My comments refer to numbered locations marked on Appendix 4 (attached - see rhs column), continuing at Section 3 before returning to the Summary & Recommendations (Section 1) at the end.

5.0 Under-estimating the impacts of logging.

9.         This caveat is only very peripherally relevant to the proposed TWC sustainable forestry plans because logging will be so fundamentally different. As the DoC Critique emphasises, the past history of logging impacts provides a general precautionary tale (the TWC plan makes the same point), but direct comparison to the new method is strained and repeated over and again in the DoC Critique. Conservation NGO's have made public statements with the same misleading implication by referring to evidence of past impacts of very different logging styles. If a fair and balanced risk assessment is to be made by the New Zealand public in their appraisal of TWC's proposals, loose generalisations based on inappropriate comparisons should not be promulgated by DoC or other conservationists, even if they are proffered for the best possible intentions of protecting the environment.

10.         These two studies are of coupes or heavily selected logging regimes, that are therefore not relevant to TWC proposals for grouped-tree removals on comparatively small spatial scales. Grouped-tree removals may still reduce nest hole and food availability to some (probably very minor) extent, so that the level of recovery of populations following predator controls may be capped to some reduced carrying capacity by the logging now proposed.

11.         O'Donnell (1991) essentially contains the same approach, data and conclusions as O'Donnell & Dilks (1987). This study is the most relevant of the publised record to predicting impacts of the type proposed by TWC, but even so it is problematical to use it in this way.
Firstly, TWC do not propose to remove the largest live trees or standing dead spars.
Secondly, it is based on data from a different forest type (the DoC Critique challenges TWC for relying on data from overseas or outside their red beech forests within New Zealand, yet now does the same; in reality the comparison by each protagonist is forced by the lack of the relevant data and the only way to remedy this is to initiate relevant research in the proposed TWC sustainable management area).
Thirdly, it tests the model by using data from forests heavily modified by historical methods not proposed for the future.
Fourthly, Spurr's data used for the test of the model compared forests logged recently c.f. long ago (so a spatial comparison was used to infer temporal change; this assumes bird abundance was the same everywhere before logging and that logging was the only important ecological change between areas);
Fifthly, the predictions of the model worked for parakeets and yellowhead, but not for kaka (we should have little confidence in any model until its predictions are upheld for a replicated series of species or place, or until a plausible post-hoc hypothesis is tested for why the model should apply to one species and not another);
Sixthly and finally, the model assumes that the preferences initiated by the bird studies demonstrates need, and so retrieves the over-riding assumption that habitat variables limit the populations. All animals will show preference of some sort or other for particular foods or places, but this does not mean that such preferences have any tangible impact on their numbers. This over-aching(sic) (over-arching ?) assumption is acknowledged by O'Donnell & Dilks (1994) but not by the DoC Critique of TWC plans reviewed here. I agree that the type of investigation proposed by the DoC Critique would be a valuable first step, and therefore urge TWC to consider doing such a study, but I caution that the study will not by itself be sufficient to clinch the key question: i.e. a conclusion about whether the proposed logging will reduce future bird or bat numbers to any significant degree.

12.         The model did not predict trajectories for changes in kaka numbers.

13.         These two sentences are a scant balance to the several above and following that imply logging impacts are likely from past evidence. The very next sentence re-iterates the same misleading implication. Equally, there is a need for detailed research of the new logging style (group-tree removals) on birds and bats. Common sense and our knowledge of forest processes suggest that these impacts are likely to be much smaller than previous methods, but vigorous research and monitoring of the new method is still needed and necessary as part of the environmentally precautionary principle embraced by TWC in its planning. Fortunately the low intensity approach proposed by TWC gives ample opportunity to do such research before major alteration is imposed on the system.

14.         Appealing to "the wide variety " of historical logging techniques as evidence for their potential to predict outcomes of TWC proposals is misleading because it omits mention that the new style of logging proposed is so fundamentally different. New Zealand has gone from large-scale clearfelling to smaller and smaller clearfells, and from logging that seriously alters species and size structure of the forest community to the current proposal to mimic natural forest structure profiles. The latter is precisely the approach taken by foresters in Europe for centuries, and just the approach that is needed to minimise potential risks to biodiversity in New Zealand (Beneke 1996, Kohm and Franklin 1997, Lindenmayer & Franklin 1997).

15.         This is precisely the approach signalled by TWC (but this was apparently not included in the draft plan critiqued by Colin O'Donnell).

16.         Proposals to predict outcomes and measure risks from uncertainties (using mathematical simulation models and a "sensitivity analysis" are planned).

17.         The assertion that logging impacts will necessarily be underestimated is assumption. It may be that the biodiversity impacts overseas may be just as extreme or worse than those expected in New Zealand; and (as partly acknowledged two paragraphs earlier) the types of logging proposed by TWC are so different from those researched in the earlier New Zealand case studies quoted by the DoC Critique that their relevance is highly questionable.

18.         All this is fair, but not relevant to the style of logging proposed by TWC. The DoC Critique's criticism here was spurred by the absence of warnings about potential impacts of logging in the Maruia Plan (these were however covered in the overview plan) and a lack of mitigation measures in the Maruia plan itself (Colin O'Donnell in litt., 5 October 1998).

6.0 Mitigation of impacts of harvesting on threatened bat species.
19.         It is agreed that the threats to birds are potentially different from those for bats. Very important in this regard is the lack of detailed information one way or the other about the importance of predation on bat numbers. If they are not vulnerable to predation their populations may be much closer to the carrying capacity set by food availability or roost hole availability than expected for the bird species, DoC's bat recovery programme (Molloy 1995) lists predation by rats, stoats and feral cats as a threat to bats. Competition for insect prey with super abundant social wasps is potentially very important in honeydew beech forest areas (Moller & Tilley 1989; Thomas et. al. 1990; Harris 1991).

20.         Colin O'Donnell points out that bats are not like birds in that they go into torpor for some months of the year and so can not fly away. A high proportion might then be killed if a key nursery tree is felled. The proposed tree-cavity model can more rigorously calculate this risk after incorporating the bats selectivity, the tree size selections used, and the months of the year in which felling is proposed. Colin O'Donnell and Jane Sedgeley and team have detected in one population in the Eglinton Valley that at least 3 separate social groups exist that restrict their roosting to a much smaller area within a very large range. Other social groups may have been present (Colin O'Donnell, in litt. 5 October 1998). Large areas of that range may have had no roosting by other social groups, or all might have been used. Replicate studies are needed to determine if this is a general pattern for short-tailed bat populations. There is no demonstration that the bats could not move or extend their areas to include new holes being formed in the sustainable forestry area, or nearby. Most importantly, the TWC protocols are striving to maintain a large number of holes by various devices. Had coupe clearfelling been proposed then a risk of hitting and seriously disrupting the roosting area would exist, but this is not proposed. TWC might still consider it prudent to mount a study to identify roosting areas in their forests, but this is a large task ( a minimum of two years work from a dedicated and experienced team) and in the end it will not prove one way or the other what the putative logging impacts might be. It represents a significant start towards assessing and potentially mitigating any putative impacts though. A formalised model of tree hole availability and the impacts of logging on it will objectively measure these risks long before any putative changes are imposed on the system. The model can also test various mitigation measures, including revising tree selection criteria, seeding of new holes by drilling starting holes and inoculating them with fungi etc.

21.         Rapid and thorough work principally by Colin O'Donnell and his team has highlighted this change of emphasis to concern for South Island long-tailed bats as well founded and in urgent need of attention.

22.         The relative importance of long-tailed bat populations on TWC land compared to surrounding DoC land, or populations elsewhere in South Island, has not been demonstrated nor measured (at least no published account of such a demonstration exists). I urge TWC to consider contracting Colin O'Donnell's team to allow a better regional perspective to be gained. Some of the data from Rhys Buckingham's surveys could help plug the gap ( they also use similar standardised techniques), but new intensified survey effort may also be called for.

23.         It is logical to expect that short-tailed bats roost hole requirements will be different, and a paper is in press with New Zealand Journal of Ecology demonstrates that they use larger diameter stems than do long-tailed bats (Colin O'Donnell, in litt., 5 October 1998). However, this evidence is very distinct from any expectation that such holes will be in limited supply, either before or after logging, or in the presence of both bats. There has not yet been an analysis of DoC's wider survey data for both bat species to test whether there are fewer or more bats in heavily cut over or coupe logged sites.

The Department of Conservation provided detailed evidence on the impacts of logging on forest wildlife to the Parliamentary Commissioner for the Environment in 1995. In summary:

(See comment 9) Logging, as done in the past has clearly caused considerable impact on wildlife, particularly threatened species (See reference list in O'Donnell, 1991 and Unpublished NZ Forest Service studies from central Westland). A review of the historical effects of logging on forest birds (O'Donnell 1991) indicated that obligate forest-dwellers disappear from logged forest. The occurrence of six key bird species in 17 comparisons between logged and unlogged forest was summarised by O'Donnell. Kaka and rifleman were absent from 10 of the 13 sites where they formerly occurred. Kaka occurred in lower numbers at the remaining three sites and rifleman at one. Parakeets were absent from five logged areas and present in lower numbers at the other seven. Although brown creeper, tomtit, and NZ robin were usually still present in logged forests, their numbers were frequently lower.

Predictive studies have indicated that coupe logging and individual tree selection techniques could cause significant impacts on native wildlife (O'Donnell & Dilks 1987,
(See comment 10) Spurr et al. 1992, Warburton et al. 1992).
.

.

.

.

.

.

.

.

.

.

(See comment 11) A model for predicting the impacts of logging on forest birds was developed by O'Donnell & Dilks (1987 & unpubl.).
The model analysed the degree of overlap between the trees preferred for foraging and those targeted for logging, and predicted the proportions of preferred habitat lost at different (tree - Ed.) extraction (felling - Ed) rates. A comparison between predicted loss of habitat and actual declines in bird numbers from logging silver beech using Spurr's 1987, 1988 data for Rowallan forest indicated that beech management would be highly detrimental to key forest birds
(See comment 11) (O'Donnell 1991).
.

.

.

.

.

(See comment 12) Selection logging of rimu, even in very small amounts would also have severe impacts on kaka and parakeet (O'Donnell & Dils 1987).
.

.

.

.

.

.

.

.

5.2 Predicting impacts of new harvesting regimes

"Avifauna remains one of the greatest unknowns in terms of the effects of any management" (BSM, p. 169).

(See comment 13) TWCL are using a new harvesting technique for North Westland forests so obviously there are no data on its impact on wildlife. Therefore, at this stage we can only use predictive approaches to assessing potential impacts.
However, the fact that a
(See comment 14) wide variety of historical harvesting techniques
have all led to significant declines in numbers of theatened species serves as a warning. Recent surveys by TWCL confirm that lower numbers of great spotted kiwi, kaka, parakeet, rifleman, and robins occur in the most recently heavily logged forests (BSM, p.70).

The BSM largely deals with effects of logging on foraging habitat for birds - but not roosting or nesting habitat, and not bat habitat. We need to go some way towards predicting impacts at this stage (see below) but if the logging is to go ahead then a
(See comment 15) rigorous research-by-management approach needs to be adopted and specific logging prescriptions need to be changed as necessary if adverse impacts are detected. Studies of wildlife foraging, breeding and roosting preferences

(See comment 16) in red beech forest will be required to make accurate predictions.

In contrast, however, the MSMP (p.29), the document which should be specifically outlining mitigation procedures to reduce these effects, chooses not to acknowledge the New Zealand research. Rather, it chooses to quote overseas papers in tropical forest types where bird communities and forest types are completely different.
(See comment 17) Thus the MSMP under-rates the potential impacts of logging.
Quoting tropical studies which show only localised loss of wildlife after logging while ignoring New Zealand studies (see above) which show significant impacts seems inappropriate.
(See comment 18) It is well known that logging can increase diversity of forest birds per se (including in New Zealand e.g. McLay 1974). However, increased diversity results from more edge species, introduced species, and honeyeaters temporarily if there is an increase in the numbers of successional plant species (especially along roads). These "increases" in diversity are at the cost of endemic, specialised and threatened species (e.g. O'Donnell 1991).

The MSMP should specifically address the conflicts raised in the BSM, rather than ignoring them.

While ranging over 50 km², including forest edges, riversides and grasslands, longtailed bats always return to traditional patches of forest as small as 2 km².
(See comment 20) The implication of this is that unless these roosting patches are identified and protected then there is a high risk of localised habitat modification wiping out a whole population.

Sampling of trees available to long-tailed bats in the Eglinton Valley, Fiordland, indicated that suitable trees were rare in the forest. Roosting trees were concentrated in lowland forests on river terraces and outwash fans on the valley floor (Sedgeley & O'Donnell in press). Few bats were present in higher altitude forests. Long-tailed bats selected trees which were > 80 cm DBH.

There is an active recovery programme being run by the Department of Conservation (Molloy 1995). The Recovery Group is becoming more concerned about the conservation status of long-tailed bats. Preliminary survey work on long-tailed bats has indicated that numbers may be much lower, at least in some regions, than was previously thought,
(See comment 21) and populations could be declining. Recent work investigating the genetics of long-tailed bat populations indicate that North and South Island populations are distinct, and there is greatest concern for the viability of South Island long-tailed bats. Intensive work has indicated that long-tailed bats are very rarely encountered in many areas, even when considerable survey effort has been made. They are very rare or absent at several South Island sites where they were widespread in the 1960's.
(See comment 22) This means that conservation of bats in the TWCL areas is of very high priority.

A second threatened bat species, the short-tailed bat may be present in North Westland forests. This species has recently been re-discovered in the Karamea (conformed) (sic) and Punakaiki (unconfirmed) areas, as well as the Eglinton Valley in Fiordland (O'Donnell et al. in press).

(See comment 23) If short-tailed bats were to be found it is likely that they would have different tree roosting requirements than long-tailed bats and different models of the potential impacts of harvesting on this species would need to be developed. For example, preliminary data on roost cavity characteristics of the Eglinton Valley show that although there may be some overlap in roost cavity size between the two species, most short-tailed bat roosts have larger entrances and internal dimensions and are located closer to the ground (O'Donnell et al in press). The agility of short-tailed bats on the ground and its maneuverability (Webb, Sedgeley & O'Donnell, in press) enable it to roost in lower cavities and basal hollows which long-tailed bats do not use.