Biological and Statistical Errors Make Inferences Circumspect: Response to Bender and Weisenberger

ABSTRACT Bender and Weisenberger (2005) reported that desert bighorn sheep (Ovis canadensis) on San Andres National Wildlife Refuge (SANWR), New Mexico, USA, were primarily limited by rainfall. However, they failed to mention, or were unaware, that persistent long-term predator control was used to enhance population growth at SANWR. Additionally, lamb:female ratios were collected throughout the year, rather than dates typically associated with assessing recruitment, and therefore influence of precipitation on lamb recruitment was unknown. Finally, model predictions forwarded by Bender and Weisenberger (2005), that carrying capacity of SANWR is zero when annual rainfall is <28.2 cm, were not supported by data, nor were their model results properly interpreted. The coefficient of determination value of 88.9% for the relationship between population size and current year's precipitation was primarily a function of serial correlation between successive years in population data, with current year's precipitation accounting for only 3.8% of this value. This suggests that precipitation was a weak predictor of population increase. These errors in concert make biological inferences reported in Bender and Weisenberger (2005) of limited value.