Several potential proximate causes may be implicated in a recent (post‐1984) decline in moose (Alces alces andersoni) numbers at their southern range periphery in northwest Minnesota, USA. These causes include deleterious effects of infectious pathogens, some of which are associated with white‐tailed deer (Odocoileus virginianus), negative effects of climate change, increased food competition with deer or moose, legal or illegal hunting, and increased predation by gray wolves (Canis lupus) and black bears (Ursus americanus). Long‐standing factors that may have contributed to the moose decline include those typically associated with marginal habitat such as nutritional deficiencies. We examined survival and productivity among radiocollared (n = 152) adult female and juvenile moose in northwest Minnesota during 1995–2000, and assessed cause of death and pathology through carcass necropsy of radiocollared and non‐radiocollared animals. Aerial moose surveys suggested that hunting was an unlikely source of the numerical decline because the level of harvest was relatively low (i.e., approx. 15%/2 yr) and the population usually grew in years following a hunt. The majority of moose mortalities (up to 87% of radiocollared moose [n = 76] and up to 65% of non‐radiocollared moose [n = 84]) were proximally related to pathology associated with parasites and infectious disease. Liver fluke (Fascioloides magna) infections apparently constituted the greatest single source of mortality and caused significant pathology in the liver, thoracic and peritoneal cavities, pericardial sac, and lungs. Mortality due to meningeal worm (Parelaphostrongylus tenuis) was less prevalent and was manifested through characteristic neurological disease. Several mortalities apparently were associated with unidentified infectious disease, probably acting in close association with malnutrition. Bone‐marrow fat was lower for moose dying of natural causes than those dying of anthropogenic factors or accidents, implying that acute malnutrition contributed to moose mortality. Blood profiles from live‐captured animals indicated that those dying in the subsequent 18 months were chronically malnourished. Relative to other populations, average annual survival rates for adult females (0.79 [0.74–0.84; 95% CI]) and yearlings (0.64 [0.48–0.86]) were low, whereas those for calves (0.66 [0.53–081]) were high. Pregnancy (48%) and twinning (19%) rates were among the lowest reported for moose, with reproductive senescence among females being apparent as early as 8 years. Pregnancy status was related to indices of acute (i.e., bone‐marrow fat) and chronic (i.e., blood condition indices) malnutrition. Opportunistic carcass recovery indicated that there likely were few prime‐aged males (.5 yr old) in the population. Analysis of protein content in moose browse and fecal samples indicated that food quality was probably adequate to support moose over winter, but the higher fecal protein among animals that died in the subsequent 18 months could be indicative of prote...

The earth is in the midst of a pronounced warming trend and temperatures in Minnesota, USA, as elsewhere, are projected to increase. Northern Minnesota represents the southern edge to the circumpolar distribution of moose (Alces alces), a species intolerant of heat. Moose increase their metabolic rate to regulate their core body temperature as temperatures rise. We hypothesized that moose survival rates would be a function of the frequency and magnitude that ambient temperatures exceeded the upper critical temperature of moose. We compared annual and seasonal moose survival in northeastern Minnesota between 2002 and 2008 with a temperature metric. We found that models based on January temperatures above the critical threshold were inversely correlated with subsequent survival and explained >78% of variability in spring, fall, and annual survival. Models based on late‐spring temperatures also explained a high proportion of survival during the subsequent fall. A model based on warm‐season temperatures was important in explaining survival during the subsequent winter. Our analyses suggest that temperatures may have a cumulative influence on survival. We expect that continuation or acceleration of current climate trends will result in decreased survival, a decrease in moose density, and ultimately, a retreat of moose northward from their current distribution.

North temperate species on the southern edge of their distribution are especially at risk to climate‐induced changes. One such species is the moose (Alces alces), whose continental United States distribution is restricted to northern states or northern portions of the Rocky Mountain cordillera. We used a series of matrix models to evaluate the demographic implications of estimated survival and reproduction schedules for a moose population in northeastern Minnesota, USA, between 2002 and 2008. We used data from a telemetry study to calculate adult survival rates and estimated calf survival and fertility of adult females by using results of helicopter surveys. Estimated age‐ and year‐specific survival rates showed a sinusoidal temporal pattern during our study and were lower for younger and old‐aged animals. Estimates of annual adult survival (when assumed to be constant for ages >1.7 yr old) ranged from 0.74 to 0.85. Annual calf survival averaged 0.40, and the annual ratio of calves born to radiocollared females averaged 0.78. Point estimates for the finite rate of increase (λ) from yearly matrices ranged from 0.67 to 0.98 during our 6‐year study, indicative of a long‐term declining population. Assuming each matrix to be equally likely to occur in the future, we estimated a long‐term stochastic growth rate of 0.85. Even if heat stress is not responsible for current levels of survival, continuation of this growth rate will ultimately result in a northward shift of the southern edge of moose distribution. Population growth rate, and its uncertainty, was most sensitive to changes in estimated adult survival rates. The relative importance of adult survival to population viability has important implications for harvest of large herbivores and the collection of information on wildlife fertility.

Sightability models are binary logistic-regression models used to estimate and adjust for visibility bias in wildlife-population surveys. Like many models in wildlife and ecology, sightability models are typically developed from small observational datasets with many candidate predictors. Aggressive modelselection methods are often employed to choose a best model for prediction and effect estimation, despite evidence that such methods can lead to overfitting (i.e., selected models may describe random error or noise rather than true predictor-response curves) and poor predictive ability. We used moose (Alces alces) sightability data from northeastern Minnesota (2005Minnesota ( -2007 as a case study to illustrate an alternative approach, which we refer to as degrees-of-freedom (df) spending: sample-size guidelines are used to determine an acceptable level of model complexity and then a pre-specified model is fit to the data and used for inference. For comparison, we also constructed sightability models using Akaike's Information Criterion (AIC) step-down procedures and model averaging (based on a small set of models developed using df-spending guidelines). We used bootstrap procedures to mimic the process of model fitting and prediction, and to compute an index of overfitting, expected predictive accuracy, and model-selection uncertainty. The index of overfitting increased 13% when the number of candidate predictors was increased from three to eight and a best model was selected using step-down procedures. Likewise, model-selection uncertainty increased when the number of candidate predictors increased. Model averaging (based on R ¼ 30 models with 1-3 predictors) effectively shrunk regression coefficients toward zero and produced similar estimates of precision to our 3-df pre-specified model. As such, model averaging may help to guard against overfitting when too many predictors are considered (relative to available sample size). The set of candidate models will influence the extent to which coefficients are shrunk toward zero, which has implications for how one might apply model averaging to problems traditionally approached using variable-selection methods. We often recommend the df-spending approach in our consulting work because it is easy to implement and it naturally forces investigators to think carefully about their models and predictors. Nonetheless, similar concepts should apply whether one is fitting 1 model or using multi-model inference. For example, model-building decisions should consider the effective sample size, and potential predictors should be screened (without looking at their relationship to the response) for missing data, narrow distributions, collinearity, potentially overly influential observations, and measurement errors (e.g., via logical error checks). ß

No abstract