Comments on the Shingle Vegetation of the Horokiwi Stream
[Received by the Editor, January 11, 1955.]
From observations on the changes in the shingle areas and the consequent changes in the vegetation, particularly of two unstable shingle banks, following floods in the Horokiwi Stream, Wellington, N.Z., the probable course of plant succession is outlined. This begins with colonization of the unstable shingle by Polygonum hydropiper and Spergula arvensis and ends with a grass dominated community similar to the adjacent pasture. A striking feature of the vegetation on this shingle is that it is composed almost entirely of introduced plants—that is, plants which are alien to the New Zealand flora.
A small open river, the Horokiwi Stream, rises at about 1,300 feet, near the summit of the Paekakariki Hills, and flows out into the north end of the Porirua Harbour. It has been described in detail by Allen (1951). The stream is subject to periodic flooding and accumulations of water-borne stones become deposited as areas of shingle at various points along the course, in particular on the flood plain of the lower reaches. The extent and position of these shingle deposits is largely determined by the volume and frequency of the floods; there may be continual alteration of the course of the stream with successive building up and breaking down of the shingle areas or they may be submerged and changed only by the heavier floods. Ten miles long, it consists of two main branches, which unite about three miles from the sea. The valleys are narrow with steep sides, but just above the junction there is a flood plain with low terraces. Below the junction the stream passes through a small gorge and emerges into the flood plain of the main valley, which is 300 yards wide at the upper end, widening to about 600 yards. In the upper part of this main valley there are low, irregular terraces and, except where the stream cuts into the old terraces, the stream banks are very low. In the lower part the stream is relatively stable, with steep or terraced banks 6 to 12 feet high, and there are occasionally small shingle beaches.
Originally the vegetation of the hills and valleys of this area was dense podocarp broadleaf rain forest. In 1849 the main road from Wellington to the Manawatu was completed, the forest being gradually cleared and burnt, until, at the end of the century, the greater part of the area was in pasture. Only a few remnants of modified forest are left on the steeper parts of the tributary gullies, but small groups of conifers and isolated trees have been planted along the stream banks of the main valley, otherwise the stream is completely open. The pasture extends to the water's edge, with small local modifications in the plant communities related to the type and height of the bank cut by the stream. The vegetation of the stream bed itself is mainly unicellular and filamentous algae. Occasionally Mimulus guttatus, Nasturtium officinale and species of Juncus are found in shallow pools.
The water supply is mainly surface run-off from the pasture and hence is quickly affected by local rainfall. Severe floods in 1941 resulted in a long period
of instability of the stream bed (Allen, 1951), and floods in 1947 caused major changes of the course in the upper part of the main valley and left local deposits of shingle of considerable depth. In this area during the next few years changes in the course of the stream were frequent. Observations on the shingle bank vegetation of the main valley were begun in early October, 1951, and on the 26th of October a flood of three feet caused considerable disturbance of the shingle and several changes of the course in the middle reaches of the main valley, while in the lower part there was a slight raising of the stream bed, some deposition of silt and shingle and a small amount of bank erosion. On the 2nd of November another flood (of 2 feet) occurred, causing movement of the shingle in the unstable areas. Several small floods occurred between November, 1951, and April, 1952, but no major changes in the shingle areas were recorded. There were no further floods before observations ended.
Shingle Areas of the Main Valley
For three and a-quarter miles the stream flows through the flood plain of the main valley; shingle deposits are a characteristic feature of this part of the course, especially in the upper part of the valley where there is a considerable depth of alluvium and course changes have been frequent during the last ten years For the detailed survey the shingle areas were divided into (1) those of the upper part and (2) those of the lower part of the main valley. Those of the upper part were further divided into (A) stable and (B) unstable shingle—this was an arbitary distinction based on the condition of the shingle at the time of observation, as there tends to be a gradation from unstable to stable, and only the apparent ends of the series have been described. Only the dominant or most characteristic plants are given in the descriptions, as a species list is appended in which is given an estimate of the abundance of each species in the various types of shingle area.
1. Upper Part of Main Valley
1. A. Stable Shingle
The areas of shingle in the upper part of the main valley that were undisturbed by the heavy floods in October were regarded as being stable at the time of the investigation. These areas occurred both as shingle beaches (banks) and as beds or islands in midstream. They were usually covered with a mixed herbaceous vegetation, although on some banks grasses were predominant. In the mixed herbaceous community Digitalis purpurea was the most prominent plant, and associated with it were the large dock Rumex obtusifolius and two species of Chrysanthemum—C. leucanthemum and C. parthenium. The most abundant grasses were Dactylis glomerata and Anthoxanthum odoratum, and the rushes Juncus bufonius and J. lampocarpus were locally abundant.
Considerable amounts of silt often accumulate between the stones and the density of the ground cover varies with the amount of silt. Where there was only a small amount of soil and the taller vegetation was relatively open, the creeping plants Geranium molle, G. Robertianum, Acaena sanguisorbae and Epilobium nummularifolium formed a close ground mat; but in areas where silt had accumulated there was a higher percentage of grasses forming a closed turf which included Dactylis glomerata, Anthoxanthum odoratum, Holcus lanatus and Poa annua. In wet depressions and swampy areas Agrostis stolonifera and
Fig. 9.—Shingle disturbed and reported on 26th October. 1951.
Fig. 10—The same area 14 months later December 1952 [ unclear: ] in flower young plants of [ unclear: ] and the grasses [ unclear: ] and [ unclear: ]
9—Shingle disturbed and reported on 26th October. 1951.
Fig. 10—The same area 14 months later December 1952 [ unclear: ] in flower young plants of [ unclear: ] and the grasses [ unclear: ] and [ unclear: ]
Fig. 1.—Summary of the changes in a shingle bank in the middle reaches of the stream, since its deposition in 1947 until 1952.
Fig. 2.—Distribution of plant communities on this unstable shingle bank before the heavy flood in October, 1951.
Fig. 3.—Changes in the shingle area and the vegetation after the flood.
Cotula coronopifolia were the most common plants with Juncus effusus and Rumex obtusifolius. In small pools and along the stream margin the characteristic plants were Glyceria fluitans, Mimulus guttatus and Callitriche verna.
In contrast, in the more stable lower parts of the stream, many of the small shingle beaches had a closed turf cover in which Agrostis alba, A. stolonifera and Anthoxanthum odoratum were prominent. After heavy floods some of these small beaches were still covered with water when most areas were dry, but they were unaltered in extent or vegetation when the water receded. These areas were regarded as examples of what is, in all probability, the final stage in stabilization of the shingle deposits.
1.B. Unstable Shingle
Most of these areas of shingle in the upper part of the main valley, however, can be regarded as unstable, as most of the banks and islands of shingle are continually undergoing changes in extent and vegetation. In mid-October, 1951, one of the larger shingle banks in this category was mapped (Fig. 2). Here the stream was confined to a narrow channel beside the cast bank (an old terrace) with a low bank of shingle (10 to 75 feet wide) at the foot of a terrace on the west. Swampy pasture extended from the top of the terrace to the hills. On the basis of the plant communities then present the bank was divided into five areas:—
The northern (upstream) end of coarse shingle had a sparse cover of Polygonum hydropiper and Spergula arvensis.
A band of vegetation in which seedlings of Rumex acetosella and the grasses Agrostis alba, Bromus mollis and Anthoxanthum odoratum were prominent among the dominant Polygonum hydropiper.
This area resembled in appearance a sparse turf with many weeds. With the grasses Dactylis [ unclear: ] , Agrostis and Anthoxanthum, other prominent plants were Trifolium repens, T. dubium, Rumex acctosclla, Ranunculus repens, Hypochaeris radicata and Stellaria media.
A narrow zone with a considerable depth of silt, here Polygonum and Spergula were dominant to the exclusion of other plants.
At the base of the terrace the vegetation was similar in composition to the pasture on the terrace, with grasses and clover dominant. Large stones were scattered over the turf.
During the October floods there were several changes of the course of the stream in the upper part of the main valley. When the water receded the degree of disturbance of the bank described above was found to have varied with the density of the previous plant cover, being least in the turfed area 5, and greatest in area 1 which was now covered with newly deposited shingle, and fresh shingle deposits had also extended this area further upstream (Fig. 9). There was a three inch deposit of silt along the stream edge of area 1. In areas 2 and 3 silt and shingle had been deposited, many of the plants were broken and in area 2 many were dead, but new shoots of Polygonum were visible 5 days after the flood. On area 4 there was a thick deposit of silt and shingle, up to 6 inches of silt had covered the Polygonum. Further large stones and debris had been deposited on the turfed area 5, but the grasses were only slightly damaged. These changes are shown diagrammatically (Fig. 3).
There was no further disturbance of the shingle of this bank during spring and summer. By mid-November area 1 was covered by Spergula in flower, and
Fig. 4.—Distribution of the plant communities and the dominant species on a bank in the lower reaches of the stream; inner part apparently stable bearing a closed turf, outer margin of unstable shingle.
Fig. 5.—Changes in this bank after the flood on 26th October, 1951.
there were fresh shoots of Mimulus guttatus along the stream edge. New shoots of Polygonum were abundant in areas 2 and 4, and young plants of Gnaphalium Keriense were common. Rumex obtusifolius and Ranunculus repens were sprouting from broken rootstocks.
By mid-December there was still an almost pure community of Polygonum and Spergula in area 1. In area 3 more species had appeared, in particular seedling grasses, also Stellaria media, Anagallis arvensis and Geranium species. The tall flowering plants of Chrysanthemum leucanthemum, Digitalis purpurea and Rumex obtusifolius were the most prominent in the very mixed community. The deep deposit of silt in area 4 was still marked by a community of Polygonum hydropiper, in which there were scattered clumps of Spergula. In January the flowering heads of the grasses overtopped all except the seed heads of the Digitalis, and the grasses were becoming the dominant members of the bank vegetation, the previous subdivision of the bank into 5 areas was no longer clearly visible. Fig. 10 shows this bank 14 months after the flooding. (Compare Fig. 9.)
Fig. 6.—Northern—(upstrcam)—end of the bank shown in Fig. 5, showing the area most affected by the flood.
Fig. 7.—Extent of recolonization six weeks after the flood
2. Lower Reaches of the Stream
Characteristic of the lower reaches of the main valley were shingle banks bearing a closed turf on the inner side, but with an unstable shingle margin. For example on the convex side of a sharp turn in the stream there was a small bank of shingle built up at the base of a steep bank. The outer margin of the shingle was unstable and had a sparse cover of Polygonum hydropiper, Juncus bufonius, Rumex acetosella and Trifolium repens, while on the inner part, which had been relatively undisturbed since May, 1947, there was a closed turf (Fig. 4).
In October, 1951, and again in April, 1952, flood waters covered this bank and small amounts of silt were deposited at the southern end, the grasses recovered rapidly and there were no changes in the vegetation apparent In October there were, however, changes in the outer margin of the shingle, and all the Mimulus plants growing along the edge were swept away. There was a slight raising of the stream bed due to deposition of shingle at the northern end and a long spit of silt and sand had been deposited at the downstream end of the bank (Fig. 5). The changes that occurred in the northern (upstream) end of this bank are shown in more detail in Fig. 6. At the end of six weeks Trifolium repens, spreading rapidly from broken rhizomes, had with Ranunculus repens covered nearly 30% of the northern end. Seedlings of Speraula. Polygonum and Rumex acetosella were numerous and Mimulus guttatus had formed
a dense mat along the edge of the water (Fig. 7). The movement of shingle and therefore recolonization by plants was, however, slight compared with the changes in the shingle area and vegetation in the upper part of the main valley.
The large, unstable shingle bank described on page 4 was deposited in 1947 when there was a complete redistribution of the shingle and a change in the course of the stream in the upper part of the main valley. A further deposit of shingle and heavy silting of the downstream areas occurred in September, 1948, and by the summer there was a scattered vegetation in which Polygonum hydropiper was prominent. The stream occupied a similar course in March, 1951, to that of September, 1948, but several changes had occurred in the intervening years. The main changes are summarised in Fig. 1. There was little change in the bank during the winter of 1951, and in mid-October the plant communities were mapped (Fig. 2). After a flood on the 26th October the changes in the shingle areas and the vegetation were recorded (Fig. 3).
The fresh deposit of shingle was colonized by Spergula arvensis (Fig. 11) and areas of silt and shingle were colonized by both Polygonum hydropiper and Spergula (Fig. 12). Polygonum, which is a much larger plant, grew very rapidly and appeared to be the dominant. In two months these two plants had formed a complete cover, and they are probably the true pioneers in the colonization of silt and shingle in this stream. Both are annual species which grow rapidly and easily from seed which germinates at any time of the year.
Most of the shingle of area 3 had been deposited in 1950, and by October, 1951, there was an open community including Rumex acetosella, Stellaria media, Plantago lanceolata and Hypochaeris radicata. The plants were broken and battered by the 1951 floods but gradually recovered, and by December, 1951, there was a dense, mixed herbaceous community in which flowering Chrysanthemum species were most prominent.
Most of area 5 had been undisturbed since 1948, and a closed turf cover had developed in which Holcus lanatus, Dactylis glomerata, Bellis perennis and Trifolium species were the most abundant; the only effect of the flood in this area was a slight silting and deposition of debris.
This suggests that on these banks three to four years may be sufficient time for the development of a closed community in which grasses predominate. From this one could postulate an increasing stability of the shingle with an increasing plant cover, in a direct time sequence.
There are, however, other factors to be considered:
The most important is the size and frequency of the floods. In the Horokiwi Stream the water supply is mainly surface run-off from the pasture and is quickly affected by the local rainfall. Allen (1951) found a very close relationship between the amount of rain and the rise in stream level Slight floods cause surface gravel movement, dislodging all vascular plants in the stream bed but having no effect on the shingle banks. Moderate floods cause a rise of up to 3 feet in stream level, usually resulting in slight marginal changes in the shingle banks in the unstable areas and building up the stable areas by silt deposition. Vegetation may be damaged but not completely destroyed. Whereas severe floods (over 3 feet) result in changes in the stream course and disturbance of the
shingle to a considerable depth. Plants are uprooted, especially on the islands, and bank erosion undermines the pasture. Even low-lying stable banks may be swept clear of vegetation, cut away and the shingle redeposited. Severe floods may completely change stable as well as unstable areas.
Mechanical settling of the shingle leads to consolidation of the banks and increases the resistance to flood waters. Thus a flood occurring while the shingle is loose and unstable will cause more disturbance than one of similar size after a long period of low levels during which the shingle has consolidated.
In this stream the position of the shingle bank in the river is also important. In the upper part of the main valley there is a considerable depth of loose shingle in the stream bed, and during the last 10 years changes in the course of the stream and changes in the shingle banks have been frequent, but the lower part of the stream is relatively stable and floods cause only small changes.
Another factor is the position of the stabilized part of the bank with a closed turf cover in relation to the direction of stream flow. The upstream ends and outer margins of the banks are most frequently disturbed and are often affected by small floods, while silting continues to build up the more stable, inner and downstream parts of the bank.
In the case of moderate floods, a factor which may be significant is the time of the year at which the flood occurs; the herbaceous community is more open in winter and less resistant to floods, becoming denser in the summer with the growth of annuals and the flowering of perennials.
There is a plentiful and continual supply of seed from the adjacent areas so colonization of the bare shingle is rapid. Slight disturbances uncover seed already buried in the shingle, and many of the perennial plants can grow from broken rootstocks or pieces of rhizome or stolon, so that the early stages of the succession may be omitted or may be of very short duration, except after heavy floods which cause complete destruction of the vegetation and redistribution of the shingle.
The interaction of these factors results in a tendency for an area of shingle to stay at the particular stage of stabilization and plant succession that it has reached. A bank of loose shingle with a sparse plant cover will be continually disturbed by small floods, whereas a bank of consolidated shingle with a dense plant cover is affected only by severe floods. A closed turf recovers quickly after inundation and small deposits of silt.
The permanency of the beds or islands of shingle in midstream, however, may depend more on the frequency of major course changes than on the stability of the shingle and the density of the plant cover. These islands are more common in the middle reaches of the main water where the stream bed is unstable, they are frequently swept clear of vegetation and the shingle cut away and redeposited. Characteristic of these islands is a mixed herbaceous community in which flowering Digitalis and Chrysanthemum species are the most conspicuous plants in summer. The final stage of a closed grass cover is rarely developed on these islands.
Banks on which grasses are dominant, often to the complete exclusion of Polygonum, Digitalis and Chrysanthemum, are a more common feature in the relatively stable lower reaches of the stream. Their vegetation is very similar to that of the adjacent pasture and where accessible they are grazed by the cattle.
Floods may cover these banks and recede without apparent change in the extent of the bank or the plant cover.
From a consideration of the various shingle banks and the changes that took place during the period of observation it is possible to suggest a sequence of plant succession beginning with colonization of the unstable shingle by Polygonum hydropiper and Spergula arvensis and ending with a grass dominated community similar to the adjacent pasture. The succession of stages may be very rapid, but is interrupted at any point and to varying extent by flood waters, depending on (a) the height and duration of the flood, (b) the mechanical stability of the shingle and (c) the density and age of the vegetation. This outline of plant succession applies only to the Horokiwi Stream, but as there are few accounts of this type of observation on river shingle it was decided to publish these notes. To form definite conclusions it would be necessary to study in detail the establishment of the plants in relation to the time and depth of silt and shingle deposition.
A striking feature of the vegetation found on the shingle of the Horokiwi Stream is that it is composed almost entirely (95%) of introduced plants— that is, plants alien to New Zealand. This very high percentage of introduced plants is emphasised by reference to White's paper (1889), in which he describes the river shingle vegetation of a Perthshire river (Scotland). Many of the species he lists as native in that habitat are to be found on the banks of the Horokiwi Stream. A large number are plants commonly associated with pastures and cultivated areas as weeds. Their presence on the shingle of the Horokiwi Stream can be attributed to the operation of biotic factors: the clearance of the forest and the introduction of pasture by man and the presence of grazing animals which encourage the development of a closed turf and prevent the establishment of woody species of the climax native forest.
I would like to thank the Fisheries Laboratory, Marine Department, Wellington, for permission to join some of their field trips to the Horokiwi Stream, the staff for their helpful criticism, and Mr. B. T. Cunningham for the information on shingle deposition.
1. Allan, H. H., 1944. Naturalized Plants of New Zealand. D.S.I.R. Bulletin 83.
2. Allen, K. R., 1951. The Horokiwi Stream. A Study of a Trout Population. Fish Bull. Wgtn., N.Z. 10. 231 pp.
3. White, F. B., 1889. The Flora of River Shingles. Scot. Nat. IV. (N.S.), p. 290-299.
|A. stolonifera||Creeping bent||f||o||f||-|
|Aira caryophyllea||Hair grass||f||f||f||i|
|Anagallis arvensis||Scarlet pimpernel||f||f||i||-|
|Anthoxanthum odoratum||Sweet vernal||f||f||a||a|
|Capsella bursa pastoris||Shepherd's purse||f||f||o||o|
|Cerastium iulgatum||Mouse-eared Chickweed||o||o||-||-|
|Chrysanthemum leucanthemum||Oxeye daisy||a||a||r||-|
|Cotula coronopifolia||Batchelor's button||a||f||r||-|
|Festuca elatior||Meadow fescue||-||r||o||o|
|F. rubra v. commutata||Chewings fescue||f||f||f||f|
|G. Robertianum||Herb Robert||f||f||-||-|
|Glyceria fluitans||Floating sweetgrass||o||i||o||-|
|Holcus lanatus||Yorkshire fog||o||o||f||a|
|Hordeum murinum||Barley grass||o||f||o||o|
|H. perforatum||St. Johns wort||i||o||-||-|
|Juncus bufonius||Toad rush||va||a||r||r|
|J. lampocarpus||Jointed rush||o||o||r||i|
|Lolium multiflorum||Italian ryegrass||-||r||o||o|
|L. perenne||Perenmal ryegrass||r||r||f||a|
|Mimulus guttatus||Monkey musk||f||o||i||-|
|P. major||Broadleaved plantain||o||f||o||r|
|Poa annua||Annual poa||a||a||f||o|
|P. pratensis||Kentucky bluegrass||f||o||f||a|
|Polygonum hydropiper||Water pepper||va||a||r||-|
|R. acetosella||Sheep's sorrel||a||a||f||o|
|Trifolium dubium||Suckling clover||f||f||f||o|
|T. repens||White clover||a||a||f||a|
|T. subterianeum||[ unclear: ] clover||r||i||f||f|
va, very abundant. a, abundant. f, frequent. o, occasional. r, rare 1, local.
Miss B. H. Croker
University of Sheffield