Long-term trends in tropical tree growth: a pantropical study

Tropical forests cover only 7% of the earth’s land surface, but harbour almost half of the world’s biodiversity. These forests also provide many ecosystem services, such as the storage of carbon and the regulation of local and regional climate, and many goods such as timber and fruits. Furthermore, tropical forests contribute disproportionately to the global carbon cycle, storing an estimated 25% of the carbon stocks on land and accounting for a third of the terrestrial net primary productivity. Therefore, changes in forest cover or in the net uptake or loss of carbon by forests directly influences the global carbon cycle. Tropical forests are under increasing anthropogenic pressure and are undergoing rapid changes due to deforestation, conversion to other land uses and logging. Additionally, there is evidence that pristine and intact tropical forests are undergoing changes due to the effects of climate change. Concerted increases in biomass and tree growth have been found in studies monitoring intact tropical forests, suggesting that they acted as considerable carbon sinks over the past decades. On the other hand, decreasing or fluctuating forest growth and biomass have also been noted. These different changes have been attributed to different climatic drivers: growth increases have been suggested to arise from growth stimulation by increasing atmospheric CO2 concentrations, while growth decreases have been interpreted to reflect the limiting effects of increased temperature on growth. As monitoring plots usually cover only a few decades, it is still unclear whether these changes are pervasive or whether they simply reflect the effect of short-term climatic fluctuations on tree growth. Assessing whether changes have occurred over centennial scales is thus crucial to understanding whether and how tropical forests are reacting to climatic changes. In this thesis we apply tree-ring analysis on a pantropical study to assess longterm changes in growth of tropical forest trees. Tree-ring analysis was used to measure long-term growth rates of ~1350 trees of different species coming from three sites across the tropics. Trends in growth over the last two centuries were then analysed using an established an a new trend-detection method. Additionally, we applied the long-term growth data from rings to improve the evaluation of forest management practices in Cameroon. All samples were collected and measured within the TROFOCLIM project led by Pieter Zuidema. The project also includes two other PhD theses and sample collection was divided among the three PhD projects and the three sites: in Bolivia (samples collected by Peter van der Sleen), Cameroon (by me) and in Thailand (by Mart Vlam). The main objectives of this thesis were: (1) to assess the potential for using treerings in a wet tropical forest in Central Africa; (2) to project timber yields in the next logging round for four Cameroonian tree species; (3) to evaluate the sensitivity and accuracy of four commonly used methods to detect long-term trends in tree-ring data; and (4) to detect whether growth rates of tropical forest trees have changed over the past ~150 years. In Chapter 2 of this dissertation, we evaluated whether growth rings are formed annually in the wood of tree species growing under very high levels of precipitation (>4000 mm) in an African tropical forest. For this purpose, we assessed whether ring structures are formed in the wood of the 22 commercially exploited tree species and found that ring structures are indeed formed by more than half of these species (in 14 species), though with varying ring clarity. On four species we proved the annual character of ring formation using radiocarbon bomb-peak dating. That rings are formed under such high levels of precipitation is surprising, as these conditions are considered improper to induce ring formation. These results suggest that the potential of tree-rings analysis is more or less similar across the tropics. Based on our results and that of other studies, we estimate that tree rings can be used to measure tree growth and ages for around a quarter to a third of tropical tree species. Worldwide, over 400 million hectares of tropical forests are set aside for timber production. Attaining sustainable use of these forests is very important, in the light of the important role of tropical forests in retaining biodiversity and storing carbon. Ensuring that timber species are not overexploited is key to ensure that forest use is sustainable and entails finding a balance between economic gains and the (ecological) sustainability of logging operations. In Chapter 3, we integrated growth data from tree-rings with logging inventory data to forecast whether timber yields can be sustained in the next harvest round for four timber species in Cameroon. Under current logging practices, future logging yields were predicted to reduce moderately to strongly for all species. These yield reductions are worrisome for forest conservation, as loss of economic value may lead to conversion of forests to other land uses. We recommend that such calculations are needed for more species and argue that these simulations should include the effects of logging and eventual silvicultural measures on the growth and survival of trees. Lifetime tree growth data – as acquired by tree-ring analysis – contains longterm trends in growth that reflect the ontogenetic development of an individual or species, i.e., these data contains an age/size signal in growth. In Chapter 4 we evaluate the sensitivity, accuracy and reliability to detect long-term trends in growth of four methods that are commonly used to disentangle these age/size trends from long-term growth trends. We applied these growth-trend detection methods to measured growth data from tree rings and to simulated growth trajectories on which increasing an decreasing trends were imposed. The results revealed that the choice of method influences results of growth-trend studies. We recommend applying two methods simultaneously when analysing long-term trends – the Regional Curve Standardization and Size Class Isolation – as these methods are complementary and showed the highest reliability to detecting long-term growth changes. In Chapter 5, we analysed long-term growth trends in tropical forest trees using a pantropical approach applying the two recommended growth-trend detection methods. We showed that growth rates for most of the 13 tropical tree species, from the three sites across the tropics, decreased over the last centuries. These species-level changes may have important demographic consequences and may eventually lead to shifts in the species composition of tropical forests. We found no strong growth changes when analysing trends aggregated per site or across sites: only weak growth reductions were detected for the Thai site and across sites. These findings contrast growth increases that would be expected if tree growth is stimulated by increased ambient CO2. These growth reductions suggest worsening growth conditions for several tropical tree species, and could result from the negative effect of temperature increases on tree growth, or reflect the effect of large-scale disturbances on these forests. If one image becomes clear from this thesis it is that long-term data are crucial to enhance the management of tropical forests and to quantify changes happening in these forests. Tree-ring analysis provides this long-term perspective for tree growth and is thus a great tool to evaluate changes in the growth of trees, including for tropical species. One of the most important finding of this thesis is that many tropical species show long-term decreases in growth. These results suggest that the commonly assumed growth increases tropical forests, based on measurements over the last couple of decades, may be incorrect. This discrepancy in results could have strong consequences, among others leading to erroneous predictions of the carbon dynamics of tropical forests under future climate change. Combining monitoring plot data (to analyse short-term changes in growth and species composition) with remotely sensed data (to accurately determine forest land cover) and with the long-term growth data from tree rings is probably the best way forward to relate recent findings of short-term changes in tree growth and forest biomass to changes over the past centuries. Such integrative approaches are needed to better quantify and understand the effects of climate change on tropical forests.