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.

.

.

.

7.0   Beech tree harvesting rates and mitigating effects on wildlife

.

.

.

.

.

. 7.1   The importance of cavity-bearing trees

24.         This misrepresents the key point. There is no dispute that some of these species are obligate tree cavity nesters and roosters. What matters is whether these holes are in limited supply and influence abundance of birds or bats in anyway (sic) ( = any way?)

.

.

.

.

.

25.         These citations cover a huge range of species and habitats, and without exception are from overseas. The DoC Critique is inconsistent in that it earlier argued that overseas studies are unlikely to be relevant to local circumstances. It is logical and expected that preferences for particular sorts of holes and limitation of numbers could result, especially in diverse faunal assemblages crowded into thin marginal strips around clearfelled areas (like the tree dwelling marsupials studied by Australian ecologists). A review on the impacts of feral honeybees (tree cavity nesters) on wildlife in New Zealand (Moller & Butz Huryn 1996) and overseas (Butz Huryn 1997) could find no evidence of tree cavities being in such short supply as to cause competition with other wildlife. As always, lack of evidence does not prove that such an effect does not occur.

.

.

.

.

.

7.2   Availability of wildlife trees in the Eglinton Valley

26.         The 95% binomial confidence intervals on a sample of 1 tree (with holes suitable for bats) out of 78 trees are 0.033% to 6.77% (Mainland et al. 1956)   i.e. the proportion of trees with bat holes may have been 5 times higher than the average used by the DoC Critique, or 39 times lower. These wide limits emphasise the uncertainty in the preliminary calculations presented here by the DoC Critique and the value of further modelling and research. No error bounds were presented on the extrapolations in the DoC report. The expected outcome could therefore be very much safer or riskier for the bats than indicated by mean value extrapolations. The value of a more rigorous and detailed risk assessment is clearly indicated. The DoC Critique is at first quite clear that it signals a potential approach rather than using the models as a firm prediction of outcomes. Unfortunately it nevertheless uses the predictions of the model to assert that logging impacts on bats and birds will occur and are even likely. Planning for a more detailed risk assessment is underway by TWC.

27.         The use of such a large number of different holes by each social group is an exciting and remarkable finding of Colin O'Donnell's and Jane Sedgeley's work. Several bats shifted roosts regularly, often at almost daily intervals. This behaviour is markedly different from most overseas bats studied, perhaps in part because tree cavity dwelling bats have received little study until the New Zealand effort. The crucial issue for our present risk analysis is whether or not they need to shift so often amongst such a large number of different holes. Does the continuous mixing reflect ecological or behavioural needs, or simply preference? Does it reflect a superabundance of holes available? It may be that the bats move regularly so that they can gain information about one another for social organisation. If this hypothesis is true, then movement may be necessitated simply because a superabundance of holes exists, forcing the bats to shift more often to stay in contact. It is not known for sure what effect reduction in number of holes in the roosting zone might have on the bats.

.

.

.

28.         Merchantable trees exclude the very large hollowed trees that will be left for birds and bats. The upper size limit of > 80 cm is claimed by the DoC Critique to be a better compromise. A formal model can test a variety of such trade-offs and consider whether the 110 cm limit proposed by TWC will be adequate to safe-guard tree-cavity availability. Twenty-two percent of bat roosts in the Eglinton Valley were in standing dead trees. These were favoured roosting sites. The calculations in the DoC Critique excluded consideration of these standing trees that will be left in situ. The calculation in the DoC Critique therefore underestimated the number of holes that will be present at Maruia, assuming that the use of Eglinton data to predict outcomes in TWC forests is valid. The larger trees undoubtedly have more holes (Elliott et al. 1996b). These authors also demonstrated that the birds added use of larger trees was not a reflection of preference per se. They used those trees more often simply because they had more holes in them, not because the quality of the holes was different in some way. This is important when predicting logging impacts because it suggests that holes in all tree sizes are equally likely to be useable (sic).

29.         Allowance should indeed be made for loss of roost trees for long-term projections, but so too must there be allowance for growth of new roost trees. The calculations presented in the O'Donnell & Dilks (1987), O'Donnell (1991) and this DoC Critique are all static. The fundamental extrapolation has been that degree of tree removal in particular size classes represents degree of habitat removal - there has been no allowance for subsequent regrowth of the trees (or compensatory mortality, growth, regeneration rates triggered by tree removal), of retention of the very largest live stems and all standing dead spars, nor TWC's proposed rotational removal of ca 15% of stems every 15 years. The models are simplistic extrapolations from which exact predictions should not be attempted, in my opinion. This same caveat was mentioned at the outset by the DoC Critique, but then it nevertheless went on to conclude that the logging impacts were likely.

30.         The logic for why the Eglinton availability figure is a "prudent upper limit" for Maruia is not spelled out. If this is a target for management or conservation safety (as urged by the environmental precautionary principle), then is the DoC Critique really argueing (sic) that over double the predicted existing number of holes in Maruia will be needed to safeguard bats? How could this be when bats are obligate tree-cavity dwellers? Where then are half the existing bats roosting if only half the necessary holes are available?

31.         Again only one side of the coin has been emphasised. Those same processes are making new holes available for the bats.

32.         The same underestimates and problems noted 28 - 31 apply to the calculations for kaka and yellow-crowned parakeets.

7.2   Predicting harvesting rates of wildlife trees

33.         This sentence and Table 1 (at the end of this paragraph) capture my fundamental concern with the whole approach taken in the DoC Critique. The logic of the prediction is not made explicit in the DoC Critique, but it implies that all the holes expected to exist in the Maruia Working Circle now are all needed by the bats and birds. Colin O'Donnell (pers. comm. 6 October 1998) confirms that this is the assumption implicit in the DoC Critique. Further overestimation of the risk comes from the way the totals for each species are added together in Table 1. In reality the species are likely to be able to share many of the holes. For example, in the Eglinton mohua and parakeets used many holes with similar dimensions (Elliott et al. 1996b), although 30% of mohua sites were too small for parakeets. Amongst the unthreatened species mentioned, only the rifleman and short-tailed bats are obligate hole users, so it is not clear whether these add risk in the way inferred in the DoC Critique. The prediction of 14.2 cavity-bearing trees as the minimum needed for birds and bats is not scientifically defensible. The flawed nature of the calculation can be illustrated for the instance of kaka. Assuming that kaka need two holes per breeding pair, the predicted minimum density of kaka would be 3.5 kaka per hectare ! Exact estimates of the absolute density of kaka are not available even for closely studied populations in high density mainland situations, but (unfortunately for conservation) the density is likely to be two orders of magnitude below this level.

34.         Compensatory changes are expected of increased tree growth rates and increased survival as a result of live-tree removals. Bigger trees have more tree-cavities. The proposed model should explore the outcomes predicted for different assumptions in compensatory changes.

.

.

.

.

.

. 35.         There has been no demonstration that these numbers of holes are required.

.

.

While tree density estimates and proposed harvesting rates have been provided for the Maruia working circle, unfortunately there have been no surveys by TWCL to determine the frequency of availability of wildlife trees in the area.

I use research findings from mixed beech forest in the Eglinton Valley, Fiordland to illustrate:
(a) how the specific effects of the logging regimes proposed might be predicted;

(b) show that proposed harvesting rates in the Maruia working circle may remove significant amounts of wildlife habitat, particularly for threatened species (long-tailed bat, kaka, yellow-crowned parakeet).

7.1   The importance of cavity-bearing trees

One of the main predicted impacts of the harvesting regimes proposed would be the reduction in the number of older cavity-bearing trees (see below).

(See comment 24) TWCL dispute that availability of cavities is in fact a limiting factor (BSM, p. 69). Their assumption is wrong. If trees which contain the cavities which hole-breeding and roosting species require are removed there are simply no alternatives available. It is extremely unlikely that threatened species such as kaka and parakeets would be able to nest on the ground or in the open.
Bats would not be able to thrive outside of cavities or in cavities which did not provide suitable microclimates because of their highly specialised thermoregulatory needs ( e.g Racey 1973, Trune & Slobodchikoff 1976, McNab 1982, Kunz 1987, Roverud & Chappell 1991, Hamilton & Barclay 1994).

(See comment 25) There is a huge literature available on how the limitation cavity of availability (sic) influences the viability of wildlife populations ( a few examples: Saunders et al. 1982, DeLotelle & Epting 1983, Nilsson 1984, Raphael & White 1984, Rendell & Robertson 1989, Lindenmayer et al 1991 Rudolph & Conner 1991, Bennett et al 1994, Newton 1994, Lindenmayer et al 1995, Rieger 1996, Vonhof & Barclay 1996, Smith 1997).

7.2   Availability of wildlife trees in the Eglinton Valley

7.1.1 (sic) Long-tailed bats: Using information from Sedgeley & O'Donnell (unpublished manuscript) and an earlier study of roost trees availability ( Sedgeley & O'Donnell, in press), it is possible to estimate how abundant suitable roost cavities may be. Firstly, 95% of roosts are found in only at (sic) lower altitudes. Of 78 random trees assessed in this study, only 17 had cavities. These 17 trees contained a total of 51 cavities which conformed to the basic assumptions of being available to bats (i.e. the entrance and internal dimensions exceeded the minimum dimensions required by a single roosting bats). However, most of the measured available cavity characteristics fell outside of the range of bat roost cavity characteristics (i.e. were hollows, were damp inside, were below 5 m from the ground, entrance and internal volume were less than minimum dimensions of roost cavities, and cavities were completely surrounded by vegetation at less than 2 m away). Only six cavities had characteristics which were within the range characteristics of cavities used by bats and these were all located on one tree.
.

.

.

(See comment 26) Thus only 1.3% of random trees contained cavities of a type which bats select to roost in.
With available trees density at 298/ha in the Eglinton Valley this extrapolates to 3.8 cavity bearing trees/ha.
.

.

.

.

.

. (See comment 27) Bats have used a minimum of 304 roost cavities in the spring-autumn months during this four year study.
The majority of bat roosts were in red beech trees and 80% percent (sic) of red beech roost trees had stem diameters of 80 cm DBH, but only 34% of available red beech trees were in this size class.

These figures give us a range of estimates for the availability of bat roost trees in the Maruia working circle. Table 3.6 (p. 32, MSMP) indicates a
.

.

.

.

.

..

.

.

.

.

. (See comment 28) merchantable tree density
of 111 trees/ha.
(See comment 28) If bat roosts occur at a similar frequency to the Eglinton (1.3%)
then we expect 1.4 bat roost trees/ha of which the majority would be in trees > 80 cm DBH.
.

.

.

.

.

. (See comment 29)Allowance must also be made for natural death of bat roost trees. A prudent upper estimate might be 3.8 trees/ha
.

.

.

.

.

..

.

.

.

.

. (See comment 30)(the availability figure for the Eglinton).

We do not know how many additional cavities are used in the winter months and this model of abundance does not include estimates for cavity turnover. Little is known about the ongoing processes of wood decay and cavity formation. Most species of bat are not known to manipulate the structural environment of their roosts,
(See comment 31) so natural structural changes and degradation can frequently exclude or cause bats to abandon their roost sites (e.g. Rieger, 1996a, 1996b).

(See comment 32)7.1.2 (sic) Kaka: Using similar extrapolation procedures we estimated that 6.4% of trees in the Eglinton Valley contained cavities suitable for nesting (19.1 trees/ha). All kaka nests in the Eglinton (and elsewhere) were in trees > 80 cm DBH.

These figures give us a range estimates (sic) for the availability of kaka nesting trees in the Maruia working circle. Table 3.6 (p. 32, MSMP) indicates a merchantable tree density of 111 trees /ha. If kaka cavities occur at a similar frequency to the Eglinton (6.4%) then we expect 7.1 nest trees/ha of which all would be in trees > 80 cm DBH. A prudent upper (See comment 30 again) estimate might be higher than this. It is realistic to expect that greater densities of cavity-bearing trees will be required for birds because they breed as pairs rather than in colonies, as bats do.

7.1.3. (sic) Yellow-crowned parakeet: Using similar extrapolation procedures (based on cavity sizes from Elliott et al. (1996a) we estimated that 5.1% of trees in the Eglinton Valley contained cavities suitable for nesting (15.2 trees/ha). Eighty-five percent of nests in the Eglinton were in trees > 80 cm DBH.

These figures give us a range estimates (sic) for the availability of parakeet nesting trees in the Maruia working circle. Table 3.6 (p. 32, MSMP) indicates a merchantible (sic) tree density of 111 trees.ha. If parakeet cavities occur at a similar frequency to the Eglinton (5.1%) then we expect 5.7 nest trees/ha of which 85% would be in trees > 80 cm DBH. A prudent upper estimate would be higher than this.

7.2 Predicting harvesting rates of wildlife trees

By combining these estimates from the Eglinton Valley (Table 1) it is possible to gauge the potential impact of harvesting on threatened hole-using species in Maruia forests. For a more definitive model, data should be collected specifically from the North Westland forests.

Harvesting rates have been set at 1.009 trees/ha/year, of which 0.392 trees will be red beech. Harvesting rates for red beech trees > 80 cm DBH have been set at 0.105 trees/ha/yr (Table 5.5, p. 80, MSMP). Red beech trees in this category occur at a rate of 17/ha in the Maruia (Table 5.1, p. 73, MSMP). Thus 1.575 trees from a pool of 17 would be felled every 15 years.

(See comment 33) We predict that a minimum of 14.2 cavity-bearing trees/ha are required for threatened species
(long-tailed bat, kaka and yellow-crowned parakeet) for breeding and roosting (i.e. 83.5% of the 17 trees > 80 cm DBH/ha present in the Maruia Working Circle). Thus the rate of felling these trees at 0.105 trees/ha/yr indicates a high probability of removing 9.3% of preferred breeding trees/ha per 15 year rotation.
(See comment 34) Over 120 years (allowing for 15 year resting between each cycle), 37% of preferred trees would be removed. This is unlikely to be sustainable for wildlife in the long term given that trees take a minimum of 450 - 600 years to reach a size whereby they are used for breeding by threatened species (see Ogden 1978, Wardle 1984) ( or ca. 320 years to reach 80 cm DBH according to Table 5.2, p. 75, MSMP)

Table 1. Examples of predicting minimum number of cavity-bearing trees/ha
(See comment 35) required by threatened species in the Maruia Working Circle.