Performance of different sampling methods in estimating species richness in semiarid grassland

Abstract

Species richness is the simplest way to describe community and species diversity, but the number of detected species is remarkably dependent on the sampling method used. Thus, it is necessary to test the effectiveness of traditional sampling methods in different habitat types. To compare the performance of various methods for sampling grassland species richness in semiarid areas, we present a comparison of species richness detection among three commonly used sampling methods. We used 48 sites within six main grassland types in the Inner Mongolia semiarid region (meadow grassland in low wetlands, hilly grassland, desert grassland, mountain meadow grassland, mountain grassland and steppe desert) as study sites. In summer and early autumn in 2017, the commonly used modified Whittaker method, Hankins method and line transect method were used to conduct species richness surveys by overlaying these methods at the same site. The number of species per subplot (1 m2), the number of species in 10 subplots (10 m2) and the total species richness were compared across all three sampling methods to evaluate the effects of subplot shape, subplot arrangement and sampling area on the detected species richness. The results regarding the number of species per subplot showed that more species were recorded in the 1:4 rectangle subplots because of the larger perimeter:area ratio than that in square subplots and because the width of the 1:25 strips is significantly smaller than the proportion of the sampled species, which caused the 1:25 strips to show the worst performance. The analysis of species richness at the 10 m2 scale supports the idea that more species are likely to be detected with a dispersed arrangement of subplots in spatially heterogeneous grasslands. The results of the total species richness analysis showed that the accuracy of the line transect method was less than 50% in all grassland types but desert grassland and steppe desert, which have low species richness. Pairwise comparisons showed no significant difference between the accuracy of the modified Whittaker method and the Hankins method, but the modified Whittaker method requires less time than the Hankins method and performed better for more grassland types. Our study suggests that the modified Whittaker method is the most efficient sampling method for surveying species richness in semiarid grassland.