Complexities of carbon, traits and tree performance in tropical forest

Abstract

Tropical forests are globally important ecosystems that absorb a large portion of anthropogenic carbon emissions. This has made global demand for tropical carbon credits high but the success of carbon credit systems will depend on the resilience of intact, planted and managed forests to novel climate regimes. To meet demand for `future forests', we need to characterise the mechanisms of long-term forest condition and better predict tree responses to the environment. This PhD presents four studies that address two key themes: how complex trait variation mediates tree performance, and the long-term, large-scale dynamics of tropical forest carbon. The first two studies use data from a seedling translocation experiment in Malaysian Borneo. I first characterise within-species trait variation in relation to ontogeny and environmental heterogeneity. I demonstrate that within-species variation is generally greater for morphological than physiological traits, with different environmental factors driving single traits versus trait \emph{co}variation. This work underscores the central role of ontogeny in driving trait variation and covariation. In the second study, I use Bayesian hierarchical models to test whether traits mediate seedling survival and growth after accounting for tree size differences and environmental heterogeneity. Although traits commonly varied across environmental gradients, few traits were clearly associated with species survival or growth. Traits that corresponded with species survival suggest that shade-tolerance mediates seedling dynamics more closely than interspecies growth differences. The final two studies use a 49-year (1971-2019) forest inventory dataset from Australian moist tropical forest. I first test forest biomass trends through time in relation to climate and cyclone incidence. I present evidence to suggest that Australia's moist tropical forests have switched from being a net carbon sink to a net source due to increasing atmospheric dryness and a lack of carbon fertilisation of woody tree growth. This work suggests the characteristics of tropical forest carbon sink decline and carbon fertilisation are divergent among continents. In the fourth study, I assess the relative importance of three sources of uncertainty in forest carbon stock estimation: i) the allometric model used to estimate tree height from diameter, ii) within-species trait variation (wood density and woody tissue carbon content), and iii) measurement errors. I show that, among these, the greatest source of uncertainty is height measurement errors, the distribution of which is reasonably uncertain. This work indicates that current efforts to collate species-level trait databases will likely provide sufficient accuracy for forest carbon estimation, though I highlight that allometric and trait variation remain under-characterised. Together, these works demonstrate that field-scale experimental manipulations and long-term ecological monitoring are critically important for characterising the complexities of tree function and forest condition. An holistic, nuanced picture of tropical forest function is sorely needed if we are to leverage the full potential of tropical forests to mitigate climate change and support our future on this planet.