Portal de Periódicos da CAPES

Demography in an increasingly variable world

2005; Elsevier BV; Volume: 21; Issue: 3 Linguagem: Inglês

10.1016/j.tree.2005.11.018

1872-8383

Mark S. Boyce, C. V. Haridas, C LEE, THENCEASSTOCHASTICDEMOGRAPHYW,

Animal Ecology and Behavior Studies

Recent advances in stochastic demography provide unique insights into the probable effects of increasing environmental variability on population dynamics, and these insights can be substantially different compared with those from deterministic models. Stochastic variation in structured population models influences estimates of population growth rate, persistence and resilience, which ultimately can alter community composition, species interactions, distributions and harvesting. Here, we discuss how understanding these demographic consequences of environmental variation will have applications for anticipating changes in populations resulting from anthropogenic activities that affect the variance in vital rates. We also highlight new tools for anticipating the consequences of the magnitude and temporal patterning of environmental variability. Recent advances in stochastic demography provide unique insights into the probable effects of increasing environmental variability on population dynamics, and these insights can be substantially different compared with those from deterministic models. Stochastic variation in structured population models influences estimates of population growth rate, persistence and resilience, which ultimately can alter community composition, species interactions, distributions and harvesting. Here, we discuss how understanding these demographic consequences of environmental variation will have applications for anticipating changes in populations resulting from anthropogenic activities that affect the variance in vital rates. We also highlight new tools for anticipating the consequences of the magnitude and temporal patterning of environmental variability. the logarithm of the population size logN(t), at time t, is approximately normally distributed when t is large, with mean and variance increasing linearly with time [3]. A diffusion approximation makes use of this fact to express the change in size in a small unit of time (growth rate) as a time-dependent random variable with a mean (an infinitesimal mean) and variance (an infinitesimal variance) [7]. the proportional change in a population property, such as growth rate or population size, given a proportional perturbation in a vital rate. For example, perturbing a rate in every environmental state such that its mean changes but its variance is constant, we obtain the elasticity with respect to the mean of that rate, Esμ (Box 2). these divide the globe into grid boxes ∼3° latitude and longitude on a side, and calculate motions of the atmosphere (potentially coupled with oceanic processes) from physical equations and parameterizations for sub-grid-scale processes. Under different scenarios of emissions, land use and regulation, simulations of these models are used to anticipate changes in the means and variances of climate variables, including temperature and precipitation, for each grid box. Regional circulation models (RCM) and statistical downscaling methods are alternatives for more fine-grained projections. a measure of the rate at which initially nearby trajectories of a system converge toward or move away from each other. The deviation at time t in the trajectories generated by two initial vectors P1(0) and P2(0) is given by |x(t)|=|P1(t)−P2(t)|, where |u| denotes the length of a vector u=(u1, u2,…uk), often taken as |u|=∑i|ui|. The (dominant) Lyapunov exponent λ1 is usually defined as the long-term average growth rate of the logarithm of this deviation when the difference |x(0)| between the two initial population vectors decreases to zero [3]. the change in a population property in response to a perturbation in a vital rate; similar to elasticity [2,3] although changes are not proportional. the study of age- or stage-structured populations in temporally varying environments, with states of the environment given by some stochastic process [2]. This process could be independent and identically distributed (i.i.d.), where independent draws from the same distribution determine the state of the process at each time. Examples of correlated processes, where the state at each time step depends on the states in the preceding time steps, include Markov processes and auto-regressive moving average (ARMA) processes [3]. here, any age- or state-specific demographic rate, such as survival or fecundity. An element of a population projection matrix (Box 2) is also a vital rate, even though it might be a function of other rates [3].