RESEARCH ARTICLE Modeling Occupancy of Hosts by Mistletoe Seeds after Accounting for Imperfect

The detection of an organism in a given site is widely used as a state variable in many meta-population and epidemiological studies. However, failure to detect the species does not nec-essarily mean that it is absent. Assessing detectability is important for occupancy (presence —absence) surveys; and identifying the factors reducing detectability may help improve sur-vey precision and efficiency. A method was used to estimate the occupancy status of host trees colonized by mistletoe seeds of Psittacanthus plagiophyllus as a function of host co-variates: host size and presence of mistletoe infections on the same or on the nearest neigh-boring host (the cashew tree Anacardium occidentale). The technique also evaluated the effect of taking detectability into account for estimating host occupancy by mistletoe seeds. Individual host trees were surveyed for presence of mistletoe seeds with the aid of two or three observers to estimate detectability and occupancy. Detectability was, on average, 17 % higher in focal-host trees with infected neighbors, while decreased about 23 to 50% from smallest to largest hosts. The presence of mistletoe plants in the sample tree had negli-gible effect on detectability. Failure to detect hosts as occupied decreased occupancy by 2.5 % on average, with maximum of 10 % for large and isolated hosts. The method presented in this study has potential for use with metapopulation studies of mistletoes, especially those focusing on the seed stage, but also as improvement of accuracy in occupancy models esti-mates often used for metapopulation dynamics of tree-dwelling plants in general