Biodiversity loss via human-related disturbances has a profound influence on ecosystem functions and services. The relationship between biodiversity and ecosystem function has attracted considerable research attention. However, the impact of disturbance on biodiversity-ecosystem function relationships remains unclear. A manipulative experiment with a strip-strip block design involving fertilizer (unfertilized, fertilized annually, fertilized every five year) and mowing (unmowed, mowed in spring, mowed in spring and fall) in a successional old field began in 1996 to examine disturbance effects on plant diversity-aboveground net primary productivity (ANPP) relationships. Diversity and ANPP data were collected in September of three years (1996, 1997, 2018). To avoid analysis of correlated variables, diversity metrics were reduced to uncorrelated taxonomic (species richness, SR) and phylogenetic (net relatedness index, NRI) diversity. Repeated-measures analysis of variance was used to test the effects of mowing, fertilizer, and their interaction on SR, NRI, and ANPP. Mixed models including generalized linear models (GLMs) and generalized additive models (GAMs) were employed to investigate the relationships between plant diversity as the response to ANPP, mowing, and fertilizer, and their interactions over time using stepwise backwards elimination of non-significant items. Acceptance of the optimal GLM or GAM was based upon Akaike Information Criterion.
Results/Conclusions
Across all years, ANPP increased by 23.0% (t61.51 = 3.54, P = 0.002) and 23.1% (t61.51 = 3.55, P = 0.002) under the spring and spring-fall mowing treatments, respectively. NRI decreased by 0.76 (t61.58 = 5.77, P < 0.001) and 0.66 (t61.58 = 5.04, P < 0.001) by spring and spring-fall mowing, respectively, which implied increased phylogenetic overdispersion (more negative NRI values) in response to these treatments. SR was not affected by fertilizer or mowing. The optimal model for SR was a GLM incorporating an interaction between ANPP and year (χ2 = 19.01, P < 0.001). Across all years, there was a humped-back relationship between SR and ANPP, but individual years showed different patterns. The optimum model for NRI was a GAM incorporating a mowing and year interaction (F2,189 = 4.74, P = 0.010). There was a negative relationship between NRI and ANPP across all years, but individual years showed different patterns. These results suggest that anthropogenic disturbances alter different dimensions of biodiversity and productivity, but do not affect diversity-productivity relationships. Our findings highlight the robust relationship between biodiversity and ecosystem function under intensified anthropogenic disturbances providing theoretical support for biodiversity conservation and ecosystem restoration.