TANE 29, 1983 THE SCIENTIFIC RESERVES OF AUCKLAND UNIVERSITY. II. QUANTITATIVE VEGETATION STUDIES

by John Ogden Department of Botany, University of Auckland, Private Bag, Auckland

SUMMARY

Forest vegetation surveys, using both plots and point-centred-quarter (plotless) methods, were undertaken by students in the University reserves at Swanson, Huapai, Oratia and Leigh over the period 1979 to 1982. The two methods gave similar estimates of species composition, total basal area and total density, but both gave wide confidence intervals on the parameters being estimated. The "leading species", in terms of both density and basal area, are defined for each area. The differences between the rankings for these different vegetation parameters are commented upon A regression of age on height of kauri ( australis) seedlings is presented. The relationship between age and trunk diameter is discussed. It is shown that in cores taken from kauri poles in ricker stands over the size range 10 to 40 cm DBH the relationship between age and diameter is very weak. When a wider size range is employed a highly statistically significant relationship is obtained. It is concluded that the highly skewed size class frequency distribution of kauri, characteristic of such stands, reflects a competitive hierarchy. The senile population structure of kanuka (Leptospermum ericoides) in ricker stands confirms their serai role in forest development. The basal area estimates are discussed in the context of other figures for forests. It is concluded that although the values are high (average 70 m2 ha 1) they are nevertheless below those commonly found in mature kauri forest.

INTRODUCTION

The natural vegetation of the Waitakere Ranges is discussed by Esler (1983) in this volume. Before the advent of European settlement the ranges were covered with temperate forest dominated by kauri (Agathis australis) or mixtures of podocarp and broad-leaved hardwood trees. Much of this bush has been milled and/or burned, and is now regenerating through Leptospermum scrubland (described by Esler and Astridge 1974). Consequently much of the vegetation is serai, and, if undisturbed, will change in structure and species composition in the future. This is particularly true of the dense kauri "ricker" stands which characterise many ridges. These stands are mostly in the process of 163 replacing Leptospermum communities (usually L. ericoides), and are recruiting seedlings and saplings of hardwood species which may ultimately form the understorey of mature kauri communities. The early stages of this process have been documented by Mirams (1957). This paper describes some of the results of University field courses on quantitative descriptive forest ecology in the University reserves at Swanson, Huapai, Oratia and Leigh in the years 1979 to 1982. The aims of the field courses have been (in part): (1) to compare plot and plotless sampling techniques for estimating forest structure and composition; (2) to provide some quantitative descriptive parameters for the areas being studied; (3) to describe the population structures of the kauri ricker communities on the sites. The detailed locations of the sites, and general descriptive and historical information, are given in the first paper in this series (Thomas and Ogden 1983 - this volume). The aim here is to introduce the field methods employed, and to present some of the results obtained. These results comprise quantitative descriptions, both of the general nature of the forest community, and of some of the ricker stands occurring within the reserves. In view of the changing nature of the vegetation the data presented may become of increasing interest, although the variety of techniques employed limits their value in this respect. Working with a class of 20 or more students poses problems of standardisation, both of detailed technique and of species identification. However, a large amount of quantitative data can be collected quickly and this, to some degree, compensates for its variable quality. The problems can be reduced by careful briefing beforehand, and by standardised data collection sheets. Sometimes inconsistencies recognised during analysis can be allowed for, or corrected, but usually individual differences in measurement or recording accuracy are reflected in large standard errors on the parameters being estimated.

Descriptive parameters in forest ecology The full description of forest vegetation requires data on the relative abundance and spatial pattern of the various tree species composing it, the quantity of material present () and its disposition through the different vertical strata. Understanding the dynamics of a forest requires also data on rates of productivity, energy flow and nutrient cycling, and some insight into past historical influences. The "abundance" of a tree species can be described by three parameters: (1) Density: Number of stems equal to or greater than some chosen minimum diameter per unit area. (2) Basal area: Sum of the cross-sectional areas of the trunks equal to or greater than the chosen minimum (at a specified height), again 164 expressed per unit area. (3) Frequency: The proportion of plots (of a specified size) in which the species occurs. All three parameters can be transformed into relative values by expressing them for each species, as a percentage of the sum for all species. Such relative values can be summed to give "importance values". Because density and basal area are largely independent, measure different attributes of species, and (unlike frequency) can be applied to the vegetation as a whole, we have concentrated on these two parameters. Further discussion can be found in most text books of quantitative ecology, (e. g. Greig-Smith 1964, Mueller-Dombois and Ellenberg 1974) and in Ogden and Powell (1979).

METHODS

Plot sampling Most of the vegetation description work was done at Swanson and Huapai in February 1980, and at Leigh in 1981. A single large (subdivided) plot was enumerated at Oratia in October 1982. Plot sampling employed 7 x 7 m plots arranged in a stratified random manner over a 1 ha area at both Swanson (24 plots) and Huapai (27 plots). At Leigh 28 10 x 10 m plots were arranged randomly along three transects running at different heights along the contours, and covering the whole of the bush patch. Plot sizes were determined by prior sampling, and represented a compromise between the "minimal area" required to ensure inclusion of most species and the practical difficulties of enumerating large plots. A minimum stem size of 10 cm "diameter at breast height" (DBH) was set. Tree ferns were recorded.

Plotless sampling In most cases point-centred-quarter (pcq) sampling was performed on a square grid of 25 points located at 20 m intervals along five parallel transects 20 m apart. Thus the total sample covered approximately 1 ha. Each point was marked temporarily with a stake, and the space around it conceptually divided into four quadrants using the cardinal compass points. In each quadrant the nearest tree or tree-fern ( > 10 cm DBH) was chosen and identified. Its diameter and distance from the point was measured. Thus the final data sample usually comprised 100 trees identified and measured (DBH) and 100 point-tree distances. The latter were used to estimate tree density as follows: mean point-tree distance = m (in metres); mean area (MA) per tree = m2; density (stems per ha) = 10 000/5i2.

165 Note that the individual point-tree distance measures can be used to calculate a standard error and confidence limits for m, and hence for the density estimate. Thus an estimate of the reliability of the density estimate can be obtained. When 100 trees are included in the sample the relative density of each species is simply the number of times it is recorded. Basal area is obtained by calculating the basal area of each tree using If r2 (r = radius at breast height) and obtaining the mean. The mean tree basal area is then multiplied by the mean density per ha to obtain an estimate of basal area per na. There is some evidence that basal areas obtained by this method overestimate the true value. This appears to arise primarily from the non-random distribution of the smaller size classes, which tend to be aggregated in the "gaps" between the larger trees (Mark and Esler 1970).

Size frequency distributions in ricker stands Detailed studies of the kauri ricker stands were made at all sites, enumerating stems ^2. 5 cm DBH in single rectangular plots divided into contiguous sub-plots. The total area sampled at each site, and the dates, are given in Table 4. As the original aim of this sampling was to obtain DBH frequency distributions for kauri and kanuka these species only were recorded at Swanson and Huapai in 1979 and 1980. Subsequently (in 1981) a repeat sample was taken in the Huapai ricker, and further samples at Leigh and Oratia. In all these latter cases all species with stems >2. 5 cm DBH were recorded.

Tree ages Ages of kauri trees were obtained from increment cores. The cores were mounted and sanded to a fine polish so that individual cells and annual ring boundaries could be clearly seen using a low-powered binocular microscope. Twenty cores were obtained from the Swanson ricker, ten from Matheson's bush, and 34 from Huapai (including cores from some older trees on the ridge at Huapai). At Oratia two trees only were cored. Age was estimated from the number of rings, with a correction for the missing portion where the core failed to transect the pith. From each site (and some others) a few seedling kauris were cut at ground level. Sections were taken from these seedlings at various heights and the number of rings counted. From these data an average height growth rate was calculated and an addition made for the number of years required for growth to coring height.

166 RESULTS

Quantitative description of forest structure Total basal area and density estimates (of trees > 10 cm DBH) are given in Tables 1 and 2. Basal area is generally considered to be a better estimate of plant biomass than is the number of individuals (Mueller- Dombois and Ellenberg 1974) but both parameters are required for an adequate description. Total basal area estimates range from 72 to 109 m2 na 1 at Swanson and Huapai (Table 1). These values are higher than those recorded in the

Table 1. Summary of basal area and density data for all stems > 10 cm DBH at all sites.

Swanson Oratia Huapai Leigh Location'1' S S S P R R G U M G Method (2) P P P pcq P pcq pcq pcq pcq pcq Basal area m2 ha 1 73 75 42 108 81 79 42 83 44 Density ha 1 1590 1258 1600 2117 1701 1556 1569 1365 772 671 95% C. L. (%), 3) 19 27 - 25 19 23 40 23 29

(1) S = slope; P = plateau; R = ridge; G = gully; U = upper slope; M = mid-slope. (2) p = plots. Note that in the case of Oratia a single large plot 20 x 50 m was used, pcq = point centred quarter method. (3) A mean and standard error can be calculated for the point-tree distances. A squaring is required to convert these values into "mean areas" and this results in the confidence limits for the density estimate being asymmetrical about the mean. The values given are the range (minimum to maximum) of the 95% confidence interval, divided by two and expressed as a % of the mean density. Note also that the true 95% confidence limits on the- basal area estimates must be equal to or greater than those given for density, but cannot be satisfactorily estimated due to the highly skewed frequency distributions of basal area per tree values.

ricker stands in the same areas, and at Oratia and Leigh (cf. Table 5). The two lowest values (42 m 2 ha1) are both from ricker communities. The low value for the gully bottom at Leigh may be an artifact arising from the difficulty of accurately measuring distances on steep terrain. The density of stems > 10 cm DBH ranges from c. l 200 to 2 200 ha 1 at Swanson, Huapai and Oratia. The density estimates have wide confidence limits, so that no significance can be attached to the slightly lower values recorded by pcq compared to plot samples. The lower values recorded at Leigh are in accord with the more open nature of this small degraded bush patch, but may have been underestimated due to the steep terrain, especially in the lower part of the gully. At both Huapai and Leigh there is a tendency for higher density values on the higher parts of the site, associated with kauri rickers and Leptospermum communities. However, the high apparent density of the kauri rickers arises primarily from stems < 10 cm DBH (cf. Table 4). Tables 2 and 3 give the leading species in each sampled area in terms of density and basal area respectively. The data on which they are based

167 are given in greater detail in the appendices. Numerical dominance (Table 2) generally goes to tree ferns (especially Cyathea dealbata) and other components of the understorey (Coprosma arborea, Rhopalostylis sapida, and Leptospermum ericoides), although some canopy species (Agathis australis, and Vitex

Table 2. Leading species » 10 cm DBH in terms of density at all sites (for details see Appendix 1).

Leading species in rank order'11 Percent of total density accounted for by these Site 1 2 3 Swanson slope Cya Aga aus 61 Oratia slope Cya Aga aus 72 Huapai plateau Cya 62 Huapai ridge, 2> Cop 49 Huapai gully sap 63 Leigh upper slope Cya Aga aus 86 Leigh mid-slope Cya Aga aus 74 Leigh gully Cya 55

(1) Abbreviations are first three letters of generic and trivial name. (2) Average of plot and pcq data. lucens) are also numerically important on some sites. Dominance in terms of basal area (Table 3), on the other hand, is shown by the larger longer-lived trees, especially kauri and (Vitex lucens). The two parameters emphasise different aspects of the vegetation, reflecting its three dimensional structure (and its temporal behaviour). Tables 2 and 3 must be studied concurrently if an impression of the vegetation is to be gained. Thus the "plateau" area at Huapai is characterised by abundant stems of silver fern (Cyathea dealbata), Coprosma arborea and nikau

Table 3. Leading species > 10 cm DBH in terms of basal area at all sites (for details see Appendix II).

Leading species in rank order 111 Percent of total basal area accounted for by Site 12 3 these 3 spp. Swanson slope Aga aus 71 Oratia slope Aga aus 79 Huapai plateau Aga aus 70 Huapai ridge (2) Aga aus 71 Huapai gully Pod dac 61 Leigh upper slope Aga aus 91 Leigh mid-slope Aga aus 91 Leigh gully 66

(1) Abbreviations are first three letters of generic and trivial name. (2) Average of plot and pcq data.

168 palms {Rhopalostylis sapida) forming a low understorey. Puriri and kauri are much less abundant, but, because they are present as large trees, they comprise a large proportion of the basal area. In no case (except the species-poor upper stand at Leigh) are the three leading species in terms of density also the leaders in terms of basal area. This is a common feature of this type of "stratified" forest community. Likewise, relatively few species are abundant, and the majority of species are relatively infrequent in their occurrence. The three leading species numerically comprise between 55 and 86 percent of all stems > 10 cm DBH (Table 2). The three leading species in terms of basal area usually exhibit even greater dominance, with values ranging from 61 to 91 percent of the total stand basal area. As the average number of species in the samples was 20 at Huapai, 16 at Swanson, 11 and Leigh and 8 at Oratia (from Appendix I) this implies that the remaining basal area (and density) is shared between many species. The apparently high at Huapai compared to Swanson is of interest in view of the less disturbed nature of this site (Thomas and Ogden 1983 - this volume) and the high overall basal area (Table 1). In Tables 2 and 3 the plot and pcq data for the ridge at Huapai have been averaged to obtain the three leading species. This procedure is justified by the close correlation between the species composition data obtained by the two methods. The correlation coefficients were 0. 944 for basal area and 0. 824 for density (n = 27; P< 0. 001). This result is in agreement with other comparisons between plot and pcq methods made by the author; the two methods usually give total basal area and total density estimates which have overlapping standard errors, and usually rank the first three (or more) species in the same order.

Kauri ages and growth rates During the course of these studies over 60 increment cores have been taken from kauri trees, usually 2 per tree. When a core bisects the pith it is possible to count the annual rings, and (assuming all rings are annual and none are missing) to arrive at an estimate for the number of years the tree has existed since it reached coring height. Where a core does not reach the pith, the number of years represented by the "missing radius" must be estimated. This is done by calculating the length of the missing radius from the tree diameter and core length, and using the average growth rate of the innermost 20 years to calculate the number of rings which it represents. Clearly, inaccuracies can easily be introduced by these procedures, and the age obtained must generally be regarded as simply the best estimate available (Ogden 1980). The relationship between seedling height and age is illustrated in Fig. 1. The 13 seedlings sampled all came from ricker stands. Fig. 1 implies that if a core from a larger tree bisects the pith at a coring height of 1. 0

169 60

Oi , . , j 12 3 4 HEIGHT IN METRES

Fig. 1. The relationship between seedling height and age in kauri seedlings taken from ricker stands, (y = 9. 58 + 0. 118x; where y = age in years and x = height in cm; r = 0. 841, n = 46, P< 0. 001). m, then on average a further 21 years must be added to obtain the germination date (y = 9. 58 + 0. 118x; where y = age in years and x = height in cm; r = 0. 841). Extrapolation from this graph contains the hidden assumption that the growth rates of the larger trees, perhaps ricker dominants, were similar, when they were seedlings, to those of current suppressed seedlings in the ricker. This assumption has not been rigorously tested, but there is no doubt that some ricker dominants were very slow growing as seedlings, and that a very wide variance in growth rates is typical of the seedlings. It is possible that errors of up to 10 years could be introduced here, but this becomes less significant with older trees. Estimated age/diameter data from kauri trees at Swanson and Huapai are presented in Fig. 2. Independently both sets of data show highly significant relationships between age and diameter (r = 0. 930, n = 21, p< 0. 001, Huapai; r = 0. 899, n = 18, p< 0. 001, Swanson). However, the relationship has a very wide variance, especially at larger diameters, making it difficult to predict age from diameter except in very general terms. Within the 10-40 cm diameter size classes there is no significant difference between the Huapai and Swanson age estimates,

170 600 O HUAPAI

• SWANSON

500

> 400

200

100

50 100 150 200

DIAMETER (CM)

Fig. 2. The relationship between trunk diameter (DBH) and age in kauri trees at Huapai (open circles; r = 0. 93, n = 25, P < 0. 001) and Swanson (solid circles; r = 0. 89, n = 16, P < 0. 001). The regression line indicated was calculated for the 10 to 40 cm diameter size classes only, from the two ricker stands; it is defined by: y = 99. 377 + 0. 823x, where y = age in years and x = DBH in cm (r = 0. 512, n = 16, P< 0. 05). Superimposed points shown once only for clarity. and the relationship between age and diameter is very weak (r = 0. 512, n = 16, P=< 0. 05).

Ricker dynamics In all four cases kauri diameter class distributions in the rickers were highly skewed, with mortality (represented by dead stems) being recorded only in the smallest size classes (Fig. 3). It seems reasonable to assume that the larger kauris in the ricker stands (excluding a few isolated individuals which are much larger) are the oldest, so that they can provide a date for the establishment of the ricker. Smaller stems include suppressed individuals of the same cohort (age class) and later arrivals. In three rickers the largest individuals were in the 30-40 cm diameter size class. In the fourth, at Oratia, a few larger individuals occurred. The weak relationship between age and diameter in the 10-40 cm DBH class suggests that the ricker kauri populations represent an initial wave of recruitment which is currently undergoing competitive thinning and developing the log-normal size frequency distribution typical of such situations (White and Harper 1970). The numerous small spindly individuals are predominantly slightly later arrivals which will be

171 KAURI KANUKA

SWANSON

HUAPAI

>• O z o . r 1 hi a C H ti fa SO

LEIGH to-

ORATIA si I1

10 20 30 40 MO SIZE CLASS - DBH (cm)

Fig. 3. Size frequency distributions for kauri and kanuka in four ricker stands. Shaded areas represent dead stems.

172 weeded out in competition with their elders. However, it is interesting to note that many of these seedlings and saplings are themselves of considerable age, and might conceivably be "released" to fill gaps in the event of damage to the canopy. The similarities between the size frequency distributions for kauri and kanuka at Swanson and Huapai (Fig. 3) are further emphasised by their similar ages. It seems safe to postulate that both rickers commenced development during the 1860—70 period at which time both areas would have been covered by manuka {Leptospermum scoparium) and/or kanuka scrub regenerating after earlier fires, probably in the 1840s and 50s. Manuka commonly preceeds kanuka in such successions (Trevarthen 1952, Mirams 1956, Esler 1967, also personal observations at Kauaeranga Valley), but even when both species commence together the proportion of kanuka generally increases due to its greater height growth and longevity (Burrows 1973). Stem density and basal area data for all the rickers are given in Tables 4 and 5. The relatively high density of small stems ( < 10 cm DBH) has been commented upon earlier. In keeping with the youth of this community, larger sized individuals are not very numerous and the total basal area is generally lower than that in mature forest (cf. Tables 1 and 5). The relative mortality figures give a crude indication of the thinning process, but clearly illustrate the different dynamic status of the two species populations. The kanuka populations in the rickers at Swanson and Huapai are composed at present of senile individuals, the largest over 130 years old, with mortality common in the smaller sizes. Although some recruitment may have occurred until c. 1900 there has been almost none since then, and the majority of individuals seem likely to die within the next 20 years. The kauri and kanuka populations in the rickers at Oratia and Leigh are similar in almost all respects to those at Swanson and Huapai, but

Table 4. Density data for four kauri rickers.

Site Swanson Huapai Huapai Oratia Leigh Area sampled m2 880 410 500 1000 600 Date Feb 79 Feb 80 Sep 81 Oct 82 Feb 81 Density ha 1 (living stems); All stems > 2. 5 cm DBH 2500 6414 6100 6020 3717 All stems > 10 cm DBH 1100 1390 2180 1610 1417 Kauri > 2. 5 cm DBH 2102 5951 3460 1790 2283 Kauri >. 10 cm DBH 636 927 820 360 900 Kanuka ^ 2. 5 cm DBH 398 463 240 480 417 Kanuka > 10 cm DBH 375 463 220 290 317 Relative mortality (%)0): Kauri 9 5 10 9 5 Kanuka 35 5 17 29 46

(1) Number of dead stems ^ 2. 5 cm DBH expressed as a percentage of the total.

173 Table 5. Basal area and age for four kauri rickers.

Site Swanson Huapai Huapai Oratia Leigh Basal area m2 ha 1: All stems > 2. 5 cm DBH 57. 4 50. 6 73. 6 49. 3 51. 7 All stems > 10 cm DBH 55. 0 43. 6 63. 4 41. 8 47. 6 Kauri > 10 cm 45. 8 32. 9 40. 8 20. 5 27. 6 Kanuka > 10 cm 9. 2 10. 7 4. 3 5. 0 13. 9 Other spp. > 10 cm nr nr 18. 3 16. 3 6. 1 Estimated age of ricker: Mean age of kauri 10-30 cm DBH 121±9 110±8 - c. 90 c. 80 Number of cores 8 7 0 2 1« 10 <>2

(1) A 24 cm DBH tree gave an estimated age of 80—110 years. The site was reputedly cleared of large kanuka in 1898 but kauri seedlings were left (Shirley 1968). (2) Additional ages from fire scars on a large kauri, and from an associated totara. the stands have lower overall basal areas and are probably younger (80—100 years old). As anticipated in a younger stand, kanuka represents a higher proportion of the basal area at Leigh. The high rate of kanuka mortality here may reflect intense competitive thinning. The lower representation of this species at Oratia may be accounted for by Shirley's (1968) comment that the area was cut over for kanuka firewood in 1898, but the regenerating kauri was avoided.

DISCUSSION

Although the basal area values recorded here may be high by world standards (Franklin 1965) they appear to be within the range recorded by other workers in New Zealand. For example Mark and Esler (1970) give values ranging from 55 to 84 m 2 na1 based on plots in various beech (Nothofagus) forests. These authors found that pcq sampling of the same stands gave generally higher basal area estimates, ranging from 55 to 107 m2 ha"1. Mark and Scott (1965) have briefly reviewed the basal area data available from Fiordland (mainly Nothofagus menziesii forest) and shown that values are often in excess of 100 m2 ha'1, and may exceptionally reach 200 m8 ha ~\ Similar high basal areas were recorded by Ogden and Powell (1979) in dominated forest in Tasmania. Other workers have recorded values ranging from 41 to 81 m2 ha 1 in beech forest (June and Ogden 1978, Westerkov and Mark 1968, Wardle 1970, Franklin 1967). Slightly lower average values, generally 40-50 m 2 ha \ have been recorded in tawa (Beilschmiedia tawa) forest (Smale 1981, West, Jeffs and Ogden 1981, Mark and Esler 1970). No basal area data appear to have been published for kauri forest, but a pcq survey of 14 sites carrying mature kauri throughout its area of distribution gave values ranging from 34 to 146 m2 ha 1 with a mean of 80 m 2 na 1 (Moinuddin Ahmed, personal communication). On this basis

174 the vegetation of the reserve at Huapai must be regarded as mature, but the remaining sites may not yet be carrying their maximum potential biomass. However, the wide, but unspecified, standard errors which should be attached to the estimates must not be forgotten. The difference between the leading species defined in terms of density and basal area deserves further comment. The forest may be conceived as being composed of relatively few large long-lived trees (kauri, podocarps, etc) and more numerous shorter lived species in the understorey. Thus, while biomass is being retained by the large trees, it is being "recycled" by the tree ferns and understorey trees and shrubs. The numerical importance of some of the canopy trees, especially kauri, reflects their abundance in the smaller size classes and may be a further indication of the serai nature of these forests. This numerical importance may be expected to decline in future, as a few individual trees rise to dominance and come to contain most of the standing biomass in any area. The successional process can, of course, be clearly seen in the kauri ricker stands, and it is partly for this reason that teaching exercises have concentrated on them. A model of ricker development, based on these studies (and others; see discussion in Ecroyd 1982) can be proposed as follows: (1) Colonisation by manuka (and/or kanuka). (2) Invasion by podocarps and/or kauri. (3) "Ricker" development (skewing of size frequency distribution). (4) Competitive thinning of the ricker allowing invasion by shade tolerant hardwoods. (5) Progressive growth of dominant kauri and failure of kauri regeneration, creating a "regeneration gap" on the site. (6) Eventual death of kauri so that the site becomes dominated by podocarp/hardwood forest. To judge from the ages of the larger kauris this whole process must take about 600—1 000 years. The latter two stages cover several hundred years and are difficult to study because so little virgin kauri forest remains. Whether kauri can regenerate in situ, with poles springing up in canopy gaps for example, or whether it requires a more massive disturbance to re-initiate the sere, is not known for certain. It is evident that once a more or less even-aged forest structure has been instigated, it can be maintained for at least one and possibly several generations. Despite Mirams' (1957) pioneering work on the early stages of the succession, the whole sequence has never been studied quantitatively. Lloyd's (1960, 1965) study of rickers at Russel Forest showed that several separate "invasions" of the kanuka stands by tanekaha, rimu and kauri can occur, so that there may be one, two or three generations (cohorts) of each of these species by the time the kanuka dies out at

175 100—130 years. The main difficulty in developing a quantitative model of the dynamics of the ricker stands lies in estimating their age. It is assumed that effective recruitment into the kanuka "nurse crop" occurs for a relatively short period of time, certainly less than 80 years. It seems likely that the first recruits will come to dominate the stand, suppressing those added later. Even if all stems started growth simultaneously a highly skewed (log-normal) size frequency distribution would be expected to develop through time, and differences in germination date would simply accentuate this trend (Harper 1977). This hypothesis is supported by the very weak relationship between age and diameter in the 10-40 cm DBH size classes reported here, and by the high mortality amongst smaller size classes. Excluding the occasional trees which are substantially larger than the others and which represent survivors from a previous generation, I have assumed that the largest trees in the ricker are the oldest. The average age estimated from cores taken from trees in the largest (ricker) size class can thus be used to estimate the "starting date" for the population. An interesting feature of the size frequency distribution for the kanuka is the tendency for dead stems to preponderate in the class immediately beneath the mode for living trees. This may imply that even in this senile population mortality is not random; the process of competitive suppression continues. The data presented in the tables and graphs in this paper give a picture of the structure of the ricker stands in the Waitakere Ranges, at the stage in which kanuka is being rapidly eliminated and the larger kauris are showing the first signs of developing the mature crown form. They may be expected to thin out further, and to develop a more diverse understorey.

ACKNOWLEDGEMENTS

The field work on which this paper is based has been mostly carried out by students under my direction during the second and third year vegetation and ecology courses in the Department of Botany, Auckland University. Thanks are due to all students who have contributed over the years, especially those at post-graduate level who assisted with class organisation, species identification and collation of results. My colleague Dr N. D. Mitchell has been jointly responsible for the field work, and his assistance and co-operation are gratefully acknowledged. Mr R. G. Serra has helped in numerous ways and was directly responsible for the preparation of cores for tree age determination. Thanks also to Mr Moinuddin Ahmed for allowing me to use unpublished results, to Ms C. Bergquist for drawing the diagrams and to Miss S. York for typing.

REFERENCES

Burrows, C. J. 1973: The ecological niches of Leptospermum scoparium and L. ericoides (Angiospermae: Myrtaceae). Mauri Ora 1: 5-12.

176 Ecroyd, C. E. 1982: Biological . 8. Agathis australis (D. Don) Lindl. () kauri. New Zealand Journal of Botany 20: 17-36. Esler, A. E. 1967: The vegetation of Kapiti Island. New Zealand Journal of Botany 5: 353-393. Esler, A. E. 1983: Forest and scrubland zones of the Waitakere Range, Auckland. Tane 29:

Esler, A. E. & Astridge, S. J. 1974: Tea tree (Leptospermum) communities of the Waitakere Range, Auckland, New Zealand. New Zealand Journal of Botany 12 (4): 485-501. Franklin, D. A. 1965: Quantitative data on forest composition. In notes and comments. New Zealand Journal of Botany 3 (2): 168. Franklin, D. A. 1967: Basal areas as determined by the point-centred quarter method. In notes and comments. New Zealand Journal of Botany 5 (1): 168-169. Grieg-Smith, P. 1964: "Quantitative Plant Ecology". 2nd edition. Butterworth and Co. Ltd. Great Britain. 256 p. Harper, J. L. 1977: "Population Biology of ". Academic Press Inc. (London) Ltd. 892 p. June, S. R. & Ogden, J. 1978: Studies on the vegetation of Mount Colenso, New Zealand. 4. An assessment of the processes of canopy maintenance and regeneration strategy in a red beech (Nothofagus fasca) forest. New Zealand Journal of Ecology 1: 7-15. Lloyd, R. C. 1960: Growth study of regenerated kauri and podocarps in Russell Forest. New Zealand Journal of Forestry 8 (2): 355-361. Lloyd, R. C. 1965: Working plan report for Russell State Forest. New Zealand Forest Service Report No. 123 (unpublished). Mark. A. F. & Esler, A. E. 1970: An assessment of the point-centred quarter method of plotless sampling in some New Zealand forests. Proceedings of the New Zealand Ecological Society 17: 106-110. Mark, A. F. & Scott, G. A. M. 1965: Quantitative data on forest composition. In notes and comments. New Zealand Journal of Botany 3 (2): 168-169. Mirams, R. V. 1957: Aspects of the natural regeneration of the kauri (Agathis australis Salisb. ). Transactions of the Royal Society of New Zealand 84 (4): 661-680. Mueller-Dombois, D. & Ellenberg, H. 1974: "Aims and Methods of Vegetation Ecology". John Wiley and Sons Inc. U. S. A. 547 p. Ogden, J. 1980: and dendroecology - an introduction. New Zealand Journal of Ecology 3: 154-156. Ogden, J. & Powell, J. A. 1979: A quantitative description of the forest vegetation on an altitudinal gradient in the Mount Field National Park, Tasmania, and a discussion of its history and dynamics. Australian Journal of Ecology 4: 293-325. Shirley, J. W. 1968: The Oratia Reserve. B. Sc (Stage III B) 3B project, University of Auckland. Smale, M. C. 1981: Growth and mortality of tawa in virgin and logged forest, Mamaku Plateau. New Zealand Forest Service, Forest Research Institute Indigenous Forest Management Report No. 32, (unpublished). 9 p. Thomas, G. M. & Ogden, J. 1983: The scientific reserves of Auckland University. I. General introduction to their history, vegetation, climate and soils. Tane 29: Trevarthen, C. B. 1952: The historical background to the vegetation of the University property at Swanson and its present condition. Tane 5: 12-17. Wardle, J. A. 1970: The ecology of Nothofagus solandri. 4. Growth and general discussion to parts 1 to 4. New Zealand Journal of Botany 8: 609-646. West, C. J.; Jeffs, A. & Ogden, J. 1981: A study of growth ring periodicity and diameter increment in tawa, Beilschmiedia tawa. Unpublished report to the New Zealand Forest Service; Department of Botany, University of Auckland 18 p. Westerkov, K. & Mark, A. F. 1968: Beech forests of the Upper Routeburn Valley. Science Record 18: 46-52. White, J. & Harper, J. L. 1970: Correlated changes in plant size and number in plant populations. Journal of Ecology 58: 467-485.

177 APPENDIX I. Relative densities (%) of species ^ 10 cm DBH at Swanson, Huapai, Leigh and Oratia. + indicates present in sample but< 0. 5% relative density. * indicates additional species recorded in an independent pcq sample at Swanson, or an independant plot sample at Leigh. Key to columns: (1) based on 24 7 x 7 m plots, ridge and slope, Swanson; (2) pcq sample (100 trees), plateau, Huapai; (3) pcq sample (100 trees), ridge, Huapai; (4) based on 27 7 x 7 m plots, ridge, Huapai; (5) pcq sample (100 trees), gully, Huapai; (6) pcq sample (44 trees), gully, Leigh; (7) pcq sample (44 trees), mid-slope, Leigh; (8) pcq sample (16 trees), upper slope, Leigh; (9) single 20 x 50 m plot, kauri ricker, south slope of ridge B, Oratia.

Swanson Huapai Leigh Oratia (1) (2) (3) (4) (5) (6) (7) (8) (9) Agathis australis 9 5 14 6 7 17 22 Alectryon excelsus 2 Beilschmiedia taraire 22 2 * Beilschmiedia tawa 1 1 Brachyglottis repanda + Carpodetus serratus * * Coprosma arborea 16 19 20 2 * Coprosma areolata 1 * Coprosma australis 2 Coprosma lucida 1 1 1 * Coprosma rhamnoides 1 Corynocarpus laevigatus 1 2 2 8 2 1 1 3 5 5 Cyathea dealbata 42 31 12 20 23 23 59 41 29 Cyathea medullaris 3 7 5 1 * * Dacrydium cupressinum 1 1 6 Dacrydium kirkii 1 Dicksonia squarrosa 10 8 Dodonaea viscosa 1 * Dysoxylum spectabile 2 1 1 1 1 Elaeocarpus dentatus + Geniostoma ligustrifolium * + Hedycarya arborea 1 2 1 Hoheria populnea 6 3 Knightia excelsa 3 7 6 3 6 5 1 Leptospermum ericoides 6 7 2 1 9 8 28 16 * Leptospermum scoparium 3 Melicytus macrophyllus 1 + 1 1 4 2 4 1 Metrosideros excelsa 1 * Metrosideros robusta Mida salicifolia * Myrsine australis 1 7 13 5 1 * Myrsine salicina Nestegis lanceolata 2 + Olearia rani 1 +

178 Swanson Huapai Leigh Oratia (1) (2) (3) (4) (5) (6) (7) (8) (9) Phyllocladus trichomanoides 7 3 2 3 2 9 21 Pittosporum tenuifolium 1 4 2 2 Podocarpus dacrydioides 1 * 2 Podocarpus ferrugineus 1 1 * 2 2 2 Pseudopanax arboreus * * + * Pseudopanax crassifolius 2 1 1 3 Pseudopanax lessonii 1 1 Rhopalostylis sapida 9 15 10 16 32 5 1 Vitex lucens 2 7 6 3 10 3

APPENDIX II. Relative basal areas (%) of species > 10 cm DBH at Swanson, Huapai, Leigh and Oratia. Column headings and symbols as in Appendix I.

Swanson Huapai Leigh Oratia (1) (2) (3) (4) (5) (6) (7) (8) (9) Agathis australis 35 8 43 38 43 27 49 Alectryon excelsus 1 Beilschmiedia taraire 25 1 * Beilschmiedia tawa 2 + Brachyglottis repanda + Carpodetus serratus * Coprosma arborea 7 8 10 1 * * Coprosma areolata + Coprosma australis + Coprosma lucida + + + * Coprosma rhamnoides + Corynocarpus laevigatus 1 2 7 + + Cordyline australis + + 2 2 2 Cyathea dealbata 26 10 4 8 9 8 8 16 14 Cyathea medullaris 1 7 1 * +* Dacrydium cupressinum + 3 4 Dacrydium kirkii + Dicksonia squarrosa 4 5 Dodonaea viscosa * + Dysoxylum spectabile 2 + 1 + 8 Elaeocarpus dentatus + Geniostoma ligustrifolium * Hedycarya arborea + 1 1 Hoheria populnea 9 + Knightia excelsa 5 4 2 2 14 2 + Leptospermum ericoides 10 5 1 1 14 4 48 13 Leptospermum scoparium 1 * Melicytus macrophyllus 1 + + + Melicytus ramiflorus 3 + + 1 + Metrosideros excelsa * + Metrosideros robusta * Mida salicifolia Myrsine australis + 3 3 3 + Myrsine salicina *

179 Swanson Huapai Leigh Oratia (1) (2) (3) (4) (5) (6) (7) (8) (9) Nestegis lanceolata + + Olearia rani + + Phyllocladus trichomanoides 5 2 + 5 . + 4 17 Pittosporum tenuifolium 2 4 + . +* Podocarpus dacrydioides 33 + Podocarpus ferrugineus +* Podocarpus totara 4 1 • Pseudopanax arboreus * * * + * Pseudopanax crassifolius 1 1 + 1 Pseudopanax lessonii + . + Rhopalostylis sapida 5 5 4 7 14 1 + Vitex lucens 52 28 14 4 27 40

180