Population dynamics of Townsend's ground squirrels in southwestern Idaho

Abstract

Townsend's ground squirrels (Spermophilus townsendii, or TGS) are a conspicuous component of the desert shrub-steppe ecosystem of southwestern Idaho. In 1991, a large-scale research project was initiated to assess the impacts of habitat alterations on TGS populations located within the Snake River Birds of Prey National Conservation Area (SRBPNCA), where TGS are the primary prey of the raptors for which the area is named. During 4 years (1991-1994), > 5,000 TGS were live-trapped on 20 study sites to answer questions about whether TGS density, survival and productivity were affected by either the recent large-scale conversion of native shrub habitat to grassland by wildfires, or the use of a portion of the SRBPNCA by the Idaho Army National Guard for military tank training. Capture-recapture methods were used to estimate abundance and survival of TGS. Abundance was estimated using the no recruitment special case of the Jolly-Seber model, with the additional feature of modeling of capture and survival probabilities. These methods have never been combined in a study of this magnitude. I was able to estimate abundance in cases with small sample sizes, where the Jolly-Seber model is generally ineffective, by pooling data from more than one study site and estimating joint capture and survival probabilities. Ground squirrel densities were highest in 1992, immediately before a population crash caused by a severe late spring drought in that year. Declines in density were greater among populations in grassland habitats than in shrub habitats. Very few juveniles born in 1992 survived to become adults. Following the crash, populations remained low, but made modest recoveries in 1994 on grassland sites. Per-day survival rates were estimated for adults in 1991-1993 during 3 seasons of the year, using Cormack-Jolly-Seber methods and parameter modeling techniques. Random effects model analyses found differences in survival rates attributable to study year in all 3 seasons. These differences were assumed to reflect varying weather conditions occurring within those years. The effects of these conditions differed by habitat during the early part of the ground squirrel active season, and by sex during the late active season. Variability in survival rates during the inactive season was due to a complex interaction among all factors: year, sex, and habitat Although lowered survival rates during and immediately following the 1992 drought were expected, survival rates remained low much later in 1993 despite apparently good environmental conditions. This was attributed to carry-over effects from the drought but the exact mechanisms for those effects are unknown. Dispersal patterns of juvenile ground squirrels were measured in a radio-telemetry study conducted in 1993-1994. Nearly 60 juveniles (38 males, 20 females) were radio-collared over the 2 years on 4 study sites selected to represent 2 habitat types: shrub and grass. Male-predominant dispersal rates, typical for Spermophilus spp., were confirmed, as nearly 60% of males dispersed whereas only 1 female did. No differences between habitats was noted in the proportion of juveniles dispersing, the fates of dispersers, nor the patterns of dispersers; low sample sizes may have been responsible for the inability to detect any such differences. Dispersal distances (which averaged 515 m) fit a truncated exponential distribution, and the direction of dispersal did not differ from random. A spatially explicit population model for TGS on the SRBPNCA was constructed to integrate the population parameters estimated in other portions of this research and to investigate potential impacts of habitat patterns on population dynamics of Townsend's ground squirrels. This population model is intended to be included in a more comprehensive model of habitat, prey, and raptors on the SRBPNCA. In the TGS population model, the model area was divided into 2.2S ha cells, within which population process of reproduction and survival take place, and between which dispersal occurs. Population parameters vary by environmental conditions (assessed 2 times during each model year) and habitat. Population parameter values were estimated from data collected during the TGS demography study. Model simulations of the TGS population model were conducted on a 10,000 cell test area of the SRBPNCA. In model simulations under random environmental conditions, TGS populations experienced population growth, and growth was higher in grass habitats than in shrub habitats. Population densities in grass habitats reached unrealistic levels (i.e., the mean of all cell populations was >50 animals/ha) at least once during model simulation in most cases, yet they remained at more modest levels in shrub habitats. In addition, populations tended to become increasingly aggregated, with highest densities concentrated in large areas of grass habitat, and lowest densities (and sometimes local extinction) in large areas of shrub habitat. If model assumptions are realistic, these patterns may indicate that intermingling of the 2 habitats is required for TGS populations to remain stable. However, an important model assumption is that there are no density dependent regulation mechanisms operating: this assumption is likely unrealistic. Therefore, future research should be concentrated on determining the means by which TGS populations are regulated, as well as on studying more closely the response of TGS population parameters to environmental conditions.