The Relative Importance of Niche Process and Neutral Process in the Community Assembly of Subtropical Karst Forest: A Perspective on the Changes of Species Distribution Patterns at Different Scales

The importance of niche processes and neutral processes to community assembly has been affirmed by 13 most studies, although their relative importance needs to be determined in many systems. Moreover, as 14 the spatial scale changes, the ecological processes that determine the community pattern may differ. We tested these ideas in subtropical karst forest in southwestern China in order to aid efforts of community reconstruction. To test the importance of niche-based and neutral mechanisms we compared the fit six models to the observed SAD of the plot at three different sampling scales (10 m × 10 m, 20 m × 20 m, 50 m × 50 m). We also used spatial autocorrelation and distance-based Moran's eigenvector maps (dbMEM) combined 21 with variation partitioning to further determine the relative contribution of the niche process and the 22 neutral process under different sampling scales. The neutral theoretical and statistical models fit the observed species abundance distribution curve best at each sampling scale. And variation partitioning showed that although the contribution of spatial 26 structure was lower at larger sampling scales, it was still important, suggesting that neutral processes drive community structure at all sampling scales. In contrast, habitat filtering and interspecies competition may lead to a net weakening of the contribution of the niche process to the species abundance pattern of the community because they act in opposite directions. inter-species relationships, geographic spatial differences considered.


38
Understanding the mechanisms that drive the variation and maintenance of species diversity in 39 ecological communities remains a fundamental question in biology. However, the effect of ecological 40 processes on the community is almost impossible to directly observe, so it is an effective way to infer 41 potential ecological processes from the observed patterns. The species abundance distribution (SAD) 42 describes the ranked abundance of species within a community, and is commonly used to describe 43 community structure and species diversity in ecological research (Preston, 1948

54
The structure of ecological communities is the result of the combined effects of various ecological 55 processes that are assumed to leave signals on patterns of species distribution (Hubbell, 2001).

56
Therefore, the SAD curve of a community has been widely used in attempts to detect the effects of 57 niche differentiation, seed dispersal limitation, species differentiation, extinction mechanisms, and 58 processes driving community structure and dynamics (Green, 2007). Dozens of models, based on 59 different theories, have been used to fit and interpret SAD patterns. In general, these models can be 60 divided into statistical models and mechanistic models. Pure statistical models provide an empirical 61 fitting of the species abundance distribution, but their biological and ecological significance is not 62 clear. For example the commonly used log-series (Fisher et al., 1943) and log-normal models (Preston, 63 1948) reflect more mathematical distribution laws than ecological significance.

64
In contrast, mechanistic models attach importance to the biological significance of the model, and 65 different models reflect different ecological processes. Mechanistic models can be further divided into 66 neutral theory models and niche models. According to the assumption of community saturation, the 67 neutral theory hypothesizes that the distribution pattern of species abundance in local communities 68 obeys a zero-sum multinomial distribution, for example the model of Hubbell (2001). This model 69 assumes that the local community has many rare species, but the specific number is affected by the 70 community size and dispersal limitation. In contrast, models based on niche theory have also been 71 developed to fit the SAD, including the broken stick, niche preemption, overlapping niche, dominance 72 decay, and power fraction models (MacArthur, 1957;Tokeshi, 1993;Tokeshi, 1996). These models

76
It is well known that niche theory and neutral theory differ in their assumptions concerning the 77 factors that influence the species composition of a community. The former assumes that species 78 composition is affected by inter-species competition and environmental variation, while the latter 79 suggests that the distribution pattern of species is affected by dispersal limitation and that therefore

91
In addition, the relative importance of these ecological processes that determine community 92 structure may vary at different spatial scales, leading to variation in the SAD across spatial scales.

93
Community habitat heterogeneity, β diversity, degree of individual aggregation, and spatial

102
We addressed these ideas in the typical karst forest community of evergreen and deciduous broad-103 leaved mixed forest in Guilin, Guangxi, southwestern China with the aim to better understand 104 mechanisms of community structure and provide a theoretical basis for the restoration and 105 reconstruction of vegetation in the area. We established a medium-sized forest plot, and compared the 106 fit of a suite of statistical, niche, and neutral models to the observed SAD. Then, we used variance 107 partitioning to assess the relative importance of environmental variation and space to community 108 structure and SAD. If niche processes primarily drive the SAD here, we should see that the pure 109 environmental components or the spatial structure of the environment occupy a larger explanatory

181
In order to assess the relative importance of niche processes and neutral processes to community 182 assembly, we used three types of model to explain and quantify the patterns and processes of the 183 species abundance distribution: two niche models (broken stick model and niche preemption model),

204
Statistical models

205
The log-normal model, proposed by Preston (1948), considers that the logarithm of the total number of 206 individuals (N) in the community conforms to the normal distribution, and the abundance of the i-th 207 species is:

209
where μ and σ represent the mean and variance of the normal distribution, respectively, and Φ 210 represents the normal deviation.

237
To test which of the above six models best fit the observed species abundance distributions, we         (Figure 1). At both the small and medium scales, the species abundance 298 distribution of the community presented a monotonically decreasing distribution (Figure 1a and b). At 299 the large sampling scale, a small peak was observed in the interval of 4-8 abundance, reflecting a 300 bimodal distribution (Figure 1c).  Note. AIC, Akaike's information criterion; D, K-S test. *p < 0.05, **p < 0.01, ***p < 0.001.

319
Spatial autocorrelation detection at different scales 320 In general, at larger sampling scales, the strength of spatial correlation was weaker (Figure 3).

381
The relative importance of niche process and neutral process 382 For a long time, the species abundance distribution pattern has been considered to reflect the species 383 diversity maintenance mechanism of the community to a certain extent (Tokeshi, 1993;Hubbell, 2001; 384 Volkov et al., 2003). However, it is worth noting that a specific species abundance distribution can be

412
The results of variation partitioning show that as the sampling scale becomes larger, the 413 interpretation rate of the spatial structure gradually decreases. This is also consistent with the result that 414 the fitting effect of the neutral theoretical model becomes worse as the sampling scale becomes larger.

434
This part may contain some biological and non-spatial structural attributes that are not observed in this 435 study. At the same time, some ecological processes such as the interaction between species have not 436 been considered in the variance decomposition. The spatial autocorrelation test shows that the 437 interaction between species has a strong force at the small and medium sampling scales, which may be 438 the reason for the higher unknown components at the small and medium sampling scales. In this study,  interests or personal relationships that could have appeared to influence the work reported in this paper.

466
Informed consent All authors gave consent for conducting and publishing the research.