Modelling of above-ground and below-ground carbon in a New Zealand native forest at a watershed scale

Abstract

Global climate change has been recognized as a major issue facing the world today and is primarily due to increasing levels of atmospheric carbon dioxide. As a result of this and due to climate change policies such as the Kyoto Protocol, there has been mounting interest in trying to characterize carbon (C) dynamics in various ecosystems, such as forests. However, there are many uncertainties surrounding large-scale forest C inventories due to the complex and heterogeneous nature of forest ecosystems. The aim of this study was to examine spatial patterns of above-ground and below-ground C storage across the North Mokihinui watershed in the west coast region of New Zealand’s South Island in relation to distinct forest communities and abiotic environmental variability. Linking environmental variables to C quantities at a landscape scale will enable the precise predictions of C storage at a larger scale through the development of local adjustment terms that include the natural spatial variability of the landscape. Data was collected during the summer of 2008/2009 from forty-two 20 m x 20 m sample plots. Measurements of tree heights and diameters, sapling number, coarse woody debris (CWD), and litter, humus and mineral soil samples were taken and used to obtain estimates of above-ground and below-ground C storage. Seven forest communities were identified in the watershed. Mean total above-ground C storage was estimated at 204.6 ± 113.9 t/ha and 30.2 ± 27.6 t/ha for the live plant and CWD pools, respectively. Live plant C quantities did not differ significantly among forest communities but were noticeably higher in the beech-dominated communities compared with the hardwood and mixed hardwood-beech communities. Forest communities did differ significantly for CWD carbon storage and were highly variable. Model selection using mixed effects models and an information theoretic approach indicated that topography was the most important influence on above-ground C quantities, followed by stand characteristics and site fertility. Mean C storage for litter and humus was estimated at 0.6 ± 0.2 t/ha and 11.0 ± 8.9 t/ha respectively. Mean C storage for three soil layers – 0-10 cm, 11-20 cm and 21-30 cm – was estimated at 58.2 ± 53.4 t/ha, 24.8 ± 12.1 t/ha and 19.8 ± 12.9 t/ha, respectively. Litter C quantities were relatively equal across all forest communities. However, for humus and the 11-20 and 21-30 cm soil layers, C quantities were much greater in beech-dominated communities than hardwood/mixed hardwood-beech communities. Soil in the 0-10 cm layer showed the opposite trend. Model selection indicated that site fertility was most important for explaining mineral soil C quantities, followed by litter quality and topography. With the exception of the litter layer, there was high spatial variability in above-ground and below-ground C pools across the North Mokihinui watershed. However, the patterns of this variability differed. On the whole, there were no strong correlations found among any of these C pools, which indicates that C storage in these pools are influenced by environmental factors in differing ways. Above-ground and below-ground forest ecosystems are highly complex in and of themselves, but they are also intricately intertwined, interacting and feeding back to each other in a variety of ways depending on the context of the site. These interactions control net ecosystem C dynamics but also make it difficult to predict how above-ground C dynamics affect the below-ground system and vice-versa. Spatial and temporal scale, along with abiotic factors, determine the nature of this context dependency. Results from this study exemplify the high spatial variability of above-ground and below-ground C, even at the relatively small scale of a watershed. They also highlight the need for further investigation of spatial variability of forest C pools in order to increase understanding of forest C dynamics and relationships of C quantities with environmental factors at a landscape-scale. This will help more precise estimations of C inventories to be made at large-scales that include the variation inherent across the landscape. The results obtained in this study will also serve as a baseline for a climate change mitigation project in the North Mokihinui watershed that will attempt to increase C sequestration through the management of ungulate browsers.