Dataset DOI: 10.5061/dryad.q2bvq840f

Description of the data and file structure

We provide two datasets in CSV format (Biodiversity_index.csv; Multifunctionality_and_functions.csv), containing data on grassland multifunctionality, 11 individual grassland functions, and 17 community diversity and trait metrics from 2021 and 2022 (results averaged across EFS and NHG sites). Moreover, all missing values in the data are represented as “null.”

We also provide the code (code_for_R.zip) and the data used for calculations (Data_for_R.zip), which demonstrate all computational steps, statistical analyses, and graphing procedures performed in R_4.2.0.

Finally, we include the figures (figures.zip), containing PDF versions of all figures presented in the Manuscript and Supporting Information.
Dominant species determine drought effects on grassland multifunctionality

Files and variables

File: code_for_R.zip

Description: It demonstrates all computational steps, statistical analyses, and graphing procedures performed in R_4.2.0.

Caculation.R: Calculation of biodiversity indicators and multifunctionality. It includes:

1. Species diversity (H-index, D-index, J-index, SR);
2. Functional diversity (FRic, FEve, FDiv, FDis) and CWM traits;
3. Phylogenetic diversity(PD, MPD, MNTD);
4. Calculation of Multifunctionality(the averaging approach and the cluster-threshold approach);
5. Calculate the correlation between MFt, MFa, and other functions.

Statistical analysis.R: All statistical analyses conducted in Section 2.4, “Statistical Analyses,” of the manuscript. It includes:

1. Liner mixed-effects(lme) results for the interaction among drought, site and year on all factors(Table S2);
2. The statistical results of Figure 1;
3. The statistical results of Figure 2;
4. The statistical results of Figure 3;
5. The statistical results of Table S3 and Figure 4.

Main figures plotted.R: R code used to generate all figures in the manuscript and Manuscript and Supporting Information. It includes:

1. Figure 1;
2. Figure 2;
3. Figure S1;
4. Figure S2;
5. Figure S5.

File: Data_for_R.zip

Description: This document contains all the raw data required for the aforementioned R execution and computational procedures. This section contains a total of 12 CSV files and 2 txt files. It includes:

2021EFS.csv, 2021NHG.csv, 2022EFS.csv, 2022NHG.csv: raw data required for species diversity (H-index, D-index, J-index, SR) in the file “Caculation.R”. Notes: none of these four datasets involve units.

Variables:

• Site:EFS and NHG
• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022

basic data_biomass.csv: raw data required for Functional diversity (FRic, FEve, FDiv, FDis) and CWM traits in the file “Caculation.R”;

Variables:

• Site:EFS and NHG
• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• Latin name:Latin names of plant species
• Biomass(g/m2):Dry weight of plants per square meter in grassland communities
• LNC(mg/g):leaf nitrogen content
• LCC(mg/g):leaf carbon content
• LPC(mg/g):leaf phosphorus content
• MPH(cm):maximum plant height
• LDMC(mg/g):leaf dry matter content
• SLA(cm2/g):specific leaf area

PD_abun.csv and PD_species.txt: raw data required for Phylogenetic diversity (PD, MPD, MNTD) in the file “Caculation.R” PD_abun.csv records species cover data, and PD_species.txt represents phylogenetic relationships among species (i.e., the phylogenetic tree).  Notes: none of these two datasets involve units.

calculation of multifunctionality.csv and calculation of multifunctionality.txt: raw data required for Calculation of Multifunctionality (the averaging approach and the cluster-threshold approach) in the file “Caculation.R”;

Variables:

• Site:EFS and NHG
• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• N_pool(g/m2):aboveground plant nitrogen pool
• C_pool(g/m2):aboveground plant carbon pool
• STP(g/kg):soil total phosphorus content
• SAP(g/kg):soil available phosphorus content
• STN(g/kg):soil total nitrogen content
• STC(g/kg):soil total carbon content
• C/N(unitless):soil carbon:nitrogen ratio
• ANPP(g/m2):aboveground net primary productivity
• BNPP(g/m2):belowground net primary productivity
• MBC(g/kg):soil microbial biomass carbon content
• MBN(g/kg):soil microbial biomass nitrogen content

functions and MFt.csv: raw data required for The statistical results of Figure 1 in the file “Statistical analysis.R” and Figure 1 in the file “Main figures plotted.R”

Variables:

• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• MFt(unitless):grassland multifunctionality
• ANPP(g/m2):aboveground net primary productivity
• BNPP(g/m2):belowground net primary productivity
• APCP(g/m2):aboveground plant carbon pool
• APNP(g/m2):aboveground plant nitrogen pool
• STC(g/kg):soil total carbon content
• STN(g/kg):soil total nitrogen content
• STP(g/kg):soil total phosphorus content
• SAP(g/kg):soil available phosphorus content
• C_N(unitless):soil carbon:nitrogen ratio
• MBC(g/kg):soil microbial biomass carbon content
• MBN(g/kg):soil microbial biomass nitrogen content

biodiversity and MFt.csv: raw data required for The statistical results of Figure 2, Figure 3, Talbe S3 and Figure 4 in the file “Statistical analysis.R” and Figure 2, Figure 3 and Figure S5 in the file “Main figures plotted.R”

Variables:

• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• H(unitless):Shannon-Wiener diversity index
• D(unitless):Simpson’s dominance index
• J(unitless):Pielou evenness index
• S(unitless):species richness
• FEve(unitless):functional evenness
• FDiv(unitless):functional divergence
• CWM.LN(mg/g):leaf nitrogen content
• CWM.LC(mg/g):leaf carbon content
• CWM.LP(mg/g):leaf phosphorus content
• CWM.H(cm):maximum plant height
• CWM.LDMC(mg/g):leaf dry matter content
• CWM.SLA(cm2/g):specific leaf area
• PD(unitless):phylogenetic diversity
• MPD(unitless):mean pairwise distance
• MNTD(unitless):mean nearest taxon distance
• MFt(unitless):grassland multifunctionality
• FRic(unitless):functional richness
• FDis(unitless):functional dispersion
• SWC(unitless):soil water content

Table S2_1.csv and Table S2_2.csv: raw data required for Liner mixed-effects(lme) results for the interaction among drought, site and year on all factors(Table S2) in the file “Statistical analysis.R”.

Variables and abbreviations are the same as above.

data_precipitation.csv: raw data required for Figure S1 in the file “Main figures plotted.R”

Variables:

• Year:For 50 consecutive years, from 1972 to 2022
• Mouth:January through December each year
• Precipitation(mm):Average monthly precipitation(mm)

File: figures.zip

Description: It contains PDF versions of all figures presented in the Manuscript and Supporting Information. Specifically includes:

1. Figure_1.pdf;

   Description: Experimental drought effects on individual grassland function and multifunctionality at two study sites combined (mean ± SE). The horizontal bars represent the 95% confidence interval. Statistical significance is depicted as * P <0.05, ** P<0.01, and *** P<0.001.

   Abbreviations: 

   ANPP(g/m2), aboveground net primary productivity;

   BNPP(g/m2), belowground net primary productivity;

   APCP(g/m2), aboveground plant carbon pool;

   APNP(g/m2), aboveground plant nitrogen pool;

   TC(g/kg), soil total carbon content;

   TN(g/kg), soil total nitrogen content;

   TP(g/kg), soil total phosphorus content;

   AP(g/kg), soil available phosphorus content;

   C:N ratio(unitless), soil carbon:nitrogen ratio;

   MBC(g/kg), soil microbial biomass carbon content;

   MBN(g/kg), soil microbial biomass nitrogen content.

2. Figure_2.pdf;

   Description: Experimental drought effects on community-weighted mean (CWM) traits and community diversity at two study sites combined. Results are presented as a response ratio (drought/control). Circle colors indicate different types of variables: species diversity (pink), functional diversity (yellow), CWM traits (green), and phylogenetic diversity (blue). The horizontal bars represent the 95% confidence interval. Statistical significance is depicted as * P <0.05 and ** P<0.01.

   Abbreviations: 

   H-index(unitless), Shannon-Wiener diversity index;

   D-index(unitless), Simpson’s dominance index;

   J-index(unitless), Pielou evenness index;

   SR(unitless), species richness;

   FEve(unitless), functional evenness;

   FDiv(unitless), functional divergence;

   LNC(mg/g), leaf nitrogen content;

   LCC(mg/g), leaf carbon content;

   LPC(mg/g), leaf phosphorus content;

   MPH(cm), maximum plant height;

   LDMC(mg/g), leaf dry matter content;

   SLA(cm2/g), specific leaf area;

   PD(unitless), phylogenetic diversity;

   MPD(unitless), mean pairwise distance;

   MNTD(unitless), mean nearest taxon distance.

3. Figure_3.pdf;

   Description: Relationships between grassland multifunctionality and community-weighted mean (CWM) traits and community diversity under extreme drought at two study sites combined. Solid line represents a significant linear relationship, with the light area showing the 95% confidence interval.

   Abbreviations are consistent with those in Figure_2 above.

4. Figure_4.pdf;

   Description: Structural equation model illustrating the effect of extreme drought on multifunctionality mediated by four composite indices representing species diversity, functional diversity, phylogenetic diversity, and community‐weighted mean (CWM) traits across grassland sites. Blue and red arrows reflect positive and negative relationships, while solid and dashed arrows reflect significant and insignificant effects, respectively. Statistical significance is depicted as * P <0.05, ** P<0.01, and *** P<0.001; standardized path coefficients are shown along the paths. Marginal R²^ ^{values (*R}2^{_m*) and conditional *R}2^* (*R2^_c*) values are given. 

5. Figure_S1.pdf;

   Description: Experimental drought shelter setup (a). The main figure shows the precipitation and probability density curves during the growing season (May to August) over the past 50 years (1972–2022). The precipitation data for the Control and experimental drought treatments from 2021 to 2022 are plotted on the probability density curves. The inset figure in the upper right illustrates the effects of drought on soil water content over two years (2021–2022).

6. Figure_S2.pdf;

   Description: Determination of functional clusters for calculating multifunctionality. (a) Elbow method used to identify the optimal number of grassland ecosystem function clusters. (b) Cluster membership and eigenvalues (x-axis) for each ecosystem function. 

   Abbreviations:

   ANPP(g/m2), aboveground net primary productivity;

   BNPP(g/m2), belowground net primary productivity;

   APCP(g/m2), aboveground plant carbon pool; 

   APNP(g/m2), aboveground plant nitrogen pool;

   STC(g/kg), soil total carbon content;

   STN(g/kg), soil total nitrogen content;

   STP(g/kg), soil total phosphorus content;

   SAP(g/kg), soil available phosphorus content;

   C/N(unitless), soil carbon:nitrogen ratio; 

   MBC(g/kg), soil microbial biomass carbon content;

   MBN(g/kg), soil microbial biomass nitrogen content.

7. Figure_S3.pdf;

   Description: Pearson correlation matrix between cluster-threshold multifunctionality (MFt), average multifunctionality (MFa), and individual ecosystem functions. Different colors represent different aspects of ecosystem functions. 

   Abbreviations are consistent with those in Figure_1 above.

8. Figure_S4.pdf;

   Description: Relationships between grassland multifunctionality and (a) functional richness (FRic), (b) functional dispersion (FDis) under extreme drought at both sites.

9. Figure_S5.pdf.

   Description: Results of the effects of experimental drought on (a) species-specific mean biomass and (b) relative abundance compared to Control (we averaged the values of biomass per year at both sites).

File: Multifunctionality_and_functions.csv

Description: It contains data on grassland multifunctionality and 11 individual grassland functions from 2021 and 2022 (results averaged across EFS and NHG sites).

Variables:

• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• Multifunctionality(unitless):grassland multifunctionality
• ANPP(g/m2):aboveground net primary productivity
• BNPP(g/m2):belowground net primary productivity
• APCP(g/m2):aboveground plant carbon pool
• APNP(g/m2):aboveground plant nitrogen pool
• TC(g/kg):soil total carbon content
• TN(g/kg):soil total nitrogen content
• TP(g/kg):soil total phosphorus content
• AP(g/kg):soil available phosphorus content
• C:N ratio(unitless):soil carbon:nitrogen ratio
• MBC(g/kg):soil microbial biomass carbon content
• MBN(g/kg):soil microbial biomass nitrogen content

File: Biodiversity_index.csv

Description: It contains data on 17 community diversity and trait metrics from 2021 and 2022 (results averaged across EFS and NHG sites).

Variables:

• Treatment:drought and control
• Block:randomized complete block design was used, with six blocks per site
• Year:2021 and 2022
• H-index(unitless):Shannon-Wiener diversity index
• D-index(unitless):Simpson’s dominance index
• J-index(unitless):Pielou evenness index
• SR(unitless):species richness
• FEve(unitless):functional evenness
• FDiv(unitless):functional divergence
• LNC(mg/g):leaf nitrogen content
• LCC(mg/g):leaf carbon content
• LPC(mg/g):leaf phosphorus content
• MPH(cm):maximum plant height
• LDMC(mg/g):leaf dry matter content
• SLA(cm2/g):specific leaf area
• PD(unitless):phylogenetic diversity
• MPD(unitless):mean pairwise distance
• MNTD(unitless):mean nearest taxon distance
• FRic(unitless):functional richness
• FDis(unitless):functional dispersion

Code/software

All calculations (biodiversity indicators and multifunctionality), statistical analyses (interactive effects, general mixed-effects models, principal component analysis and piecewise structural equation models), and figure generation in this study were performed in R 4.2.0, and we have provided all the code.