recaptures), late November–early December, 2003–2006. Last, we captured and processed seven calves each summer, 2004–2005, and an additional four calves in December 2005.

Body fat varied by year (F_3,128  =  29.7; P < 0.001), season (late winter v. late autumn [F_1,128  =  20.1; P < 0.001]), and lactation status (F_1,128  =  13.7; P < 0.001 [Table 1]). Body fat levels were higher in November than April regardless of lactation status (P ≤ 0.009). Among years, BF of lactating cows was lower (P < 0.001) in 2005 and 2006 than in 2003 and 2004 (Table 1). Similarly, dry cows were in better condition in 2003 and 2004 (P ≤ 0.001). Condition levels did not vary by year at the low of condition in early April except for April 2003 (P ≤ 0.052), when sample sizes were small (4 vs. 14–22; Table 1).

Mean body mass of ≥ 2.5-y-old cows (H₃  =  0.2; P  =  0.983), yearling cows (H₃  =  2.7; P  =  0.436), or calves (U  =  −0.7; P  =  0.219) did not differ among years (Table 2). Because of small sample sizes in 2003 and 2004, we pooled yearling cows from 2003 and 2004, when condition of elk was significantly higher than in 2005 and 2006, and compared these years against 2005 and 2006, but yearling mass still did not differ (H₂  =  2.3; P  =  0.311).

We documented eight mortalities during this project: one female from harvesting (2004); one female from illegal harvest (2006); three females from unknown causes for which we could not exclude capture as the ultimate cause of mortality; and three that were capture-related. In addition, two calves (one each in 2004 and 2005) and one adult female lost their collars and one adult female was never redetected using telemetry. The latter 10 cases were censored from survival analyses. Survival of adult females for the biological year (June–May) was similar among years (Z < 1.0; P ≥ 0.318) and ranged from 0.94 (SE  =  0.06) to 1.00 (SE  =  0.00; Table 3).

None of the individually identifiable calves (those radiocollared or associated with a collared cow) died in 2004–2005, 2005–2006, or 2006–2007, resulting in a maximum overall calf survival rate of 1.00 (SE  =  0.00) for each year. Using pregnancy and lactation status from individual adult cows, estimated preweaning survival was lowest (Z ≥ 3.4; P ≤ 0.0006) in 2006–2007 (0.38 vs. 0.80 and 1.00 for 2004–2005 and 2005–2006, respectively) except for 2003–2004 (0.50), which was similar (Z  =  0.4; P  =  0.719; Table 3). All radiocollared calves or calves observed at heel that were alive at weaning were verified by lactation in the mother, and all subsequently survived until the following June, resulting in postweaning calf survival of 1.00 (SE  =  0) for each year. Thus, annual survival rates were identical to preweaning (Table 3).

Calf survival was positively related to maternal MAXFAT, girth, mass, rBCS, and BF the previous autumn (Table 4). Maximum subcutaneous fat thickness was the best individual predictor of maternal attributes on calf survival and correctly classified the fate of 71.6% of calves. Odds ratios (3.48; 95% CI  =  1.37–8.86) indicated that probability of survival for a calf increased about 3.5 times for each 1-cm increase in MAXFAT of their mother. Among precipitation periods, calf survival was related (χ²  =  6.9, P  =  0.008; β  =  0.402 [SE  =  0.153]) only to cumulative annual precipitation through parturition (i.e., January–June). Odds ratios (1.50; 95% CI  =  1.11–2.02]) indicated that probability of a calf surviving increased about 1.5 times for each 1-cm increase in cumulative precipitation through June. When MAXFAT and cumulative precipitation through parturition were modeled together, MAXFAT was dropped from the final model (χ²  =  0.1; P  =  0.712) while cumulative precipitation was retained (χ²  =  6.5; P  =  0.011) indicating that cumulative precipitation through parturition had the dominant effect on calf survival (Figure 2).

For calves whose mothers were assessed for size and condition in late winter, calf survival was positively related to maternal age, girth, mass, and depth of the loin muscle in late winter (early April; Table 4). The strongest relationship of maternal attributes during late winter on calf survival was maternal body mass, which correctly classified the fate of 74.4% of calves. Odds ratios (1.08; 95% CI  =  1.01–1.16) indicated that probability of survival for a calf increased about 1.1 times for each 1-kg increase in maternal body mass. For this subset of calves, survival was also only related (χ²  =  9.1; P  =  0.003; β  =  0.354 [SE  =  0.118]) to cumulative annual precipitation through parturition. Odds ratios (1.42; 95% CI  =  1.13–1.79) indicated that probability of a calf surviving increased about 1.4 times for each 1-cm increase in cumulative precipitation through June. When maternal mass and cumulative precipitation through parturition were modeled together, both maternal mass (χ²  =  3.7; P  =  0.054) and cumulative precipitation (χ²  =  5.9; P  =  0.015) were retained in the final model, but odds ratios (1.078; 95% CI  =  0.999–1.163) that included one suggested that the effect of maternal mass was weaker than the effect of cumulative precipitation (odds ratio  =  1.42; 95% CI  =  1.07–1.90).

Pregnancy rates ranged from 0.00 to 0.50 for yearling and 0.69 to 1.00 for ≥ 2.5-y-old cows, respectively, and were similar within age classes (Fisher's exact P  =  1.000 and 0.251 for yearlings and ≥ 2.5-y-old cows, respectively) among years (Table 5). Lactating cows (including yearlings) comprised 22–56% of cows and proportions were similar (Fisher's exact P  =  0.109) among years (Table 5). For ≥ 2.5-y-olds only, lactation rates were 0.29, 0.63, 0.67, and 0.30 for 2003–2006, respectively (Table 5).

Logistic models including age, girth, and rBCS were positively related to the probability of a ≥ 2.5-y-old cow conceiving (Table 6). Probability of pregnancy of ≥ 2.5-y-old cows was best predicted by body mass, which correctly classified the pregnancy status of 70.0% of the elk. Odds ratios indicated that probability of pregnancy increased about 1.06 times (95% CI  =  1.008–1.111) for each 1 kg increase in body mass. Pregnancy in yearling cows was unrelated to any measure of size or condition.

We classified and documented a minimum of 43, 51, and 53 elk during autumn composition surveys, 2004–2006, respectively. Age and sex composition (SE) of elk was 60 (4) bulls and 55 (4) calves, 50 (4) bulls and 44 (4) calves, and 59 (3) bulls and 24 (2) calves/100 cows, 2004–2006, respectively. Bull ratios were similar (P ≥ 0.035 at α_exp  =  0.10) but calf ratios varied (P ≤ 0.030) among years (Table 5). The maximum estimate of λ was 1.22 ( =  0.94 + 0.5 × 0.55), 1.22 ( =  1.00 + 0.5 × 0.44), and 1.08 ( =  0.96 + 0.5 × 0.24) for 2004–2006, respectively. If we substituted observed lactation rates for calf ∶ cow ratios, estimates of λ were 1.11, 1.22, 1.25, and 1.09 for 2003–2006, respectively.

Minimum population size in 2006 was ≥ 53 elk for an approximate elk density of 0.15 elk/km² (Table 7). Similarly derived estimates for 2004 and 2005 were ≥ 43 (0.13 elk/km²) and ≥ 51 (0.15 elk/km²), respectively. Assuming that the initial January 2000 population estimate of 20 elk was correct, the population has grown at a mean rate of 15%/y (λ  =  1.15) from colonization in 2000 through December 2006. Realized rate of population increase based on minimum population estimates declined from λ  =  1.18 (2000–2005) to λ  =  1.04 (2005–2006).

Neither minimum population size (Z ≤ 1.4; P ≥ 0.318) or number of cows and calves (Z ≤ 1.4; P ≥ 0.303) were related to cow ∶ calf ratios, ≥ 2.5-y-old cow pregnancy rates, yearling pregnancy rate, adult survival, calf survival, proportions of adult cows lactating, or proportions of all cows lactating.

Body fat level of lactating cows indicated that cows in CCNHP were able to acquire diets ranging from poor (2005 and 2006) to high marginal–low good (2003 and 2004) nutrition, based on thresholds from nutritional studies of penned elk (Cook et al. 2004). For free-ranging elk, nutrition effectively ceases to be limiting when lactating cows are able to accrue about 13.7% BF (Piasecke and Bender 2009). Lactating cows in CCNHP never achieved > 13.2% BF, indicating that habitat quality was always limiting to some degree. This was further evidenced by lactating cows being unable to accrue condition levels comparable to dry cows (Table 1). However, strong effects on individual and population performance are unlikely to be seen in elk populations until BF levels of lactating cows drop below 7.9% (Piasecke and Bender 2009). Lactating cows were below this threshold only in 2005 and 2006 (Table 1), and this eventually resulted in decreased calf survival (0.38 vs. 0.77 for 2003–2005) and consequently lowered (λ  =  1.04 vs. 1.18 for 2000–2004) population rates-of-increase.

In CCNHP, accrual of BF in cows was most strongly positively related to cumulative precipitation from September to November (i.e., from the end of the primary lactation to forage senescence; Bender 2007). Declines in nutritional condition were, thus, a result of density-independent effects (i.e., low precipitation), highlighting the sensitivity of elk in our arid study area to pronounced annual variation in habitat quality related to spatially and temporally patchy precipitation characteristic of arid environments (Sowell 2001; Bender 2007).

Production and survival of calves is the most variable and influential component of elk and other ungulate population dynamics (Clutton-Brock et al. 1982; Gaillard et al. 2000; Eberhardt 2002). Past research has shown that conception, calf survival, and recruitment were dependent on condition of adult cows (Clutton-Brock et al. 1982; Cook et al. 2004), results that we also observed in CCNHP. Population density also can affect population productivity because of competition for resources, and productivity can also be limited by density-independent factors such as poor forage quality and precipitation (Clutton-Brock et al. 1987; Smith and Anderson 1998; Piasecke and Bender 2009). We found that productivity in CCNHP was affected by each of these factors except for density dependence, because neither maternal condition nor population demographics were related to density.

Body mass most strongly affected pregnancy in ≥ 2.5-y-old cows in CCNHP, similar to results seen in other free-ranging elk populations (Piasecke 2006). In contrast, pregnancy in yearlings was unrelated to size or condition, perhaps because small sample sizes (n  =  13) precluded a sensitive analysis of factors affecting yearling pregnancy. Both absolute (Grier 1968; Cook et al. 2004) and relative (Haigh and Hudson 1993) body mass have been implicated as important variables for yearling pregnancy in elk. Yearling cows in CCNHP were able to achieve about 80–85% of the mean ≥ 2.5-y-old cow body mass (Table 2), a greater proportion than reproductive thresholds delineated for penned (70%; Haigh and Hudson 1993) and free-ranging elk (50%; Piasecke 2006), but pregnancy rates were still low (x¯  =  0.33). In terms of absolute mass, yearling pregnancy exceeded 25% only when body mass exceeded 163–169 kg in Yellowstone National Park (Greer 1968) and about 170 kg for multiple free-ranging elk populations (Piasecke 2006). Yearling mass in CCNHP ranged from 170 to 180 kg (Table 2); thus, pregnancy in yearling cows may have been related to a threshold in absolute body mass, rather than relative mass.

Calf survival ranged from 0.38 to 1.00 in CCNHP, and the calf survival rate of 0.38, calf ∶ cow ratio of 24∶100, and lactation rate of 0.25 in 2006 all reflected the decreased and low productivity of elk following the decline in condition first seen in 2005. From 2003 to 2005, productivity in CCNHP was comparable to the highest productivity rates seen (51–56/100 for elk in Michigan; Bender et al. 2002) and with the fastest growing elk herd documented (λ  =  1.22; Eberhardt et al. 1996; Bender et al. 2002). Similarly, the highest documented elk calf survival through summer was 0.95 for a population in Michigan (Bender et al. 2002), and elk in CCNHP approached that level (0.80–1.0) in 2 of 4 y. This high productivity was reflected in the mean historic growth rate (λ  =  1.18) seen prior to 2006, despite inclusion of the lower growth potential seen in 2003 (λ  =  1.11). In contrast, habitat limitations associated with unfavorable precipitation patterns in 2005 and 2006 resulted in lower (λ  =  1.04) growth rates in 2006.

Maternal condition positively affected calf survival in CCNHP, similar to responses found elsewhere (Wegge 1975; Thorne et al. 1976; Guinness et al. 1978; Clutton-Brock et al. 1982). However, most of the causes of preweaning mortality these studies found were classed as density independent (i.e., drowning, accidents, predation), suggesting that chance also played a strong role in preweaning survival in addition to maternal influences. We also saw this in CCNHP, where cumulative annual precipitation through gestation, a density-independent effect, was also positively related to calf survival. Additionally, all calf mortality occurred at or near parturition, a period primarily associated with density-independent mortality (Harper et al. 1967; Arman et al. 1978; Clutton-Brock et al. 1987; Gogan and Barrett 1987; Smith and Anderson 1998) and when calves typically experience their greatest mortality (Guinness et al. 1978; Clutton-Brock et al. 1988). Because pregnancy was also related to cow condition, most aspects of productivity in CCNHP (i.e., conception, preweaning calf survival) were, thus, influenced to some degree by both maternal condition and density-independent effects (i.e., precipitation). The only exception was postweaning calf survival, which was uniformly high (Table 3).

The importance of cumulative precipitation through gestation in CCNHP was likely due to effects on spring green-up and, hence, quantity and quality of forage. Successfully raising a calf came at a strong cost to cows; lactating cows were unable to accrue condition at the rate of dry cows, and consequently entered winter in poorer condition (Table 1). Because condition is largely a result of nutrition (Cook et al. 2004), and condition also affected calf survival, the influence of precipitation, a density-independent effect, was likely at least partially mediated through maternal condition, which is most generally associated with density dependence. Productivity of elk can, thus, be influenced by factors other than resource competition (i.e., density dependence), highlighting the importance of density-independent effects on elk population dynamics.

Adult female survival (0.94–1.00) in CCNHP was comparable to the highest rates seen for free-ranging elk (Bender et al. 2008) even in 2005 and 2006, when elk condition was extremely low. This indicates that even given conditions such as elk faced in 2005–2006, habitat conditions were still suitable to allow high adult female survival, the last demographic affected by resource stress (Gaillard et al. 2000). Elk are protected from harvest while on CCNHP, but off the Park they are subject to regulated tribal harvest during the autumn and early winter. Thus, although the population is lightly hunted (the only non–capture-related mortalities for 2003–2006 we observed were one cow harvested in 2004 and one cow illegally killed in 2006), adult survival rate was high even compared to unhunted populations and was at or near the expected maximum for adult cows (x ¯  =  0.98, 2003–2006 [Eberhardt 2002; Bender et al. 2008]).

Because of the low number of mortalities during this study, we were unable to model survival and, thus, identify whether condition or other factors influenced adult cow vulnerability. Given the extremely high survival seen, however, it is unlikely that any factor was significantly affecting adult cow survival in CCNHP. However, the relative lack of disturbance associated with humans may also minimize energy losses of individual elk, which may be important for elk to persist in arid habitats during unfavorable years (Strohmeyer and Peek 1996).

Condition and other demographic data indicate that arid Southwestern grassland–scrublands are suitable habitat for elk, at least at low population densities. Because elk can survive and show high productivity in these habitats, it opens the option of establishing and managing for elk in these habitat types, heretofore generally considered unsuitable. However, annual changes in habitat conditions driven by highly variable precipitation patterns can be very pronounced in these habitats. This necessitates the importance of annual monitoring of condition of elk, because monitoring of other parameters (pregnancy rates, lactation rates, calf ∶ cow ratios) would not have demonstrated the lessening of habitat capability to support elk seen in 2005 until the following year. Similarly, potential impacts of elk on other large wild herbivores are unknown, and elk may adversely affect habitat at high population densities (Bender 2007). Monitoring of trends of other herbivores and range condition is warranted if elk populations continue to increase in arid grassland–scrubland where community-level grazing capacity is limited.

We thank the U.S. Geological Survey, U.S. National Park Service (Chaco Culture National Historical Park), and the New Mexico State University, Agriculture Experiment Station for funding this project.

We thank H. Halbritter, T. Kamienski, L. Lomas, J. Ramaka, B. Shattuck, and E. Watters for field and logistical assistance, and two anonymous reviewers and the journal editors for their efforts to improve this manuscript. All activities were in accordance with NMSU IACUC Permit NM-2002-027.