Description

Monitoring rare species often fails to account for imperfect detection. Corncrake occurrence is only indicated by calling males, which are highly mobile within the breeding season, making it difficult to determine actual breeding population sizes.
We conducted nocturnal counts at intervals of approx. 10 days during the breeding season over multiple years. A multi-state occupancy model for open populations was used to estimate numbers of occupied calling sites alongside with immigration and departure probabilities for each interval, while accounting for imperfect detection. A hierarchical formulation of the model enabled us to estimate also the number of occupied calling sites in subareas, where only two surveys per season took place. Thus a cumulative detection probability taking into account inter-seasonal movements could be calculated for the first time for Corncrakes.
Based on an average detection probability of 0.60 (CrI: 0.56-0.65) estimated numbers of males present during the two annual surveys was substantially higher (up to 50 %) than the numbers counted. Numbers of calling males peaked in mid and late May, when also most females started first broods. We found a high degree of turnover in site occupancy. Low detection probability associated with constant departure strongly limited the proportion of birds that can be detected even with frequent counts at 10-day intervals. Because timing of mowing is based on calling site occupancy in the study area, effective protection depends on information on Corncrake occurrence. Repeated nocturnal surveys during May and June are recommended to increase encounter probability for conservation purpose, but reliable population estimates require an analysis with open population models, especially at breeding sites where only few counts are feasible.