A comparison of growth, structure and diversity of mixed species and monoculture reforestation systems in the Philippines
Le, Dinh Hai; Herbohn, John; Smith, Carl; Nguyen, Huong (2020), A comparison of growth, structure and diversity of mixed species and monoculture reforestation systems in the Philippines, Dryad, Dataset, https://doi.org/10.5061/dryad.2547d7wnd
Forests in the Philippines, and many other developing countries in the tropics, have been extensively cleared over recent decades. There have been increasing efforts to reforest these cleared lands to achieve both socio-economic and environmental objectives. To date, planted forests have been dominated by monocultures. There has however been increasing calls to use mixtures of species, although there is limited evidence to support mixed species plantations being a better or win – win approach to reforestation. To address this, we compared the tree growth, forest structure and tree species diversity performance of monoculture and mixed species tree plantations across 168 sites (251 survey plots) on Leyte Island, the Philippines. Our results indicate that mixtures of fast-growing exotics species had better growth performance compared to monocultures of fast-growing exotics species, and also better tree species diversity performance at both the plot and landscape scale. Our results suggest that mixtures of exotic or native species can provide benefits over monoculture plantations. Mixtures of productive exotic species are most suited to situations where the production function of the forest is most important, while mixtures of native species are most suited to situations where the biodiversity function of the forest is most important.
Data collection methods
We collected data from a circular plot located within the center of at least one forest block within each of the MONO, MIS and MNS sites (for some sites data was collected from only one plot, while for large plantations there were several forest blocks, so data was collected from several plots). Each circular plot had a 5m radius, however if a plot did not contain at least seven trees with a diameter at breast height (DBH) of at least 10cm then the plot radius was extended to 10m. For each plot, canopy cover (% projective foliage cover), understorey and ground cover (%), diameter at breast height (DBH cm) of all trees ≥5cm DBH, total height (Ht m) of all trees ≥10cm DBH, height of tallest tree (m) and tree species name and age of all trees were recorded. Projective foliage cover (PFC) was measured at the center of each plot using digital photos (Kanowski et al., 2008). Only vegetation >2m above ground level was included in PFC measurements. Tree DBH was measured using a DBH tape and tree heights were measured using a digital hypsometer (LaserAce 150 Hypsometer). Understorey, shrub, vine and scrambler, coarse woody debris and litter cover was measured at three 1x1m quadrats within each plot, located at the center of the plot and 4m either side of the center along a transect.
Within the natural forest reference site, we collected data within a one hectare (10,000m2) permanent plot that contained 100 subplots, each 100m2. The permanent plot was located at least 100m from the edge of the Mt. Pangasugan rainforest reserve. DBH of trees ≥5cm DBH, Ht of trees ≥10cm DBH, height of tallest tree (m) and tree species name were recorded for all trees in all 100m2 subplots. Two opposing diagonal transects running through the center of the permanent plot were used to measure PFC, understorey, shrub, vine and scrambler, coarse woody debris and litter cover. PFC was measured at the center of each 100m2 subplot along the diagonal transects using digital photos. Understorey, shrub, vine and scrambler, coarse woody debris and litter cover was measured within one 1m2 quadrat located at the center of each 10x10m subplot along the diagonal transects.
Data analysis methods
We calculated a number of indices to compare forest growth performance, forest structure and tree species diversity between the different plantation types (see Table 4 for a description of the indices). These selected indices are based on the conceptual model for assessing reforestation success in tropical developing countries developed by Le et al. (2012). For MONO, MIS and MNS, these indices were calculated for each circular plot and extrapolated for a hectare, however, for the natural forest reference site these same indices were calculated for the entire one hectare permanent plot (not for the subplots).
The indices were analyzed using mixed effects models developed in IBM SPSS Statistics 20 (2011) software. We modelled the indices as a function of revegetation method (MONO, MIS, MNS), and stand age class (≤10, 11-19 and ≥20 years) (fixed effects) with a random effect structure of site. Prior to modelling, we used diagnostic plots and tests of normality (Kolmogorov-Smirnov test) to check if the indices of continuous variables satisfied the assumption of normality (Pinheiro & Bates, 2000). We found several indices were not normally distributed (DBH, basal area (BA), total volume (TV), above ground biomass (AGB), mean annual increment (MAI) DBH, MAI Ht, MAI BA, MAI TV, MAI AGB and density of woody stems in the classes DBH≥5cm, DBH 5-20cm and Ht 2-10m) and data were subsequently transformed into normal distributions. After transformation, we used a linear regression for the target distribution in the mixed effect models. For indices of ordinal variables (grass, herb, fern, litter, shrub, vine and scrambler, and coarse woody debris cover indices) we used a multinomial logistic regression for the target distribution and a cumulative logit link function in the mixed effects models. To compare the predictive power of the models with different fixed effects we used information-theoretic model selection procedures (∆AIC) (Burnham & Anderson, 2002) (for details see Supplementary Tables S1 and S2).
We also conducted pair-wise comparisons (using Hochberg’s GT2 test because the sample sizes for subgroups are very different) (Field, 2017) on the mixed effects models to identify statistically significance differences (level of rejecting the null hypothesis was fixed at α = 0.05) in forest growth performance, forest structure and tree species diversity between the three revegetation methods (MONO, MIS, MNS) across three age classes (≤10, 11-19 and ≥20 years). Forest structure, tree species diversity and some forest growth performance indices were also compared with the natural forest reference site.