Background

In 1939, wildfires across Victoria severely burnt vast tracts of forest including Melbourne’s water supply catchments in the Central Highlands. A significant proportion of these catchments were occupied by even-aged mature Mountain Ash (Eucalyptus regnans) forests originating from past fires in the1800’s. Mountain Ash is a fire sensitive species, but once reaching seed bearing age, they naturally regenerate from seeds held in the capsules of their crowns. This generally results in exceedingly dense regeneration due to a receptive burnt seedbed, with thousands of germinates per hectare that rapidly occupy the burnt sites with fast growing regrowth. Streamflow gauging indicated a decline in water yields post-fire in the Mountain Ash water supply catchments. Accordingly, The Melbourne Metropolitan Board of Works (MMBW - the Catchment Manager) initiated some thinning trials to investigate if such a treatment could reduce the observed decline in water yields.

Early MMBW Thinning Studies

Early studies by the MMBW on the prospects of reducing basal areas by thinning to increase water yields were undertaken in the Black Spur group of catchments. The treatments tested were uniform thinning to remove 40% and 60% of the initial basal area, and patch cutting (80m diameter patches) to remove 60% of the basal area. The three treatments were investigated using relatively small sub-catchments. Operationally, it was found that the latter treatment was simpler to implement due to the need to confine damage to retained stems to a minimum during a uniform thinning from below.

Preliminary results indicated marked increases in streamflow for all three treatments, and these persisted for at least several years (O’Shaughnessy et al. 1981). It was recommended that similar treatments needed to be tested on larger catchments. O’Shaughnessy et al. (1993) reported that the uniform thinning treatment that reduced the basal area of a 37 year old stand by around 50% increased streamflow by about 25% (equivalent to 100mm per annum in streamflow) over the following ten years, after which time the increase commenced to decay. From a wood production perspective this decay would be expected given the very nature of uniform thinning which aims to concentrate increased growth on retained trees through total site occupancy.

Root configuration plays an important role in site occupancy and hence transmission of rainfall to groundwater and eventually streamflow. Root systems are also strongly linked to soil type (eg. deep krasnozem soils to shallow duplex soils) and the habits of individual tree species. Incoll (1979) undertook a detailed study in a 29 year old unthinned Mountain Ash stand near Narbethong to provide fundamental information on root systems of this species. He used high pressure pumps to expose root systems, along with a complete excavation of the root systems of a dominant and a suppressed tree. A direct relationship was found between root spread and tree diameter at breast height. The total estimated length of roots greater than 5mm in diameter was about 59km/ha. As would be expected in a krasnozem soil, roots were recorded to a depth of 2.6m close to tree boles. Importantly, inter and intra tree grafts were detected, with 61% of plot trees being involved in inter-tree grafting. Clearly, this assists dense Mountain Ash regrowth to rapidly gain full occupancy of sites.

The Kuczera Curve

As noted above, using an extensive network of high quality gauged sub-catchments and catchments to monitor streamflow, it soon became obvious to the Catchment Manager that water yields from catchments dominated by Mountain Ash were rapidly declining soon after the wildfire. In 1987, George Kuczera published a water yield model, now commonly called the Kuczera Curve, which predicted annual water yield from the 1939 Mountain Ash fire regeneration for a period of over 100 years post-fire.

The Kuczera Curve describes annual water yield with the age of fire regrowth. The Curve is adjusted for variation around an average annual rainfall of 1600mm for the Mountain Ash water supply catchments and importantly assumes sub-catchments to be comprised solely of 1939 Mountain Ash regrowth. The model recognizes that water yield varies substantially between catchments, which is largely determined by the density of post-fire Mountain Ash regrowth. The model also applies largely to krasnozems which are deep, well-structured soils with high infiltration rates and high water holding capacity. For such soils and climatic features, mature Mountain Ash forests typically transpire around 70% of rainfall due to soil factors and the ability of Mountain Ash root systems to occupy large volumes of soil.

As expected, water yield temporarily increased immediately after fire due to a lack of transpiration. Many studies (eg. clearing of native forests for conversion to pine plantations) have observed such temporary increases. The Curve then predicts that water yield (streamflow) declines sharply post-fire as dense regeneration develops into rapidly growing regrowth forests. The Curve shows that this rapid decline in water yield continues to around 30 years of age, coinciding with a predicted reduction in streamflow of around 50%. The Curve then predicts a slow recovery as the natural thinning process commences, though yields were still depressed to around 40% at an age of around 60 years. This slow recovery continues until the regrowth reaches more than 100 years of age.

The model (see figure below) has been the subject of considerable scrutiny. It is clear from these analyses that the Curve cannot be extrapolated to other ecosystems such as dry sclerophyll forests because water yield and response to wildfire varies substantially according to forest type (species, silvics, structure and age), soil attributes (including fertility, structure, depth, infiltration rates and water holding capacity), and rainfall patterns (total and seasonal variation). Dry sclerophyll mixed species forests for example are largely fire resistant and any post-fire regeneration will be dependent on the silvics of individual species. Soils are also often relatively shallow and can be subject to post-fire erosion. Clearly, the Curve has no application to such forests, and a major decline would not be expected under such conditions, but as discussed later there may be exceptions.

The strategic importance of the Curve predictions now and in the future in terms of Melbourne’s water supply cannot be over emphasized. Clearly, such predictions are disturbing given that Mountain Ash catchments at the time of the Crotty Creek study covered 53% of Melbourne’s total water supply catchment area, and these Mountain Ash forests contributed 70-80% of the total water catchment yields. As already noted, this led to research on the relationship between density of Mountain Ash fire regrowth and water yield. From a Curve perspective, it is noteworthy that MMBW scientists found that in the North Maroondah Experimental Area, the 1939 fires which converted old growth Mountain Ash forest to a regrowth forest resulted in a decline of up to 30% in streamflow, and that this decrease was sustained between 1947 and 1981 (O’Shaughnessy et al. 1981). They also found that water yield from small sub-catchments varied significantly according to the density of regrowth.

The Crotty Creek Project

In 1976, the MMBW and FCV initiated a joint study to further investigate alternative treatments to reduce basal area and increase water yield. The study was undertaken in the 122ha Crotty Creek catchment in State Forest near Narbethong and the North Maroondah Experimental Area. This was the first study to use a relatively large catchment to investigate the impacts of reduced basal area to increase water yields and simultaneously maintain a viable forest for future timber production. A high-quality stream gauging weir was constructed and monitoring of rainfall and streamflow commenced in 1976. The Crotty Creek catchment was being managed for a range of values including water quality and yield, wood production and conservation of flora and fauna values. The thinning treatment comprised removal of 50% of the basal area by alternate 35 m of cut and uncut strips. This was undertaken between 1979 and 1985.

Rainfall and streamflow monitoring continued from 1976 to 1990. This was carried out meticulously to the highest possible standards by MMBW scientists. Complete confidence can therefore be guaranteed in the findings.

Prior to imposing the strip thinning treatment, the logging contractor tested the feasibility of applying the technique over a large catchment with some steeply sloping sections. The 40 year old regrowth was also a challenge in terms of damage to the retained trees on strip edges. Landscape issues were also given careful consideration, including the ability of the contractor to make strips curved where appropriate.

The study recognized that it is not possible to simultaneously optimize wood and water yields using the strip thinning method. Of further note is that early streamflow measurements showed that a significant portion of the total streamflow comprised baseflow. This reflected the deep and well-structured soils which had a high water holding capacity. Regeneration of cut strips with understory species and Mountain Ash was monitored along with response of retained Mountain Ash to the thinning treatment.

Interim Findings from the Crotty Creek Project

Monitoring of streamflow at the 122ha Crotty Creek catchment commenced in 1976 for calibration purposes using a control catchment in the nearby North Maroondah Experimental Area (O’Shaughnessy et al. 1981). Strip thinning was then undertaken over a six year period between December 1979 and March 1985 to reduce the basal area by 50% using 35m wide cut and uncut strips. Interim findings were reported by O’Shaughnessy et al. (1993). This second progress report comprised chapters by P. J. O’Shaughnessy (Introduction), G. W. Beach (Treatment Overview), W. D. Incoll (Treatment Effects on Forest Growth), S.R. Haydon (Hydrometerological Data Collection Program, Modelling of Annual Streamflow Yield Curves, and Treatment Effects on Streamflow Characteristics) and R. Benyon (Discussion and Conclusions).

Measurement of eight growth plots scattered across the catchment together with control plots located in nearby unthinned stands showed that five years after treatment, gross stand growth was reduced by 33%. This confirmed that the site had not been fully re-occupied by the retained trees. Natural regeneration of Mountain Ash on the cut strips was found to be of low density and had poor vigour. This is a desirable result from wood and water yield perspectives, particularly as damage to retained trees on strip edges was minimal. In the absence of another wildfire, no further Mountain Ash regeneration would occur because the disturbed seedbed created during the harvesting operation only remains receptive to seed germination for the first 2-3 years after harvesting.

In 2007 the regeneration and composition of the understorey vegetation in the cut and uncut strips were assessed using 48 plots in each. (Tran, 2007) The low density of Mountain Ash in the cut strips was confirmed with 17% (4 plots) of the cut plots and 96% (46 plots) of the uncut plots containing Mountain Ash. The cut strips had slightly but significantly greater species richness, with 23 species compared to 20 species (Additionally Persoonia arborea, Lomatia fraseri and Pittosporum bicolor). Three seeding species were advantaged by the cut strips, Acacia dealbata, Prostanthera lasianthos and Ziera arborescens, while resprouting species of initial concern (Dicksonia antarctica, Olearia argophylla and Hedycarya augustifolia) appeared not to be adversely affected by strip thinning. The increased flora heterogeneity in species may provide significant biodiversity benefits, such as increased food and habitat for fauna like the endangered Leadbeater’s possum.

Three models were used to determine streamflow responses to treatment, these being the Soil Dryness Index (SDI), the Climatic Index (CI) and REGMOD. An examination of longterm rainfall records at the nearby Black Spur station showed that rainfall was below average during the calibration period, whilst the post-treatment period was marked by a severe drought in 1982/83. In contrast, the 1989/90 period experienced very high rainfall. The strip thinning treatment increased annual yields by 39% using the SDI model, 16% using the CI model and 23% using REGMOD over the 5 year post-treatment period. The author considered that agreement between all three models was “quite good” in both absolute and percentage terms. However, after adjustment for the wet and dry periods, the SDI increase was reduced to 31%, whilst predicted yields for the CI and REGMOD models were 21% and 26% respectively using the pre-1989/90 streamflow data. This makes the estimates from the three models a “lot closer.” Overall, it was concluded that strip thinning conservatively increased streamflow by 26% (an average of the three predictions), and that these increases persisted for at least ten years.

Increased water yields were associated with significantly reduced wood yields. Economic analyses showed however that strip thinning has the potential to increase the combined economic output from the catchments using realistic values for wood and water. It was also concluded that the water yield increase measured for the 43 year old regrowth stands should apply to similar Mountain Ash regrowth stands in the North Maroondah catchments and nearby State Forests.

The strip thinning treatment was also found to be a practical method for reducing stand basal area and could be conducted with similar operational ease and productivity to a clearfelling operation.

Extrapolation of the Kuczera Curve to other Forest Types

There has been recent consideration whether the Kuczera Curve can be extrapolated to other forest types. However, there is a strong consensus that the Curve applies strictly to pure Mountain Ash regrowth catchments with a history of fires in the 1800’s and 1939. It has been speculated that wildfire in some dry sclerophyll forest types may in fact be associated with increased water yields for an unknown period of time. Unlike wet sclerophyll forests, including Mountain Ash, dry sclerophyll forests can be subject to massive erosion following a high intensity wildfire and post-fire storm events. As an example, six days after the 1983 Ash Wednesday fires, a small catchment in a dry sclerophyll forest near Warburton experienced an intense storm event that resulted in gross erosion of catchment slopes and gullies (Leitch et al. 1983). Soils in the locality were observed to be hydrophobic for at least three months post-fire. Williams et al. (1973) observed less dramatic erosion following a wildfire in the Macalister catchment. There is little doubt therefore that water yields will increase for a short period after wildfire in some dry sclerophyll forests that are prone to erosion. Krasnozem soils on the other hand have a low propensity to erode in their natural state.

Some dry sclerophyll forests show a similar pattern of post-fire regeneration to Mountain Ash. They include the Boola Boola State Forest and a significant portion of the East Gippsland coastal and foothill mixed species forests. For the latter forests, exceedingly dense E. seiberi (Silvertop Ash) regeneration is common following both wildfires and clearfell timber harvesting. The management of this regrowth for timber production has been intensively studied, with a focus on early thinning (Flinn and Mamers, 1991). Whilst such thinning will almost certainly increase water yields by an unknown amount and response period, any increase and its duration will not match the Curve for a number of reasons including post-thinning coppice growth, climatic patterns and the properties of the soil types in these eastern Victorian forests.

To summarize, recognizing that many water supply catchments across Victoria comprise a medium to high proportion of a contrasting range of dry sclerophyll forests and conditions (species composition, climatic patterns and soil types), and acknowledging that research is still emerging on this subject, it is reasonable to assume that wildfire, in the short term, will lead to increased water yield dependent on soil type and post-fire storm events, but any increase in water yield is likely to be associated with mid-severe catchment erosion in the event of post-fire storm events. In other cases, dense regrowth of similar characteristics to Mountain Ash is likely to decrease water yield. The magnitude of this depression has not been investigated nor has the duration of any decline. And the Kuczera Curve does not provide guidance on this matter. The major impacts of the 2003 and 2006/07 fires that collectively burnt 10% of Victoria’s land area on water quality, which persisted for at least three years post-fire due to erosion-based sediment stores in burnt river systems, are instructive.

Implications of Crotty Creek Findings for the Future Management of Fire Regrowth

The Crotty Creek project has provided water supply catchment managers with a silvicultural system for advanced Mountain Ash fire regrowth that results in significant and sustained increases in water yield over the period of the study. Now that timber harvesting is no longer practiced in Victorian native forests, there are new opportunities to revise the parameters of a strip thinning treatment to further increase water yields. Such revisions could for example aim at reducing the standing basal area by around 60% using wider cut strips and narrower uncut strips. To reduce the extent of root invasion by retained trees on the cut strips, the uncut strips could be simultaneously thinned from below, excluding any habitat trees across the catchment. Whilst O’Shaughnessy et al. (1981) noted that a strip width above 40m or more might “carry considerable regeneration”, the second progress report provides results of the monitoring of understory species and Mountain Ash on the cut strips. Regeneration was not an issue for 35 m strips, despite the maturing trees at Crotty Creek being tall and this provided opportunity for seed dispersal to the disturbed seedbed created during the harvesting operation. Ideally, strip thinning should be undertaken when trees are small to capitalize on the predictions of the Kuczera Curve (sharp declines in yield in first 30 or so years) and further reduce the opportunity for natural regeneration on the cut strips. The earlier the intervention, the greater the yield response and longevity will be.

Climate change is predicted inter alia to progressively reduce annual rainfall and increase the frequency of catastrophic fire weather events. The probability of widespread wildfires in Melbourne’s water supply catchments over the next 2-3 decades is increasing, which would result in reduced water yields to a similar quantum to that predicted by the Kuczera model. In the case of a future fire event, very early silvicultural intervention could be considered to significantly reduce the basal area of Mountain Ash regrowth. The gains in water yield would be substantially greater than for maturing stands that are approaching the plateau of the Curve. The two most viable options appear to be patch cutting and strip thinning. If the latter is done, then early intervention before seed crops develop has many advantages. Uniform thinning should not be considered unless the basal area reduction is at least 70% to delay full site occupancy by retained stems. Such a reduction could however be associated with natural regeneration.

Conclusions

In conclusion, an expected increase in the population of greater Melbourne over the next two or so decades, a likely reduction in annual rainfall and changed weather patterns, and an increasing probability of more frequent days of catastrophic fire weather due to climate change, are all important considerations for Melbourne's water supply. The Crotty Creek project has provided an insight into silvicultural options that can focus solely on water yield without the need to consider wood yields in order to reduce the impacts of fire-induced Mountain Ash regeneration and regrowth on water yield. This project and kindred studies by MMBW in nearby catchments indicate that the primary aim of any silvicultural intervention should focus on more or less permanently reducing site occupancy by the Mountain Ash regrowth through a significant reduction in stand basal area, thereby allowing some rainfall to infiltrate directly to groundwater. Patch cutting with an 80m diameter also achieves this objective. The East Gippsland regrowth study provides guidance for early thinning techniques.

Irrespective of which silvicultural intervention may be adopted following a future wildfire event that impacts a significant proportion of the Mountain Ash water supply catchments, O’Shaughnessy et al. (1993) rightly stress that it will be critical to ensure that flora and fauna impacts can be predicted with confidence before any operational thinning is undertaken. The retention of all overmature trees of any species (including veteran Mountain Gums) should be adopted across all catchments. Indeed, from operational and environmental viewpoints, a mixture of the two most promising thinning treatments, as modified along the above lines, could be adopted to cater for variation in topographic (particularly slope) and stand features and for environmental, landscape and water quality considerations. Importantly, to maximize water yield responses, thinning treatments should be imposed well before the “bottoming out” of the Curve and not when the Curve is in recovery phase.

Note: From a research findings perspective, with one exception, this article is confined to some relevant unpublished and published scientific papers up to and including the 1993 Second Progress Report on the Crotty Creek Project.

Acknowledgements: The author is grateful to Peter Farrell and Simon Murphy for highly valued comments on the first draft.

References

Flinn, D. W. and Mamers, H. (Eds) (1991). Management of Eucalypt Regrowth in East Gippsland. Executive Summary of Research Findings. Dept. Conserv. and Env., Melbourne, 43pp.

Incoll, W, D. (1979). Root system investigations in stands of Eucalyptus regnans. For. Tech. Paper 27: 23-32.

Leitch, C. J., Flinn, D.W. and van de Graaff, R.H.M. (1983). Erosion and nutrient loss resulting from the Ash Wednesday (February 1983) wildfires: a case study. Aust. For. 46:173-180. Also - Res Branch Rep. 251.

O’Shaughnessy, P. J., Moran, R. J. and Flinn, D.W. (Eds) (1981). The Crotty Creek Project. The Effects of Strip Thinning Eucalyptus regnans on water yield. First Progress Report - Pre- treatment phase. For. Comm. Vic., Misc. Pub. No 9 and MMBW Report No MMBW-W-0013, 78pp.

O’Shaughnessy, P.J., Benyon, R. and Lucas, A. (Eds) (1993). The Crotty Creek Project. The effects of strip thinning Eucalyptus regnans on forest growth and water yield. Res. Rep. No 358, 74pp.

Tran, P. (2007). The effects of strip thinning on the regeneration and composition of the understorey in a Eucalyptus regnans forest. Univ. of Melb. Bachelor of Forestry/Bachelor of Science Honours Research Project Thesis, 67pp.

Williams, D. F., Bren, L. J. and Craig, F. G. (1973). Effects of wildfire on overland flow and erosion in the Macalister watershed. Res. Branch Rep. No 29, 19pp.