Data and code from: Volatile organic compounds diversity mediates tree diversity-insect herbivory relationships

The effects of plant diversity on insect herbivory are inconsistent in the literature, highlighting the need for further mechanistic understanding. Plant volatile organic compounds (VOCs) are influenced by plant diversity and related to insect herbivory. However, the role of VOCs in mediating plant diversity effects on insect herbivory remains poorly explored. Here, we investigated chewing insect herbivory of 128,818 leaves from 2606 trees, measured leaf VOCs from 137 tree individuals, and assessed tree growth across a tree species richness gradient in a Biodiversity-Ecosystem Functioning experiment in southern China. We aimed to explore the mechanisms underlying tree diversity effects on insect herbivory by examining the roles of plant VOCs.

Methods
Based on the VOC profiles of the 137 samples [VOC_abundance_data. csv; group_info. csv], variations in the composition and concentration of samples’ VOCs across different plant diversity levels were evaluated by the permutational multivariate analysis of variance (PERMANOVA) and the principal coordinate analysis (PCoA). We first constructed a sample-VOC abundance matrix for all samples, where peak areas from chromatograms represented the abundance of each compound (Massad et al., 2017). Before analysis, we standardized the VOC abundance by dividing each emission by the row sum, converting the data to relative abundance. In the PERMANOVA, the standardized VOC matrix was considered as the response variable, while plant species richness and species identity as independent variables. The analysis was based on the Bray-Curtis dissimilarity and we set the number of permutations to 999. We used “vegan”  (Oksanen et al., 2022) and “ape” (Paradis & Schliep, 2019) packages to perform PERMANOVA and PCoA analyses separately. Additionally, we used a Mantel test to evaluate whether differences in VOCs between individuals were responsible for differences in their insect herbivory or RGR, using the same permutation number and distance matrix as in the PERMANOVA. To test whether trees with more distinct VOCs suffer greater insect herbivory, for each of the 137 sampled trees, we quantified its VOC distinctiveness as the mean Bray-Curtis distance between its VOC profiles and those of its neighbors, using the “vegdist” function in the “vegan” package. This metric represents the degree of VOC differentiation of an individual from its surroundings (Gaüzère et al., 2023). Finally, we used the generalized linear mixed effects model (GLMM) with a beta distribution and logit link to examine the relationships between insect herbivory and VOC distinctiveness, as the insect herbivory was proportional data. The model was fitted using the “glmmTMB” function in the “glmmTMB” package (Brooks et al., 2017). To account for spatial autocorrelation and the non-independence of tree individuals originating from the same plot or species, we included plot nested in block and species as random intercepts.

We constructed a binary species-VOC matrix by combining conspecific samples [VOC_SP_01_data. csv], where each row represents a species and each column a unique VOC. The matrix denotes the presence (1) or absence (0) of each VOC in a given species. Based on this matrix, we quantified the dissimilarity of VOCs between each pair of species using the Jaccard distance. We also calculated the relative abundance of each species within each plot [species distribution. csv]. Using these data, we calculated the community-level VOC diversity for all plots as Rao’s quadratic entropy index (Ricotta & Moretti, 2011; Salazar et al., 2016) with the “dbFD” function in the “FD” package (Grenié & Gruson, 2023).

To examine the effects of tree species richness, community VOC diversity and RGR on insect herbivory, we fitted a GLMM with a beta distribution and logit link [LAR and RGR data. csv]. In the model, we also included plot nested in block and species as random intercepts. This analysis included the insect herbivory and growth data of all blocks and plots. Insect herbivory was considered as the response variable, where tree species richness, community VOC diversity, their interaction term and RGR were included as independent variables. In addition, we fitted a linear mixed effects model (LMM) to investigate the effects of tree species richness and community VOC diversity on RGR  by the “lmer” function in the “lme4” package (Bates et al., 2015), which also included plot nested in block and species as random intercepts. Lastly, the relationship between plant species richness and community VOC diversity was also tested by an LMM, with VOCs diversity as the response variable and block as random intercepts.

We constructed structural equation modeling (SEM) using the “psem” function in the “piecewiseSEM” package (Lefcheck, 2016) to further elucidate the direct and indirect relationships among tree species richness, community VOC diversity, RGR, and insect herbivory [LAR and RGR data. csv]. We hypothesized that tree species richness would directly reduce insect herbivory, with these effects mediated by the changes in community VOC diversity and RGR along richness gradients (Appendix: Figure S2). In the SEM, we used GLMMs with a beta distribution and logit link to construct the pathways of tree species richness, community VOC diversity and RGR on insect herbivory, with plot nested in block and species as random intercepts. LMMs were used to construct the pathways of community VOC diversity on RGR, with random effects structured as in the GLMMs. We also used an LMM to analyze the effects of tree species richness on community VOC diversity, with block being random intercepts. All independent variables were z-scored standardized before model construction.

References

Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67: 1–48.

Brooks M E, Kristensen K, Benthem K J ,van, Magnusson A, Berg C W, Nielsen A, Skaug H J, Mächler M, Bolker B M. 2017. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal 9: 378.

Gaüzère P, Blonder B, Denelle P, Fournier B, Grenié M, Delalandre L, Münkemüller T, Munoz F, Violle C, Thuiller W. 2023. The functional trait distinctiveness of plant species is scale dependent. Ecography 2023: e06504.

Grenié M, Gruson H. 2023. fundiversity: a modular R package to compute functional diversity indices. Ecography 2023: e06585.

Lefcheck JS. 2016. piecewiseSEM: piecewise structural equation modeling in R for ecology, evolution, and systematics. Methods in Ecology and Evolution 7: 573–579.

Massad TJ, Martins de Moraes M, Philbin C, Oliveira C, Cebrian Torrejon G, Fumiko Yamaguchi L, Jeffrey CS, Dyer LA, Richards LA, Kato MJ. 2017. Similarity in volatile communities leads to increased herbivory and greater tropical forest diversity. Ecology 98: 1750–1756.

Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Solymos P, Stevens MHH, Szoecs E, et al. 2022. vegan: community ecology package.

Paradis E, Schliep K. 2019. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35: 526–528.

Ricotta C, Moretti M. 2011. CWM and Rao’s quadratic diversity: a unified framework for functional ecology. Oecologia 167: 181–188.

Salazar D, Jaramillo A, Marquis RJ. 2016. The impact of plant chemical diversity on plant–herbivore interactions at the community level. Oecologia 181: 1199–1208.