Title page
Title :Structures of intransitive competition network affect functional attributes of plant community under nitrogen enrichment
Sun Qingqinga,1, Yang Junjieb,1, Yang Fengyana, Zhao Yuyinga, Zhang Guangmingc, Wei Cunzhengb, Han Xingguob,*, Li Jinshana,1,*
a College of Science, Beijing Forestry University, Beijing 100083, China
b State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China
c Department of Pharmaceutical Science, Changzhi Medical College, Changzhi 046000, China
* Corresponding author : Han Xingguo, e-mail:xghan@ibcas.ac.cn; Li Jinshan, e-mail:lijinshan@bjfu.edu.cn
Acknowledgements: This work was supported by the National Key Research and Development Project of China (2016YFC0500705) and the Fundamental Research Funds for the Central Universities (2015ZCQ-LY-01).
Conflict of interest : The authors declare they have no competing interests.
Authors contributions:Sun Q.Q. and Li J.S. conceived the ideas and designed methodology; Yang J.J. collected the data; Sun Q.Q., Yang J.J. and Li J. S. analyzed the data and wrote the manuscript with feedback from Zhang G.M., Yang F. Y. and Zhao Y. Y. All authors contributed critically to the drafts and gave final approval for publication.
Abstract
Atmospheric nitrogen (N) deposition is a potential danger factor for grassland ecology, and will cause unpredictable consequences to plant communities. However, how plant species interactions response to N enrichment and then affect ecological functions are not fully known. We investigated how intransitive competition network was related to the functional attributes of plant community under a 13-years N-deposition experiment. Results showed that intransitive competition network was not a single structure, but a complexly interwoven structure of various simple structures. Nested work was more common, accounting for 76.96%, and gained new species at a higher colonization rate than short network did. The network had a long-term mechanism to maintain the small-scale Alpha diversity, and a significant lag effect on the large-scale Gamma diversity. Under the conditions of N ≥ 2 g N·m-2·year-1, without mowing and under high fertilization frequency, the increase of network complexity significantly decreased plot biomass gradually. The relationship between biomass and network complexity is quadratic curves, also between abundancy and the complexity, but with the opposite bending directions, which indicated that biomass and abundance were complementary to each other, which may be a mechanism of maintaining the relative balance of species competition. In addition, the decrease of species asynchronism changing with the increase of N-enrichment gradually destroyed ecosystem stability. However, at medium N enrichment, intransitive network counteracted the negative effects of N enrichment and maintained or even improved the biomass ecosystem stability. Our results suggested that intransitive competition network is an internal mechanism of self-restoration of a grassland ecosystem. Under nitrogen enrichment conditions, competitive networks complexity is reduced, leading to a reduction in species diversity. These analyses emphasize the important role of intransitive network structure to stabilize grassland ecosystem. In order to achieve sustainable development of grassland, it is indispensable to control nitrogen addition rate.
Key words: biodiversity, complexity, ecosystem stability, nested network, species asynchrony

1. Introduction

Nitrogen (N) fertilization management on natural grasslands and pastures can significantly increase yield of plants and animals, which has brought great convenience and abundant benefits to modern agriculture and animal husbandry (Tilman 1987; Zhang et al. 2015). Nutrient enrichment generally increases primary production but reduces biodiversity. Theoretically, nutrient enrichment can destabilize ecosystems (Rosenzweig 1971). In grassland ecosystem, the species extinction and community composition are caused by overfertilization or atmospheric Nitrogen deposition (Clark and Tilman 2008; Socher et al. 2013; Zhang et al. 2019).
Some studies (Freckleton and Watkinson 2001; Wang et al. 2020)) clearly indicate that changes in competitive intensity are also key drivers of community composition. The intensity is a kind of quantification of competition structure, while competitive structure changes with the variation of environment (Allesina and Levine 2011) and is susceptible to the influence of human factors (Soliveres et al. 2015). Competition network is an important determinant for community functions, with a gradient from weak competition community to strong competition community (Maynard et al. 2017b). Structure of networks is a driver of biodiversity and coexistence in natural communities (Sun et al. 2021). In addition, the nature and intensity of competitive interactions also affect growth, productivity, and biomass of surviving individuals. Therefore, structure of competition networks, i.e., the result of pairwise species competition (Allesina and Levine 2011), should play an important role in community attributes.
The immediate cause of existence for competition is that the species composition in the focus community, since species have different functional and life-history characteristics, the particular combination of which maintains intransitivity or hierarchical competition (Aschehoug et al. 2016). Intransitive competition is more common in vascular plants, and there is a significant positive relationship between species richness and intransitivity (Soliveres et al. 2015; Gallien 2017a), which reveals complex competitive interactions among species (Allesina and Levine 2011; Engel and Wetzin 2008) and greatly promotes the theory of species symbiosise. Empirical analysis and simulation study (Ulrich et al. 2017) both showed that intransitive competition, as mechanism for maintaining high alpha diversity (Laird and Schamp 2006), can promote symbiosis of competitors. Therefore, understanding the good competition is important to understand the way that community is structured and how coexistence of species occurs.
In addition, ecosystems are always exposed to environmental perturbations that tend to alter temporal stability of plant communities (Ma et al. 2017; Douda et al. 2018). Species asynchrony occurs in natural communities due to interspecific competition and environmental stresses. Empirical observations have shown that there are positive covariances among species biomass in many communities, which do not indicate absence of competitive interactions. Besides the effects of diversity itself on community function, the interaction of diversity and competition, as two key attributes, determines biomass production, respiration, and carbon-use efficiency. On other hand, availability of abiotic resources varying over time, potentially breaks down inherent species competition. However, it is unclear to understand how species competitive interactions respond to N enrichment and affect plant community function (Gross et al. 2017; Mouillot et al. 2011).
We tested the following hypotheses: (1) the intransitive network with nested structures is more common in vascular plants; (2) at the small scale, the function of nested network to maintain species richness was more significant; (3) the intransitive network complexity could maintain the relative balance between biomass and plant number; (4) the negative effect of nitrogen can be mitigated by intransitive network.

2. Materials and methods

2.1 Experimental sites and sampling

The field experiments were carried out on the temperate steppe (43 ° 32 ’51 ”N, 116 ° 40’ 23” E) in Xilinhot, Inner Mongolia Autonomous Region, China, where belongs to the typical temperate semi-arid climate and has relatively flat terrain. Since 1999, 50 hectares of the experimental area had been fenced off to eliminate the impact of large-scale grazing, and no fertilizer had been applied before the beginning of the experiment. The mean annual temperature in 2008-2020 was 1.45°C, the mean temperature in the growing season was 17.04°C, the annual precipitation was 329.9mm, and the precipitation in the growing season was 227.6mm. Two C3 perennial herbaceous plants, Stipa grandis and Leymus chinensis, are the dominant species in the vegetation community.
The nitrogen used in the experiment was ammonium nitrate (NH4NO3,>99%), which was added to the experimental plots during the growing season and the non-growing season by wet deposition (ammonium nitrate dissolved in distilled water) and dry deposition (ammonium nitrate mixed in fine sand), respectively. Species traits such as biomass, plant abundance et, al., as the sample data, were measured on a 0.5m×2m strip plot in every experimental block of 8 𝑚2. The experiments designing with completely randomized block, and setting up with 9 nitrogen annual adding rates (0, 1, 2, 3, 5, 10, 15, 20, 50 g N·𝑚-2·𝑦𝑒𝑎𝑟-1), 2 kinds of fertilization frequency (2 times a year, 1 time a month), and 2 processings of mowing (yes or not), in addition to two control groups, a total of 38 different treatments were performed, each of which was repeated 10 times.

2.2 Methods

2.2.1 Species richness

Based on theoretical work (Wang and Loreau 2014), as well as previous empirical studies (e.g., Wilcox et al. 2017; Zhang et al. 2019), the Alpha diversity was defined as the species richness of 1m2 plot (small scale), and the Gamma-diversity as the total richness of 10 repeated plots of 1m2 (large scale) under a same treatment. In addition, soil ammonium toxicity is greater under N enrichment, leading to low colonization rate and even high species extinction rate (Zhang 2015). In order to explore the effect of intransitive competition network on the community attributes, species gained or lost have also become the key point of the discussion.

2.2.2 Spatial variability

Lloyd’s central theory (1967) demonstrated that average density is difficult to explain ecological causality between competition and density regulation. At individual level, Lloyd designed average clumping as a measurement of spatial patches to capture spatial aggregation associated with intraspecific competition for resources. In order to explore the influence of N enrichments on spatial differences in species richness, we used Lloyd’s variance-to-mean ratio,\(I=\frac{\sigma^{2}}{\mu^{2}}-\frac{1}{\mu+1}\). Where 𝜇 and 𝜎2 are the mean and variance of species richness under different treatments or variables. When I<1, the spatial difference of species richness was small and the distribution was more uniform; when I>1, the spatial difference of species richness was great; and when I = 1, it is the expectation of the Poisson’s random distribution.

2.2.3 Measurement of intransitivity

Nitrogen enrichment can reduce environmental heterogeneity in small communities (Western 2001; Fraterrigo et al. 2005), therefore, similar environment is likely to promote similar community dynamics (Wesche et al. 2012). We introduced the Apriori algorithm to mine the species dominant assembly in spatial clumping and ensured that the inference of a species competition network was based on at least 5 plots in 10-repeated ones (Wang et al. 2020; Sun et al. 2021).
Starting from the dominant assembly, the ‘reverse engineering’ and colonial competition model proposed by Ulrich et al. (2014) was used to simulate the best competition matrix, entries of which are competition coefficients between species. Simulations were done to produce a large number of competitive matrix C , as well as transfer probability matrix P , obtained from C by colonial competition model. Then Spearman tests between PU and the relative abundance vectorU of experimental plots under same treatment were done. The average of Spearman coefficients (𝑟𝑠) from the tests was used as a selection index for best C and P . Spearman coefficients acquired from the experimental data were generally low, showing that the role of interspecific competition mechanism was relatively small, while environmental heterogeneity (Dufour et al. 2006) or other more important mechanisms (Ulrich et al. 2018) maybe play important role. Therefore, only when \(r_{s}\geq 0.6\), interspecific competition is the dominant factor for a assembly, and then the matrix C was going to be the optimal competition matrix. Species-competition network was built up by the matrix C, in which species were vertices and edge weights were the entries of C.
We used deformation of the Flow Hierarchy (Luo 2011; Sun et al. 2021) to measure the complexity of intransitive network.
\begin{equation} \text{Complexity}=\left\{\begin{matrix}0,\text{\ \ \ \ \ \ \ \ }\text{if}\ \frac{h*c}{N_{\text{node}}}>1\\ 1-\frac{h*c}{N_{\text{node}}},\text{if}\ \ 0\leq\frac{h*c}{N_{\text{node}}}<1\\ \end{matrix}\right.\ \nonumber \\ \end{equation}
where h is the measure of flow hierarchy, \(N_{\text{node}}\)is the species number of the focal network, and c is a constant. When complexity is 0, there is only chain structure (e.g., A>B>C). When complexity is 1, there are a large loop structures that all member of network in it.

2.2.4 Biomass stability and species asynchrony

Following previous work (Tilman 1999), biomass stability was calculated as \(S_{\text{community}}=\frac{\mu_{T}}{\sigma_{T}}\),where\(\mu_{T}\) and \(\sigma_{T}\) are the interannual mean and standard deviation of community biomass over the 13 years respectively.
For each 1-m2 strip, the degree of asynchrony in the population dynamics of constituent species of its local community was quantified as 1-\(\frac{\sigma_{T}^{2}}{{(\sum_{i=1}^{M}{\sigma_{i})}}^{2}}\), where\(\sigma_{T}^{2}\) is the temporal variance of community biomass, and\(\sigma_{i}\) is the standard deviation of biomass of species i in theM -species community (Loreau and de Mazancourt 2008).

2.2.5 Statistical analysis

Linear or quadratic polynomial least squares regression was used to study the relationships between variables, such as plot biomass and N-addition rates, plant abundance of plot and network complexity etc. The independent sample T-test was used to test the significance of differences among diversity of different kinds of network. The Python packages used in this paper were ”StatsModels”, ”NextWorkX” and ”SciPy”.

3. Results

3.1 Network structure affects richness