Thirteen years of monitoring an alpine short-lived perennial: Novel methods disprove the former assessment of population viability

Abstract

Our former assessment of viability in a local population of Androsace albana, an alpine short-lived perennial plant species (Logofet et al., 2020b), relied upon 10-year observations of the population structure on permanent sample plots and the corresponding, time-inhomogeneous, matrix model of stage-structured population dynamics. We applied a concept of the population in a random environment and its stochastic growth rate (λS) estimated by a Monte Carlo technique under the so-called “i.i.d.” mode of randomness, a simple model popular in the literature, meaning independent, identically distributed environments. All the Monte Carlo tests resulted in the estimates of λS < 1, meaning a negative forecast of population viability. A recent expansion of the monitoring period up to 13 years has now been combined with a more realistic model of randomness, namely, a Markov chain of changes in the environment that is recovered from a longer time-series of a local weather index correlated with variations in the asymptotic growth rate (λ1) of the model population within the 13-year period. The former design of Monte Carlo tests with the updated time series of λ1s and the realistic model of randomness have principally changed the viability assessment to λS > 1, whereas the i.i.d. tests have still resulted in λS < 1, i.e., a qualitatively opposite forecast of population viability.An alternative approach to the viability assessment guides to the original concept of pattern-multiplicative average of nonnegative matrices and reduces to calculating λ1(G), the dominant eigenvalue of the average matrix G. Unexpectedly, λ1(G) turns out less than 1, thus suggesting a conceptual revision of how the local population viability should be predicted.