Greater Prairie-Chicken (Tympanuchus cupido) Demographics in Fragmented Wisconsin Landscapes: Examining Limiting Vital Rates

Abstract

The North American Great Plains biome has experienced precipitous human-induced habitat losses exceeding 90% of its original distribution over the last century. These alterations are due to European expansion and concomitant land use practices. As a result, numerous grassland bird species have experienced continued declines throughout their endemic ranges. Not surprisingly, prairie-grouse have experienced range-wide declines and some extant populations persist in highly fragmented landscapes uncharacteristic of their evolutionary history. Such landscapes may limit populations by constraining vital rates (i.e., nest and brood survival). Furthermore, management practices may have unintended negative consequences. The Greater Prairie-Chicken (Tympanuchus cupido) population in Wisconsin persists in a highly fragmented landscape where management practices are under review (i.e., disturbance types and intervals); however, current demographic information regarding limiting vital rates for the Wisconsin Greater Prairie-Chicken (GRPC) population is lacking. Aside from apparent nest success estimates provided for the Buena Vista Grassland Management Area (Hamerstrom 1939, Hamerstrom and Hamerstrom 1973, Golner 1997, Toepfer 2006), minimal data on nest or juvenile survival is available. Thus, we sought to investigate the relative effects of environmental characteristics at local-, edge-, and patch-levels on GRPC nest and brood survival to inform future management of GRPC in Wisconsin. During the breeding season (March-May) of 2014 and 2015 we captured and radio-tagged 62 individual female GRPC using walk-in style traps. We subsequently located and monitored 74 GRPC nests of which 23 successfully hatched (i.e., ≥ 1 hatched egg). We used information theory (IT) with Akaike’s Information Criterion adjusted for small samples (AICc) and the nest survival option in Program MARK to analyze the relationship between nest daily survival rate (DSR) and visual obstruction of the nest, residual cover at the nest and nest area, distance to nearest edge, patch shape, patch size, and percent core area in grass cover. We considered models competitive if ΔAICc values were ≤ 2 AICc units of the top model. The most parsimonious model, S(residual), considered nest DSR a function of residual vegetation at the nest and surrounding area (Ŝ = 0.9763, SE = 0.003, [95% CI = 0.9685, 0.9822], Wi = 0.28) and was ~ 2.5 times more likely than the second competing model (S(.)). The mean residual cover at successful and unsuccessful nests was 31.9% and 44.1%, respectively. There was some model uncertainty because the null (S(.); Ŝ = 0.9762, SE = 0.003, [95% CI = 0.9684, 0.9820], Wi = 0.11) was competitive (ΔAICc = 1.83). The top two models estimate success of ~ 43 percent. We tracked females of successful nests ≥5 days/week (n = 23) via radio telemetry from hatch to 90 days of age and performed weekly flush counts of broods until 70 days of age, collar shedding, transmitter failure, hen mortality, or brood loss. We considered a brood lost if 2 consecutive flushes yielded zero counts. We used the Young Survival model in Program MARK which is based on the Lukacs model of young survival. This particular model accounts for imperfect detection during counts. Using this model we employed an IT approach with AICc to compare the importance of each a priori model. GRPC survival was best explained when detection (p) and survival (Phi) estimates varied across 3 and 4 intervals, respectively. Detection probability was considered different across 3 intervals: p1 (count 1), p2 (counts 2-7), and p3 (count 8-10). Conversely, survival (Phi) was best explained using 4 age intervals: Phi1 (0-14 days), Phi2 (15-21 days), Phi3 (22-28 days), and Phi4 (29-70 days). Juvenile survival estimates were best explained by additive effects of residual and forb cover estimates at flush locations (Phi1 = 0.96, SE = 0.02; Phi2 = 0.84, SE = 0.04; Phi3 = 0.66, SE = 0.08; Phi4 = 0.82, SE = 0.03). We estimate survival probability during the study at ~ 0.16 (95% CI = 0.11, 0.22). Contrary to other studies, juvenile survival at our sites declined from hatch until 28 days of age which suggests predation may be a limiting factor. Vegetation characteristics best explained juvenile survival in our study and may exacerbate or mediate mortality from abiotic (i.e., extreme weather events) or biotic (i.e., predation) events. We recommend spatially and temporally altering current management practices (e.g. prescribed fire and grazing) that best mimic historical disturbances and promote desirable local-level vegetation characteristics.