1. TWCL's goals of beech management system are to maintain a healthy forest in terms of stand structure and species composition, including the presence of some dead and dying trees. The resultant managed forest should continue to sustain the full range of natural values that occurs within similar forests in the conservation estate
   
2. Small groups of trees that are expected to die or be killed by falling trees within the next 30 years are harvested at a spacing and time scale that mimics the natural forest processes. Helicopter logging is the essential factor in ensuring the success of the system.
   
3. The way the sustainable yields are calculated in the Beech Plans represents an advance in forestry management practice. Yields are determined from a size-staged model, which deals in changes in numbers and sizes of trees rather than timber volumes.
   
4. The data inputs for the model came from forest inventory involving over 800 sample plots and separate surveys of seedling and tree growth rates. During annual operations, distributions for individual forest types and areas are obtained by more localised inventory and used to determine the annual compartment yields.
   
5. The sustainable yield is determined by assigning a defined proportion of the expected mortality determined above to a harvest. This proportion is set at a precautionary 50% or less of the size classes above 30-cm diameter to various maximum diameters depending on individual species. Trees selected for harvesting are those expected to die or be killed within the period of 30 years
   
6. The yield projection period for the model is limited to 35 years to coincide with the term of the Resource Consents.
   
7. There is an auditable record of the yield of all trees felled within the forest. Random checks can be made of the record system and the forest operational areas to verify that the harvest was within the specified sustainable yields.
   
8. It is recommended that Council set certain criteria on the number of retained large trees as a condition of the consents. I recommend that consents be reviewed in the extremely unlikely event that the numbers of large trees (> 70 cm) fall to 80% of that found in equivalent undisturbed forest. These criteria are simple, clearly auditable, and limit the risk to the forests in the unlikely event that beech management proves to be unsustainable.
   
9. The risk of harm to the forest environment from the level of harvesting proposed is completely reversible if logging was halted.
   
10. I believe that the TWCL beech management system provides a reasonable balance between the interests of forestry and ecological interests. Timber harvesting and ecological values are not mutually exclusive. The low rates of timber removal and all of the safety margins incorporated in the TWCL's proposal reflect a precautionary approach and will ensure that imposed changes on the forests are minor and sustainable.
    
11. Landcare criticisms of the model are falsely based on an assumption that harvesting will add to tree losses without any reduction in the natural losses. Consequently, their predicted trade-off between timber yields and harvesting impacts is completely unrealistic.
    

The way the sustainable yields are calculated in the Beech Plans represents an advance in forestry management practice. Yields are determined from a size-staged model, which deals in changes in numbers and sizes of trees rather than timber volumes.

1. DIAMETER SIZE CLASS DISTRIBUTION: for each tree species the number of stems present in the existing forest within fixed (10cm) diameter size classes.
   
   
2. RECRUITMENT: the number of seedlings that establish, survive and grow to enter the smallest tree size class.
   
   
3. SURVIVOR GROWTH: the amount by which live trees within size classes increase in diameter between measurement periods.
   
   
4. HARVEST: the numbers of trees within size classes that are removed or destroyed during the harvesting process within a measurement period.
   
   
5. MORTALITY: the numbers of trees within size classes that die through competition, disease or other external causes within a measurement period.
   

In a natural forest, over large areas and long periods of time, the mortality rate must equal the growth rate. If it did not, either the number and size of trees would get larger and larger or the forest would slowly disappear. However tree growth occurs at a regular annual rate, whereas tree deaths vary widely in scale and frequency. Consequently, in the short to medium term on any particular site, net growth and mortality may not be in balance.

The infrequent nature of mortality makes it difficult to measure the true mortality rate without fieldwork extending over several decades, perhaps centuries. Lincoln University is involved in such a study in the Maruia Valley (Stewart, pers comm.)

The forest growth model can be used to derive the long-term mortality rates that must be expected if the forest is to maintain its existing tree numbers and structure. Beginning with the smallest size-class, the mortality rate for each size-class is set by iteration so that the same number of trees remains in that size-class at the end of a set period (usually a single 15-year felling cycle). This results in the number of trees and shape of the diameter distribution remaining relatively constant through time.

For some individual species, the presence of cohorts means that strict adherence to this process is impossible and biologically unrealistic. In these circumstances mortality rates above and below the cohort are adjusted to replicate natural process of the cohorts passing through the forest structure.


3.4     Determining Sustainable Yields


The sustainable yield is determined by assigning a defined proportion of the expected mortality determined above to a harvest. This proportion is set at a precautionary 50% or less of the size classes above 30-cm diameter to various maximum diameters depending on individual species. Therefore, the harvest is set within the limits of expected mortality and is implemented by selecting trees in the forest that would be expected to die within the period of the next one or two felling cycles.

The harvest is initially determined for each species over the whole Working Circle and then recalculated for each forest operational area based on resource data specific to that area. Any trees felled but not extracted, wind-fallen trees harvested, or trees harvested along roadsides, are included within this yield.

---

Ian L James,     Assessment of Environmental Effects                                                               Page 8 of 12

Within each 15-year cycle, complete re-measurement of the permanent sample plot system as well as other data collected over the period on actual harvest rates will be used to update the growth model projections. The results of this monitoring will enable measured levels of growth, mortality and forest status to be input into forest models to periodically adjust yields. A decrease in yields would occur as part of TWCL's Environmental Management Systems. Any increase in yields would require a variation of the Resource Consent.

3.4     Units of Yield


Sustainable yield is specified solely in terms of tree numbers over a range of diameter classes and then converted to merchantable volume of the day for commercial management purposes. This approach differs from traditional forestry practice where the sustainable yield is determined and controlled on the basis of merchantable timber volume measured in cubic metres. The difficulty with volumetric measures is that they are potentially inconsistent because volume specifications of what constitutes a marketable product can change over time.

Tree numbers provide a better basis for measuring any ecological impacts on forest composition, structure and factors important to wildlife habitat and the landscape. This information is required to adequately meet the Resource Management Act 1991. Furthermore, non-forestry personnel and the general public find a yield expressed in tree numbers easier to understand than a yield based on timber volumes.

3.6     Projection Period


The yield projection period for the model is limited to 35 years to coincide with the term of the Resource Consents. It is recognised that this is a relatively short time for a forest where some of the large trees can survive for several centuries.

To make reliable predictions over longer periods requires much more biological data and complicated growth models that take into account the compensatory responses of the forests to harvesting. Recruitment, growth, and mortality rates are highly responsive processes. They are not easily calibrated and vary unpredictably over long time periods.

Long term predictions cannot make allowances for management responses to forest conditions. The Plans contain regular reviews and audits at several levels. Both the quantum of the sustainable yields and management actions such as tree selection, planting, and harvesting systems will be adapted to the measured changes in the forest.

Accordingly, it is totally misleading and unreliable for Landcare to make predictions of forest structure over several centuries using the initial data inputs and the relatively simple size-staged model.

3.7     Audit of Yield


Expressing the yield in tree numbers has the advantage of allowing for a precise accounting of the yield during marking and harvesting operations, and ensures greater transparency in the audit of the harvest. Every tree felled has its stump tagged. Records are kept of its tag number, diameter, height, species, location (by a GPS reading), year, health status, group size, volume, and point of sale. These data provide an auditable record of the yield of all trees felled within the forest. Random checks can be made of the record system and the forest operational areas to verify that the harvest was within the specified sustainable yields.

---

Ian L James,     Assessment of Environmental Effects                                                               Page 9 of 12

3.8     Limitations of the Model


No model perfectly describes all ecosystem interactions and the outputs are dependent on the quality of the input data. There are clearly some gaps in the data on which the model is based and further time is required to fill these. As stated earlier the precision of input data on structure growth and recruitment is limited to working circle averages. Further site-specific information will be collected as operations begin in individual forest types and localities. Information on tree recruitment into the smallest tree diameter size-class is another area where more information is required. It is planned to have more complete information following measurement of the working circle inventory plots after 10 years

A second problem arises in determining the proportion of expected mortality that will be harvested (presently this has been set at 50%). The determination of the percentage harvested is critical for biodiversity conservation as dying and dead trees are essential for many species within the forest, especially forest birds and bats (see below) and too high a harvest level could have adverse effects on biodiversity. At present this has been set on a precautionary basis and further research is required to better estimate the harvest level.

The 50% level will not be exceeded unless a variation of consent is obtained in the future on the basis of research that determines that a higher percentage level of harvest is sustainable (and if the Councils or the Environment Court accept that variation). In this particular case consent conditions should expressly provide for a capability to seek a variation for this purpose.

Third, the model is essentially an equilibrium model in that it aims to maintain the current structure of the forest. Clearly these forests are non-equilibrium systems that are structured by a mixture of frequent and infrequent disturbance events. The potential of the model to force the forest into a static condition is reduced by ensuring that the forest growth model is sensitive to disturbance-induced changes in forest structure. Regular updating of inputs and re-calculation of the growth model (e.g., every ten years) are essential to ensure the sustainable yield follows natural trends in the forest environment. By doing this, the risks from incorrect assumptions can be mitigated long before they can have any significant impact on the forest ecosystem.

There are several other areas of uncertainty associated with using the forest growth model. One is in determining which trees are likely to die within the next 15 years and therefore correctly selecting them for harvest. The probabilities of mortality associated with varying degrees of senescence have not yet been quantified. In the meantime, new field procedures are being used to identify but not fell the potential "gap maker" trees. While they are included as yield, leaving them standing will further reduce the immediate impacts of harvesting.

3.9     Model Sensitivities and Monitoring


The arguments presented so far give directions for monitoring of these Resource Consents. There is a consensus amongst all parties that a key aspect of sustainability is the fate of large trees (greater than 70cm diameter). Their importance to the forest structure, landscape, wildlife and other values is well recognised and understood.

The model is completely deterministic and provides no statistical error limits to the sustainable yield estimates. However, it is informative to examine the effects of the probable limits of error of inputs on large trees.

Almost all the trees greater than 70-cm diameter in the Maruia Working Circle are red beech. With logging, their numbers will fall by approximately 10% immediately after logging and then recover after 15 years (Diagram 2). Lowering of the recruitment rate will have no impact on the remaining large trees for at least 100 years. Reducing growth rates by 20% would lower the number of trees over 70cm diameter by 15% of in 35 years. In the situation where growth rates are reduced and mortality rates increase by 20%, i.e. negative errors are compounded,

---

Ian L James,     Assessment of Environmental Effects                                                               Page 10 of 12

numbers could fall by 19% after 35 years logging (section 3.8). Landcare's worst case scenario where logging is entirely additional to current natural tree losses is for a loss of 21% after 35 years (Diagram 1). In the Grey and Inangahua Working Circles, several species including rimu, red, hard and silver beech are found over 70cm diameter.

I believe it is reasonable for Council to set certain criteria to ensure that a desired proportion of large trees is retained as a condition of the consents. I would recommend that consents be put "on watch" if numbers of large trees fall to 80% of those found in equivalent undisturbed forest over the same time period and that the consents be revoked if numbers fall to 75%.

Monitoring of Council's conditions should be based on the permanent inventory plots that are randomly located throughout all three working circles. The loss of large trees can then be compared against the proportion of large trees measured in additional permanent plots located in surrounding conservation lands.

The criteria are simple, clearly auditable, and will limit the risk to the forests in the unlikely event that beech management proves to be unsustainable.

It is recognised that there is a need for more precise estimates of recruitment, growth and mortality and a process whereby sustainable yields are revised on a regular timetable. New input data will be available once the permanent inventory plots are remeasured within 10 years. It is logical that yields are recalculated at that time.

3.10     Risk management


It is important for sustainable management to retain the ability of the forest to adapt to change and to keep open options for the future (Palmer, 1991). Another key issue is that impacts are kept low enough to be reversible if management is halted.

Several alternative projections of the potential forest impacts were made earlier. Those of most interest in this hearing are the ranges of negative outcomes possible. I have taken Landcare's assumptions in diagram 2 as a worst case scenario; that is, no growth response, and all trees felled are additional to the normal tree losses; there will be 79% of the trees remaining over 70 cm diameter after 35 years.

Clearly, if this scenario eventuated moves would be made to either rectify the problems or halt management. If the latter course were imposed, the forest would re-form 95% of the original trees over 70-cm diameter within 60 years. The exact original number of large trees would be recovered in 75 years after logging ceased (figure 5 and 6). These (Landcare) forecasts assume no growth or recruitment responses.

On the other hand, any positive response by the forest to management would permit higher sustainable yields. There is ample evidence to show that growth of trees surrounding canopy openings will increase (potentially it will more than double) from the rates used in the TWCL model. Similarly, regeneration triggered by gap formation is likely to greatly increase the number of recruits assumed. Tree deaths should be reduced by the removal of the trees most at risk.

Ministry of Forestry silvicultural trials in the Maruia and Grey Valleys, where trees have been carefully heli-harvested in small groups, record few tree deaths and a marked growth responses. After 4 years increment on edge trees has exceeded the volume of trees lost around coupes (Beneke and Baker, pers comm.). If any one of these responses occur then the sustainable yield forecasts in these Plans will be seen as very conservative.

---

Ian L James,     Assessment of Environmental Effects                                                               Page 11 of 12

Figure 5. Comparison of alternative predictions; no harvesting, TWCL and Landcare outcomes in terms of trees greater than 70-cm diameter (s.p.h.) Also the recovery from Landcare outcome (worst case scenario) if harvesting ceased. (Mortality rates during the re-growth period reduced in proportion to remaining tree numbers, Growth response 0.25 and recruitment rate doubled).

4     Conclusions

I believe that the TWCL beech management system provides a reasonable balance between the interests of forestry and ecological interests. I remain confident that while a managed beech forest will be slightly different from one that is never managed, the important ecological functions can be maintained. In my opinion timber harvesting and ecological values are not mutually exclusive and the "tradeoffs" as claimed by Landcare are greatly exaggerated..

TWCL has recognised that its chosen model could not answer all questions to all people; few models can, particularly at an early stage. However, the model and data limitations will be managed within a conservative set of starting assumptions, and integrated into an adaptive management framework that controls risk. The plans specifically note the need for regular review of the model and input data. Resource consent conditions can be imposed, monitored and enforced to ensure the proposal remains within the stated bounds and is adapted if necessary to maintain sustainability.

TWCL'S plans specifically note that more complex ecologically based models will be developed, as more monitoring data is available (in approximately 10 years time). Until then, there is little justification for developing more sophisticated models for which there are no reliable data. I believe that the low rates of timber removal and all of the safety margins incorporated in the TWCL's proposal reflect a precautionary approach and will ensure that imposed changes on the forests are minor and sustainable.

---

Ian L James,     Assessment of Environmental Effects                                                               Page 12 of 12

5     References

Buongiorno J. and Mitchie, B.R., 1980: A Matrix Model of Uneven-Aged Forest Management. Forest Science, Volume 26, Number 4, (609-625).

Bruenig E.F., 1996: Conservation and Management of Tropical Rainforests. An integrated approach to sustainability. Published by CAB International.

Benecke, U., 1996. Ecological silviculture: The application of age-old methods. New Zealand Forestry 41, 27-33.

Davis, L.S. and Johnson, K.N. 1987: Forest Management. McCraw-Hill Inc. New York USA.

Efford M. 1999: Analysis of a model currently used for assessing sustainable yield in indigenous forests: Journal of the Royal Society Volume 29, Number 2, (175-184)

Franklin, J.F., Cromack, K., Jr., Denison, W., Mckee, A., Maser, C., Sedell, J., Swanson, F., Juday, G., 1981. Ecological Characteristics of Old-growth Douglas-Fir Forests. General Technical Report PNW-118, USDA Forest Service, Pacific Northwest Forest and Range Experiment Station, Portland, Oregon, U.S.A.

Harcombe P.A., Allen R.B., Wardle J.A. and Platt K.H., 1998: Spatial and Temporal Patterns in Stand Structure, Biomass, Growth, and Mortality in a Monospecific Nothofagus solandri var. cliffioides (Hook. F.) Poole forest in New Zealand. Journal of Sustainable Forestry, Volume 6 Number ¾ (313-345).

McClunie W. 1997: Beech Growth Rate Surveys. Internal Timberlands Report.

Myers C. A. and Martin E. C., 1963: Mortality of Southwestern Ponderosa Pine Sawtimber after Seconf Partial Harvest. Journal of Forestry, February, (128 –130).

Norton, D.A. 1983: Modern New Zealand Tree Ring Chronologies 1. Nothofagus menziesii.
Tree Ring Bulletin 43, pp. 1-17.

Ogdon J., 1971: Studies on the vegetation of Mt Colenso, New Zealand. 2. The population dynamics of red beech. Proceedings of the NZ Ecological Society, 18, (66-75).

Palmer, J.R., 1991: What should sustainability look like? Communications to WRI's Colloquium on Sustainability in Natural Tropical Forest Management. Washington. D. C. 8p.

Seydack A.H.W., et. al., 1995: An unconventional approach to timber yield regulation for multi-aged multispecies forests. II. Application to a South African forest.
Forest Ecology and Management, Volume 77, (155-168).

Smale M. C., Beverage A.E. and Herbert J.W., 1998: Selection Silviculture Trials in North Island Native Forests: Impacts on the Residual Forest and their Implications for Sustainable Forest Management.
NZ Forestry, November 1998 (19-29).

Steward G.A., 1999: Windfall and mortality within the Okurapoto dense podocarp selection logging trial, Whirinaki Forest Park. Unpublished Report for the Ministry of Agriculture and Forestry.

Usher, M.B., 1969: A Matrix Model for Forest Management. Biometrics, June (309-314).