Plant invasions alter soil biota and microbial activities: A global meta-analysis

Plant invasion is a major component of global environmental change that can significantly alter soil biota, which are essential for organic matter decomposition and nutrient cycling in terrestrial ecosystems. However, the particular linkage between invasive plants, soil biotas, and enzymatic activities remains unclear. Here, we conducted a comprehensive meta-analysis using 688 paired observations from 107 studies to evaluate to the impact of plant invasion on soil biota, enzymatic activities and nutrient cycling. Our results showed that plant invasion significantly reduced herbivores (45%), detritivores (27%) and omnivores (45%) abundance, while allelopathy, woody invaders and forest ecosystems significantly reduced predator abundance by 47, 46, and 32%, respectively. Invasive plants also slightly reduced bacterial and fungal biomass, but significantly increased arbuscular mycorrhizal fungal (AMF) biomass, MBC and MBN by 29, 19 and 32%, respectively. Moreover, invasive plants had inconsistent impact on the activities of C-decomposing enzymes, but invertase, phenol oxidase and β-glucosidase were found significantly higher in invaded than uninvaded sites.  Invaded sites had significantly higher activities of N- and P-releasing soil enzymes ranged from 18 to 27% than uninvaded sites. Soil microbial biomass, N-mineralization, soil respiration, available (N, P) nutrients, NH4+-N and nutrient stocks were all higher in invasive than native plants rhizosphere soils. Our findings suggest that invasive plants had negative impact on soil functional groups, however, they enhanced soil nutrient-releasing enzymes and soil available nutrients. This in turn accelerates nutrient cycling and promote the persistence and success of invasive plants.

Methods
We selected peer-reviewed journal articles using Web of Science and Google Scholars and screened the identified records using Preferred Reporting Items for Systematic Review and Meta-analysis procedure (PRISMA). Primary studies that satisfied the following selection criteria were included: (1) studies specifically described the identities of invasive and native plants and investigated either in the field observational or experimental studies, (2) at least one of the soil functional groups, or microbial composition or enzymatic activity was measured simultaneously in the invasive and native plants using the PLFA method (Kong et al., 2011), (3) invaded sites and control or native plants were found in the same biotic and abiotic conditions, (4) the means and sample sizes of the selected variables were available or could be calculated from the related publications. In our meta-analysis, we only used the soil microbial community composition and structure, soil microbial biomass and their enzymatic activities measured with PLFAs, the most used method for microbial measurements (G. Zhang et al., 2019). We applied the following criteria to identify independent case studies:  (1) If multiple independent pairs of invasive and native plants from different ecosystems were reported in the same article, we treated each pair as a separate case study that complied with other meta-analyses' criteria and (2) if the study article reported one native plant and two more invasive plants, each invasive-native combination was considered as a separate case study. The data presented in the form of figures were extracted by using webplotdigitizer 4 https://automeris.io/WebPlotDigitizer.     

For each pair of response variables in invasive (i) and native (n) plants, the log response ratio (LRR) was calculated based on mean values (X) as a measurement of effect size (Bakbergenuly et al., 2020; Lajeunesse, 2011). The LRR is a unit-free index that estimates the size of the impact and its directions. The values of LRR = 0, showed no difference in the measured response variable between invasive and native plants, a negative value indicates that the invaded site has a lower value than the noninvaded sites and a positive value presents that the invaded site has a greater value than the noninvaded site (Luo et al., 2006). The effect of plant invasion was estimated (Eq. 1) using the succeeding natural LRR (Bakbergenuly et al., 2020).

 ---------------------------------------------------- (1)

Where  and , are the results of the means in the treatment and control group, respectively. We calculated the variance (VR) that related with the effects computed by Eq. 2: -

-------------------------------------------------------------------------- (2)

Where St and Sc are the standard deviations of the experimental and control treatments, respectively, where as  and  are the data points of the experimental and control treatments, respectively. The standard error of the log response ratio calculated (Eq. 3) as follows: -

SE (lnR)=  ----------------------------------------------------------------------------- (3)

We used Eq. 4, the weighted mean response ratio (R++) to identify the overall effects of the experimental versus the control group treatments, respectively (Hedges et al., 1999).

  ---------------------------------------------------------------------- (4)

Where  is the weighting factor,  and  are th data points in the th dataset in each category group, ,  and  are the number of datasets and data points in each dataset, respectively, in the category groups. The SE of R++ was calculated by Eq. 5: -

------------------------------------------------------------------ (5)