Main drivers of plant diversity patterns of rubber plantations in the Greater Mekong Subregion

. The Greater Mekong Subregion (GMS) is one the global biodiversity hotspots. However, the diversity has been seriously threatened due to environmental degradation and deforestation, especially by expansion of rubber plantations. Yet, little is known about the impact of expansion of rubber plantations on regional plant diversity as well as the drivers for plant diversity of rubber plantations in this region. In this study, we analyzed plant diversity patterns of rubber plantations in the GMS based on a ground survey of a large number of samples. We found that diversity varied across countries due to varying agricultural intensities. Laos had the highest diversity, followed China, Myanmar, and Cambodia. Plant species richness of Laos was about 1.5 times that of Vietnam. We uncovered latitudinal gradients in plant diversity across these artiﬁcial forests of rubber plantations and these gradients caused by environmental variables such as temperature. Results (RDA), and forest demonstrated and


Introduction
Many tropical regions contain hotspots of biodiversity (Myers et al., 2000), especially for the Great Mekong Subregion (GMS), which is threatened by agriculture (Delzeit et al., 2017;Egli et al., 2018;Shackelford et al., 2014;Kehoe et al., 2017). Much of the land has recently been converted from forest to agriculture (Li et al., 2007), and rubber plantations have quickly expanded throughout the region (Ziegler et al., 2009;Li et al., 2015;Ahrends et al., 2015) due to a surge in the global demand for natural rubber, driven largely by the growth of tire and automobile industries. For example, 23.5 % of Cambodia's forest cover was destroyed between 2001 and 2015 to make way for crops such as rubber (Fig. S1h in the Supplement) and palm oil (Grogan et al., 2019). In southwest China, nearly 10 % of the total area of nature reserves had been converted to rubber monoculture by 2010 (Chen et al., 2016). At present, the GMS is a globally important rubber-planting region (Xiao et al., 2021).
Agricultural land uses can exacerbate many infectious diseases in Southeast Asia (Shah et al., 2019) and reduce biodiversity (Xu, 2011;Warren-Thomas et al., 2018;Fitzherbert et al., 2018;Zabel et al., 2019;Singh et al., 2019). Previous studies have shown that rubber cultivation not only affects plant diversity (Hu et al., 2013), but also affects the soil fauna (Chaudhuri et al., 2013;Xiao et al., 2014), bird diversity (Aratrakorn et al., 2006;Li et al., 2013), and bat diversity (Phommexay et al., 2011). There is also a large body of literature on the effects of forest conversion from tropical forest to rubber plantations on soil microbial composition and diversity (Tripathi et al., 2012;Schneider et al., 2015;Kerfahi et al., 2016, Lan et al., 2017a, b, c, 2020aCai et al., 2018). However, the impact of expansion of rubber plantations on regional plant diversity as well as the drivers for plant diversity of rubber plantations in the GMS are still unclear.
Latitudinal gradients in species diversity are well known (Mccoy and Connor, 1980), which holds that there is a fairly regular increase in the number of species of some higher taxon from the poles to the Equator. It has been suggested that the latitudinal diversity gradient could be caused by environmental variables such as temperature and precipitation. Previous studies have also demonstrated that temperature (Nottingham et al., 2018) and soil nutrients (Soons et al., 2017) as well as water resource utilization efficiency (Han et al., 2020) were the dominant drivers of plant diversity. However, whether latitudinal gradients in species diversity exist in rubber plantations, which are greatly affected by management measures, is still unknown.
In addition, rubber plantations have lower biodiversity than natural forests (Chaudhary et al., 2016). Generally speaking, species-rich zones showed a higher proportion of alien plant species in their flora (Stadler et al., 2000); thus, exotic plants are ubiquitous in rubber plantations which indicate that. Although the fact that exotic invasive species significantly decrease plant diversity (Xu et al., 2022) is universally known, we still do not know whether exotic species are the main driver for the sharp decline in plant diversity in rubber plantations. Thus, we hypothesize that (1) latitudinal gradients in plant diversity would not exist in rubber plantations due to the strong intensity of management and (2) exotic plants will result in a sharp decline in the plant diversity of rubber plantations because areas of low plant species richness may be invaded more easily than areas of high plant species richness (Stohlgren et al., 1999) and exotic species may results in loss of plant diversity (Xu et al., 2022). To test these hypotheses, we surveyed a large number of plots on rubber plantations in the GMS to investigate plant diversity and analyzed the associated drivers. Our study provides an empirical case for understanding the effect of rubber plantations on plant diversity in the Greater Mekong region and the restoration and protection of biodiversity in this region.

Study area
The Mekong River basin has a total length of 4880 km and a drainage area of 795 000 km 2 , with 326 million people liv- ing in the basin. The GMS encompasses a variety of climate types and geographical characteristics and is rich in water and biological resources (Wu et al., 2020). Rubber plantations are one of the most widespread vegetation types in the region and are distributed throughout the south of Yunnan province, almost all states of Thailand and Laos, the southern half of Vietnam and Myanmar, and the eastern half of Cambodia.

Sampling methods
Before the field investigation, we first determined the investigation route according to the distribution of rubber plantations in this regions. Then, plots were randomly selected approximately equidistant from each other (every 10-20 km according to the actual situation) along the investigation route (Yaseen, 2013). We did not deliberately select plots according types of rubber plantations, and thus these plots were independent from each other. Consequently, a total of 240 plots, each with an area of 100 m 2 (10 m × 10 m), were selected in the GMS, with 32 plots in Vietnam, 24 in Cambodia,15 in Laos,73 in Thailand,47 in Myanmar, and 49 in China (Fig. 1).
We started the investigation only after the guide (local people) asked the farmer's consent. Plot measurements, such as longitude, latitude, elevation, slope degree, slope aspect, rubber tree height, and canopy density, were recorded in detail (Table S1 in the Supplement). Annual and perennial plant species, shrubs, trees, and lianas as well as theirs seedlings were recorded. We do not investigate bryophytes, but ferns were investigated. Species information, such as species name, height and coverage, and life form (nonwoody, shrub, liana, or tree) (Lan et al., 2014), from each plot in the rubber plantations was also recorded. We visually assigned a cover value to each species in each quadrant of the plot, using an ordinal cover class scale with class limits 0.5 %, 1 %, 2 %, 5 %, 10 %, 15 %, 20 %, and thereafter every 10 % up to 100 %. The cover values for each species in the plot were then averaged across the four quadrants (Sabatini et al., 2016). Climate data, including annual average temperature and annual average precipitation, were obtained from WorldClim 2 (Fick and Hijmans, 2017) (http://worldclim.org, last access: 8 March 2021) based on the geographic coordinates of each sample site.

Data analysis
Relative height (RH), relative dominance (RD, using coverage), and relative frequency (RF) were calculated for each species to estimate the importance value (IV). Importance value, as defined here, differs from previous studies (e.g., Mcintosh, 1950, 1951;Greig-Smith, 1983;Linares-Palomino and Alvarez, 2005) because most understory species are herbs, which make the precise measure of abundance difficult. We define the importance value as the relative frequency as RF j = 100 × F j / j F j , the relative height as RH j = 100 × H j / j H j , and the relative dominance as where F j is the number of plots containing species j , D j is the coverage of species j , and H j is the height of species j . For the local community, there were no frequency data; therefore, the importance value is defined as Species richness and the Shannon index were used to measure the α diversity of each plot. It should be noted that the importance values of each species were used to calculate the Shannon diversity (i.e., replace "abundance" or "number of individuals" with "important value"). Principal coordinate analysis (PCoA) based on Bray-Curtis distance of species IVs (importance values) was performed to compare plant species composition across countries using the R package "amplicon". Analysis of similarity (ANOSIM) was used to test for differences in diversity indices among countries. Multiple linear regression was used to find whether there were positive or negative correlations between diversity (richness) and environmental variables including latitude, longitude, elevation, rainfall, temperature, slope degree, tree age, tree height, and canopy density. A machine learning algorithm, random forest (Breiman, 2001), was used to model α diversity (richness) and rank the feature importance of environmental factors with 999 iterations. In order to understand how plant compositions are structured by environmental factors, a redundancy analysis (RDA) for the importance value of species was carried out using the vegan package (version 2.5-7) (Oksanen et al., 2020) in the R (version 4.04) environment (R Core Team, 2021). Statistical significance was assessed using Monte Carlo tests with 999 permutations.

Plant composition of rubber plantations
A total of 949 plant species, representing 550 genera and 153 families, were recorded across rubber plantations of the six countries (Tables 1 and S2 in the Supplement). Our results also showed that 445 (46.89 %) were herbs, with a large number of Compositae (Table 1). Plant communities of rubber plantations tended to be dominated by Fabaceae, Euphorbiaceae, Poaceae, Rubiaceae, and Compositae (Table S3 in the Supplement). The five most common species observed were Cyrtococcum patens, Chromolaena odorata, Asystasia chelonoides, Axonopus compressus, and M. pudica (Table S4 in the Supplement). From 237 plots containing exotic plant species, most of them were from tropical America. A total of 121 (12.75%) species were identified as exotic (belonging to 45 families and 91 genera). The five most common exotic species were C. odorata, M. pudica, Axonopus compressus, Ageratum conyzoides, and Borreria latifolia. C. odorata and M. pudica were recorded in almost every plot (Fig. 2).
PCoA and ANOSIM were used to reveal the difference in plant compositions among these six countries. And the results showed significant differences (R = 0.383, P = 0.001) in species composition among these countries (Fig. 3a-b). Meanwhile, the first and second axes of RDA explained 5.95 % and 3.11 % of variation of species compositions, respectively (Fig. 4a). All environmental factors explained 18.65 % of the total variation (Fig. 4b). Countries, latitude, longitude, canopy height, and elevation all significantly impacted plant compositions of rubber plantations in the GMS and explained 5.62 %, 3.37 %, 3.14 %, 1.11 %, and 1.10 % of the total variations (Table 2).

Plant diversity of rubber plantations
Species richness of rubber plantations in Laos was the highest among the six countries, followed by China and Myanmar, while the richness of Thailand, Cambodia, and Vietnam  was relatively lower (Fig. 5a). The same was true for Shannon diversity (Fig. 5b).
The results of multiple linear regression (R 2 = 0.3227, P <0.001) showed that temperature (P <0.001), tree height (P <0.001), latitude (P <0.01), and slope degree (P <0.001) were positively correlated with the species richness (Fig. 6a). Among these factors, temperature (with the highest intercept 1.3) is the most important factor affecting plant diversity. Random forest results showed that high mean squared errors of latitude, temperature, and countries were the top three features affecting plant diversity of rubber plantations (Fig. 6b).

Effects of exotic species on plant diversity of rubber plantations
The exotic species richness of rubber plantations was relatively higher in Cambodia, Vietnam, and Myanmar compared to China, Laos, and Thailand (Fig. 3c). In order to clarify whether exotic species can reduce plant diversity, we analyzed the relationship between the dominance of exotic  species and the species richness in the plot. In view of the fact that C. odorata and M. pudica are the two most common exotic species in rubber plantations (Fig. 7a), the two species were selected for analysis. The importance values of exotic species C. odorata (Fig. S2a in the Supplement) and M. pudica (Fig. S2b) were negatively correlated with species richness, suggesting that exotic species with high dominance will reduce rubber plantation diversity. However, exotic species richness was positively correlated with species richness (Fig. 7c). Richness of communities where C. odorata (M. pudica) was present was not lower than those where it was absent (Fig. 7b). In sum, diversity of the community was reduced only when the dominance of exotic species was high.

Main drivers for plant composition and diversity of rubber plantations
Rubber plantations constitute one of the most important agroecosystems of tropical regions and play an important role in their carbon budgets (Chen et al., 2020). For plant composition, latitude ranks second ( Table 2) in terms of its impact on plant composition, indicating that latitude is an important driver of plant composition of rubber plantations. For plant diversity, both multiple linear regression and random forest showed that temperature was the most important factor for plant diversity of rubber plantations. Our results are consis-  tent with a previous study which revealed that temperature is the main driver for plant diversity (Nottingham et al., 2018). We were surprised to find that understory plant diversity of artificial rubber plantations increased with latitude, similar to that of the global diversity patterns (Rohde, 1992;Perrigo et al., 2013) that latitudinal gradients are known for in which maximum diversity does not occur near the Equator (Mccoy and Connor, 1980). One study suggest that the diversity of plant communities was directly affected by latitude . Our results showed that elevation was not as important as other factors, which is different from a previous study in which elevation significantly affects plant species diversity .
Plant diversity of north Laos and south China was relatively higher than other countries. This observation may be due to the large variation in elevation in these areas, which translates into greater environmental heterogeneity. In addition, greater slope may increase environmental heterogene-ity and expand niche space (Morrison-Whittle and Goddard, 2015). In any case, temperature could largely contribute to explaining the latitudinal diversity gradient patterns of rubber plantations.

Not all exotic plants cause the loss of plant diversity in rubber plantations
Rubber plantation expansion and intensification has occurred in many regions that are key for biodiversity conservation. Monoculture plantations have been promoted to restore the world's forested areas but have done little to slow the loss of biodiversity . It has been hypothesized that exotic species might more easily invade areas of low species diversity than areas of high species diversity (Stohlgren et al., 1999). A recent study shows exotic plants account for ∼ 17 % of the total importance value for natural ecosystems and ∼ 35 % of the total importance  value for human-modified ecosystems (Chandrasekaran et al., 2001). Here, in rubber plantations, exotic plants made up roughly 12% of the total recorded species and 22.80 % of the coverage. C. odorata is a noxious perennial weed in many parts of the world (Kushwaha et al., 1981), and it is unsurprising that it was recorded in almost all plantation plots in our study. These indicated that invasion by exotic species has either already occurred or is inevitable in many systems (Stohlgren et al., 1999). M. pudica, the "sensitive plant", is a worldwide, pan-tropical invasive species (Melkonian et al., 2014). M. pudica, as many tropical grasses and herbs, is tolerant of low pH (Humphreys, 1997;Paudel, 2018), which explains its ubiquity in acidic rubber plantation soil.
More importantly, our study demonstrated that the diversity of the community was reduced only when the importance value of exotic species is large enough; not all exotic species cause the loss of plant diversity in rubber plantations, which follows the theory that many species can coexist in spatially heterogeneous areas as long as nutrients and light are not limiting (Huston and DeAngelis, 1994). Our results also were consistent with the idea that inhibition of plant diversity by exotic species invasion gradually weakened with increased precipitation (Xu et al., 2022) due to higher precipitation in the GMS. In addition, management of rubber plantations reduces the dominance of exotic species to a great extent, thus providing space for the survival of other plants.

Plant composition and diversity is largely affected by management
Forests that are intensively managed for production purposes generally have lower biodiversity than natural forests (Chaudhary et al., 2016), and this is especially true for rubber plantations (He and Martin, 2016). In artificial forests such as rubber plantations, there is no doubt that management measures and agricultural intensity are two of the most important factors affecting plant diversity. The application of herbicides and sprout control causes low diversity of understory plants; this is especially true of rubber plantations of Vietnam (Fig. S1f) In poor areas, we cannot just talk about ecological goals without first understanding local cultures and economies. Well-managed forests can alleviate poverty in rural areas, as outlined by the United Nations Sustainable Development Goals (Lewis et al., 2019). A previous study conducted in India demonstrated that a no-weeding practice in mature rubber plantations did not affect rubber yield (Abraham and Joseph, 2016). A similar study conducted in China also showed that natural management strategies can improve biodiversity without reducing latex production (Lan et al., 2017d). There is strong evidence that adopting more natural management strategies improves plant diversity without reducing latex production (Lan et al., 2017d). More innovative management measures, such as cease of weeding and herbicide application (He and Martin, 2015), must be implemented to improve the biodiversity of rubber plantations, so as to promote the biodiversity of the region.

Conclusion
We provide a large regional study on the plant diversity of rubber plantations in a global biodiversity hotspot. Plant diversity followed global trends with respect to latitude and temperature. Exotic species were very common in rubber plantations, especially where agricultural intensity was strong. However, not all exotic species directly drive the loss of biodiversity. Only higher dominance of some exotic species was associated with a loss of plant diversity within rubber plantations. We must make greater efforts to balance agricultural production with conservation goals in this re-gion, particularly in Vietnam and Cambodia, to minimize the loss of biodiversity.
Data availability. Data generated in this study are available from the corresponding authors upon reasonable request.
Author contributions. GL contributed to the conceptualization, methodology, writing, review, and editing. BC contributed to the methodology, review, and editing. CY, RS, BC, ZW, and XZ performed the investigation.
Competing interests. The contact author has declared that neither they nor their co-authors have any competing interests.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.