Data from: Climate interacts with the functional trait structure of tree communities to influence forest productivity
Data files
May 24, 2024 version files 12.64 MB
-
README.md
-
RFdata_DupontLeducetal2024.csv
Abstract
Tree functional diversity can increase forest productivity by enhancing species interactions and providing greater growth stability. However, very few studies have examined the influence of tree community trait structure on survivor growth, recruitment, and mortality simultaneously, which are the main drivers of forest population dynamics. Here we explore the interactions among functional diversity, productivity, and climate to investigate the role of the trait structure of communities on forest productivity and to determine under what circumstances functional diversity should be promoted to ensure forest adaptive capacity under future climate. Using random-forest modeling and a network of permanent sample plots covering a broad gradient of climatic conditions, we isolated the effects of functional diversity—described as the distribution of trait values in a community—and climate variables on net forest productivity (NFP), survivor growth, recruitment, and mortality. Based on our findings, community-level trait structure affects forest productivity in different ways. NFP was influenced by three traits from three different plant strategy dimensions, whereas survivor growth and recruitment were strongly correlated with leaf and resource acquisition traits, and tree mortality with a mix of traits reflecting various plant strategies. We also observed climate interactions with the functional trait structure of tree communities. For instance, we observed an interaction between drought tolerance and mean annual temperature: at low temperatures, NFP biomass accumulation increased with the value of the drought tolerance trait; however, at higher temperatures, the opposite pattern was observed. However, we found contrasting patterns of population response to climate variability, depending on their functional diversity. Greater functional diversity does not necessarily increase biomass accumulation under different climatic conditions.
Synthesis. As all components of forest productivity contribute to NFP, studies on forest productivity should not only consider survivor growth but also recruitment and mortality. Each component responds differently in terms of biomass changes to climatic variation, according to the trait structure of tree communities. This study provides a framework to identify the trait structure that should be targeted under different climate scenarios to anticipate change and help strengthen forest response capacity to climate change.
README: Data from: Climate interacts with the functional trait structure of tree communities to influence forest productivity
https://doi.org/10.5061/dryad.n5tb2rc41
No local data collection was conducted in this study, as it relied on a meta-analysis of secondary data. The team of authors and collaborators, however, is representative of the main regions covered by the meta-analysis, ensuring appropriate interpretation of the data and results from the studied regions.
Data were derived from the following resources available in the public domain: Québec permanent sample plot forest inventory (https://www.donneesquebec.ca/recherche/dataset/placettes-echantillons-permanentes-1970-a-aujourd-hui/resource/ccf8d0d7-85fe-49b0-a965-e914b2395fb7, Newfoundland and Labrador Department of Fisheries, Farming and Natural Resources (https://www.gov.nl.ca/ffa/programs-and-funding/forestry-programs-and-funding/managing/inv-plan/. Functional diversity indices and traits data were derived from the TOPIC database with the permission from TOPIC (Aubin et al., 2020). Traits data from Paquette and Messier (2011) were also used. The data derived from other sources can be published under CC0.
Description of the data
RF modeling was performed using the recursive feature elimination (rfe) algorithm from the caret package (Kuhn, 2020) in R. The RF was calculated to find the best subset of predictors—from a model having a single explanatory variable to a model using all explanatory variables—that generates the lowest root mean square error (RMSE) with a tolerance of 3% (3% more error than the model with the lowest RMSE) and with 50 repetitions of 10-fold cross-validation. We used this set of 37 explanatory variables describing forest structure, environment, and trait structure to predict NFP and the aboveground biomass increment attributed to survivor growth, recruitment, and mortality. The optimal model was recalibrated using the randomForest package (Liaw and Wiener, 2002) in R, with the number of features sampled at each split set to the default value for regression (i.e., p/3 where p is number of variables) and the number of random trees set to 20,000.
Explanatory variables | Description |
---|---|
ID_PE_MES | Plot id and no measurement |
Biomass_NetP | Net forest productivity biomass increment (Mg ha−1 yr−1) |
Biomass_Grow | Survivor growth biomass increment (Mg ha−1 yr−1) |
Biomass_Mort | Tree recruitment biomass increment (Mg ha−1 yr−1) |
Biomass_Recr | Tree mortality biomass loss (Mg ha−1 yr−1) |
Biomass_cov | Plot total biomass (Mg ha−1) |
BAmerch | Basal area (m2 ha−1) |
DomHeight | Dominant height (m) |
Compo_DomSp | Dominant species |
Compo_prop | Basal area of the dominant species (m2·ha−1) |
TPI_20m_5c | Topographic position index |
TWI_20m | topographic wetness index |
Pmean_int | Mean total annual precipitation (mm) |
Tmean_int | Mean annual temperatures (°C) |
Tmax_int | Maximum annual temperatures (°C) |
Pmin_int | Minimum total annual precipitation (mm) |
nbsp | Species richness (N) |
shannon_true | True Shannon-Wiener diversity index |
simpson | Gini–Simpson index calculated with all functional traits |
FDis | Functional dispersion index calculated with all functional traits |
FGR | Functional group richness calculated with all functional traits |
CMW_SeFreq | Community-level weighted means of the SeFreq trait |
CMW_TolD | Community-level weighted means of the TolD trait |
CMW_LMA | Community-level weighted means of the LMA trait |
FDis_Pb | Functional dispersion of the Pb trait |
FDis_Nmass | Functional dispersion of the Nmass trait |
FDis_WDR | Functional dispersion of the WDR trait |
simps_tree | Gini–Simpson index based on the mix of four traits from the tree stature group (maxH, GR, WD, WDR) |
FDis_leaf | Functional dispersion index of four traits from the leaf group (LL, LMA, Nmass, LS) |
raoQ_Eco506 | Rao’s quadratic entropy index based on the mix of three traits from the plant strategy dimensions (EM, FrostFMin, SeM) |
raoQ_Eco242 | Rao’s quadratic entropy index based on the mix of three traits from the plant strategy dimensions (Nmass, RainMin, SeM) |
simps_Eco1 | Gini–Simpson index based on the mix of three traits from the plant strategy dimensions (LS, maxH, Veg) |
redon_Eco5 | Functional redundancy index based on the mix of three traits from the plant strategy dimensions (LS, maxH, SeFreq) |
FDis_Eco296 | Functional dispersion index of three traits from the plant strategy dimensions (GR, TolW, AsexM) |
raoQ_Eco277 | Rao’s quadratic entropy index based on the mix of three traits from the plant strategy dimensions (GR, TolS, SeFreq) |
FDis_Eco457 | Functional dispersion index of three traits from the plant strategy dimensions (EM, WDR, Veg) |
FDis_Logi310 | Functional dispersion index of five traits from the functional group dimensions (maxH, SeFreq, TolD, RootD, LL) |
simps_Logi1 | Gini–Simpson index based on the mix of five traits from the functional group dimensions (maxH, Veg, TolS, AM, LS) |
redon_Logi1252 | Functional redundancy index based on the mix of five traits from the functional group dimensions (WD, SeM, TolW, AM, Nmass) |
redon_Logi394 | Functional redundancy index based on the mix of five traits from the functional group dimensions (maxH, SeOptP, TolW, RootD, LL) |
raoQ_Logi354 | Rao’s quadratic entropy index based on the mix of five traits from the functional group dimensions (maxH, SeFreq, FrostFMin, EM, LL) |
FDis_Logi2106 | Functional dispersion of five traits from the functional group dimensions (WDR, SeOptP, TolD, EM, LL) |
Methods
No local data collection was conducted in this study, as it relied on a meta-analysis of secondary data. The team of authors and collaborators, however, is representative of the main regions covered by the meta-analysis, ensuring appropriate interpretation of the data and results from the studied regions.
Data were derived from the following resources available in the public domain: Québec permanent sample plot forest inventory (https://www.donneesquebec.ca/recherche/dataset/placettes-echantillons-permanentes-1970-a-aujourd-hui/resource/ccf8d0d7-85fe-49b0-a965-e914b2395fb7), Newfoundland and Labrador Department of Fisheries, Farming and Natural Resources (https://www.gov.nl.ca/ffa/programs-and-funding/forestry-programs-and-funding/managing/inv-plan/). Functional diversity indices and traits data were derived from the TOPIC database with the permission from TOPIC (Aubin et al., 2020). Traits data from Paquette and Messier (2011) were also used.