Spatial autocorrelation at multi-scale of soil collembolan community in farmland of the Sanjiang Plain, Northeast China

Abstract

Multi-scale autocorrelation is one of the fundamental factors in soil animal spatial heterogeneity and spatial co-occurrence patterns. In August and October 2011, collembolan communities were sampled in farmland of the Sanjiang Plain with a history of more than 30 years of cultivation. The sampling field was divided into 100 subsamples through intensive sampling at the nodes of a 10 ×10 regular grid with 5 m inter-sample distances. Global Moran’s I and Local Moran’s I indices were used to test the multi-scale autocorrelation for abundance of the collembolan community and for abundance of each species. Both indices were calculated with the Open Geoda software platform.We collected 10603.62 individuals/ m2, belonging to 17 collembolan species, in August, and 38698 individuals/ m2, belonging to 13 collembolan species, in October. In terms of one-way ANOVA analysis, species richness of the collembolan community in August was non-significantly different from that in October, and the abundance of the collembolan community in August was significantly lower than that in October (P<0.001). According to the results of Global Moran’s I indices, abundance of collembolan community showed significant positive autocorrelation at 5—50 m in August, whereas the abundance of the collembolan community only showed significant positive autocorrelation at 5 m in October. At the same time, most of the collembolan species showed significant autocorrelation at the multi-scale. Oligaphorura ursi showed significantly positive autocorrelation at 5—50 m in August and at 5, 35 and 40 m in October. Protaphorura sp.1, Allonychiurus songi, Sminthurinus sp.1, Tullbergia sp.1, Desoria sp.1, Entomobrya sp.3, Hypogastrura sp.1 and Lepidocyrtus felipei showed significant spatial autocorrelation in both August and October, and most of the significant autocorrelations were positive at the multi-scale. Folsomia sp.2 and Folsomia sp.1 only showed obvious spatial autocorrelation at a few scales in August. The other six collembolan species did not show significant spatial autocorrelation in either August or October. According to the results of Local Moran’s I indices, abundance of the collembolan community and of all collembolan species showed significant local autocorrelation (P<0.05). Locally, abundance of the collembolan community formed “high-high”and “low-low”spatial aggregations in August. These “high-high” and “low-low” spatial aggregations persisted in the experiment plot in October, but the size and distribution of the spatial aggregations were different from those in August. The abundance of the collembolan community also formed one “low-high” spatial outlier in August, whereas no significant “low-high” or “high-low” spatial outliers were detected in October. All of the collembolan species formed “high-high” and/ or “low-low” spatial aggregations, accompanied with “ high-low” and/ or “ low-high” spatial outliers, which formed a horizontal mosaic of “patches” and “gaps”. From summer to autumn (August to October), the horizontal mosaic structure showed temporal variation, with the size and spatial distribution of these patches being different between the two seasons. This study revealed that the abundance of the collembolan community and most of the collembolan species showed obviously multi-scale spatial autocorrelations. Spatial aggregation is a general rule for the collembolan community in farmland of the Sanjiang Plain, forming a horizontal mosaic of “patches” and “gaps”. This mosaic structure showed temporal variation from summer to autumn.