Predicting species distributions based on incomplete survey data: the trade‐off between precision and scale

摘要: Systematic species surveys over large areas are mostly not affordable, constraining conservation planners to make best use of incomplete data. Spatially explicit distribution models (SDM) may be useful detect and compensate for information. SDMs can either based on standardized, systematic sampling in a restricted subarea, or - as cost-effective alternative data haphazardly collated by "volunteer-based monitoring schemes" (VMS), area-wide but inherently biased heterogeneous spatial precision. Using capercaillie Tetrao urogallus, we evaluated the capacity generated from survey localise unknown inhabited predict relative local observation density. Addressing trade-off between precision, sample size extent area, compared three different strategies: VMS-data collected throughout whole study area (7000 km 2 ) using 1) exact locations 2) aggregated grid cells an average individual home range, 3) transect counts conducted within small subarea (23.8 ). For each strategy, two sizes modelling methods (ENFA Maxent), which were cross-validation independent Models (strategies 1 performed equally well predicting density localizing "unknown" occurrences. They always outperformed strategy 3-models, irrespective method, partly because provided more comprehensive clues setting discrimination-threshold presence absence. Accounting potential errors due extrapolation (e.g. projections outside environmental domain potentially biasing variables) reduced, did fully observed discrepancies. As they cover broader range species-habitat relations, achieved better model quality with less a-priori knowledge. Furthermore, highly mobile like resolution corresponding individuals' lead good predictions locations. Consequently, when effort is necessary, precise unsystematically representative region preferable systematically sampled region.