from these sites to the hill crests since, as already suggested, it is but a recent entrant into the Groveburn. But both mountain and silver beech are, and probably for a long period of time have been, established along the terrace margins bordering the Waiau River. Their absence from the hill forests is attributable to the biotic barrier of relatively stable podocarp forest which, previous to logging, occupied the main gravel river terraces.

Both species did penetrate the barrier in part. They spread slowly across the terraces occupying pockets of swampy ground where competition with the podocarps was not severe or, temporarily, local wind-fall areas; but they did not penetrate any distance up the lower hill slopes where light sandy soils carried vigorous podocarp stands with dense understories of ferns, shrubs and small broad-leaved trees. They were unable to reach the hill crests where site conditions were most suited to them and, with the virtual elimination of the podocarps following the climatic change, the hill crest stands have degenerated into scrub forest of stunted rata and kamahi. On comparable sites elsewhere, seed sources being available, either one or both of the beech species invade, suppress and eliminate the rata and kamahi stands with great rapidity.

In the forests immediately to the west of the Waiau River and in the Long-woods, migration patterns are relatively straight-forward. There are comparatively few complications arising through the operation of biotic or other effective barriers to migration In other forests, however, such factors assume greater importance. They may even mask entirely the normal processes of change, distorting out of recognition the usual patterns of forest re-adjustment.

The gravel river terraces flanking the Wairaurahiri River carry a complex admixture of forest types though the basic factors of site are remarkably uniform throughout. Near the coast, the terraces carry rimu stands similar to those developed on the Rowallan coastal gravel terraces as already described. Inland, the terraces may carry rimu stands, silver beech stands or mountain beech stands, or the terrace stands may be composed of any two or all three of these species in any proportion. At first sight there would seem to be no good reason for this marked heterogeneity but it is possible to account for it in terms of the hypothesis of recent climatic variation.

It has been seen that rapid invasion of an area of podocarp forest by the beech species, following a climatic change adverse to the podocarps, is dependent on water carriage of seed of the incoming species. Now the watershed flanking the Wairaurahiri Valley, the Hump Ridge, is composed of two main types of rock, basal coal measure sandstones and coarse conglomerates; and, on the upper hill slopes, silver beech is the dominant or only species on the conglomerates with mountain beech the dominant or only species on the sandstones. Where streams crossing the river terraces drain conglomerate slopes. therefore, the seed brought in is largely silver beech seed and the forest on the terrace will be silver beech/podocarp forest. Where streams drain sandstone slopes the terrace stands are mountain beech/podocarp stands Mixed mountain beech/silver beech/podocarp stands occur wherever seed of both beech species was immediately available and the pure podocarp stands occupy the centres of the larger terraces remote from the streams as well as all terraces untraversed by any streams.

To the north, near Lake Hauroko, the position is complicated by recent faulting which has resulted in slight headward tilting of tributary streams with con-

sequent secondary changes in forest development following changes in terrace drainage, and recent river deposits have been colonised by a wide range of species. These secondary effects superimposed on the main trends in forest development have produced such apparently abnormal forest types as the kahikatea/mountain beech/kowhai (Sophora microphylla)/manuka (Leptospermum scoparium) type, but the main trend remains clear, a progressive replacement of podocarp forest by beech forest. Local variations on the theme depend upon the operation of purely local factors of site but sometimes, also, upon the operation of factors wholly extraneous to the site. In the present instance these extraneous factors were geological factors governing the type of seed supply. The type of forest now established on any particular river terrace depends not on any of the regular basic factors of site but upon the extraneous factor, the nature of the terrain drained by the terrace streams.

In many cases, therefore, it is largely a matter of accident which species arrives first, and in many instances the incoming species may not be that which is best suited to the site. In the Wairaurahiri Valley the terrace podocarps are on many terraces, in process of replacement by mountain beech though, on sites of this quality, silver beech would seem to be the obvious successor species. In all such cases even those forest types now in course of development will not be the ‘climax’ types for the site under existing climatic conditions. Some of them may be but it will definitely be unsafe, in any instance, to assume that this is so. The many processes of forest evolution are conditioned and controlled not only by climatic factors, but by accidents of geology and geomorphology.

One further example from the forests of the Hump Ridge might be described In general, on the Ridge, the midslope podocarp stands have been eliminated leaving a midslope zone of mountain beech/silver beech forest containing a strong podocarp element; but at the extreme southern end of the Ridge near Sandhill Point, a distinct type of podocarp forest survives at comparatively high altitudes. 1,500 to 2,000 ft. above sea level. The type is characterised by the presence of small, stunted rimu with stunted Hall's totara co-dominant. Its survival is possibly attributable to some peculiarly favourable local site factor complex but if Diagram 7 be consulted, it will be seen that the hill-crest occupied by the stands lies, at its highest point, below the postulated ancient altitudinal limit of the podocarp forests. Invasion of the type by either of the beech species must proceed uphill from the saddle, a slow process possibly rendered slower by local operation of podocarp favourable factors.

The lower slopes below the ridge crest podocarp stand carry podocarp forest of a type similar to that developed on the Longwood midslopes and these stands merge into coastal gravel terrace stands developed near the shore. The ridge crest stands could not, therefore, be invaded by other of the beech species directly from below. Invasion on the east (the aspect illustrated in the Diagram) could only proceed downstream from sandstone-conglomerate areas on the main upper slopes of the Hump Ridge followed by slow movement along the coastal cliffs and up into and through the coastal terrace podocarp stands Such a movement can be demonstrated on the ground but it is very slow in its latter stages.

To the west of the Hump Ridge, silver beech could enter by a circuitous path from conglomerate slopes below the main summits, a rapid series of movements down secondary streams linked by intervening slow movements across the sharp

Diagram 7 Hump Ridge

ridges between the streams, and this movement, also, can be demonstrated on the ground. Fingers of silver beech forest extend down all streams draining from the main summits and, spreading out from these riparian stands, small pockets of young silver beech extend upslope into the high altitude podocarp stands. Mountain beech, however, is unable to invade the podocarp stands from the west since it is effectively sealed off by a barrier of silver beech forest on soils derived from the conglomerates.

The real problem in the case of these high altitude ridge crest podocarp stands is not the problem of their presence but the problem of their condition. Why have these stands persisted as podocarp stands without replacement, as usual under such circumstances elsewhere, of the podocarps by rata, kamahi or other scrub hardwood species? The answer is not known but it would seem that an edaphic factor is responsible. Podocarp forest of this unusual type has only been found on soils derived from this same series of coal measure conglomerates.

Silver beech forest on the conglomerates forms a barrier through which mountain beech can spread only by water carriage of seed. Comparable conditions obtain on many lime rich soils. Outcropping limestones or lime rich sandstones, e g., Helmet Hill, Dean Hill, and Diggers Ridge, to cite West Waiau examples, commonly carry on their dip slopes pure stands of silver beech though mountain beech may be dominant or co-dominant on adjacent sandstones, siltstones or alluvials and may even be present on the dry limestone escarpments. Distribution and migration patterns are distorted by these lime rich soils and, failing penetration by water carriage, effective barriers to the further spread of mountain beech are created.

Up to this point in the discussion it has been tacitly assumed that the beech species did not survive the podocarp forest period save in high altitude forests. This was undoubtedly the general rule but there were one or two exceptions to be noted In discussing the mid-slope podocarps of the Longwood Range it was

suggested that, throughout the era of maximum podocarp forest expansion, there may have existed ‘evanescent silver beech communities of pioneer character’ on riparian sites. Now it would be possible for these ‘evanescent communities’ to outlive the remnant high altitude seed sources from which they took their origin. Through continued establishment and re-establishment on fresh sites, eroded stream banks and fresh alluviums, it would be possible for them to persist long after the disappearance of the primary seed sources. There would be a continued tendency toward downstream drift of these detached riparian communities with the final survivors maintaining a precarious foothold in the hostile climatic environment at that date obtaining in the lowlands.

No definite examples of the operation of a survival mechanism of this type have been traced in the forests of Western Southland though the operation of such mechanisms is suspected in several instances. In Otago, particularly in the Dunedin District, several of the small silver beech forests do appear to have taken their origin from casually surviving riparian stands and to be without connection with surviving high altitude silver beech stands. But clear cut examples are not, to-day, to be found in the South Island though many good examples are encountered in North Island forests where the events of the more distant past have not been so thoroughly masked by rapid recent changes in forest type and species distribution. Two North Island examples, only, might be instanced. Firstly, there is the case of the isolated riparian silver beech stand on the eastern (Lake Taupo) slopes of the Hauhangaroa Range as described though not explained by Poole (1950); and secondly, there are the riparian stands, several beech species being represented, of the Mamaku plateau to the north and east of Rotorua. These North Island riparian beech communities admit no other interpretation of their origin and continued existence than that advanced herein; and it must therefore be anticipated that comparable survival mechanisms were once operative in the south. Even in the case of the Alton and Rowallan Reversals, there may have been some survival of beech seed sources along the streams throughout the podocarp forest period. It is not necessary to assume such survival in order to explain the situation developed to-day but it was possible. It would not invalidate the account given herein of the processes of forest re adjustment but would mean, only, that the present account has been somewhat over-simplified. Survival of beech seed sources on riparian sites within a podocarp region would, when coupled with entry, by water carriage of seed, of the beech species possibly result in the speeding up of all re-adjustment processes. The basic patterns of change would not be greatly affected.

Mountain and silver beech could also survive on lowland sites throughout a climatic era favourable to podocarp forest development as species of bog forest. One possible instance has already been described from the forests to the east of the Pourakino River, Longwood Range, and many other cases will be described later. Survival of beech seed sources in lowland bogs is by no means an infrequent phenomenon but is one which commonly provides the key to an understanding of present type distributions.

Not every item of forest type distribution will be wholly explicable in terms of the one simple basic hypothesis; straightforward examples such as those described from the Longwood and West Waiau forests will inevitably be few; secondary hypotheses must frequently be invoked before distribution patterns can be understood in detail. The basic patterns will commonly be distorted or modified