Main content area

Trait correlation network analysis identifies biomass allocation traits and stem specific length as hub traits in herbaceous perennial plants

Kleyer, Michael, Trinogga, Juliane, Cebrián‐Piqueras, Miguel A., Trenkamp, Anastasia, Fløjgaard, Camilla, Ejrnæs, Rasmus, Bouma, Tjeerd J., Minden, Vanessa, Maier, Martin, Mantilla‐Contreras, Jasmin, Albach, Dirk C., Blasius, Bernd
Thejournal of ecology 2019 v.107 no.2 pp. 829-842
allometry, carbon, cluster analysis, data collection, dry matter partitioning, economics, herbaceous plants, leaves, nitrogen content, perennials, phenotype, plant height, plant organs, reproduction, tissues, water uptake, Northern European region
Correlations among plant traits often reflect important trade‐offs or allometric relationships in biological functions like carbon gain, support, water uptake, and reproduction that are associated with different plant organs. Whether trait correlations can be aggregated to “spectra” or “leading dimensions,” whether these dimensions are consistent across plant organs, spatial scale, and growth forms are still open questions. To illustrate the current state of knowledge, we constructed a network of published trait correlations associated with the “leaf economics spectrum,” “biomass allocation dimension,” “seed dimension,” and carbon and nitrogen concentrations. This literature‐based network was compared to a network based on a dataset of 23 traits from 2,530 individuals of 126 plant species from 381 plots in Northwest Europe. The observed network comprised more significant correlations than the literature‐based network. Network centrality measures showed that size traits such as the mass of leaf, stem, below‐ground, and reproductive tissues and plant height were the most central traits in the network, confirming the importance of allometric relationships in herbaceous plants. Stem mass and stem‐specific length were “hub” traits correlated with most traits. Environmental selection of hub traits may affect the whole phenotype. In contrast to the literature‐based network, SLA and leaf N were of minor importance. Based on cluster analysis and subsequent PCAs of the resulting trait clusters, we found a “size” module, a “seed” module, two modules representing C and N concentrations in plant organs, and a “partitioning” module representing organ mass fractions. A module representing the plant economics spectrum did not emerge. Synthesis. Although we found support for several trait dimensions, the observed trait network deviated significantly from current knowledge, suggesting that previous studies have overlooked trait coordination at the whole‐plant level. Furthermore, network analysis suggests that stem traits have a stronger regulatory role in herbaceous plants than leaf traits.