Study site
Our study site is the species-rich sub-alpine meadows located in the eastern part of the Qinghai-Tibetan plateau, Hezuo, China (34°55′N, 102°53′E) with mean elevation approximately 3000 m above sea level. Although the Tibetan Plateau Monsoon and Asian Monsoon28 brings rain, the study region has cold and dry climate, with mean annual temperature of 2.4° C and mean annual precipitation of just 530 mm23. The vegetation is dominated by herbaceous species such as Elymus nutans Griseb (Poaceae), Kobresia humilis (C.A. Mey.) Serg. (Cyperaceae) and Thermopsis lanceolata R. Br. (Fabaceae)23. Human impacts include agricultural exploitation and pastoralism are the primary current land use, which in places have caused serious land degradation. In response, local governments have stopped further agricultural exploitation and constructed fences to restrict livestock grazing. These efforts gave rise to successional chronosequences, such as the ones we use in our study.
We identified a chronosequence of fields that had been undisturbed for 4-, 6-, 10-, 13-, and 40-years (the control)19,23. All our sample sites, except for the control meadows, had been used for agriculture to grow highland barley in the recent past, with cessation of cultivation within the last 4–13 years. The time since last agricultural use was determined by interviews with local farmers. There are 1–10 km apart among the five meadows and all meadows possessed comparable topographic characteristics (e.g., orientation and slope), soil types and climate (Fig. 1A). This chronosequence is one of the same chronosequence in our previous work23 and we have observed that species richness increased from 61 to 82 species during succession, with 50 species sharing among all five successional meadows. Species composition was similar between 4-year and 6-year meadows, with 60 species sharing between these two meadows. Similar patterns were found in late successional meadows, with 70 species shared among 10-year, 13-year and undisturbed meadows.
Location map of our study sites and our quadrat sampling design. (A) locations of five sites representing each of the five successional ages (4-, 6-, 10-, 13-year and undisturbed grassland), (B) the 30 0.5 × 0.5 m2 quadrats sampling design in each of the five successional meadows. The map of Fig. 1A was obtained from Google Earth online version (https://earth.google.com/, access on 12/10/2018). Figure labels on the map were added using Google Earth online toolkit and text labels using Windows image processing software Paint.
Field sampling
The vegetation in each field was sampled in August 2013. An area of 120 × 120 m2 was randomly selected in each meadow. Within this area, thirty 0.5 × 0.5 m2 quadrats were regularly arranged in six parallel transects, with 20 m intervals between each two adjacent quadrats (detail please see Fig. 1B). To determine species richness and abundances, in each quadrat we recorded all the aboveground ramets and identified them to species.
To determine aboveground biomass, we removed all the ramets in each quadrat and took them to the laboratory, where they were oven-dried at 100℃ for 2 days and then weighed. Productivity is typically the amount of carbon fixed per unit time, not standing biomass. Here we follow methods of previous diversity–productivity studies in grasslands29,30, which have used aboveground biomass as proxy for productivity.
Functional trait data collection
We quantified the carbon economy of leaves by measuring specific leaf area (SLA, cm2 g−1). We quantified light capture strategy via photosynthesis rate (A, u mol−1). We estimated resistance to abiotic stress via leaf proline content (Pro, mg/kg), seed mass (SM, g) and seed germination rate (SG, %). Importantly, the functional traits for the same species at each successional age separately if they occurred in multiple meadows were measured to ensure that successional age-related intraspecific variation was appropriately incorporated into our analyses. All functional traits were determined as described in our previous work19,22,23 and the detailed procedures were given in the Supplementary Material.
Statistical methods
First, we compared variation during successional change in the proportion of total biomass for the three main functional groups of plants: forbs (dominant in early succession), legumes, and graminoids (both dominant in later succession) to check whether there are significant turnovers in the dominant plant taxa from early to late succession. Then, we used Spearman correlation analysis to quantify whether significantly positive correlations between empirical species diversity (S, numbers of species richness per square meters) and productivity (aboveground biomass per square meters, P) can be observed in each successional meadow.
For each of the five functional traits (SLA, A, Pro, Sm, and SG), we calculated two functional diversity indices: the community-weighted mean (CWM) and functional diversity (FD) represented by Rao’s quadratic entropy (RaoQ).
The two indices were calculated as follows:
$$ {CWM} = sumlimits_{i = 1}^{n} {p_{ij} times t_{ij} } $$
(1)
where pij is the relative abundance of the species i in each 0.5 × 0.5 m2 quadrat j, and tijis the mean trait value of the species i in each successional meadow j.
$$ RaoQ_{i} = sumlimits_{i = 1}^{n} {sumlimits_{i = 1}^{n} {p_{i} times p_{k} times d_{ik} } } $$
(2)
where pi and pkare the relative abundance of species i and k in each 0.5 × 0.5m2 quadrat j respectively and dik is the dissimilarity coefficient based on Euclidean distance between two species i and k in the multivariate trait space of each successional meadow j.
Then, a variance partitioning analysis was used to test the relative contributions of species richness, the CWM and FD represented by RaoQ of these five traits to productivity in each successional meadow. We also used variance partitioning to allocate changes in productivity in each successional meadow arising from four complementary components: (a) variation explained by species richness, (b) variation explained CWM of each of the five traits, (c) variation explained by FD of each of the five traits only, and (d) “unexplained variation”31. Across all successional meadows, species richness, and aboveground biomass, CWM and FD of all five traits (SLA, A, Pro, SM, and SG) were strongly right-skewed, so we log-transformed species richness, and aboveground biomass, CWM and FD of all five traits to meet the assumption of normality required by variance partitioning. At each successional meadow, variance partitioning was done using the function of “varpart” in “vegan” package in R32. All analyses above were performed in R (R Core Team 2019).
Source: Ecology - nature.com
