A major challenge for groundwater research is to consider the complex relationships amongst groundwater resources, vegetation (function and structure) and climate. Transpiration from vegetation (especially trees when present) is the principal pathway for discharge of water from vegetated landscapes. Whilst it is known that the ability to access groundwater can help plants survive drought conditions and in particular the importance of groundwater in arid and semi-arid areas is well documented, there have been few studies that compare ecophysiological (e.g. leaf water relations), structural (e.g. basal area, leaf area index) and functional (e.g. rates of tree water-use, above-ground net primary productivity; water-use-efficiency) attributes of trees along a naturally occurring gradient in depth-to-groundwater, especially in mesic environments.
The aim of this research was to establish whether differences in groundwater depth along a transect influences ecophysiological, functional and structural attributes of remnant woodlands in southeast of Australia growing in a region with relatively high annual rainfall. The study area was located in the Kangaloon bore-field area, NSW, where depth-to-groundwater varies from 2.4 m to 37.5 m. To address this aim; seasonal measurements were made at seven sites at three scales (leaf-, tree- and stand-scales).
Structural attributes of woodlands, above-ground productivity were significantly different across sites. The three shallowest sites with 2.4m, 4.3 m and 5.5 m depth-to-groundwater had significantly larger above-ground biomass and productivity than the four deepest sites (sites where depth-to-groundwater was more than 9.8 m). A significant shift occurred in all measured variables when depth-to-groundwater increased from 5.5 m to 9.8 m. This result was found consistently for each structural trait examined (LAI, tree height, BA, ANPP, AGB, stem density). There were no differences in three structural traits (BA, height and LAI) nor ANPP across the four deepest sites. All these traits were significantly smaller at these deepest sites compared to the three shallowest sites.
Rates of stand transpiration rate tended to be smaller at all sites compared to many other studies conducted in a similar environment. This result was attributed to larger-than-average rainfall received across all sites across the entire study period, with concomitant reductions in solar radiation, temperature and VPD compared to long term means. Despite this, there were significant differences across sites and these differences were not consistent with my initial hypothesis: namely that as depth-to-groundwater decreases stand transpiration rates will increase. Rates of stand transpiration at the shallowest groundwater site (2.4 m) were the same as those at the deepest groundwater site (37.5 m) despite significantly larger tree density, BA, LAI at the shallowest site compared to the deepest site. Rates of stand transpiration were consistently the largest at the 4.3 m site compared to all other sites.
Tree hydraulic architecture was the least affected by depth-to-groundwater. Hydraulic architecture of trees was examined by measurement of the following traits: Huber value, branch hydraulic conductivity (leaf and sapwood specific), xylem sensitivity to embolism and sapwood density. Huber value (Hᵥ) increased significantly as depth-to-groundwater increased, in agreement with my initial hypothesis: namely that HV is larger at drier sites (deeper groundwater sites) than sites with shallow groundwater. Neither sapwood density nor branch hydraulic conductivity (sapwood and leaf area specific) varied significantly across sites, in contrast to expectations. Xylem vulnerability to embolism was assessed in summer and winter by determining the water potential associated with both a 50% and 88% loss of conductance (PLC₅₀ and PLC₈₈ respectively). PLC₅₀ in both seasons was significantly and negatively correlated with depth-to-groundwater.
Leaf-scale measurements showed that trees occupying sites with the shallowest water table were more sensitive to drought stress than those growing at sites with the deepest water-tables. There were significant changes across some leaf traits, including: leaf turgor loss point, osmotic potential at full turgor and the relative water content at turgor loss point (RWCTLP). All of these traits declined as depth-to-groundwater increased. In contrast, leaf volumetric elasticity was independent of depth-to-groundwater. The form of the relationship between depth-to-groundwater and leaf-scale and stand-scale structural traits differed between the two sets of data. In the former a negative and linear response to increasing depth-to-groundwater was observed but in the latter an exponential decay response was observed.
When all leaf-scale, tree-scale and stand-scale traits were normalized (zero to one) to produce a single, average response across all traits, as a function of depth-to-groundwater, a significant step-function response to increase in depth-to-groundwater was observed. For the three shallowest sites, there were minimal changes as a function of depth-to-groundwater but as depth-to-groundwater increased from 5.5 m to 9.8 m there was a significant reduction in mean normalised trait value. When depth-to-groundwater was larger than approximately 9-10 m a consistent reduction in normalised trait value occurred, with no significant difference across the four deepest sites.
This thesis has demonstrated that even in a mesic environment, groundwater can have an important impact on ecophysiological, structural and functional traits of trees. Understanding how trees respond to changes in groundwater availability is a crucial knowledge gap in our current understanding about groundwater and vegetation interactions. Determining this response function has management and conservation applications which indicate potential changes in ecosystem function, structure, growth and ultimately survival. It can also potentially determine the safe limit threshold for groundwater drawdown.
Whilst it is acknowledged that variation in traits along environmental gradients in the field are, by definition, correlative (that is correlated with variation in the environmental variable identified along the gradient), the use of multiple sites across a very small spatial gradient strongly supported the conclusion that trait variation was associated with variation in groundwater depth and not with variation in climate. Furthermore, consistency in trends across the gradient in groundwater depth also support the conclusion that trait variation identified in this thesis can be attributed to variation in groundwater depth, rather than randomly occurring in space.