An extract from Ecological Society News letter: Nr 93; November 1999.


Presented to the Transport & Environment Select Committee 1/9/99 by Murray Efford. Feel free to send replies for inclusion in the next newsletter or to Murray Efford.

Thank you for inviting me to appear before the committee.

I am a scientist employed by the crown research institute Landcare Research in Dunedin. I have some expertise in forest ecology and resource management through my work on possums. My submission draws heavily on my experience in the computer modelling of animal and plant populations.

I would like to address four questions:
        What are "models" and how can they help us to manage resources?
        What do models say about the trade-offs involved in low-intensity beech harvest?
        How has modelling informed recent decisions on beech management?
        What are the implications for the Forests Amendment Bill?

What are "models" ... ?
A "model" distils what we know about processes in nature into a compact and rigorous statement. Usually it is a set of mathematical equations or a computer program. Models are used routinely in ecology to explore the "behaviour" of dynamic systems - how they might unfold over time given different conditions. Models are most useful when they are simple and have explicit biological assumptions.

... and how can they help us manage resources ?

Models for sustainable forestry have two fundamental aims:
What do models say about the trade-offs involved in low-intensity beech harvest?
Timberlands West Coast Ltd (TWC) plans to establish a low-intensity harvest regime for various beech species
(2) .

Most of the beech to be harvested by TWC is in old-growth forests that have not been logged before. This is especially true in the Maruia Working Circle, where most of the wood is in trees 200 to 400 years old
(3). Logging exposes these forests to many additional risks (weed spread, increase in possum numbers, long-term nutrient depletion from removal of wood, changing species composition) (4). However, the central issue is the direct, long-term effect of logging on forests, especially the loss of large trees. Large trees are essential to the aesthetic and ecological qualities of a forests: they provide physical structure and simple grandeur, roost sites for kaka, kakariki and bats, niches for perching plants, etc. Their conservation importance is generally accepted.

Models help us appreciate the long-term effects on management. What rate of wood extraction can be continued in perpetuity? What will be the effect on the number of large, old trees? Sensibly, Timberlands chose a style of model that, if properly used, could help answer both these questions.

A size-structured forest model (Fig. 1) allows us to predict timber yield, from the numbers of trees in millable size classes, and to predict the future number of large trees. Such models are widely used in population ecology. More complex forestry models, on which I do not claim expertise, include site characteristics (terrain, soil fertility, etc.) to make more specific predictions. TWC justifiably excluded this detail from their calculations of yield at the level of the working circle (10 000 - 20 000 hectares).

Figure 1. Size-structured tree population model.

TWC used their model to calculate a harvest per size class that they believed to be "sustainable" in the strong sense of not changing the size structure of the forest, i.e the number of trees in different size classes.
This use of the model was without scientific precedent
(5) and was correspondingly risky.

Unfortunately the model contained two significant errors, one essentially mathematical and the other logical. The formulae for growth from one size class to the next were unintentionally biased in a way that exaggerated the rate of increase in the number of trees, and hence the timber yield. For example, a forest in a steady state (natural mortality = recruitment) appeared in the TWC model to be increasing at about 20% every 15 years, which was about the size of the intended harvest

The logical error is that the TWC model assumes what it sets out to prove: that logging does not change forests. It does this by presupposing that harvesting "preempts" natural mortality, or in other words that low-intensity logging does not increase the total mortality rate in a forest. Sustainable harvesting of steady-state populations relies on the observation that deleted stocks tend to increase their per capita biological productivity. This is what ecologists call compensatory or "density-dependent" population growth. The assumptions in the TWC model amounted to a belief that they could obtain a density-dependent response from the forest without reducing its density.

I documented these concerns in a scientific paper published in late June
(7). The challenge was then to adapt the model so that it might be used in Carl Walters' words " ... to make predictions about the impacts of alternative policies". The result was a new, interactive model that has been available on the Landcare Research web site(8) since 10 August. The new model explicitly includes compensatory diameter growth (the intensity of the effect can be varied by the user). In other respects it is a mathematically refined version of the one used by TWC and uses the same input data.

We have yet to explore systematically the behaviour of the model - in fact, no such work is currently scheduled. However, two general conclusions can be supported:
The rate and extent to which harvesting reduces the number of large trees depends on how many trees are felled in each 15-year cycle.

Compensatory mortality (increased survival at lower stand density) was not allowed in the version on our website because it is unlikely to occur. Small coupe logging has previously been observed to weaken forests and to increase wind-throw and pinhole borer infestation. It is still unclear whether the net effect of logging on the survival of remaining trees in positive or negative. Nevertheless, preliminary trials show that even optimistic levels of compensatory mortality do not change our general conclusions.

How has modelling informed recent decision on beech management?

Did the decision process that ended in the approval of the beech scheme last week benefit from the improved transparency that modelling can bring to decisions of indigenous forest management? Rather little, it appears. To a large extent this was predictable from the differing professional and personal values of the various parties - interest groups, experts, TWC and the Ministry of Agriculture and Forestry(MAF). Models addressing goals specific to one group (for example timber yield) do not carry much weight with other parties.

Drafts of the TWC plans had been reviewed by a panel of forestry experts for the Ministry of Forestry (MOF). Their December 1997 reports were released under the Official Information Act in November 1998. Their comments are noteworthy in two respects.

Firstly, Dr G. Whyte identified as a "major deficiency" of the plans in general "lack of clarity in stating management objectives and how conflicts among them are to be resolved"
(9). Specifically (and I think rhetorically) he asked:
"[How] can the impact of sustained yield management on natural ecological processes be minimised and at the same time develop the forest resources in a manner that contributes to the economy of the West Coast region?"

Secondly, the experts were dismissive of the idea that TWC can or should retain the present forest structure. This reflected their professional enthusiasm for ":good forestry practice" emphasising timber yield. For example, Dr Nora Devoe's report
(11) stated:

"The New Zealand public must recognise that a managed forest is not the same thing as a nature preserve. A well-managed forest differs from an unmanaged one in several respects, not least of which are species composition, age structure, and growth rates, and for sound biological and economic reasons."

TWC adopted a strategy emphasising environmental sensitivity, and chose to "place politics first" by avoiding deliberate stand enhancement
(12). This was moderated slightly in the final plans by the inclusion of "improvement felling" (thinning). However, the final plans did not allay the fundamental concerns about incompatible goals. In November 1997(13),

"TWC was invited to indicate in plans the likely future status of the old (untouched) component of the stands, as portrayed by the model over a time sequence, to provide some basis for comparing the present and future structure of the forest".

This invitation was not taken up by TWC, and the final plans presented no time projections and little indication of how conflicting goals were to be reconciled

I believe that the public policy process surrounding the beech scheme decision has been quite unsatisfactory. The published accounts of the TWC modelling failed to depict the likely long-term outcomes of low-intensity logging. Modelling known to be inadequate
(15) was promoted by TWC as showing that logging could proceed without changing the forest structure. MAF published a Policy Discussion paper that implicitly endorsed the TWC modelling and omitted any reference to changing forest structure from the list of environmental effects(16). When inadequacies were revealed in public during the consultation process, TWC dismissed the criticisms as "asinine"(17)(b). MAF in turn dismissed the relevance of modelling inadequacies to the decision because they believed from other models that the volume yield was achievable(18). This opinion significantly omitted any reference to the ecological constraint of maintaining the old-growth forest structure. TWC have subsequently asserted that other as yet unpublished models justify their optimism(19).

Implications for the Forests Amendment Bill

My expertise does not extend to the legal issues surrounding forest management.
There does appear to be an opportunity through this Bill to promote more robust and transparent procedures for indigenous forest management.
This is likely to require the vigorous development and application of models.
Conservation groups generally do not have the resources to pursue technical issues, yet they de facto have substantial responsibility for safeguarding the public interest through the resource consent process in the Environment Court.
This situation might be partially corrected by requiring MAF to take a more active role in promoting and resourcing open and effective debate.

  1. Walters, C.J. 1997. Challenges in adaptive management of riparian and coastal ecosystems. Conservation Ecology (online) 1 (2):1.
  2. The plans published in September 1998 are available at
  3. This statement is based on the growth rates and current size distribution in Section 5 of the TWC Maruia Working Circle plan.
  4. Some of these risks are acknowledged in the TWC plans: others are documented in various submissions of November 1998, including those of the Royal Society of NZ, and of G. Elliott and others.
  5. A "Review report" on the draft TWC plans dated 5 December 1997 was prepared by Udo Beneke and Alan Griffiths for the Ministry of Forestry. It includes individual reports by six experts, and notes of a meeting on 27/11/97 between TWC representatives and some of the panel. At the meeting "(Ian James) noted that unfortunately this use of the model has not been published".
  6. TWC wrote to various MP's on 12 July and claimed of my published scientific paper that it "considerably down rates his original results". I suspect this misconception is based on a misreading of the paper, which in fact provides the basis for my present argument.
  7. Efford, M.G. 1999. Analysis of a model currently used for assessing sustainable yield in indigenous forests. Journal of the Royal Society of New Zealand 29: 175-184.
  8. The web address is The new model was presented at the annual meeting of the NZ Ecological Society in Blenheim on 2 July. Online documentation describes its technical differences from the TWC model. Except for compensatory growth these are derived from Efford (1999).
  9. Review report p. 51.
  10. Ibid.
  11. Review report p. 61.
  12. Review report p. 68.
  13. See also Hager, N. and Burton, B. 1999. Secrets and lies. Craig Potton Publishing.
  14. Review report P. 69.
  15. This is probably because the TWC model was not able to make such projections. It assumed a tree population kept at equilibrium.
  16. The modelling expert used by MoF, Dr Whyte, was consistently dismissive of the TWC modelling (e.g., Review report p. 54). Although some of his requests for more explanation were met in the final version, he later stated "My initial reaction is that you will never be able to do the calculation of sustained yield yourself, as the methodology is subsumed within a model which Ian James and TWC will not reveal algorithmically." (email to M. Efford 13/10/97)
  17. MAF Policy Discussion Paper 8. October 1988. Pages 14 - 16.
  18. Christchurch Press 27/1/98
  19. The ministry's response to submissions critical of the modelling was:
  20. "When MAF's Indigenous Forestry Unit (IFU) reviewed the plans, it was aware of the shortcomings in the model. The IFU was satisfied that the current volume projections for harvest were well below broadly estimated volume estimates providing a considerable safety margin while models are improved as part of adaptive management".
  21. "An analysis of public comments on the sustainable management plans for the beech/podocarp beech production forests of North Westland." Section 2.4 24/12/98.
  22. Kit Richards on Kim Hill Show National Radio 25/8/99.
  23. The amount of detail we include in a model depends partly on how much is known, but also on its purpose. Simple models that abstract the key elements of a system are often the most informative.
  24. Rather than follow through on its public commitment (press release) to have an independent scientist review the submissions.

Murray Efford
Landcare Research
Private Bag 1930
03 477 4050