(April 19), due to recent rains, and this was succeeded by a foot of perfectly dry sand. At the 4-feet level there was a stratum of fine damp gravel, which fell away easily, presenting no difficulty for penetration by roots.

The plant possessed a main taproot about 1½ inches diameter at the top. The branches of the plant joined the root below the ground level, forming a multicipital crown. At a depth of 18 inches, the taproot divided into two equal branches, one of which could not be followed. The other continued downwards to a depth of 4 feet, where it turned abruptly in a horizontal direction. From the taproot in its uppermost 18 inches arose six main laterals which all pursued an almost horizontal course for about two feet, when they turned downwards abruptly, taking a vertically downward direction to a depth of 4 feet, when all took a horizontal course. This sudden deviation of the roots at the 4-feet level is striking, and is coincident with the stratum of damp sand which occurred at this level. This damp layer was doubtless the capillary fringe of the water-table. (The capillary fringe is defined as that stratum of soil immediately above the level of the standing water-table, into which the water rises by capillary attraction). At a distance of about a hundred yards from the position of the plant was situated a small artesian well, which by its constant flow may have maintained the water-table at a fairly constant height. The coincidence of the horizontal extension of the roots with the damp layer might be interpreted as a response on the part of the roots to the special conditions obtaining. Possibly a response to the stimulus of water was in evidence, or lack of aeration at greater depth may have determined the condition.

All the upper portions of the main roots showed little branching. The main mass of absorbing roots was at the 4-feet level, where profuse bunches of roots branched to the third order were situated. A number of small roots were, however, produced from the taproot in its upper portion. The roots are rich brown in colour, quite tough and flexible. The horizontal root at the left of Figure 11 was followed for a further distance of 3 feet before it was finally lost.

In Figure 12 is shown the root-system of another specimen growing in a similar type of soil. A glance at the figure will show that the lateral extension of the roots is far greater than in the first specimen, for here they were found to extend horizontally to a distance of 9 feet. A ground plan of this development is seen in Figure-13. A taproot is again present, but here it turned sharply to the left at the 2-feet level, a smaller branch of the main root continuing vertically downwards to a maximum depth of 5 feet 6 inches, after having divided into a number of smaller branches. Although the lateral root to the right of the figure showed considerable horizontal extension, it eventually turned abruptly downwards at a distance of 9 feet from the base of the plant. The horizontal extension of the taproot to the left was lost at a distance of 9 feet from the crown of the plant. There was found to be no definite damp stratum in this case, the soil becoming imperceptibly damper from a depth of 4 feet 6 inches downwards. Possibly in this spot the level of the

Light.—No exact instrumental data are available, but it is clear that the isolation to which the district is subjected is generally considerable. In summer, bright sunny days follow one another with regular monotony. The sky is generally free of clouds and the heat is often intense. During winter and spring, however, though sunny days are common, dense mists may hang over the valleys and tops of the ranges, completely obscuring the sun sometimes for days at a time. In the hot season, the light is made more intense on sandy bare areas by the reflection which occurs from them.

Rainfall.—The limiting factor of the environment is certainly the moisture available in the soil. The mean monthly and annual precipitation for Clyde is given in Table I, while similar data for Earnscleugh is given in Table II.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table I.—Mean monthly and annual precipitation and mean monthly and annual number of rain days at Clyde. (These figures supplied by Govt. Meteorologist).

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table II.—Mean monthly and annual precipitation at Earnscleugh over a period of seven years (1920–1926).

It will be seen from these figures that the monthly precipitation is low, and apparently more or less evenly distributed over the whole year. But consideration of these figures alone would give an entirely misleading conception of the amount of moisture in the soil at different times of the year. Considering the Clyde data, it is precisely in those months which show the lowest rainfall, viz., May to September, that the soil is wettest. The reason for this is to be found in the wide difference between the amounts of evaporation in winter and summer. During winter and spring, i.e., from about May to the end of October, the soil is generally quite damp to a depth of several feet, and it seems remarkable to an observer at this period that the plant covering is so sparse on the soil which appears eminently suited to carry a much more abundant vegetation.

The penetrating power of the moisture is considerably lessened by other factors. On the steep Dunstan slopes, the surface is baked hard, and in places is free of soil altogether. Much of the rain which falls comes in short, sharp, and heavy downpours, which runs off extremely rapidly. The effect of these downpours is to convert dry watercourses into torrents which scour out the soil, exposing the underlying rock. After a storm, a few hours of the hot summer sun causes the ground to take on its previous parched appearance.

On the flat terraces, however, there is little or no run-off. But the underlying formation is a lightly compacted gravel into which moisture sinks very rapidly. One of the difficulties with which farmers must contend is the serious loss of water which takes place from the water-races dug in the porous ground.

Another characteristic of the rainfall is that the same months of different years often show considerable variation in the amount of precipitation. Table III gives the monthly rainfall at Earnscleugh for the period 1921–1927. A glance at the table will show that October, generally the wettest month of the year, was in 1922 easily the driest. Although January (generally the hottest month) has an average of 1.57 inches, yet in 1928 it had only .39 inches. Study of the data of Table III will show further variations of a similar nature.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table III.—Monthly and yearly rainfall (in inches) at Earnscleugh, 1921–1928.

The average number of days upon which rain falls during the year at Clyde is 75, a figure which gives an indication of the preponderance of dry days.

On the area considered in this study, only very occasional falls of snow are recorded. These are at most very light, and never lie on the ground more than a few days. On the ranges, however, which reach an altitude of more than 5000 feet, snow lies continuously above the 3000-foot level during the winter and spring months. This region is not considered here.

Humidity.—Information on this point is available from general observation only. Collection of exact observations would have required constant presence in the locality. It is apparent, however,

that the air humidity in summer is generally very low. The prevailing westerly winds have already deposited their moisture upon the mountains to the west, and by the time they blow over Central Otago they are warm and dry. During the winter and spring, however, owing to the prevalent mists, evaporation is slight and air humidity must frequently be high.

As a factor correlated with the study of root-systems, data bearing upon the soil water-content at different times and different depths would be important. This also requires constant presence upon the scene, and ideally, a laboratory on the spot. The only information that can be given on this point is that gained by visual examination while excavating the root-systems. As Weaver points out (1919, p. 24), root variations are probably due to a number of factors, among which water content of the soil and its penetrability probably stand first in importance.

Temperature.—New Zealand is commonly said to enjoy an insular climate, and for the most part this is the case. But in Central Otago, the widest region of the South Island, there is found an approach to continental range of temperature. The Cromwell district does not experience the lower temperatures of some other parts of Otago during the winter, but nevertheless the temperature occasionally reaches 15° Fahr. On the other hand, in summer, shade temperatures of over 100° Fahr. are not uncommon.

Table IV gives the mean monthly maximum and minimum temperatures at Earnscleugh over a period of four years, 1921 to 1924, inclusive. Additional data are not available for the extraction of a more accurate mean.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table IV.—Mean monthly maximum and minimum temperatures in degrees Fahr. at Earnscleugh from 1921 to 1924.

It will be noted from these figures that even in the summer months (November to February) while the maximum temperature is high; the minimum temperatures recorded at night are often low, so that the vegetation is subjected to considerable diurnal and nocturnal extremes.

Going into the question more fully, and taking the year 1924 as an example, the records show that in January there occurred one day with a maximum temperature of 102° Fahr., while there were eighteen days over 80°. In February, there were twenty days over 80°, and in March, nine days over 80°. In the following November, there were fourteen days over 80°, and in December, eight days over 80°. These figures show that the high temperatures reached are the rule rather than the exception during the summer months.

Figures are available from Ophir giving the number of frosts recorded on the grass monthly. At this place a frost is regarded as any record below 30.4° Fahr. Table V gives these figures, and they are instructive in showing the duration and frequency of the low temperatures, especially during the winter months.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table V.—No. of frosts recorded on the grass monthly at Ophir.

It will be seen that during the winter months frosts occur regularly—on almost every night. Moreover, the maximum temperature on many days of the winter months does not rise above 32° Fahr., and consequently the ground is frequently frozen hard for weeks at a time.

Wind.—During the hot summer especially, the area is subjected to violent winds from the north-west. These occur not infrequently and with great velocity, and are usually warm and dry. In conjunction with the intense insolation and low humidity, their effect is desiccating in the extreme, and evaporation must reach a maximum. These strong winds have been instrumental in moulding the sandy flat which characterises that part of the valley adjacent to Cromwell township. The sand is believed by some to have been brought down by the Clutha in the great flood of 1878, but Park (1908, p. 35) suggests a more likely source. He states that “this would doubtless account for a large proportion of the sand, but the terrace on which Cromwell is built contains a large amount of drift-sand mixed with the gravels, and a constant supply of this sand, derived from the terrace faces between Lowburn and Deadman's Point, is carried by the wind across Cromwell Flat.” (Fig. 20).

As will be shown later, the distribution of the sand has had an important effect in modifying the vegetation of this flat.

Strong winds from the south-west also occur at all seasons, but these are generally moisture laden, and if they do not bring rain, they cause the higher levels to be shrouded by mists, especially in winter.

The effect of the wind in conjunction with the sand particles, is also seen in the disintegrating action which it has upon the dying Raoulia patches. These patches are a characteristic feature of the landscape, and lie close to the surface, where the sand blast action is greatest. At a certain age in the development of these circular patches, the central region dies, while the peripheral portion remains alive. The resistance of the central portion is lessened, and the constant bombardment of the sand particles has the effect of eroding the dead portion away (Fig. 5).

II.—Physiographic Factors:

The whole area is characterised by strong topographic relief. According to Park (1908, p. 15 et seq.), the valley had its origin as an area of subsidence involved in the greater “block mountain” system of Central Otago. Lacustrine and subsequent fluviatile deposition and erosion, combined with further faulting caused the valley to assume its present topography. It is flanked to the north-west by the Pisa Range, rising to 6300ft., and to the south-east by the Dunstan Range, rising to a maximum of 5320ft. At the lower end of the valley, to the south-west, lies the Carrick Range, about 5000ft. in height. The Clutha River, formed by the junction of the Upper Clutha and Kawarau Rivers, flows through the Dunstan Mountains in a narrow, deep gorge which turns off from the main valley about three miles from its south-west end. The floor of the valley is occupied by a system of terraces which clearly indicate the intermittent lowering of the base level of erosion.

The north face of the Dunstan Range is remarkable for its apparent barrenness. It is an excellent example of the influence of physiographic features in determining plant covering. The slopes are exceedingly steep, and characterised by immense outcrops of the mica-schist rock, of which the range is composed. The soil varies in thickness in different places. On the ridges and slopes it may be absent altogether, the surface being covered by angular stones weathered from the rocks above, owing to the daily extremes of temperature to which these rocks are exposed. On the other more level places, and especially at the bottom of the steep gullies, the soil may attain a depth of many feet. All gradations between these extremes may be present.

The flank of the range in question faces in a northerly direction and is thus exposed to the full glare of the sun. In summer, the heat of the surface of the ground is intense, often so hot that the stones cannot be touched with the bare hand. Buxton (1924) has made actual measurements of the heat attained on the actual surface of deserts, where it may be supposed that seeds must lie. He obtained temperatures of 60° C. (140° F.), where the shade temperature was 38° C. (100° F.). He points out that “reflected radiant heat is a factor with which desert fauna and flora have to reckon, especially in valleys.” As shade temperatures of between 90° F. and 100° F. are common in summer in the Cromwell district, then it is reasonable to suppose that the surface of the ground would probably attain temperatures corresponding to those cited by Buxton—i.e., 140° F.

On account of the deep dissection of the main range by lateral valleys, certain slopes are much more exposed to the sun than others. The character of the plant covering in the shady slopes is consequently different from that of the sunny slopes. On the sunny slopes, the Raoulia patches are often the only occupants, with the addition of certain small annuals in winter and spring. On the shady slopes, the Raoulia patches are much closer together, and the annuals appear more abundantly. Comparison of the Dunstan side of the valley with the Pisa side, whose slope is towards the south, gives some idea

of the influence which the direction of the slope has had upon the plant covering. Both slopes in the past have been subjected to, the same treatment as regards burning, stocking, and depredations of rabbits, but the Pisa side shows little of the depletion which so characterises the Dunstan side. The nature of the underlying rock is the same, the climatic factors are similar. Presumably, the direction of the slope has been chiefly instrumental in causing the difference in vegetation.

III.—Edaphic Factors:

The Dunstan and Pisa Ranges are composed of the same micaschist rock which characterises a very large area of Central Otago. The soil derived from it is very fertile, as evidenced by the excellent crops which are obtained when irrigation is intelligently practised. The floor of the valley is largely occupied by fluviatile and lacustrine gravels. These are covered by a fertile soil of an average thickness of about a foot, which requires only sufficient irrigation to bring it into productivity. The soil covering in certain places is, however, very thin, and on the Cromwell Flat itself, is composed of drifting sand, which makes cultivation difficult.

The extreme permeability of the gravel beds causes the moisture that is precipitated to be quickly absorbed. The difficulty of getting water to the higher terraces, combined with the thirsty nature of the sub-soil has been instrumental in discouraging cultivation on these terraces. The soil itself is light in colour and texture, containing little humus, and consequently dries out very quickly.

On the lower terraces bordering the river, certain small areas of ground occur which are almost entirely bare of vegetation except for two species of plants, Chenopodium glaucum and Atriplex buchanani. These areas are composed of a very fine alkaline soil, derived as a concentrated decomposition product of the mica-schist rocks. On the surface a whitish crust forms in dry weather, the composition of which is given in Table VI.^* The composition of the soil at a depth of six inches is given in Table VII.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table VI.—Composition of surface crust of alkali patches.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table VII.—Composition of soil of alkali patches at a depth of six inches.

It will be seen that the concentration of salts in the upper crust is excessive (22%), and far higher than that in the deeper soil (0.34%). A similar concentration of salts in the upper crust has been found in more extensive alkali soils in other countries. Nevertheless, even a concentration of 0.34% in the deeper soil, where it may be supposed that the plant has its absorbing system, is a higher concentration of salts than is found in ordinary soils. Only plants which can tolerate such an amount of salts in the soil, can thrive in such places.

On the Dunstan Range, below the 3000-foot level, the soil which remains on the surface, has probably derived the humus it now possesses from the decay of plants which formerly covered this part of the range. At the present time, almost the only plant which can be considered to contribute appreciably to the humus is Raoulia lutescens, but it cannot be said that the addition so made occurs in any quantity.

The Water Supply of the Soil.—On the slopes of the ranges, the water available for plant life depends almost entirely upon the scanty rainfall. The nature of the underlying metamorphic rock precludes any possibility of the existence of springs or water-table. Of course, fissures in the rock hold a certain amount of moisture, and it is upon this water that most of the very occasional perennials which occur can be reckoned to depend. As mentioned in the section on climate, the rain which falls comes in sudden, heavy showers, most of which runs off rapidly into the precipitous gullies, and subsequently into the Clutha River. The lack of plant covering must also lessen the amount of water retention. The surface of the soil is generally baked hard, so that there is no mulch to retard evaporation under a hot sun.

On the terraces, on the other hand, the presence of a water-table at varying distances from the surface allows deep-rooting plants to obtain water from its capillary fringe. Examination of the actual depth of the water-table in various places was too extensive a task to carry out, but certain observations were made while excavating root-systems. In one case especially, the response of the roots to the presence of the capillary fringe was found to be very striking (Fig. 11). The well-known water-holding capacity of fine sand has also been partly responsible in determining plant distribution. This point will be elaborated in a later section.

IV.—Biotic Factors:

The valley of the Upper Clutha, especially upon the Dunstan side, forms along with other regions of Central Otago, a very striking case of the influence of biotic factors in determining the character of the plant covering. The communities which now exist below the 3000-foot level (or thereabouts) can definitely be stated to owe their present nature largely to the destructive influence of fire, sheep, and rabbits, indirectly through the influence of man.

This conclusion can be arrived at not only by a critical study of existing vegetation, but is amply substantiated by the known history of the area as described by settlers of sixty and seventy years ago.

The lower slopes of the Dunstan Range before the occupation by the sheep farmer were covered by a community dominated by the tussock grass. Now, tussocks (Poa caespitosa and Festuca Novae-Zelandiae) grow only in shady or inaccessible places and alongside water-races. The ground is irregularly dotted with the intensely xerophytic cushions or patches of the close-growing Raoulia lutescens. A more complete description of the community is given later.

As Cockayne points out (1922, p. 328), the methods of sheep farming on other tussock grassland communities of New Zealand were, and still are, much the same as those used in the area under consideration. Burning, stocking, and rabbit infestation were as wholesale elsewhere as here. Yet only those parts of Central Otago which are subjected to such a semi-arid climate, the hottest and driest in New Zealand, have undergone depletion. That climatic and physiographic factors finally have the determining influence on the existing vegetation can be shown by the fact that, above 3000 feet, where the rainfall increases, although sheep and rabbits have access to these altitudes, depletion ceases. On sunny faces of gullies in the depleted zone, depletion is complete, while on the opposite shady faces, there are often many tussocks.

The harsh leaves of the tussocks do not afford very palatable feed for sheep, and so the early sheep farmers were in the habit of running indiscriminate fires through the grassland, to take advantage of the more succulent green leaves produced in spring. This burning was carried out year after year, the cumulative effect of which has proved disastrous.

In addition, the runholder stocked his land beyond its capacity, with little regard for the future. Finally, the coming of the rabbits fifty years ago added to the destruction. Their propensity for rapid reproduction and their voracious appetites are well known—“With an eminently favourable climate, abundant food, and a soil suitable for burrowing or rocks in plenty for their homes, these animals increased enormously, so that with them and the sheep, the country became greatly overstocked. Every plant at all palatable was eaten to the ground; the depleted area ascended higher and higher; those perennials alone could survive which either were not eaten at all or were furnished with far-extending underground stems, and

possessed the power of rapid growth after being cropped close. Then there were annual species, which possessed great and rapid powers of increase by means of seed, or, in the case of such plants as die yearly to the ground, by far creeping subterranean stems. It is the addition of the last three categories to the pasture—mostly foreign plants, with sorrel and wild geranium the most important as feed, whose advent in quantity has been made possible by the new ground—which has rendered the sheep runs still productive, bringing in far more good feed than the general aspect of the depleted areas would lead one to imagine.” (Cockayne, 1922, p. 329).

The ubiquitous Raoulia, with its compacted habit, offered no food whatever to the rabbit or sheep, nor was it affected by burning. Foweraker (1917) has shown that its leaf surface, reduced to a minimum, protected by woolly hairs, gives it ample protection against excessive transpiration, and the water storage cells in its leaves gives it power to tide over drought periods. Its abundant winged fruits (it is a Composite) confers upon it efficient powers of dispersal. Its extensive, fibrous, and deep system of roots, occupying the greatest possible amount of soil, gives it power to withstand long periods of draught. All these characteristics combined enabled it to take possession rapidly of the ground made bare by the destruction of the tussock covering. It is dominant now on account of the inability of other plants to withstand the concerted influence of unfavourable factors, not because it occupies ground which might otherwise support tussock-grassland.

Cockayne (1922) has shown that fencing of small areas from stock and rabbits resulted in the rapid regeneration of the original plant covering only when the stumps or underground parts of these plants remained alive in the ground. Where all the plants were quite dead, regeneration took place very slowly by natural spread of seeds on wind-swept areas. But where seeds of both native and introduced plants were sown and raked in, these were able to germinate and establish themselves on the experimental areas.

At the present time, due to more intelligent farming, stocking is not quite so heavy, and rabbits have been greatly reduced in numbers. But, except on the Pisa south-facing side of the valley, no very appreciable difference in the amount of tussock can be observed.

While examining the locality, an interesting piece of evidence bearing upon the composition of the original vegetation was discovered. As previously mentioned, the range is characterised by great outcrops of rock, of very varying shapes and sizes. One such rock was discovered which possessed a more or less flat top. This flat top of the rock was obviously out of reach of sheep and rabbits, was also safe from fire, and could only be scaled with difficulty. On its summit were found eleven species of plants, excluding several mosses and lichens, growing in close association on a layer of soil. These were as follows:—

• Poa Colensoi Hook (tussock growth form).
• Poa caespitosa Forst (tussock growth form).
• Pimelea aridula Cockayne (small shrub).

of stones quite bare of vegetation in some places, and in others slowly being colonised by certain species. The succession on such places would be a study in itself. Again, there are a few places near the river which are subject to periodic flooding and which show a typical swamp association—as evidenced by consociations of Typha augustifolia and Juncus spp. Such localised places do not enter into this discussion.

I.—The Dunstan Lower Slopes:

An account of the severe factors of the habitat acting upon the association which covers this region has already been given.

The dominant species is Raoulia lutescens Beauv. The silvery green flat cushions of this plant (locally known as “scabweed”) comprise the striking feature of the landscape. At all seasons of the year they retain the same appearance, seemingly unaffected by the changes in temperature or humidity. On steep and gentle slopes alike, their deep, fibrous roots afford ample anchorage and a sufficiency of water, while their reduced leaf-surface brings transpiration to a minimum. As mentioned above, the frequency with which the patches occur differs on sunny and shady slopes. Measurements of the diameter of a number of patches were taken during March, and these were checked again in August—six months later. It was found that during this period one patch had increased in average diameter by about one inch while others had increased by ¼-inch to ¾-inch. It is unfortunate that the observations could not have extended over a longer period—at least a year—so as to obtain more reliable results. Moreover, these measurements represent the growth during the slower growing period. However, supposing that one takes a conservative estimate of an average annual growth of one inch in diameter, one might well suppose, considering the numerous patches present, that in a very few years they would coalesce to form a continuous carpet. This is actually found in some south-facing slopes, but over the greater part it never occurs. The reason seems to be that after a certain maximum is reached, the patch begins to die away in the centre. Thus its resistance is lowered, and the wind aided by small stones and sand, soon erodes the dead portion away, blowing away also the humus which had collected beneath. Thus the original large patch becomes broken up sometimes into two, three, or four smaller patches, representing the peripheral living portions, and new ground is laid bare.

Considering the open exposed ground, that is, excluding such places as shady crevices between the large outcrops of rocks, and the rocks themselves—different seasonal aspects can be distinguished in the association.

Prevernal and Vernal Aspect.—As noted above, the Raoulia patches are present at all seasons, and maintain much the same appearance. But during the period between June and September, the ground between the patches becomes quite green with thousands of seedlings of such plants as Poa Maniototo Petrie, a small tufted

grass, Agrostis alba Linn. (introduced), Stellaria gracilenta Hook., a wiry herb, and Mysosurus aristatus Benth., a small herb with radical leaves.

Early Aestival Aspect.—From November to January the annuals which had germinated previously, develop and produce their seed. The abundant white starry capitula and winged fruits of Raoulia lutescens are produced during December and January.

Late Aestival and Autumnal Aspect.—From February to May the effect of the hot summer period shows itself, and the annuals are for the most part either dead or severely desiccated. Poa Maniototo persists as dead or almost dead tufts between the scabweed. Only the scabweed appears to be alive, and the soil between the patches is more or less bare. Small cushions of Colobanthus brevisepalus T. Kirk, one to two inches in diameter, appear to persist actually on the scabweed patches all the year round.

There are a number of shrubs which grow on the slopes. However, they are found only dotted here and there, and never occur in such numbers as to lend any distinct facies to the association. The commonest of these is Hymenanthera dentata var. alpina T. Kirk, an almost leafless, densely interlaced and low-growing shrub. Olearia odorata Petrie, O. lineata Cockayne, and also Sophora tetraptera Mill, are generally to be found in the more shady gullies, where they may attain the height of shrubby trees.

The moss and lichen flora is extensive, the abundance of bare, jagged rocks offering an excellent holdfast. The most abundant moss is Grimmia trichophylla, growing in small rounded cushions, its “leaves” covered with small silky hairs.

II.—The Pisa Lower Slopes:

It has been observed previously that these slopes face in a southerly direction, and that there is little of the depletion which so characterises the opposite side of the Clutha Valley. With the exception of but a few, all the plants growing on the Dunstan side are also present on the Pisa side of the valley, but they are present here with others, in far greater amount. Except on the barest tops of the ridges between the lateral gullies, Raoulia species are rare. The tussock grasses (Poa caespitosa and Poa Colensoi) are specially abundant, and in most places the silver tussock (P. caespitosa) forms a continuous covering, becoming the dominant species. It has already been stated that this difference in vegetation is considered to be in direct relation to physiographic features.

III.—The Terraces:

The terraces occupy the greater part of the floor of the Clutha Valley at this part of it (Fig. 20). They are composed of great thicknesses of fluviatile and lacustrine gravels which for the most part are covered by a layer of light-coloured soil of an average depth of about one foot. The gravels are absorbent and quickly take up the moisture which is precipitated.

The association, except where man's activities have completely changed it, shows definite seasonal aspects. During the prevernal and vernal periods the dominant species are Erodium cicutarium L'Herit, and Rumex acetosella Linn.; both introduced plants. The latter is present almost the whole year round. Associated with these species is a very abundant lichen which has not been identified. This lichen is peculiar in being, quite unattached, and in cartwheel tracks and slight depressions it may be scooped up in handfuls. No apothecia could be discovered upon it. Almost as frequent as the Erodium are dead stumps of the tiny Poa Maniototo Petrie. Patches of a species of Polytrichum are also common.

During the early aestival period, the continued hot weather causes the lichen to dry up and disappear. Erodium cicutarium becomes less dominant, and Rumex acetosella to a large degree persists, giving to the ground a reddish tint.

Over the greater part of the terraces, especially wherever the top soil becomes thinner, the patches of Raoulia lutescens and R. australis occur dotted here and there. In places where the top soil is very thin or even absent altogether, these plants may be the sole occupants, but for the most part they do not occur to nearly the same extent as on the lower slopes of the Dunstan Range.

In certain places Geranium sessiliflorum Cav. may become an important member of the turf. The character of the gravel is apparently very suitable for the development of its characteristic surface and deep sets of absorbing roots. Its comparative rarity on the Dunstan slopes may be explained on similar grounds—i.e., that a sufficiently great depth of soil is not present to allow of the development of the deeper roots of this type of root-system.

Hymenanthera dentata var. alpina occurs as compact low shrubs (Fig. 22).

Poa caespitosa is conspicuous by its absence. This would seem to be related to the character of its root-system. The species is distinctly perennial and requires a certain minimum of moisture in the surface soil all the year round, owing to its shallow root-habit. The absorbent nature of the sub-soil causes the top soil to dry out in summer very easily, and thus the moisture content is too low to allow this tussock to live, except in shady places, or where for any other reason the amount of surface moisture is maintained. It will be shown in considering the Cromwell sandy flat that other reasons do here permit the establishment of Poa caespitosa.

IV.—The Cromwell Sandy Flat:

It has been shown under the section on Climatic Factors that this area is for the most part covered with a layer of drifting sand derived from the sandy faces of the terraces near Deadman's Point. The area which is so covered can be seen by reference to the map (Fig. 20).

A combination of factors has here brought about the dominance of Poa caespitosa. The root-habit of this species is a shallow one,

and it can be fairly certainly asserted that it would not occur except in soils which would allow such penetration of the taproot. Throughout New Zealand, Poa caespitosa thrives under a variety of climatic conditions. But in the semi-arid district, its moisture requirement in the surface soil is just at the point of balance between an adequacy and a deficiency. Any factor which tends to bring the soil moistur in the uppermost two feet of soil above a certain minimum, will favour the dominance of the tussock. A knowledge of its root habit gives a clearer insight into its distribution.

The presence of plants having a system of roots absorbing from the capillary fringe indicates the possibility of using on the terraces crop plants which are known to have a similar habit, e.g., lucerne. The difficulty with regard to this is that the young plants would need to be supplied with water by irrigation for two or three years until the deeper roots had been established. This has been the experience in certain parts of the U.S.A. (Meinzer 1927, p. 89). It appears that provided the lucerne is tided over the critical period in the first few years of its life, then satisfactory crops are returned without further irrigation. However, the author adds that “on lands where natural subirrigation has proved feasible, the soil and subsoil down to the water-table is a dark grey clay loam or sandy loam and a black loam derived largely from decomposed peat. Attempts to extend cultivation of subirrigated lucerne to adjoining areas where the subsoil is gravelly have not proved successful, although the depths to the ground water in these areas are no greater than in the area where success has been attained.”

It may be added, however, that the depths to ground water in the above mentioned region were from 9 to 15 feet. From the few observations made upon the soils of the Cromwell area, the depths of the capillary fringe at least, seems to be somewhat less than these figures, in one case 4 feet. This relative shallowness of the watertable might well discount the difficulty of the presence of a gravelly subsoil.

Three of the species examined were found to have distinct surface and deeper systems of absorbing roots, viz., Carmichaelia Petriei, Geranium sessiliflorum, and Hymenanthera dentata var. alpina. This character seems to be a distinct development in relation to the water supply of the soil.

The true definition of the term “xerophily” has been much discussed. (Delf, 1915). The older view was to class as xerophytes those plants with form and structure apparently adapted to reduce water loss by transpiration. But it has been proved that not all plants which grow in such places are so modified, and plants with thickened cuticle, sunken stomata, or succulent leaves are not exclusively characteristic of desert places. Two examples can be quoted from the present study, viz., Hymenanthera dentata var. alpina and Geranium sessiliflorum, neither of which has leaves highly modified to reduce transpiration.