PT  - JOURNAL ARTICLE
AU  - Nao Takashina
AU  - Maria Beger
AU  - Buntarou Kusumoto
AU  - Suren Rathnayake
AU  - Hugh P. Possingham
TI  - A theory for ecological survey methods to map individual distributions
AID  - 10.1101/137158
DP  - 2017 Jan 01
TA  - bioRxiv
PG  - 137158
4099  - http://biorxiv.org/content/early/2017/05/12/137158.short
4100  - http://biorxiv.org/content/early/2017/05/12/137158.full
AB  - Spatially-explicit approaches have been widely recommended for various applications of ecosystem management. In practice, the quality of the data involved in the management decision-making, such as presence/absence or habitat maps, affects the management actions recommended, and therefore it is a key to management success. However, available data is often biased and incomplete. Although previous studies have advanced ways to effectively resolve data bias and missing data, there still remains a question about how we design the entire ecological survey to develop a dataset through field surveys. Ecological survey may inherently have multiple spatial scales to be determined beforehand, such as the spatial extent of the ecosystem under concern (observation window), the resolution to map the individual distributions (mapping unit), and the area of survey within each mapping units (sampling unit). In this paper, we develop a theory to understand ecological survey for mapping individual distributions applying spatially-explicit stochastic models. Firstly, we use spatial point processes to describe individual spatial placements drawn using either random or clustering processes. An ecological survey is then introduced with a set of spatial scales and individual detectability. Regardless of the spatial pattern assumed, the choice of mapping unit largely affects presence detection rate, and the fraction of individuals covered by the presence-detected patches. Tradeoffs between these quantities and the resolution of the map are found, associated with an equivalent asymptotic behaviors for both metrics at sufficiently small and large mapping unit scales. Our approach enables us to directly discuss the effect of multiple spatial scales in the survey, and estimating the survey outcome such as the presence detection rate and the number of individuals the presence-detected patches hold. The developed theory may significantly facilitate management decision-making and inform the design of monitoring and data gathering.