National?scale predictions of plant assemblages via community distribution models: Leveraging published data to guide future surveys

Species distribution models (SDMs) have been widely used to create maps of expected species incidence, often using citizen science (CS) occurrence data as inputs. Environmental policy is informed by knowledge community distributions, but there fewer attempts utilise the potential (CDMs) predict these. Many countries vegetation classification systems which include phytosociological information on individual species. Within Great Britain, National Vegetation Classification (NVC) primary standard for communities, and whilst produced at regional scales, cost-effective techniques are required national scales. Published NVC records 22 upland communities in England Wales were observed occurrences (presence-only data). Predictors CDMs enhanced index (EVI), elevation, slope, aspect, temperature rainfall. Five modelling methods investigated: generalised linear (GLMs), support vector machines (SVM), random forests (RF), maximum entropy (MaxEnt) likelihood (MaxLike). Model quality was assessed via bootstrapping area under curve (AUC), true skill statistic (TSS) Kappa index. There only small differences accuracy (median TSS model 0.742; range 0.280 0.873) with RF best overall CDM method. Across all summer winter temperatures annual rainfall most important predictor variables. NVCs spatially disjunct distributions both lowlands uplands, or that responded localised management environmental conditions, poorly predicted. Synthesis applications. can be reliably predicted large spatial scales from extant datasets. Management practitioners use community-level predictions design targeted field surveys typically associated specific communities. Most existing CS survey schemes focus rather than Hence future development new similar Plant Monitoring Scheme (NPMS), aligns NVC, will enable generate up-to-date