摘要：Assessing patterns of species distribution and abundance is important to understand the driving processes of, and predict future changes in, biodiversity. To this date, ecological studies have been mainly designed to investigate the effects of the mean magnitude of predictor variables, although ecological factors naturally vary in space and time. In a nine month long field experiment, we tested the effects of different temporal patterns (regular, lowly and highly irregular) in biomass removal (=disturbance event) on the diversity, species composition, and biomass accrual of macrobenthic assemblages grown on 15 × 15 cm2 PVC-panels. For each pattern of disturbance, disturbance events were timed at three sequences to control for possible confounding effects with recruitment patterns. Disturbance intensity was kept identical among treatments. Assemblages developed in the absence of disturbance for 3 months prior to a 150-day manipulation period, during which the biomass from 20% of the panel area was removed at each of ten disturbance events. Additional undisturbed settlement panels were deployed in the field to assess monthly recruitment rates and species succession over a one year period. Disturbance (i) reduced biomass and total species cover, (ii) changed species composition during the first half of the manipulation period significantly, and (iii) was without effect on species richness and evenness. Irregular disturbance regimes enhanced the abundance of the ascidian Ciona intestinalis, biomass accrual, and total species cover of assemblages relative to the regular disturbance regime, but had either no or only transient effects on diversity and species composition, respectively. Neither the degree of irregularity in disturbance nor the sequence of disturbance events affected any of the response variables significantly. Recruitment of species was strongly seasonal with almost only diatoms recruiting during winter, while recruitment was most intense during summer. Our results suggest that the temporal patterns of predictor variables might be of low explanatory power for the variance of responses in communities with seasonal recruitment patterns that are exposed to a high level of disturbance. Thus the need to include temporal patterns of predictor variables in experimental designs may depend on community dynamics and the characteristics of the process under investigation.
