rimu in Southland. In other words, rimu maintains a place within developing red or hard beech stands with greater facility than it does in silver or mountain beech stands, replacement of rimu by red beech or hard beech seldom, therefore, proceeding to completion. And finally, as may have been the case in Southland forests to the west of the Waitutu River, there is, in North Westland and in Western Nelson, frequent super-imposition of two distinct processes, the one, invasion of various types of bog forest by rimu and the other, invasion of the rimu stands by one or other of the beech species.

These forests have not yet been studied in as great a detail as have been the simpler Southland forests so that it is not yet possible to present a full account of them. All that can be attempted is the selection and brief description of a few significant items.

In the forests lying to the south of the Ahaura River, to the east of the Grey River, and to the north of the Taramakau, migration patterns similar to those described for Southland forests are traceable though they have been somewhat obscured by the effects of a full century of gold-mining and timber logging. Primary seed sources for silver beech and for red beech appear to have been high altitude stands toward the headwaters of the Grey and Ahaura Rivers. For mountain beech the primary seed source would appear to have been the bog forests of the lowland pakihi. For all three species additional seed sources on the Paparoa Range and in the hill country to the south of Reefton were tapped by western and northern tributaries of the Grey River. Entry of species into the region would, therefore, have been rapid but their subsequent spread to the south has been extremely slow for, latterly, migration routes have no longer paralleled stream and river courses. Broadly speaking, there has been a slow infiltration of the beech species through the lowland podocarp stands, a movement across interfluves accompanied by periodic rapid fingering out of the beech stands down secondary streams, a pattern somewhat distorted by the residual pakihi around which (or from which, since, in certain cases, they were in themselves beech seed sources) mountain and silver beech have spread more rapidly. By the time the line of the Arnold River is reached, the beech species are present on riparian sites only and all interfluves are still occupied by Westland type podocarp stands. To the south of the Arnold the forests are typical Westland podocarp forests and contain little beech of any description save for a few small pockets, principally or entirely of red beech, occupying sites suggestive of stand origin from seed sources along the lower reaches of the Grey River.

Red beech is again present in the Taramakau Valley with the main stands upstream from the confluence of the Taramakau and Otira Rivers. These stands appear to be of entirely different provenance to those of the Grey and Arnold Valleys from which they are separated by a broad belt of uninterrupted podocarp forest. The only traceable connection they have with other stands of red beech is by way of the Taramakau River itself. In other words it is suggested that the primary seed source for the Taramakau red beech is to be sought in the beech forests to the east of the main divide, the Hurunui Valley and Lake Sumner red beech stands. That red beech is but a recent entrant into the Taramakau is strongly suggested by the forest type distribution patterns in evidence about the Otira-Taramakau confluence, typical altitudinal reversal patterns of the Southland type with podocarp and podocarp associate scrub hardwood forest types lying above the red beech stands.

In the Ahaura-Taramakau forests we have, therefore, all the standard Westland forest types, the relict rimu/rata/kamahi stands of the hill slopes, the lowland rimu stands of varying age, the kahikatea swamps and the kahikatea/rimu stands of partially swampy soils, the kahikatea/matai stands of the recent alluviums, the silver pine pakihi, and all the many intergrading types of forest. And, superimposed on this basic pattern, are the migration patterns of the incoming beech species with additional new types of forest produced through long continued survival of mountain and silver beech in association with silver pine and kaikawaka around the pakihi.

One final point remains for mention: the precise part, if any, played by hard beech in this forward movement of the beech forests into the Westland podocarp forest region, is not known. Hard beech is present in abundance in the forests to the north of the Ahaura but its distribution in the forests to the south has not yet been studied.

To the north of the Ahaura River, as already mentioned, the various beech species are generally dominant. In the Grey and Inangahua Valleys the forests of the valley floor alluviums are variable in type but are mainly kahikatea/silver beech/red beech/rimu stands containing, locally, a few matai, and any one species, save matai, may be dominant. Red beech, however, is universally dominant, typically in association with scattered rimu and small kamahi, on the lower valley slopes. These red beech stands pass upwards into hard beech stands which occupy the steeper and drier slopes and ridge crests, rimu again being present though the understory species changes to Quintinia. A few rata occur on these hard beech ridges and there is local development of rimu/rata/kamahi stands.

The hard beech stands, in turn, wherever the valley slopes continue upwards unbroken, pass once more into red beech stands but these upper slope red beech stands contain few rimu. They persist upslope, with increase in abundance of rata, to 2,000–2,500 ft. at which point silver beech again appears, assuming dominance close to the timber line. Silver beech is the common timber line species throughout the region.

This is the basic species distribution pattern for these forests as a whole. It is a simple pattern and in it there are few indications of any radical changes in progress. The local ridge crest rimu/rata/kamahi stands are suggestive of such changes as is also the general disproportion in the numbers of old and young rimu in the stands; but generally the forest types within the pattern appear fully stable with species distribution governed by temperature factors (coupled with altitude) and soil moisture factors (coupled with angle of slope and exposure).

This basic pattern is, however, seldom developed in such a simple form. The foothills lying between the main rivers and the mountains are carved from immense thicknesses of gravels. And on these gravels, wherever level or near level ground occurs, a pakihi or pakihi type bog forests are found. In other words, a complex topographically controlled pattern of bog forest types is, over the greater part of the region, super-imposed on the basic regional pattern; and it is in these bog forest types that the greater interest for present purposes lies.

On the Westland coastal gravel plains, as already described, the open pakihi are marginally invaded by manuka, silver pine and rimu, in that order. This gives rise to a series of forest types portraying all the various stages in the development

of rimu stands. In Southland, on the gravel terrace lands to the west of the Waitutu River, these processes of pakihi colonisation by rimu were rendered more complex by the presence, around the pakihi, of silver and mountain beech or by their incoming at an early stage. Similar complications are usual in the forests to the north of the Ahaura River. Rarely, in these forests, simple colonisation of the pakihi by manuka, silver pine and rimu is encountered but this is usually followed by invasion of the developing rimu stands by red beech or by hard beech or by both, giving rise to a series of forest types marked, in a sub-mature stage, by stands of rimu over an understory of younger red beech or hard beech, the stands frequently containing relict silver pine, kaikawaka or even large old manuka. In the majority of cases, however, mountain and/or silver beech are also present from the start in which case these species advance with the manuka, or but a short distance behind it, across the pakihi. On such sites both species remain stunted with production of dense thickets through which silver pine aid kaikawaka appear. Rimu then marginally invade the stands finally giving rise to rimu stands over a relict underwood of mountain beech, silver beech, silver pine and kaikawaka. And at a later stage these stands in turn, may be marginally invaded by red beech or by hard beech.

In other cases, notably at higher altitudes, rimu, red beech and hard beech may play no part in the sequence of events and the final stage in forest development is then marked by little more than the development of stunted stands of silver beech or of mountain beech with, at these altitudes, some bog pine or yellow-silver pine. The possible permutations and combinations are almost endless and no two bog forest stands, in this region, are quite the same in specific composition. Even rata and kamahi may, in certain cases, enter the stands, usually at a fairly late stage in stand development.

To cite one specific example: on a broad flat-topped gravel terrace above the Snowy River a few miles to the north of the Grey River, there is a stand characterised by the presence of widely spaced rimu, the trees being of considerable girth with big canopies though they are of short total height. These short rimu over-top a stunted underwood of mountain beech and silver beech through which are scattered kaikawaka, silver pine, stunted rata and kamahi, and dead stems of manuka. And, peripherally, the stand also contains some red beech and hard beech.

The origin of this stand is thought to have been, firstly, the establishment, by bird carriage of seed, of rimu within manuka thickets dotted over a pakihi bordered by a fringing stand of mountain beech, silver beech, silver pine and kaikawaka. The rimu soon over-topped the manuka thickets with consequent development of spreading canopies and short massive boles. Secondly, the fringing stands closed in over the pakihi with later incoming of rata and kamahi and active suppression of manuka; and, thirdly, with continued lowering of water-tables, red beech and hard beech have marginally invaded the stand, spreading upwards from the terrace slopes below.

Confirmation of this hypothetical scheme of stand development was later obtained by the finding, in another area not many miles away, of an earlier stage in stand development. In this case the greater part of the gravel ridge crest was still occupied by open pakihi but, on local raised mounds, there were manuka thickets containing heavily branched rimu saplings. And around the pakihi

there were fringing stands, in active course of advance over the open ground, composed of silver beech, mountain beech, silver pine and kaikawaka.

In these two stands we have an adequate explanation of the age structure of certain of the lowland rimu stands of the Westland coastal plains. In one such stand the age of the rimu dominants varied from 150 to 700 years but with the bulk of the crop in the 300–400 year age class. The older trees might well be of comparable origin to the rimu of the Snowy River stand, scattered trees established across the old pakihi on local raised mounds or on pockets of otherwise dry ground. By the time these were 300 years of age, occupation of the remainder of the open pakihi by rimu proceeded fairly rapidly, the last few gaps being filled 150 years ago. Unfortunately no note was taken, at the time, of the form of the older trees in comparison with that of the younger.

But to return to the Snowy River stand: why did it not develop over its entire extent as a rimu pole stand? This stand lies close to the altitudinal point at which, in the basic regional forest type distribution pattern, the midslope hard beech/rimu stands give place to the red beech/rata stands of the upper slopes, i.e., at about the altitudinal limit for rimu. Rimu is a shade tolerant species but not on marginal sites. Where climatic conditions border on the unfavourable, rimu requires almost full light for adequate growth. This it obtained, in the present case, in the open manuka thickets but rimu seedlings and saplings do not long survive under the dense canopy of bog forest silver or mountain beech. Once these two species had closed in over the old Snowy River pakihi, the stand could no longer develop as a rimu stand.

Unfortunately, opportunities for study of undisturbed pakihi are rare. All but the most inaccessible pakihi have been burnt and re-burnt in the course of their exploitation for silver pine produce, even long fallen logs being eagerly sought for the durable post timbers they contain. The above account of the processes of pakihi colonisation has been built up mainly through analyses of older stands and reconstruction of their probable modes of origin, deductions supported, in a few instances, by examination of virgin pakihi. But, by and large, the open pakihi typical of Westland to-day have been so strongly modified by fire that we no longer deal with natural but with induced plant communities. It might also be noted that, in at least one instance, the Mai-Mai pakihi of the Grey Valley, a fire history dating back to pre-European days is thought probable, forest types bordering this pakihi bear traces of very ancient fires. This burning of the pakihi by the Maori is by no means improbable for the pakihi were the only open spaces within heavily forested regions and pakihi vegetation, during even short spells of fine weather, is highly inflammable; and the Maori used fire.

Degradation processes with forested lands reverting to pakihi also occur though on a purely local and minor scale. In almost every case so far examined, the initiating factor has been a local change in direction or freedom of stream flow. On the other hand, and on a major or regional scale, degradation of forest to pakihi has occurred in the past, possibly many times, for most of the larger pakihi contain buried timbers; but the process of to-day is the colonisation process and this process of pakihi invasion by forest has been in full operation on a regional scale for many centuries.

Before abandoning the topic of the pakihi, it must be emphasised that not all forest types extraneous to the basic regional forest type distribution pattern

are of pakihi origin. Some stands, though very closely resembling the forest types developed on old pakihi lands, owe their condition to other factors, e.g., to the shallowness of forest soils on glacially over-ridden granite spurs or to the general infertility of certain sandstone soils which, particularly when subjected to excessive ground water seepage, typically carry stunted stands of silver beech or mountain beech with many common bog forest species. In fact, in respect to extensive areas of forest in North Westland and in Western Nelson, lithological factors of site assume very great importance in the control of forest type distribution and may distort the basic regional distribution pattern almost out of recognition.

If the forests of the Paparoa Range to the west of the Grey and Inangahua. Valleys were here to be described, it could only be in emphasis of this point. Limestones, sandstones, siltstones, shales, conglomerates, breccias, coals, greywackes, granites and gravel beds may all outcrop within an area of radius but a mile or two. With additional type variations due to variations, frequently gross, in aspect and in altitude the forests become a veritable jig-saw puzzle of miscellaneous forest types. And for such forests, if they are to be understood, it will first be necessary to establish the validity, or otherwise, of the hypotheses advanced herein, later applying the knowledge so gained toward disentanglement of the more complex.

Studies of the Paparoa forests are in progress as are also studies of the forests of the lithologically complex, topographically rugged mountains to the north of the Buller River. The forests are mainly beech forests with local development of podocarp stands on lowland sites, with beech podocarp stands on the lower hills, and with extensive development of bog forest types at a variety of altitudes. Hard beech plays an increasing part from the northern Paparoas, where extensive hard beech/rimu stands are developed, northwards to the West Haven Inlet; and, from south to north, there is a steady incoming of forest species more typical of North Island forests. Thus, in the far northwest, such species as northern rata (Metrosideros robusta) and pukatea (Laurelia novae-zelandiae) enter the lowland stands, together with many northern shrub species. In the far north-west, also, even matai regenerate freely on deep lowland alluvial soils, e.g., in the Aorere Valley; and the matai seedlings and saplings display fair growth. But this is simply to say that site conditions, in these northern valleys, now aproximate those conditions which must have once been normal for hill sites in the far south.

For the moment, but one further matter in connection with the forests of North Westland and Western Nelson remains for discussion. It has already been remarked that silver beech is the typical regional timber line species. Mountain beech is universally represented in the forests of the lowlands and, as a species of bog forest types or as a species of stands developed on infertile sandstone soils, it may closely approach the timber line; but, save in rare cases, it does not occur at the timber line. On the other hand, in one or two instances, red beech is the timber line species, despite Cockayne's emphatic disbelief in such a possibility (Cockayne, 1926, p. 34).

On Mt. Harata near the confluence of the Grey and Clarke Rivers, red beech, silver beech, hard beech and mountain beech are all present in the forest on the lower slopes. The mid-slopes carry red beech/hard beech stands, the upper slopes red beech, only, with rata achieving co-dominance toward the