Physiological responses to drought of Eucalyptus globulus and Eucalyptus nitens in plantations

Eucalyptus globulus Labill. and E. nitens (Deane and Maiden) Maiden are the predominant hardwood plantation species in southern Australia. This thesis describes some physiological strategies exhibited by these species in response to drought as a means of determining their suitability for establishment on water limited sites. To this end a 2 ha experimental plantation was established on a low rainfall site (ca. 5 1 5 mm a-1) in August 1990. The plantation was divided into irrigated and rainfed blocks so that the effects of soil drought could be separated from those of diurnal and seasonal climatic variation. Between November 1991 and April 1993 pre-dawn leaf water potential (˜í¬Æmax) was significantly lower in the rainfed than the irrigated treatment for six defined periods or stress cycles. At the end of this period, when ˜í¬Æmax was -2.37 and -2.34 MPa in the rainfed E. globulus and E. nitens respectively, leaf osmotic potential and bulk elastic modulus were still not significantly affected by water stress. At this time a significant interspecific difference was evident in the shape of the desorption isotherms. In the region of positive turgor the mean slope of these relationships was significantly greater in E. nitens (14.5 MPa) than E. globulus (9.3 MPa) resulting in turgor loss at a significantly higher relative water content in E. nitens (0. 86) than E. globulus (0. 79). This interspecific difference in leaf water relations was independent of soil water status. Allometric relationships between leaf area and sapwood area were developed by destructive sampling in July 1992 and July 1993. These relationships were used with monthly growth data to plot the course of leaf area index (L *) between August 1991 and April 1993. L* was significantly lower in E. nitens than E.globulus throughout the study and was significantly reduced by water stress after November 1992. During the 1992/93 growing season the L* of the rainfed E.globulus increased rapidly after rewatering, resulting in a stepped pattern of leaf area development which was not observed in E. nitens. At the end of the experiment, in April 1993 , L* of the irrigated and rainfed treatments were, respectively, 8.3 and 6.1 in E. globulus and 6.9 and 4.3 in E. nitens. Daily maximum stomatal conductance of both E. globulus and E. nitens was significantly reduced when ˜í¬Æmax was <-0 . 5 5 MPa. Reduction in stomatal conductance (gs) as ˜í¬Æ max decreased was greater in E. nitens than E.globulus. After rewatering, stomatal conductance was slower to recover in E. nitens than E.globulus. These differences caused a significant species by water stress interaction. A phenomenological model of stomatal conductance of upper canopy foliage (gsu) was developed for the irrigated trees. The maximum gsu observed (387 mmol m-2 S-1) was attenuated with normalised functions of total solar radiation (Q), air temperature (T) and vapour pressure deficit (D). This model explained 70% of the variation in gsu. The effect of water stress on the daily total of gsu was accommodated by predicting the ratio of total daily stomatal conductance of rainfed to that of irrigated trees as a function of the cumulative water stress integral for the preceding x days (S˜í¬Æx), where x was eight days for E. globulus and 20 days for E. nitens. Using S˜í¬Æx as an independent variable accommodated the longer residual effect of water stress in E. nitens than E. globulus. The model explained 75 and 73% of the residual variation in the daily total of gsu of E. nitens and E. globulus after the direct effects of Q, T and D were removed. Two estimates of canopy conductance were compared in the irrigated treatment of both species. The first (gc) was a summation of the parallel conductances of three canopy layers and the second (gc') involved integration of the light response function of a leaf level model with respect to cumulative leaf area index. At a daily time step gc' explained 79% of the variation in gc when L* was less than 6, but underestimated gc by as much as 41 % at leaf area indices above 6. This underestimation indicated that at high L*, E. globulus and E. nitens canopies may be weakly coupled to ambient atmospheric conditions. The degree of coupling was investigated in the irrigated E. globulus by simultaneous measurement of transpiration (using sapflow sensors) and stomatal conductance over three days in March 1994. The mean value of the dimensionless decoupling coefficient (˜í¬©) was 0.63. Despite this, a linear relationship with D explained 85% of the variation in transpiration. The vertical profile of ˜í¬© and absorbed net radiation were examined and ˜í¬© was highest where the lowest amount of radiant energy was absorbed. It is argued that transpiration of irrigated and rainfed E. globulus and nitens may be calculated by treating the canopy as a single layer with a surface conductance of gc'. E. globulus and nitens responded to drought by stomatal closure rather than by adjustment of the osmotic or elastic properties of their leaves. This response was stronger in E. nitens than E. globulus so that on sites where water stress is moderate and seasonal, E. globulus will probably grow more rapidly than E. nitens. Another major outcome of this thesis is quantification of the different stomatal responses of the two species at the canopy scale in a way that allows transpiration to be calculated from relatively easily obtained variables. Calculation of canopy conductance as gc' predicts a much lower value of gs for foliage at the bottom of high L* canopies than was measured . This suggests that as L* increased the irrigated canopies of both species became more water use efficient. It is argued that if the effects of water stress on gs and L* are considered together then long term water stress will reduce the water use efficiency of E. nitens more than that of E.globulus.