An improved understanding of ungulate population dynamics using count data: insights from western Montana

Understanding the dynamics of ungulate populations is critical given their
ecological and economic importance. In particular, the ability to evaluate
the evidence for potential drivers of variation in population trajectories
is important for informed management. However, the use of age ratio data
(e.g., juveniles:adult females) as an index of variation in population
dynamics is hindered by a lack of statistical power and difficult
interpretation. Here, we show that the use of a population model based on
count, classification and harvest data can dramatically improve the
understanding of ungulate population dynamics by: 1) providing estimates
of vital rates (e.g., per capita recruitment and population growth) that
are easier to interpret and more useful to managers than age ratios and 2)
increasing the power to assess potential sources of variation in key vital
rates. We used a time series of elk (Cervus canadensis) spring count and
classification data (2004 to 2016) and fall harvest data from hunting
districts in western Montana to construct a population model to estimate
vital rates and assess evidence for an association between a series of
environmental covariates and indices of predator abundance on per capita
recruitment rates of elk calves. Our results suggest that per capita
recruitment rates were negatively associated with cold and wet springs,
and severe winters, and positively associated with summer precipitation.
In contrast, an analysis of the raw age ratio data failed to detect these
relationships. Our approach based on a population model provided estimates
of the region-wide mean per capita recruitment rate (mean = 0.25,
 90% CI = 0.21, 0.29), temporal variation in
hunting-district-specific recruitment rates (minimum = 0.09; 90% CI =
[0.07, 0.11], maximum = 0.43; 90% CI = [0.38, 0.48]), and annual
population growth rates (minimum = 0.83; 90% CI = [0.78, 0.87], maximum =
1.20; 90% CI = [1.11, 1.29]). We recommend using routinely collected
population count and classification data and a population modeling
approach rather than interpreting estimated age ratios as a substantial
improvement in understanding population dynamics.