HETEROGENEITY AND POWER TO DETECT TRENDS IN MOOSE BROWSE UTILIZATION OF WILLOW COMMUNITIES

Abstract

Monitoring of browse utilization of plant communities is consistently recommended as an important component of monitoring moose (Alces alces) populations across regions. We monitored winter browse utilization by moose within a willow (Salix spp.) -dominated winter range of Montana in 2008–2010. We sought to improve our understanding of: 1) spatiotemporal heterogeneity of intensity of moose browsing across the winter range, 2) species-specific selection of willow by moose during winter, and 3) appropriate sample sizes, placement, and stratification of monitoring sites for estimating browse utilization. During 3 consecutive winters we monitored 108–111 transect segments, each 50 m in length, in a systematic distribution across willow communities and assessed the effects of covariates potentially predictive of variation in browsing. Mean annual estimated browse utilization across all segments was 11.5% of sampled twigs in 2008 (95% CI = 9.4 – 13.7%), 8.0% in 2009 (95% CI = 6.2 – 9.8%), and 8.3% in 2010 (95% CI = 6.5 – 10.1%). Modeling of variation in browse utilization revealed positive relationships with the proportion of preferred species (β = 0.44,P = 0.05) and previously browsed willow plants (β = 3.13, P < 0.001), and a negative relationship with willow patch width (β = 0.002, P < 0.001). We found that planeleaf (Salix planifolia), Wolf ʼs (S. wolfii), and Boothʼs willow (S. boothii) were the most consistently preferred species, whereas Drummondʼs (S. drummondiana) and Geyer willow (S. geyeriana) willow were moderately preferred; Lemmonʼs willow (S. lemmonii) was used less than expected. Power analyses indicated that detecting a 10% increase in browse utilization with 95% confidence in consecutive years required measuring 38–41, 50-m segments. Because systems with low and heterogeneous browse utilization of willow present challenges for efficient monitoring, we encourage power analyses as a means of evaluating sampling protocols, in addition to consideration of covariates predictive of spatiotemporal heterogeneity.