Primroses, ca 1000 species from the family Primulaceae, are mostly distributed in high mountains and attract people's attention with their high ornamental value. However, they are increasingly threatened by human activities and climate change in recent decades. China harbors more than half of global primrose species, most of which are endemic. To strengthen studies on their conservation, we established a distribution database for primroses in China, including 12 genera and 535 speceies. Based on this database, we explored the geographic pattern of the richness of primrose species in relation to environmental factors and identified conservation priority areas. The results showed that the primrose richness was highest in the mountain areas from the eastern Himalayas to the Hengduan Mountains, with habitat heterogeneity and temperature seasonality together shaping that pattern. Furthermore, the richnesses of endemic and threatened primroses were strongly affected by the long-term climate stability since the Last Glacial Maximum. On average, national nature reserves (NNRs) covered 29.6% of the distribution areas of primroses. By analyzing less protected species, we proposed that nature reserves need to be established in several conservation gaps mainly located in Southwest China to protect the primrose species in China.

Distribution data of primroses in China

Data on the distribution of primrose plants in China were obtained from six sources: 1) national, provincial and local floras in China, 2) occurrence of endemic species derived from the Diversity and Geographical Distribution of Endemic Seed Plants in China, 3) scientific articles related to field investigations for primrose species, 4) checklists of nature reserves, 5) specimen records available from the National Specimen Information Infrastructure (NSII, http://nsii.org.cn), 6) records from the Global Biodiversity Information Facility (GBIF, https://www.gbif.org/). All species names were standardized using the Flora of China (Hu and Kelso, 1996) and the Catalogue of Life (http://www.catalogueoflife.org/).

Among all these sources, distribution records from the first four resources are county-level occurrence, which usually leads to overestimating the distribution range of species. Therefore, we also collected information on the occurrence elevations and habitat types of each species. By overlapping the county-level records with a digital elevation model (DEM) (with a resolution of 30 m from http://reverb.echo.nasa.gov/reverb /redirect/wist) and a vegetation map of China (1: 1,000,000) (Editorial Committee of Vegetation Map of China, 2007), we selected the grids of suitable habitats of the corresponding species in its occurring county as the refined distribution areas. Then, these distribution data were uniformly converted into grids with a resolution of 25 x 25km.

Environmental factors

Contemporary climate data were obtained from the WorldClim website (http://www.worldclim.org/) with a spatial resolution of 30 arc-second. The climatic variables included mean annual temperature (MAT, °C), mean temperature of warmest quarter (MTWQ, °C), mean temperature of coldest quarter (MTCQ, °C), mean annual precipitation (MAP, mm), temperature seasonality (TSN, unitless) and precipitation seasonality (PSN, unitless). In addition, we calculated the potential evapotranspiration (PET, mm), actual evapotranspiration (AET, mm), moisture index (Im, unitless) and water deficit (WD, defined as the difference between PET and AET) using mean monthly temperature and precipitation (Thorthwaite and Hare, 1955; Francis and Currie, 2003).

Four variables were used to represent habitat heterogeneity: range of altitude (RALT, m), number of vegetation types (VEG, unitless), range of MAT (RMAT, °C) and MAP (RMAP, mm) within each 25 km grid. RALT is the difference between the highest and lowest elevations in each grid, based on the abovementioned DEM production. VEG was obtained by overlapping the vegetation map of China (1:1,000,000) with the 25 km grid system. RMAT (or RMAP) represents climatic heterogeneity, reflected as the difference between maximum and minimum MAT (or MAP) within each grid.

Long-term anomalies in temperature (TANO) and precipitation (PANO) were represented by the absolute values of the difference in MAT (and MAP) between the Last Glacial Maximum (LGM) and the present. The climate data in LGM were also derived from the WorldClim website (Watanabe et al., 2011).