Plant traits and carbon storage in freshwater wetlands
AuthorWindecker, Saras Mei
AffiliationSchool of BioSciences
Document TypePhD thesis
Access StatusOpen Access
© 2019 Saras Mei Windecker
Freshwater wetlands are an important part of the global carbon cycle due to their role in sequestering carbon in the soil. Despite covering less than 8% of global land area, freshwater wetlands are significant reservoirs of soil carbon due to long-term storage in the soil. Soil carbon stocks are high because wetlands are highly productive and can have low rates of decomposition. Plant communities capture carbon during photosynthesis and deposit it as litter. My thesis broadly examines the influence of vegetation communities on soil carbon storage across a range of scales, from the landscape down to the leaf. To examine the relative contribution of vegetation to carbon stock on a landscape-wide scale, I modelled carbon density of soil cores collected in 100 inland wetlands across Victoria, Australia. I determined that the nonlinear relationship of carbon with soil depth varies between wetlands, which has important implications for extrapolating carbon density necessary in wide-scale comparisons. Using a hierarchical model with continuous spatial data, I found that we could predict soil carbon stock using these globally available datasets and that intermediate inundation was a strong predictor for high soil carbon stock. At the wetland level, plants take up carbon dioxide during photosynthesis and contribute to soil carbon through deposition of litter. Litter quality, such as nitrogen level, affects how much litter material remains in the soil and for how long. Leaf economics spectrum theory suggests that a range of functional traits correlate along a spectrum of nutrient level that can be used to generalise litter quality. Despite relevance to a range of ecosystem processes including decomposition, direct measures of carbon composition of litter are not included among the economics spectrum traits. To test the generality of the trait correlations to carbon composition, I measured seven traits of litter in 29 wetland plant species: litter area per mass, dry matter content, nitrogen content, carbon content, and relative composition of hemicellulose, cellulose, and lignin. I tested thermogravimetric analysis, an analytic technique for estimating these carbon components. To replace proprietary software currently used in the modelling step of this estimation technique, I developed an open-source statistical R package called 'mixchar'. In general, species invested predominantly in either lignin- or cellulose-based tissue, and variation in this division was not correlated with variation in the other morphologic and chemical traits. Decomposition of litter impact long-term storage of carbon in the soil and is driven by litter quality. I conducted a mesocosm decomposition experiment of 29 wetland plant species, and evaluated the predictive performance of functional traits related to both nitrogen level and carbon complexity. I prepared a series of Bayesian decay models to test which traits affected rate of overall decay. I evaluated the models by comparing the cross-validative predictive performance on a new species. I found that trait models that include both litter nitrogen and carbon characteristics ranked among the best at predicting decomposition. However, differences among trait models were marginal, suggesting perhaps no single 'best' model for litter decomposition rate based on traits alone. My research improves our understanding of how plants contribute to soil carbon, from the scale of a leaf to the wider landscape, and clarifies how plants decay in wetlands and contribute to carbon storage in wetland soils.
Keywordsfunctional traits; decomposition; soil carbon; freshwater wetlands; hierarchical modeling
- Click on "Export Reference in RIS Format" and choose "open with... Endnote".
- Click on "Export Reference in RIS Format". Login to Refworks, go to References => Import References