Patterns and processes in tropical forests: an analysis of forest structure and function from long-term permanent plots in Venezuela, Northern South America

Abstract

The differential responses of trees to climate and resource gradients, coupled with historical and current disturbances affect forest dynamics and creates complex landscape patterns in forest assemblages and structures. Long-term data collected from a wide range of forest conditions provides an excellent opportunity to examine the underlying processes behind these patterns, which ultimately drive variations in species diversity, aboveground carbon stocks and forest dynamics in general. In addition, long-term data contains valuable information on the potential effects that climate change may have for terrestrial ecosystems. Thus, management policies, including conservation and restoration might be largely benefited from the results of this type of research. This study is based on long-term data collected from tree populations in different forest-types with contrasting environmental conditions in Venezuela, Northern South America. The Instituto de Investigaciones para el Desarrollo Forestal (INDEFOR) affiliated to Universidad de Los Andes in Mérida-Venezuela, maintains the longest-running monitoring plots in mature forests anywhere in Latin America and perhaps the tropics, including some sites that were established more than half a century ago. Since 2004, I have been part of four field campaigns (2004, 2009, 2012, 2016) to continue the monitoring of these sites. The information collected is now part of a larger network of similar sites under the umbrella of the Amazon Forest Inventory Network (RAINFOR project) where data has been stored and curated into the Forestplots platform to serve the global scientific community with relevant information from multiple tropical sites in Asia, Africa and South America. Three main studies compose this dissertation project, all connected through a common theme with a special focus on stem turnover, growth, productivity, and carbon cycle. This dissertation broadly tried to answer the following questions:  What are the patterns of tree turnover (i.e. recruitment and mortality) across different forest types?  What are the factors controlling Woody productivity (AGWP) across a diverse environmental gradient?  What are the relationships between turnover, productivity and aboveground biomass (AGB)?  Have turnover rates and AGWP changed over the last three to four decades?  What are the principal causal mechanisms behind the observed trends in turnover rates and AGWP?  How predictable are these processes using species functional traits? Part 1: Data from 50 long-term permanent plots from across Venezuelan forests was used to describe the patterns of stem turnover, aboveground biomass (AGB) and woody productivity (AGWP) over contrasting tropical ecological conditions, and the relationships between them and potential climatic drivers. I used a principal component analysis coupled with linear and mixed effects models to analyze the relationship between climate, forest structure and stem dynamics. Two major axes aligned with orthogonal temperature and moisture gradients effectively described more than 90% of the environmental variability in the dataset. Average turnover was 1.91 ± 0.10 % year-1 with mortality and recruitment being almost identical, and close to average rates for other mature tropical forests. Nevertheless, dynamic rates were significantly different (p < 0.001) among regions, with the lowland forests in Western alluvial plains being the most dynamic, while Guiana Shield forests are showing the lowest turnover rates. A weak positive relationship between AGB and AWGP was found with Guiana Shield forests having, on average, the highest values for both variables (204.8 ± 14.3 Mg C ha-1 and 3.27 ± 0.27 Mg C ha-1 year-1), but AGB was much more strongly and negatively related to stem turnover. Moisture was a fundamental driver of turnover, with longer dry seasons apparently favoring greater rates of tree turnover and thus lower biomass. Regional variation in AGWP among Venezuelan forests strongly reflects resource availability, with greatest AGWP when both precipitation and temperatures are high. Overall, forests in wet, low elevation sites with low turnover, stored the greatest amounts of biomass carbon. While faster dynamics are closely associated with lower carbon storage, turnover rates and woody productivity were weakly correlated, indicating that stem dynamics and carbon dynamics might be decoupled from one another. Part 2: Long-term data from 44 plots was used to analyze the temporal dynamics of stem turnover (i.e. mortality and recruitment), aboveground biomass (AGB) and woody productivity (AGWP) across different ecological regions in Venezuela over the last four decades. By means of generalized additive models to analyze temporal trends and linear models to test for a significant change in slopes, I found that 21 plots had a positive slope for recruitment rates (average slope 0.02 % year-1, p < 0.001), while 32 plots showed a positive slope of mortality over time (average slope 0.02 % year-1, p < 0.001). Regardless of region, elevation, soil fertility or climate seasonality, the increase in tree mortality was highly significant (p < 0.001). A significant increase in the amount of biomass from dead trees (average AGBloss slope = 0.02 Mg C ha-1 year-2, p < 0.001) was also found combined with a modest but steady increase in AGWP (average AGWP slope = 0.02 Mg C ha-1 year-2, p < 0.001). Although a decline in AGB change was found, the magnitude was not sufficient to significantly affect the forest carbon sink (average slope for net AGB change = - 0.003 Mg C ha-1 year-2, p < 0.117). Using generalized mixed effects models, I detected that for the monitoring period of each plot, competition factors appears to be more relevant explaining observed trends in recruitment rates, while climate, particularly the steady and significant increase in mean annual temperatures was largely responsible for the increase in mortality rates. There was a clear regional effect with each driver contributing in different proportions to the observed trends in turnover, AGWP and net biomass change. The results of this study highlight the importance of including stand development and competition factors to improve our understanding of forest dynamics over time. This study can be used as an input for other more complex models aiming the analysis of effects of climate change on forest ecosystems in the tropics. Part 3: The objectives of this study were: 1) to describe the functional trait space in two contrasting tropical environmental settings in Venezuela; 2) analyze the relationships between relative growth rates (RGRs) and the average traits values for different tree species; 3) analyze the potential links between the community weighted means (CWMs) of traits with four main response variables at the community scale: rates of recruitment (r) and mortality (m), aboveground carbon (AGB) and woody productivity (AGWP). For this, I employed a principal component analysis combined with linear mixed models and data from 10 functional leaf and stem-level traits that are known to be important attributes of tree species. Additionally, four functional diversity indexes, soil nutrient content and chemistry, along with plot stem density were included into the analysis as potential predictors. I used data from long-term permanent sample plots in two contrasting mature forest sites in western Venezuela. I test the existence of a slow–fast continuum where lowland and fertile forests would be generally dominated by acquisitive traits (e.g. high Specific Leaf Area – SLA) influencing a much faster dynamics and higher productivity, while high-elevation forests can be better described in terms of conservative traits (e.g. high leaf mass per area - LMA) driving lower turnover rates and productivity but a higher AGB. Results show that both sites are noticeably different (p < 0.001) with regards to their structure, dynamics and function, with lowland forests having higher annual recruitment (2.67 ± 0.55 % year-1 – Mean ± Standard deviation), mortality (3.09 ± 0.63 % year-1), and AGWP (2.95 ± 0.63 Mg C ha-1 year-1), but overall lower AGB (130.99 ± 44.71 Mg C ha-1) when compared to montane forests (r = 2.06 ± 0.26 % year-1; m = 1.59 ± 0.21 % year-1; AGWP = 2.39 ± 0.36 Mg C ha-1 year-1; AGB = 178.49 ± 27.95 Mg C ha-1). The species’ traits in montane forests were mostly associated with adaptation to high soil moisture, greater shade tolerance and thus lower growth rates (e.g. high leaf thickness, high LMA, low foliar nutrient content), whereas for lowland species, traits were more associated with higher resource availability (e.g. high SLA, high nutrient concentration in leaves). I found that different CWMs of traits have effects over different responses with CWM of leaf nitrogen simultaneously involved in all of the studied response variables. Yet, functional traits were more important when predicting AGB and AGWP compared to stem turnover highlighting the importance of other factors such as soil fertility and environmental conditions when studying stand dynamics. Venezuela is one of the most biologically diverse countries on earth. Yet, a high proportion of its ecosystems, especially forests, are threatened by deforestation and degradation that may compromise the services these areas currently provide while enhancing climate change through the emissions of carbon. The findings of this project represent a body of knowledge regarding the ecology and dynamics of different forest-types, some of which seem to be already experiencing the effects of increasing temperatures resulting in higher rates of tree mortality and a reduction in productivity. However, in some cases, these areas are not only still acting as carbon sinks mitigating the effects of climate change, but they also constitute large reservoirs of carbon and habitat for biodiversity making them highly important for a sustainable future. I strongly believe that the results presented in this dissertation project may have important connections and implications to initiatives such as reducing carbon emissions from deforestation and forest degradation, sustainable forest (or land) management for timber and nontimber forest products and increasing forest cover by forest landscape restoration. Each chapter in this dissertation contains information to improve decision making processes at different levels, and I see the results and conclusions of my research as important elements for a better future of Venezuelan tropical forests, its research community and overall the country.