Demographic modeling and monitoring cycle in a long-lived endangered shrub

Abstract Demographic monitoring is a well established tool for conservation managers. But, monitoring programs typically do not offer methods to explore the relationship between projected population trends and the probability of trend detection. Moreover, conservation biologists need to evaluate and incorporate the effect of variability and habitat perturbation on the efficiency of monitoring programs. We studied population demography of an endangered shrub endemic to East Central Spain, Vella pseudocytisus subsp. paui in order to understand its present demographic performance, how it is affected by increasing variability and perturbation and how these two factors are related to monitoring and monitoring thresholds. Using Lefkovich demographic matrices on six years of data, we produced 19 different population projections, and we compared these modeled projections against observed projection. We designated three conceptually important detection thresholds: an early warning; an unequivocal signal; and, a quasi-extinction threshold. Based on calculated detection times of all models for every threshold, we produced an averaged monitoring cycle (minimum visit frequency) to provide managers with a tool to design a consistent monitoring program for this plant. Our results indicate a relatively stable current population along with large detection times for critical threshold changes of modeled populations when considering small management changes (low population variability and low perturbation intensity). In contrast, model combinations incorporating high population variance and high perturbation produce disproportionally short times to detection of a significant population trend. In terms of designing studies to maximise results while minimising monitoring effort, designing a monitoring program capable of providing an early warning system of potential population failure requires more effort than detecting an unequivocal signal of decline. Finally, we propose two alternatives when calculating monitoring cycle (MC) values. One of them implies demographic MC if population perturbation and variation are high. For that, an optimal monitoring cycle is estimated based on a range of possible scenarios for population trends. Alternatively, we propose environmental MC, when low values of perturbation and variation are expected.