Rapid structural and compositional change in an old-growth subtropical forest: using plant traits to identify probable drivers

Abstract

Recent studies have shown directional changes in old-growth tropical forests, but changes are complex and diverse, and their drivers unclear. Here, we report rapid net structural and compositional changes in an old-growth subtropical forest and we assess the functional nature of these changes to test hypothetical drivers including recovery from past disturbances, reduction in ungulate browsing, CO2 fertilization, and increases in rainfall and temperature. The study relies on 15 years of demographic monitoring within 8 ha of subtropical montane forest in Argentina. Between 1992 and 2007, stem density markedly increased by 50% (12 stems ha(-1) y(-1)) and basal area by 6% (0.13 m(2) ha(-1) y(-1)). Increased stem density resulted from enhanced recruitment of understory treelets (Piper tucumanum, Eugenia uniflora, Allophylus edulis) into small size classes. Among 27 common tree species, net population growth was negatively correlated with maximum tree size and longevity, and positively correlated with leaf size and leaf nutrient content, especially so when initial population size was controlled for. Changes were inconsistent with predictions derived from past disturbances (no increase in shade-tolerant or long-lived late-succesional species), rainfall or temperature increase (no increase in evergreen or deciduous species, respectively). However, the increase in nutrient-rich soft-leaved species was consistent with exclusion of large herbivores two decades before monitoring started; and CO2 fertilization could help explain the disproportionate increase in small stems. Reductions in populations of large vertebrates have been observed in many otherwise undisturbed tropical forests, and our results suggest they can have important structural and functional repercussions in these forests.

Figure 1. Rainfall and temperature records registered at Obispo Colombres Meteorological Station, Tucumán.

a) Five year moving averages of annual rainfall (solid line), rainy season rainfall (October – March, dotted line) and dry season rainfall (April – September, dashed line) from 1900 to 2010; b) annual average of maximum and c) minimum temperatures during October – March (summer, dotted line), and April – September (winter, solid line) from 1972 to 2010. Rainfall and temperature data were registered at Obispo Colombres Meteorological Station, Tucumán (ca.500 m lower and 15 km east of the study site).

Figure 2. Structural changes in 15 years.

a) Changes in density of individuals, b) density of stems, and c) basal area in all four census. Letters identify non-different groups from Friedman ANOVA and posteriori Wilcoxon match pair tests. Error bars correspond to standard error. All data are the per-hectare average of 8 ha.

Figure 3. Basal area for 1992 and 2007 across 15(numbers indicate the upper limit of the class).

Figure 4. Changes in stem density and basal area across species in 15 years.

Slope of the regression (rate of change) calculated for a) stem density, and b) basal area for each species across 15 years. Error bars correspond to standard error.

Figure 5. Relationship between stem density changes in 15 years and species life histories and morphological traits.

Coplots between the trend of change (b) in stem density and demographic and morphological principal components (PC), for low (left) and high (right) initial stem density in 1992. Plotted values are ranks. To aid interpretation, ranked axes were rescaled so that negative b in stem density matched with negative ranks. Correlations in black correspond to partial Spearman correlations; values in grey represent standard Spearman correlations (that did not control for initial stem density). Regression lines are for graphical reference only. See Table S3 for species code abbrevattions.

Figure 6. Relationship between area basal changes in 15 years and species life histories and morphological traits.

Coplots between the trend of change (b) in basal area and demographic and morphological principal components (PC), for low (left) and high (right) initial basal area in 1992. Plotted values are ranks. To aid interpretation, ranked axes were rescaled so that negative b in basal area matched with negative ranks. Correlations in black correspond to partial Spearman correlations; values in grey represent standard Spearman correlations (that did not control for initial basal area). Regression lines are for graphical reference only. See Table S3 for species code abbrevattions.