Biodiversity
Measures of species, functional, phylogenetic, and structural diversity were each represented by three to five diversity indices commonly found in the literature. The indices selected represented either the richness, evenness, or the variation of each diversity measure. Every index was calculated using relative basal area (BA) of the species within a plot from either the 2001-2003 Bartlett Experimental Forest (BEF) cruise plot inventory (Leak et al. 2017) or the 1995-1998 Hubbard Brook Experimental Forest (HBEF) valley plot inventory (Battles and Fahey 2019). Indices were calculated using packages from R version 3.6.0 (R Core Team 2019).
Species Diversity
Species diversity was represented using three indices: the Shannon’s entropy index (H’; Shannon 1948), species richness (S; McIntosh 1967), and species evenness (J; Pielou 1966). H’ was calculated using the fractional basal area of each species. S was determined by counting the number of unique species found on a plot and J was calculated using Pielou’s evenness index.
Functional Diversity
Functional diversity was represented by five indices: functional richness (FRic), evenness (FEve), divergence (FDiv; Villéger, Mason, and Mouillot 2008), dispersion (FDis; Laliberté and Legendre 2010), and Rao’s Quadratic Entropy (RaoQ; Botta‐Dukát 2005). Indices were calculated using the R package FD (v. 1.0.12; Lalibert, Legendre, and Shipley 2014).
Seven species-specific functional traits that link plant functioning to the availability of light, water, and nutrients were selected to calculate the functional diversity indices. Six of these traits: mycorrhizal fungi type, leaf type, xylem type, waterlogging tolerance, drought tolerance, and shade tolerance, were obtained from the TRY plant database (Kattge et al. 2020). Values for the seventh trait, mean, mass-based foliar nitrogen (N), were obtained from the Northeastern Ecosystem Research Cooperative (NERC) database (Northeastern Ecosystem Research Cooperative (NERC) 2010).
Phylogenetic Diversity
Five indices were selected to represent phylogenetic diversity: Faith’s phylogenetic diversity index (PD; Faith 1992), phylogenetic species variability (PSV), phylogenetic species richness (PSR), phylogenetic species evenness (PSE), and phylogenetic species clustering (PSC; Helmus et al. 2007). To calculate these indices, a dendrogram was created based on the species’ taxonomic rankings, and branch lengths were given a value of one unit per segment. These indices were calculated using the R package picanate (v.1.8; Kembel et al. 2019).
Structural Diversity
Several indices were used to characterize structural diversity. For both sites, structural diversity was represented by the variation in DBH based on measurements from the inventory data. Indices used include the quadradic mean of DBH (DBHq), standard deviation of DBH (DBHsd), the Shannon’s entropy index (DBH_H’), richness (DBH_S), and evenness (DBH_J). The DBHq represents the average stand diameter per plot (Curtis and Marshall 2000; Storch, Dormann, and Bauhus 2018) while the DBHsd represented the variation in diameter (Storch, Dormann, and Bauhus 2018). For DBH_H’, the number of species was replaced with the number of DBH classes grouped into 3 cm bins while proportions were based on the basal area per DBH class. Richness was based on the number of DBH classes and evenness of the DBH classes was calculated using Pielou’s evenness (Park et al. 2019).
For BEF, additional indices of structural diversity were calculated from measurements of height and canopy surface roughness using discrete return Light Detection and Ranging (LiDAR) data collected in the summer of 2017 by the National Ecological Observatory Network (NEON). LiDAR points were returned at a density of 1-4 points/m2 and processed to create a 1 m resolution canopy height model (CHM). Indices derived from the LiDAR data include standard deviation of height (ht_sd), canopy surface roughness (rumple; Parker and Russ 2004; Kane et al. 2010), the Shannon’s Entropy Index (ht_H’), richness (ht_S), and evenness (ht_J). LiDAR-based structural diversity indices were not calculated for HBEF, as comparable leaf-on LiDAR data collected around the same time frame as the BEF LiDAR was not available. Similar to DBH_H’, ht_H’ was calculated using 3 m bin height classes from the LiDAR points. Lastly, rumple was calculated from the CHM using the R package lidR (Roussel et al. 2019) and represents the ratio between the canopy surface area to the ground surface area (Kane et al. 2010).
Productivity
Forest productivity was calculated using sequential measurements of tree diameter on a subset of 38 plots at BEF and 18 plots at HBEF. The subset of plots was chosen for productivity because they were measured at a high temporal frequency and included (1) individually tagged trees (which allowed determination of individual tree growth and mortality) and (2) additional plant trait data, such as mass-based foliar nitrogen (N). Foliar N measurements were conducted using methods described in Ollinger et al., (2008). Subset plots were previously chosen and measured by Smith et al., (2002) to collect repeat wood growth measurements. DBH measurements for every species, living and dead, were converted to wood biomass using species specific allometric equations (Jenkins 2004). Woody biomass was scaled up to the plot level by summing the biomass of all species on a plot and dividing by plot area, making sure to carry over trees that died between remeasurement periods. Finally, wood growth was calculated by subtracting the previous year’s measurements from the most recent measurements and dividing by the number of years between measurement periods.
Site Characteristics
In addition to the biodiversity indices and productivity estimates, some site characteristics were also generated for the sites. From a digital elevation model (DEM) and data products developed by Fraser, McGuire, and Bailey (2019), measures of elevation, aspect, and maximum slope were extracted for each of the plots. Forest type was determined by using species composition information from inventory data, where species type was used to determine the percentage of broadleaf species on a plot. Plots dominated by broadleaf species (≥ 80% broadleaf coverage) were classified as broadleaf while plots dominated by conifer species (≤ 20% broadleaf coverage) were classified as evergreen. Plots classified as mixed forest contained a combination of conifer and broadleaf species (20% < x > 80% broadleaf coverage).
For BEF only, management history for the plots was categorized into two classes: managed and unmanaged. These classifications were based on the compartment acreage treatment list provided in the metadata of the inventory data. Plots classified as unmanaged were in compartments that have not been manipulated by any human activity, while managed plots were any plots that have been exposed to at least one treatment (i.e., shelterwood, clear cut, or patch cut).