J Plant Ecol ›› 2016, Vol. 9 ›› Issue (3): 272-284.

• Research Articles •

### Estimating tree and stand sapwood area in spatially heterogeneous southeastern Australian forests

Dominik Jaskierniak1,*, George Kuczera2, Richard G. Benyon1 and Arko Lucieer3

1. 1 Department of Ecosystem and Forest Science, University of Melbourne, Parkville, Victoria 3010, Australia; 2 School of Engineering, University of Newcastle, Callaghan, New South Wales 2308, Australia; 3 School of Land and Food, University of Tasmania, Sandy Bay, Tasmania 7005, Australia
• Received:2015-01-06 Accepted:2015-07-25 Published:2016-05-25
• Contact: Jaskierniak, Dominik

Abstract: Aims Natural and anthropogenic changes in forests can have important influences on transpiration and water production. Understanding the effects of increasing disturbances, due for example to climate change and forest harvesting, requires detailed information on how forest density and structural attributes relate to transpiration. Mean annual transpiration of eucalypt forest communities is often strongly correlated with total cross-sectional sapwood area. Our aim was to test an efficient method for estimating sapwood area at 1.3 m height (SA 1.3) in a large number of trees to understand the spatial heterogeneity of tree and stand sapwood area within and between forest communities, and develop allometric relationships that predict SA 1.3 with forest inventory data. We also apply tree competition models to determine the degree to which the relationship between SA 1.3 and tree basal area at 1.3 m height (BA 1.3) is influenced by competition.
Methods We visited 25 recently harvested southeastern Australian forest sites consisting of 1379 trees and 5 Eucalyptus species to evaluate a new efficient data collection method for estimating SA 1.3 with tree taper and stump dimensions data using mixed effects models. The locations of 784 stumps within one 5-ha site were accurately mapped using an unmanned aerial vehicle (UAV), and four distance-dependent tree competition models were applied across the site to explain within-stand variation in the ratio of SA 1.3 to BA 1.3. Data from 24 additional sites, consisting of ten 15 m radial plots per site, were used to analyse within-site variation in R Ha (the ratio of stand sapwood area SA Ha to stand basal area BA Ha). The radial plots were merged within each site to evaluate between-site variations in R Ha across the landscape. For predicting SA Ha with forest inventory data, we computed the relationship between SA Ha and a new index of total stem perimeter per hectare, defined as ? B A H a N T, where N T is tree stocking density.
Important findings Our 1379 measured stems represent the most comprehensive measure of sapwood area, surpassing the 757 measured stems in native eucalypt forests published in literature. The species-specific R Ha varied considerably across sites and therefore extrapolating SA Ha with spatially distributed BA Ha maps and a generalized R Ha would introduce local uncertainty. We found that the species-specific stem perimeter index was more effective at capturing variability in SA Ha across the landscape using forest composition, structure and density data (R 2 : 0.72–0.77). The strong correlation between tree SA 1.3 and BA 1.3 improved slightly using tree competition models (R 2 increased from 0.86 to 0.88). Relating SA Ha to routinely measured forest inventory attributes within permanent plots and Light Detection and Ranging (LiDAR) data may provide opportunities to map forest water use in time and space across large areas disturbed by wildfire and logging.

Aims Natural and anthropogenic changes in forests can have important influences on transpiration and water production. Understanding the effects of increasing disturbances, due for example to climate change and forest harvesting, requires detailed information on how forest density and structural attributes relate to transpiration. Mean annual transpiration of eucalypt forest communities is often strongly correlated with total cross-sectional sapwood area. Our aim was to test an efficient method for estimating sapwood area at 1.3 m height (SA 1.3) in a large number of trees to understand the spatial heterogeneity of tree and stand sapwood area within and between forest communities, and develop allometric relationships that predict SA 1.3 with forest inventory data. We also apply tree competition models to determine the degree to which the relationship between SA 1.3 and tree basal area at 1.3 m height (BA 1.3) is influenced by competition.
Methods We visited 25 recently harvested southeastern Australian forest sites consisting of 1379 trees and 5 Eucalyptus species to evaluate a new efficient data collection method for estimating SA 1.3 with tree taper and stump dimensions data using mixed effects models. The locations of 784 stumps within one 5-ha site were accurately mapped using an unmanned aerial vehicle (UAV), and four distance-dependent tree competition models were applied across the site to explain within-stand variation in the ratio of SA 1.3 to BA 1.3. Data from 24 additional sites, consisting of ten 15 m radial plots per site, were used to analyse within-site variation in R Ha (the ratio of stand sapwood area SA Ha to stand basal area BA Ha). The radial plots were merged within each site to evaluate between-site variations in R Ha across the landscape. For predicting SA Ha with forest inventory data, we computed the relationship between SA Ha and a new index of total stem perimeter per hectare, defined as ? B A H a N T, where N T is tree stocking density.
Important findings Our 1379 measured stems represent the most comprehensive measure of sapwood area, surpassing the 757 measured stems in native eucalypt forests published in literature. The species-specific R Ha varied considerably across sites and therefore extrapolating SA Ha with spatially distributed BA Ha maps and a generalized R Ha would introduce local uncertainty. We found that the species-specific stem perimeter index was more effective at capturing variability in SA Ha across the landscape using forest composition, structure and density data (R 2 : 0.72–0.77). The strong correlation between tree SA 1.3 and BA 1.3 improved slightly using tree competition models (R 2 increased from 0.86 to 0.88). Relating SA Ha to routinely measured forest inventory attributes within permanent plots and Light Detection and Ranging (LiDAR) data may provide opportunities to map forest water use in time and space across large areas disturbed by wildfire and logging.