This Wang_2023_DATA_README.txt file was generated on 2023-11-21 by Jiang Wang

GENERAL INFORMATION

1. Title of Dataset: Data from: Herbivory and allelopathy contribute jointly to the diversity-invasibility relationship.

* *

2. Author Information

Corresponding Investigator

Name: Dr Xiaoyan Wang

Institution: School of Life Science/Zhejiang Provincial Key Laboratory of Plant Evolutionary Ecology and Conservation, Taizhou University, China

Email: wxy3470117@163.com
Co-investigator 1

Name: Dr Jiang Wang

Institution: School of Life Science/Zhejiang Provincial Key Laboratory of Plant Evolutionary Ecology and Conservation, Taizhou University, China

Email: wangjiang@tzc.edu.cn
Co-investigator 2

Name: Dr Song Gao

Institution: School of Life Science/Zhejiang Provincial Key Laboratory of Plant Evolutionary Ecology and Conservation, Taizhou University, China

Email: lygs2008@163.com
Co-investigator 3

Name: Hefang Hong

Institution: Linhai Branch of Taizhou Ecological Environment Bureau,Linhai 317000, China

Email: 6598958@qq.com
Co-investigator 4

Name: Wei Xue

Institution: School of Life Science/Zhejiang Provincial Key Laboratory of Plant Evolutionary Ecology and Conservation, Taizhou University, China

Email: x_wei1988@163.com
Co-investigator 5

Name: Jiwei Yuan

Taizhou Ecological and Environmental Monitoring Center, Taizhou, China

Email: 46785047@qq.com

 

Co-investigator 6

Name: Mark van Kleunen

School of Advanced Study, Taizhou University, Taizhou 318000, China;4 Ecology, Department of Biology, University of Konstanz, 78464 Konstanz, Germany

Email:mark. vankleunen@uni-konstanz.de

 

Co-investigator 7

Name: Junmin Li

School of Advanced Study, Taizhou University, Taizhou 318000, China;4 Ecology, Department of Biology, University of Konstanz, 78464 Konstanz, Germany

Email: limtzc@126.com

3. Date of data collection: 2018 and 2019

4. Geographic location of data collection: Taizhou, Zhejiang, China

5. Funding sources that supported the collection of the data: National Key R&D Program of China (2016YFC1201100), the Natural Science Foundation of Zhejiang Province (No. LY22C030001, LTY22C030004), the National Natural Science Foundation of China (No. 31870504), the Taizhou University National Funds for Distinguished Young Scientists (No. 2017JQ005, 2019JQ005).

6. Recommended citation for this dataset: Wang et al. (2023), Data from: Herbivory and allelopathy contribute jointly to the diversity-invasibility relationship, Dryad, Dataset

* *

DATA & FILE OVERVIEW

1. Description of dataset

To construct experimental communities of different species richness (1, 2, 4 and 8 species), we used eight native species (Arctium lappa L., Patrinia villosa (Thunb.) Juss., Achyranthes bidentata Bl., Reynoutria japonica Houtt., Taraxacum mongolicum Hand.-Mazz., Plantago asiatica Ledeb., Platycodon grandiflorus (Jacq.) A. DC., Antenoron filiforme (Thunb.) Rob. et Vaut.). These species were chosen because they are common in natural shrub-grassland communities in the Zhejiang province, China, and two of them, P. villosa and A. bidentata, are usually very dominant in those communities. Solidago canadensis L. was selected as the invader.

In May 2017, we prepared 188 plots of 2 m × 2 m in a previous agricultural field that had been abandoned seven years earlier, in Luqiao (121°23′46.16″ E, 28°31′21.62″ N). During October to December 2016, the top 10 cm of the soil was removed to deplete the seed bank, and the remaining soil was ploughed and repeatedly rotovated to produce a fine soil. In March to April 2017, the plots were weeded three times to further deplete the seed bank. The experimental area was divided into four blocks of 47 plots. In each plot, we constructed monocultures for each of the eight species with one replicate each, six two-species mixtures of different compositions and with three replicates each, six four-species mixtures with different compositions and three replicates each, and one eight-species mixture with three replicates. Each native species was used in an equal number of plots at each species-richness level. When the invasion treatment started, we also established three monoculture plots of the invader (S. canadensis) to assess potential side effects of activated carbon application

2. File:

Name: Wang_2023_data.xlsx

Description: All data of this study.

METHODOLOGICAL INFORMATION

Treatments

In the herbivory-reduction treatment, we sprayed the plots with an insecticide. To avoid that the insecticide spray would drift into plots of the non-insecticide treatment, we randomly assigned the four blocks, each with 47 plots, to one of the four experimental treatments (i.e. herbivory reduction, allelopathy reduction, herbivory and allelopathy reduction, no herbivory and allelopathy reduction). Within each block, the monocultures and species mixtures of the native plants as well as the three monocultures of S. canadensis were randomly assigned to different plots. In the blocks with herbivory reduction, we sprayed each plot every 10 days with a 0.5-L solution of Abamectin (1:500, v:v; Hongke Biochemical Co., Ltd, Shijiazhuang city, Hebei province in China), which is a broad-spectrum insecticide. The other plots were sprayed with tap water instead, and the plants in those plots were considered to be fully exposed to herbivory. In the blocks with allelopathy reduction, we had, prior to planting, mixed 8 L (~4 kg) of activated carbon (HG/T 3491-1999, Wuxi Yatai United Chemical Co., Ltd) into the top 20 cm of soil in each plot, resulting in a concentration of 1: 100 (v:v). Activated carbon can adsorb chemicals, including allelopathic substances produced by plants, and thereby supposedly neutralizes or reduces allelopathic interactions (Inderjit and Callaway, 2003; Prati & Bossdorf, 2004; Ridenour & Callaway, 2001; Yuan et al., 2022). However, as activated carbon may have undesired side-effects on plant growth (Lau et al., 2008; Kabouw et al., 2010; Weißhuhn & Prati, 2009; Wurst et al., 2010), we tested for side effects in the additional S. canadensis monoculture plots. We found no significant effect of activated carbon on biomass production of S. canadensis (Appendix S1: Fig. S1), which suggests that side effects of activated carbon were absent or minimal in our study. Therefore, plots without activated carbon were considered to have the full strength of allelopathic interactions among the plants.

Measurements

To quantify the likelihood of resource competition in each plot, we measured light interception by the canopy and soil nutrient contents. Light interception in each plot was measured twice on cloud-free days (September 25–28, 2018 and September 21–24, 2019). Photosynthetically active radiation (PAR) was measured at three randomly selected points within the central 1.6 m × 1.6 m area of each plot using a PAR ceptometer (GLZ-C, Zhejiang Top Instrument Co., Ltd, China). This was done at mid-day (between 11:00 and 14:00), when the solar angle was maximal. Light-interception proportion was calculated as (PAR above the canopy - PAR at ground level)/PAR above the canopy.

To determine nutrient contents, we took three soil cores (6.4 cm in diameter and 20 cm in depth) from the central 1.6 m × 1.6 m area of each plot, and thoroughly mixed the three cores to obtain one composite sample per plot. This was done on the days that we took the light measurments. Samples were air-dried for two weeks and then sieved through a 2-mm mesh. After that, 20 g of each soil sample was ground into powder and used to measure soil total nitrogen and total phosphorus contents with an autoanalyzer (Autoanalyzer 3, BRAN+LUEBBE, Germany). We also determined the total organic carbon content using the method of Nelson and Sommers (1982).

Before biomass harvest in 2018 and 2019 years, in each plot of herbivory and allelopathy and that of herbivory and allelopathy reduction treatment, 30 leaves (10 leaves was randomly selected in each of upper, middle and lower layers of canopy) of each species living in each plot were surveyed for damage by herbivores. If the total number of leaves of a species was less than 30, and the leaves were all surveyed. We counted the number of leaves with herbivore damage (e.g. holes). We then quantified the herbivory ratio as the number of leaves with herbivory damage divided by 30. In addition, we calculated for the damaged leaves on each plant species a herbivory intensity (an herbivory-severity index) as the average proportion of leaf area lost due to herbivory. To measure foliar flavonoid content of native speies and S. canadensis, five plots that having the species (if the species have died and another plot was reselected) were randomly selected in each of four treatments. Leaf samples (5-10 number) of the species were randomly gathered in each of five plots. The method of Shen et al. (2005) was used to measure the foliar flavonoid content.

To determine the biomass production of plants, and whether this changed over time, we did two harvests. The first one was done in October 2018, one year after the invaders had been planted, and the second one was done in October 2019. At the first harvest, we collected aboveground biomass of all living plants in one half (1 m × 2 m) of each plot, and at the second harvest, we did this for the remaining halves. At both harvests, the plants were sorted to species, dried at 80°C for 48 h and weighed.

DATA-SPECIFIC INFORMATION FOR: Wang_2023_data.xlsx

1. Number of variables: 12

2. Number of cases/rows:

Sheet 1 'data': 377

Sheet 2 'biomass': 1121

3. Variable List:

plot_name: Plot name.

year: 1 2018 year, 2 2019 year.

Species_richness: Species number in plot

community_composition: Communities with different composition, 1-21 indicate 21 types communities with different compositions.

allelopathy: 0 allelopathy reduction, 1 allelopathy.

herbivory: 0 herbivory reduction, 1 herbivory.

total_organic_carbon: total organic carbon concentration in soil of each plot.

total_nitrogen: total nitrogen concentration in soil of each plot.      

total_phosphorus : total phosphorus concentration in soil of each plot.

light_interception: light interception of plant communities of each plot.

invader_biomass: the biomass of invader in each plot.

complementarity_effect: Complementarity effect among native species. 

selection_effect: Selection effect among native species.

species name: latin name of native species in each plot.

biomass: biomass of each native species originally planted in each plot.

4. Missing data codes:

N/A: not applicable

5 Script of R

library(nlme)

library(lme4)

library(lmerTest)

setwd("E:/wei.xue/Jiang Wang")

HDR <- read.csv("data_name.csv", header = TRUE, stringsAsFactors = TRUE)

head(HDR)

str(HDR)

HDR$Y <- as.factor(HDR$Y)

HDR$H <- as.factor(HDR$H)

HDR$richness <- as.numeric(HDR$richness)

HDR$A <- as.factor(HDR$A)

HDR$species.composition <- as.factor(HDR$species.composition)

####analysis of richenss effects

mod1 <- lme(biomass~ scale(richness,scale = FALSE)  + A + H +

              scale(richness,scale = FALSE):A + scale(richness,scale = FALSE):H + A:H +

              scale(richness,scale = FALSE):A:H

            ,random = ~ 1 | species.composition,

            na.action = na.exclude,

            data = subset(HDR, Y=="1"))# change "1" to "2" to test effects in year 2

summary(mod1)

anova(mod1)

intervals(mod1)

var <-mod1$apVar

var

par<-attr(var, "Pars")

par

#get variance explained by the random factor

vc<-exp(par)^2

vc

###analysis including diversity effects

mod3 <- lme(biomass~ H + A + complementarity.effect.transform + selection.effect.transform +

             H:complementarity.effect.transform + A:complementarity.effect.transform + H:selection.effect.transform + A:selection.effect.transform +

             H:A:complementarity.effect.transform + H:A:selection.effect.transform

            ,random = ~ 1 | species.composition,

            na.action = na.exclude,

            data = subset(HDR, Y=="1"))# change "1" to "2" to test effects in year 2

summary(mod3)

anova(mod3)

intervals(mod3)

var <-mod3$apVar

var

par<-attr(var, "Pars")

par

vc<-exp(par)^2

Vc

6. Phyton code

 # Import dataset

import pandas as pd

# file can be replaced with SEMdata2019.csv

file = r'F:\wj\20231003\SEMdata2018.csv' 

data = pd.read_csv(file)
data.head()

# correlations

corr_matrix = data.corr()

corr_matrix["invader_biomass"].sort_values(ascending=False)

# Spliting target variable and independent variables

X = data.drop(['处理','lable','year','invader_biomass','invader_invader_native'], axis = 1)

y = data['invader_biomass']

X.head()

# Splitting to training and testing data

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X,y, test_size = 0.3, random_state = 4)

# Import Random Forest Regressor

from sklearn.ensemble import RandomForestRegressor

# Create a Random Forest Regressor

reg = RandomForestRegressor()

# Train the model using the training sets

reg.fit(X_train, y_train)

# Model prediction on train data

y_pred = reg.predict(X_train)

from sklearn import metrics as skmet

# Model Evaluation

print('R^2:',skmet.r2_score(y_train, y_pred))

print('Adjusted R^2:',1 - (1-skmet.r2_score(y_train, y_pred))*(len(y_train)-1)/(len(y_train)-X_train.shape[1]-1))

print('MAE:',skmet.mean_absolute_error(y_train, y_pred))

print('MSE:',skmet.mean_squared_error(y_train, y_pred))

# Predicting Test data with the model

y_test_pred = reg.predict(X_test)

# Model Evaluation - Test set

acc_rf = skmet.r2_score(y_test, y_test_pred)

print('R^2:', acc_rf)

print('Adjusted R^2:',1 - (1-skmet.r2_score(y_test, y_test_pred))*(len(y_test)-1)/(len(y_test)-X_test.shape[1]-1))

print('MAE:',skmet.mean_absolute_error(y_test, y_test_pred))

print('MSE:',skmet.mean_squared_error(y_test, y_test_pred))

num_vars = len(X.columns)

num_vars

names = list(X.columns)

bounds = list(map(list,zip(X.min()-X.min()*0.95,X.max()+X.max()*0.95)))

names

# print("min:",X.min())

# print("max:",X.max())

# pip install SALib

from SALib.sample import saltelli

from SALib.analyze import sobol

from SALib.test_functions import Ishigami

import numpy as np

import SALib

# Define the model inputs

problem = {

    'num_vars': num_vars,

    'names': names,

    'bounds': bounds

}

# Generate samples

param_values = SALib.sample.saltelli.sample(problem, 2**14)

# Run model (example)

Y = reg.predict(param_values)

# Perform analysis

Si = sobol.analyze(problem, Y)

total_Si, first_Si, second_Si = Si.to_df()

total_Si, first_Si

import matplotlib.pyplot as plt

import matplotlib.colors as mcolors

import random

x=list(total_Si.index)

y=list(total_Si['ST'])  

 

colors=list(map(list,list(mcolors.cnames.items())[10:10+len(x)]))

colors=[i[1] for i in colors]

colors

# # for i in range(len(x)):

# #     colors.append()[random.randint(0,10)][1])

 

def autolabel(rects):

    for rect in rects:

        height = rect.get_height()

        print(height)

        plt.text(rect.get_x()+rect.get_width()/2.-0.2, 1.03*height, '%s' % round(height,2), size=10, family="Times new roman")

 

cm = plt.bar(x, y, color=colors,)  

plt.xticks(rotation=45)

autolabel(cm)

 

# # plt.grid(True,linestyle=':',color='r',alpha=0.6)

plt.show()

 

#################################

import shap

plt.figure()

shap.initjs()

model = reg

# explain the model

explainer = shap.TreeExplainer(model)

shap_values = explainer.shap_values(X[0:120])

# visualize the impact of each features

7. shap.summary_plot(shap_values, X[0:120])