4. PASTURE ESTABLISHMENT

When the cover beneath coconuts has to be modified or replaced with exotic pasture species, the methods used will depend upon several factors: intensity and scale of development, nature of the existing vegetation, soil type and topography and characteristics of species to be sown or planted (Whiteman et al., 1974; Whiteman, 1980). Where cropping has been practised under young coconuts little land preparation will be necessary, whereas the tangled mass of undergrowth found in many older coconut areas may require considerable clearing (Reynolds, 1980).

Successful pasture establishment is largely a matter of common sense and practical experience (Teitzel and Middleton, 1978). While the different nature of each site may require different methods of establishment, the basic principles and sequence of steps to be followed are similar. The principles and techniques of pasture establishment have been described by a number of authors: Anon., 1974; Anon., 1993a; Campbell, 1972; Evans et al., 1992; Guzman and Allo, 1975; Humphreys, 1978; Jones, 1975; Jones and Jones, 1971; Lane, 1981; MacFarlane et al., 1991; Partridge, 1977; Plucknett, 1979; Roberts, 1974; Reynolds, 1978h; Skerman, 1977; Skerman and Riveros, 1990; Skerman et al., 1988; Steel et al., 1980; Swain, 1968; Teitzel, 1992; Teitzel and Middleton 1978; Teitzel et al., 1974; Toledo and Morales, 1979; Whiteman et al., 1974; Whiteman, 1980; Williamson and Payne, 1978.

Pasture establishment has been defined by Gramshaw et al. (1993) as “the sequences of seed germination and seedling development that normally permit the persistence of the introduced species in the longer-term” or “the conversion of seed or other propagating material into production or resource-maintenance benefits”.

4.1 Preliminary considerations

Good planning is a major prerequisite of successful pasture establishment. All available information on soil type and fertility, native grass and legume species, age of coconut palms, degree of shade and seasonality of rainfall should be collected and analyzed. MacFarlane et al. (1992) suggest that the agricultural extension worker should check that: the farmer has enough money or labour for the project; there is a suitable site with a strong fence available; the right pasture species to use are determined; the best time of year for establishing the pasture has been determined, and a decision is made whether seeds or cuttings will be used for establishment. Only then should a plan be prepared. Teitzel and Middleton (1978) suggest a twelve point procedure for pasture establishment. Two important principles to follow are: to plant in each sowing season only that area which finances and resources will allow to be done thoroughly, and to decide on priorities for pasture establishment. Where rich and poor land is available for development (providing no other crops are to be planted) the richer land should be developed first, as the return on investment is likely to be higher (Williamson and Payne, 1978). Under coconuts improve first the areas where the available forage is poorest and carrying capacity lowest, leaving the areas with good local pasture.

4.2 Clearing land for planting

The principal aim of the farmer is to provide a favourable environment in which seeds germinate, seedlings emerge, grow and survive. Competition for light, water and nutrients must be eliminated or reduced to a point where productive pastures can be established and can survive with other vegetation (Skerman, 1977). In South-East Queensland, Cook and Ratcliff (1992) noted that the control or management of plant competition during the first three months of establishment was a key factor in determining the final pasture composition and, ultimately, the production of the different pastures that developed.

4.2.1 Clearing trees and bushes

- hand clearing or mechanized methods can be used. Where there is a plentiful supply of labour then mattock, machete, panga or bushknife and axe can be used on trees, saplings and bushes. In certain places use of the chain saw reduces clearing time. Once cut the debris (including fallen coconut trees, fronds and trunks) can be piled into heaps and burnt; this ought to be done a few weeks before the rainy season is expected (Skerman, 1977). Bulldozers with rake blades and root rippers are sometimes used to remove stumps and window debris, but considerable care should be taken not to remove valuable top soil because soil fertility in many tropical soils is closely related to soil organic carbon status (Chase, 1977, Friend and Birch, 1960: Kalpage and Thenabandu, 1969; Reynolds, 1973, 1976b; Setzer, 1967; Seubert et al., 1977; Turenne, 1970; Webster and Wilson, 1966: Yamaya, 1966; see Table 63). In Vanuatu, MacFarlane (1993) notes that the use of tractor drawn heavy chain or angled bar (see Figure 106) has proven a low cost and effective method of remov-ing Solanum torvum from pastures, reducing the need for herbicide usage on plantations and large commercial smallholdings. Evans et al. (1992) suggest that for any farming system to be sustainable it should maintain soil organic matter levels of at least 3%.

When few weeds remain after clearing, surface preparation can begin.

Figure 106. - Tractor drawn bar and chain used to remove Solanum torvum in pasture rehabilitation in Vanuatu.

Table 63. - The relationship between various soil properties and soil organic carbon status at Togitogiga, Western Samoa (Reynolds, 1972)

P = 0.001

4.2.2 Clearing grasses and weeds.

Where there is only a light cover of unwanted material, there are at least three methods that can be used:

1. Fence the area and establish a system of paddocks

   • with a high stocking rate, graze the first paddock using the animals as weeders to consume or trample existing grasses and weeds (Partridge, 1977). If possible dry cows or mature steérs should be used at a stocking rate of as many as 50 animals per hectare (care should be taken to ensure that poisonous weeds like lantana (Lantana camara) and milkweed (Asclepias curassavica) are cleared first. For the effects of poisoning with Lantana on cattle, goats and sheep refer to Frisch et al. (1984); Obwolo et al. (1990); Pass et al. (1978) and Seawright (1963). In Vanuatu, MacFarlane et al. (1992) suggest that if an area is covered by Merremia (big leaf vine) cattle at 5 or 10 ha^-1 will defoliate or trample the vine after 4 to 6 weeks so that the area is ready for planting or improved pasture without the need for ploughing if grazed during the wet season. This saves a considerable amount of labour with bushknives, possibly 20 to 30 man days ha^-1. Mullen (1993) indicates that seeding should take place as soon as the ground is bare and before pioneer native grasses emerge. To allow for seed burial by hoof action, cattle should be left in the paddock for one to two days after seeding. Some animal weight loss is likely to occur as the grasses and weeds are grazed out.

   • once one paddock is grazed/trampled so that little surface green vegetation is left, the animals are moved to the next paddock, leaving the first paddock ready for the last preparation phase. Mullen (1993) suggests that if the grazing and trampling is done in the wet season the paddock can be seeded one day prior to removal of stock, or hand planted with cuttings.

   • where bushes and shrubs occur (e.g. Castilloa elastica, Psidium guajava), stem painting with a diesel oil: 2.4.5 - T mixture (2–3 percent 2.4.5 - T) may be necessary to prevent regrowth (Reynolds, 1978k).

2. Where no animals are available then weeds will have to be slashed with machete or sprayed with chemicals.

3. Where the covering vegetation is dry and brown at the end of the dry season, fire may be used to expose the soil for the surface preparation phase. Care should be taken to ensure that coconut palms are not damaged by fire (Plucknett, 1979). In Vanuatu, planting sites are often prepared in native pastures by burning coconut husks or coconut fronds in strips (strips are also prepared by spraying glyphosate).

4.3 Land preparation

4.3.1 Methods

The type and extent of land preparation will depend to a large extent on available resources and whether bunch or stoloniferous grasses, seeds or cuttings are to be used.

1. Tractor and disc plough followed by disc harrowing (see Figure 107) prepares the cleanest seedbed (see Figure 108), but is expensive. Recommendations for ploughing and harrowing twice followed by sub-soiling, based on practical experience in the Philippines, have been made by McEvoy (1974), Guzman and Allo (1975). In Western Samoa one pass with the disc plough followed by one or two passes with the disc harrow was found to be adequate and in Malaysia Chen et al. (1978) reported that a mixed sward of P. maximum and S. guianensis was sown under five year old oil palms after the area had been lightly worked over twice with disc harrows. Rika et al. (1981) established grass-legume pastures under coconuts in Bali after the area had been prepared with three ploughings, while Watson and Whiteman (1981a) used a tractor mounted rotary cultivator and off-set disks, taking care to avoid deep cultivation and damage to coconut roots. In Sri Lanka Liyanage (1986) recommended a light ploughing followed by discing and then a second ploughing (and discing) at right angles to the first. In Vanuatu, disc harrows with heavy chain harrows following behind can produce an adequate, low cost seed bed in one pass (MacFarlane, 1993).

   

   Figure 107. - Seedbed preparation with tractor and discs, Vanuatu (D. MacFarlane).

   

   Figure 108. - Seedbed prepared by disc ploughing and harrowing in two directions.

2. Tractor and chisel plough or ox-plough should be used when soil is stony or where less expensive methods, more rapid or less complete cultivation (for cuttings) are required.

3. Heavy grazing might be sufficient to remove most of the surface vegetation. Herbicides or hand hoes can be used to spray or prepare strips or circles for planting. If heavy grazing has churned and exposed the soil then little additional work will be required.

4. If burning has removed all or most of the surface vegetation, then either points i), ii) or even direct planting/sowing can be applied.

4.3.2 Depth of cultivation

Disagreement exists as to the effect of tillage on coconuts. Because some coconut roots are close to the surface and may be damaged by tillage implements it has been suggested that cultivation should not take place under coconuts (Wijewardene, 1954). However, Liyanage (1986) recommended light ploughing (and discing) to a depth of 15 cm. In the Philippines it was recommended that the first ploughing should penetrate 15 to 20 cm, the second to 25 cm and sub-soiling should reach 50 cm, however, this operation should not be carried out closer than 1.5 m to the base of the palms to avoid root damage (McEvoy, 1974). Rognon et al. (1990) reporting on coconut smallholdings in Indonesia indicated that intercropping should be undertaken with care to avoid damaging the root system of the coconut palms. Ploughing and hoeing should not be done within two metres of the palms. Reports indicate some coconut yield depression in the Solomon Islands where soil was cultivated to a depth of 15 cm (Plucknett, 1979). Although a similar temporary depression was noted in Western Samoa, it is suggested that cultivation to a depth of about 15 cm is likely to have no more than a very temporary depressive effect on coconut trees (Reynolds, 1981). There is also evidence that ploughing and sub-soiling can have beneficial effects on coconut yields (see Chapter 7). It is interesting to note that studies at CPCRI in South India showed that the roots of a mature palm, planted in a medium textured sandy loam soil, were concentrated within a radius of 2 m around the base, with about 85 percent of the roots between 30–120 cm depth from the surface and virtually no roots in the top 30 cm layer of soil (Kushwah et al., 1973). According to Anilkumar and Wahid (1988) 80 percent of the active roots were located 2 m from the tree stem at a depth of about 0.60 m, with virtually no roots in the surface horizon.

4.3.3 Time

Although cultivation can be done at any time, all methods of land preparation are best carried out as close as possible to the start of the wet season (optimum time), to take advantage of the long wet spell for the establishment phase. If the period between surface preparation and planting is too long then competing vegetation may regrow (Plucknett, 1979; Reynolds, 1978a, 1978h, 1978j, Teitzel and Middleton, 1978), so once bush is cleared planting should be done immediately (see Figure 109).

4.4 Planting time in relation to age of coconut trees

From the time coconut trees are planted until they are about 8 years old (particularly during the first 5 years), trees may suffer physical damage from grazing cattle (see Figure 43). Therefore, when establishing coconuts it is probably best to raise catch crops of vegetables, root crops, pineapples, etc., during the first 3 or 4 years before establishing pastures in the fourth or fifth year. Cattle can then begin to graze the pastures as the first nuts appear and as the trees grow out of their reach. Judicious fertilizing of young coconuts should ensure rapid growth allowing early grazing of pastures. Payne (1985) suggests that cattle should not be grazed in young palms for the first 4–5 years after coconut planting and Steel at al. (1989) indicate that grass should not be planted or the area grazed until coconuts are 3½–5 years old (or more, depending on the growth of the coconuts). In Tonga, Havea (1980) suggested that a coconut palm must be at least 5 years old, preferably even older, “before it is tall enough to withstand the reach of cattle”. If pastures are established in the first year then rapid grass growth may require expensive weeding to prevent setback of young coconuts. Where grasses such as Napier grass are planted for cut-and-carry systems, planting in the first year is possible. Intercrops other than pastures are discussed by Plucknett (1979) and Denamany et al. (1979) and see Section 1.4. Vandermeer (1989) stresses that the principal agronomic goal of a young plantation is to obtain the maximum growth rate of the trees, a goal that might be hindered by the presence of an annual crop in the system. The problem is thus to determine the balance between production of the annual crop and reduction in growth rate of the tree crop. More specifically, the producer must know how much time will be lost in getting the plantation to its productive size. Often the annual crop may have a negligible effect on the growth of the perennial, however, there are a number of examples where intercrops have been shown to have a significant negative effect on the growth of young plantation species (e.g. Redhead et al., 1983; Agamuthu and Broughton, 1985).

Figure 109. - Pasture establishment should be initiated immediately after bush clearance and seedbed preparation, before weeds regrow (Photo D. MacFarlane).

The importance of good weed control for young trees was demonstrated by Ryan and Lewty (1984) with hoop pine (Araucaria cunninghamii) in Australia. Trees should be kept free from weed competition for at least the first year after planting or until they reach about 1.5 m in height. A weed free circle of 1.25 m radius around each young tree ensured that young trees developed without competition for soil moisture and nitrogen. The effect on tree growth was dramatic (see Figure 110).

Various techniques for growing pines and pasture together have been researched in USA (Lewis, 1955; Lewis et al., 1983) and New Zealand (Percival and Knowles, 1983a; Tustin et al., 1979) and although in the majority of studies pine seedlings were found to be very susceptible to competition from pasture (necessitating herbicide strip or spot spraying or cultivation) Lewis (1985) reported on a successful trial to establish slash pine (Pinus elliottii) directly into a dense pasture sod.

If catch crops or hand/machine harvested forage crops are not grown, then cover crops such as Calopogonium mucunoides or Pueraria phaseoloides can be used to suppress weed growth, enrich the soil serving as pioneer legumes into which grasses and other legumes can be established later.

In Vanuatu the inclusion of the annual (or weakly perennial) Silk Sorghum in improved pasture mixes for areas expected to have major competition from Solanum torvum in the newly establishing pasture is a proven strategy to reduce weed control costs (MacFarlane, 1993). Not only does the Sorghum shade the weeds, but also it provides early feed for cattle prior to the full establishment of the pasture (see Figure 111).

Figure 110. - Growth of hoop pine under different grass control schedules (after Ryan and Lewty, 1984).

Child (1974) stressed that it is essential to keep young coconut seedlings free of grass, with hoe, bush knife or best of all with appropriate herbicides. In Nigeria, Remison and Mgbeze (1987) demonstrated the competitive effects of weeds and the benefits of mulching on coconut seedlings. Kasasian and Seeyave (1969) suggested that the early stages of a crop growth cycle are the most critical in terms of weed competition and demonstrated the effects of early weeding on yields. For coconuts Liyanage (1984) indicated that the most critical stage for weed control is from planting up to the fourth or the fifth year. Fawole and Agboola (1986) identified the problem of weed control in new tree crop plantation establishment as one of the main reasons for yield decline.

Figure 111. - Early growth of Silk Forage Sorghum (Sorghum spp. hybrid cv. Silk) in Signal-legume pastures under coconuts, Malekula, Vanuatu.

In Sri Lanka, Chandrasekera (1984) demonstrated that where well-weeded intercrops were planted between young rubber trees growth of the rubber was not affected, while Fernando (1989) reported on the construction of a low-cost chopping roller to control cover crops and weeds in small coconut holdings. In Tanzania Romney (1988) demonstrated a relationship between the size of the weed-free circle and the growth and early yield of hybrid coconuts on a sandy soil with average rainfall of 1,100 mm. It was suggested that complete weeding should be carried out for trees older than 4 years. The minimum size of circle increased from 1.5 m for trees aged 0–4 years to 4.0 m for trees aged 3–4 years. A measure of the effects on yields and profits of early weed control in cocoa growing was demonstrated by Bonaparte (1979); (see Table 64).

Table 64. - The effects of early weed control in cocoa on net profits (Bonaparte, 1979)

In Sri Lanka (see Table 65) where young rubber was grown with a ground cover of legumes (Pueraria phaseoloides and Desmodium ovalifolium) growth and yield was superior to trees grown with a natural cover of weeds and grasses (Yogaratnam et al., 1984).

Table 65. - Effects of ground covers on growth and yield of immature Hevea at the end of 10 years (Yogaratnam et al., 1984)

* P = 0.05

4.5 Seed and planting material nurseries

When planning to undertake a pasture improvement programme the first step should be to establish a nursery (see Figure 112a and b), i.e. an area of grass or legume from which planting material can be taken for large scale plantings (Chin and Tay, 1993; Guzman and Allo, 1975; Plucknett, 1979; Reynolds, 1977c, 1978g; Steel et al., 1980). The nursery serves as a multiplication area and should be established between six months to one year ahead of the time at which planting material is required. It is important that nurseries are well fertilized and are regularly weeded so that plants are weed free (Liyanage, 1984), vigorous and strong (see Figure 113). The ratio of land area devoted to nursery and land area to be planted varies considerably according to the species and growth rate of the grass or legume itself, as well as the planting system used (Plucknett, 1979). For their low density planting WSTEC in Western Samoa establishes one hectare of B. brizantha for every 200 ha of pasture to be sown. For high density planting a one hectare nursery of B. brizantha could supply sufficient material at each cutting to plant a further 15 to 20 hectares (Reynolds, 1978g; Steel et al., 1980). In Vanuatu MacFarlane et al. (1992) suggest that a nursery area of 25 m × 20 m is suitable for planting an area of one hectare of improved pasture from cuttings. Separate grass and legume nurseries makes collection of planting material easier. Plucknett (1979) suggests that for Dichanthium aristatum a one hectare nursery may provide planting material for only 7 ha, Guzman and Allo (1975) indicate that one hectare of Brachiaria mutica should be sufficient to plant 15 ha of new ground and Williamson and Payne (1978) calculated that one hectare of guinea grass spaced at 50 × 50 cm should provide sufficient planting material for 24 to 30 hectares of land if each parent plant is split into four tufts. Jayawardana (1985) and Liyanage (1986) indicate that in Sri Lanka a 0.4 ha nursery will provide sufficient grass cuttings to plant 6–8 ha. It has been recommended that planting of hetero in grass nurseries is the best way of assisting its spread in pastures (Steel et al., 1980). However, in Vanuatu on at least one large 1,000 ha. property the manager established a separate hetero nursery (as most grasses were seeded) and after slashing (see Figure 114) the hetero was either hand planted at 1 m spacing or was dropped as cuttings behind a tractor which were then pressed into the soil by a roller attachment at the same time as seeding took place. Trellises may be used for producing seed for legumes like C. pubescens (see Figure 115) and M. atropupureum (Castillo and Siota, 1978; Kowithayakorn and Humphreys, 1987); leucaena seedlings can be raised in the nursery for later planting out (Wijewardene and Waidyanatha, 1984) or seed collected from leucaena blocks established for seed production and seeded directly into the pasture area. Production of planting material and seed by the farmer may ensure a ready supply as well as cost reduction. In Vanuatu some villagers on Tanna and Santo are being encouraged to produce legume seed (e.g. Siratro) as a means of increasing their income (Eberhard and Robinson, 1993). In Ethiopia, the ILCA seed unit is evaluating simple equipment capable of achieving a seed sample quality acceptable to farmers (Hanson, 1993) while Peterson et al. (1988) describe a prototype design for a small hand-drawn grass seed harvester constructed from bicycle components. For full details of tropical pasture seed production reference should be made to Humphreys and Riveros (1986). MacFarlane et al. (1992) stress that nursery grasses should be slashed or grazed to a height of 10 cm prior to transplanting, especially if conditions are likely to become dry after planting.

Figure 112a. - Establishing a B. brizantha nursery from cuttings.

Figure 112b. - The nursery after six months.

Figure 113. - A vigorous weed free nursery stand of Signal grass in Zanzibar, Tanzania.

Figure 114. - A hetero nursery with hetero (D. heterophyllum) ready for planting).

4.6 Planting material

4.6.1 Types of planting material

Five main types are used: seeds, cuttings (stems and stolons), rhizomes, divided root stocks or pieces, stakes and bare stem seedlings.

The choice of which type to use will depend upon: species being used, resources available, type of establishment being undertaken and availability of the planting material. Where sufficient funds are available seed may be purchased for both grass and legume establishment. Seeds may also be produced as described in Section 4.5. For some species where there are problems related to poor quality seed, seed dormancy or hard seededness, cuttings or pieces may be preferred. Seed propagation can be unreliable in many tropical areas due to the short life of the seed unless good storage facilities exist. Root stocks or tufts of the large species can be subdivided and hand planted; if spaced at about 2 m by 0.5–1 m some 5,000–10,000 splits will be required ha^-1 (Bogdan, 1977). If these are planted in wet weather then the splits survive reasonably well. For small-holders and low cost establishment, cuttings can be slashed from the nursery by bush knifing an existing stand of grass (plus a few legumes like puero and hetero) a few centimetres above soil surface. The cuttings will root from the nodes when planted or scattered and disced in. The best type of planting material is stems (or stolons) with numerous nodes (Steel et al., 1980). MacFarlane et al. (1992) suggest that rooted cuttings are better than stolons for Koronivia in Vanuatu but in Western Samoa rooted cuttings and stolons of B. brizantha established equally well especially in wet conditions. Leafy material is not suitable as only the stems produce roots and start growing. The grasses vary in how easily they strike. B. brizantha, B. mutica and I. aristatum are usually established from cuttings and all strike easily. With B. humidicola (see Figure 116a) and B. decumbens cuttings may not strike as well, so stems lying close to the ground (stolons - where the rooted stem lies on the soil surface) should be pulled up or cut as these are likely to have existing root systems. Buffalo grass should always be planted using rooted cuttings as it is very intolerant of low soil moisture during establishment (MacFarlane et al. 1992).

Figure 115. - Centro (Centrosema pubescens) for seed production growing on a wire mesh trellise, Zanzibar, Tanzania.

Rhizomatous grasses produce stems or rhizomes below the soil surface (e.g., Guatemala grass - Tripsacum laxum). Rhizomes can be obtained by digging them up and cutting them into 5–10 cm length with at least two nodes on each rhizome (Aminah et al., 1989). Sen and Qin (1987) emphasize that grasses propagated from rhizomes grow rapidly in the first year as they have more vigorous buds, thicker stems and stronger tillering abilities than seed propagated grasses.

When an existing plant is divided to produce a number of pieces or root stocks (with roots attached) the majority of the plant may be dug out of the ground leaving only a part to regrow and replenish the nursery. Pieces are used mainly for bunch grasses like Guinea. Tree legumes like L. leucocephala and G. maculata or sepium can be propagated from stakes or stem/branch cuttings which subsequently root and sprout. Chadhokar (1982) indicated that rooting of stem cuttings takes place about six weeks after planting; nodulation about ten to twelve weeks after planting. The bare stem method of propagating leucaena involves removing seedlings from the nursery when at about 1 m in size, stripping all leaves, branches and trimming roots to leave only 15–25 cm of straight bare tap root (Anon., 1980a). These are then planted either in deep ploughed soil or individually prepared holes and within a month should be firmly established.

Figure 116a. - A stolon of Koronivia grass (Brachiaria humidicola) rooting at the nodes.

Although Growder and Chheda (1982) estimated that four men can collect in one day sufficient vegetative materials for 1 ha (and six men can transplant 1 ha in 1 day) experience in Western Samoa suggests that, using bush knives to slash planting material in the nursery, four men should be able to collect in one day sufficient vegetative materials for several ha.

4.6.2 Planting material size

Often cuttings of B. mutica and I. aristatum are scattered and disced into the soil. Where cuttings are hand planted, as with B. brizantha, or B. decumbens, a small handful of cuttings is placed into a hole, covering part of the cutting with soil (see Figure 116b). An alternative method is to double fold the cuttings, push the bottom of the U into the hole and at the same time cover with soil thus ensuring good soil-plant material contact. Jayawardana (1985) and Liyanage (1986) suggest that cuttings should be 30 cm in length. Jawawardana also mentions that some 4,000 cuttings of pasture grasses and 16,000 cuttings of fodder grasses will be required to plant 1 ha while Liyanage suggests 20,000 cuttings of pasture grasses and 15,000 for fodder grasses! The effects of partial (at least one node exposed) and complete covering of stolons and of compaction after planting of B. decumbens and B. humidicola are shown in Figure 117. It can be seen that complete coverage of the stolons was very detrimental in both species. Compaction after covering has a large positive effect in the case of B. humidicola, but little effect on B. decumbens (Aminah et al., 1989a). In Hawaii, Jiffy (R) pots of peat compound plus nutrients (into which stems or stolon sections with one or two nodes are pushed) are scattered from horseback (or even from the air), a method which broadcasts cuttings without the labour input of digging (Aminah et al., 1989; Partridge, 1977). Trampling of the pots by cattle into wet soil may assist rooting.

Figure 116b. - Establishing stoloniferous grasses by hand under coconuts.

Figure 117. - Effect of partial v. complete coverage and of compaction after planting on stolon growth of B. humidicola and B. decumbens (Anon. 1981c).

Preston (1992) presented data showing the effect of method of establishment (seed or cutting) and plant density on the yield of Gliricidia sepium (see Figure 118). At 1 × 1 m there was little difference between seed and cuttings.

With King grass (see Figure 119), P. purpureum and T. laxum, stem cuttings or canes of 20–30 cm in length (or pieces of rhizome for T. laxum) with at least 2 or 3 nodes taken from mature stems are pushed into the soil at an angle (like sugar cane planting; see Figure 120). In Florida Sollenberger et al. (1990) found that successful establishment of dwarf elephant grass was most likely if undefoliated stems are planted.

Mtengeti and Wilman (1993) recently reported from Tanzania that horizontal planting of stem sections (buried below 60 mm of soil) produced higher leaf yields 12 weeks after planting (35–39% higher) than the traditional method in which stem sections were planted with two nodes covered with soil and a third note exposed. It was noted that horizontal planting is easier to mechanize and saves time.

Figure 118. - Effect of plant density and method of establishment on yield of Gliricidia sepium (after Preston, 1992).

Figure 119. - Stem cuttings of King grass (P. purpureum × P. americanum) in a nursery in Vietnam ready for distribution to farmers.

Figure 120. - An area of Napier grass (P. purpureum) established from stem cuttings.

In Cuba the best establishment of King grass was achieved with 3 buds or more on seed pieces and a 5 cm planting depth (Ayala et al., 1983). Root stocks or pieces can vary in size but must contain sufficient material to regenerate. The effect of planting piece size of tall guinea on early regrowth is shown in Table 66. Although bigger pieces were vigorous and initially produced more regrowth, differences were less by second harvest. For easy machine handling pieces of 2.5 cm diameter and 30 cm length were chosen. The removal of excess leafy material improves handling and also reduces evaporation losses. For L. leucocephala and G. sepium sticks or stakes of 30–60 cm in length are commonly used. Studies have shown that the establishment of both L. leucocephala and G. sepium is affected by stem size (Chadhokar, 1982; Guevarra et al., 1978; see Table 67). Recently Adejumo (1991) demonstrated that the dry matter yield of gliricidia increased with increasing length and girth of sticks or cuttings (see Table 68) and recommended that gliricidia be propagated from cuttings 100 cm long and not more than 10 cm in girth to ease establishment, reduce labour and produce optimum digestible crude protein. Wiersum and Dirdjosoemarto (1987) suggested that if cuttings are taken from stems at least 6 months old and have a minimal length of around 50 cm, good results are normally obtained provided that the cuttings are planted under favourable humid conditions in holes at least 10–15 cm deep. Stylosanthes guianensis can be established vegetatively from stem cuttings (Miles, 1982).

4.6.3 Planting material storage

Grass and legume seeds require careful storage for good preservation (Harty, 1980; Steel et al., 1980); otherwise the seed deteriorates and loses viability (Hacker and Williams, 1993). In a review of seed longevity Humphreys (1979) and Humphreys and Riveros (1986) indicate that inattention to proper seed drying, packaging and storage requirements has been a primary cause of the failure of sowings of high priced tropical grass seed. Chin (1980) has reviewed the basic principles for storage and testing of forage seeds in the tropics.

Table 66. - Effect on early regrowth caused by size of tall guinea (P. maximum) planting pieces (Reynolds, 1978c)

¹ - Planting date 10/1/75, first harvest 24/2/75, second harvest 8/4/75.
² - Standard length of 30 cm, diameter measured when lightly compressed between thumb and forefinger.
³ - Standard diameter of 2.5 cm.
⁴ - Values with one or more common letter(s) are not significantly different at P = 0.05 (Duncan's New Multiple Range Test).

Table 67. - Effect of size and stem part on establishment of G. sepium (Chadhokar, 1982)

Table 68. - Total dry matter yield (t ha¹) of gliricidia planted vegetatively from cuttings of different lengths and girths - 4 Jan. to 5 Dec., 1988 (Adejumo,1991)

Length (cm)	Girth (cm)
	5	10	15
25	1.52	3.64	3.53	2.90d
50	2.21	4.86	5.04	4.04c
75	3.13	6.54	6.60	5.42b
100	5.04	6.01	7.08	6.31a
	2.98b	5.46a	5.56a	4.67

Means in the last column or bottom row having the same subscript letters are not significantly different (P<0.01).

Research by Delouche et al., (1973), Harrington (1963, 1971), Harrison (1971), Peel and Prodonoff (1970), Roberts (1973), Roe and Williams (1969) reported by Humphreys (1979) and Humphreys and Riveros (1986) clearly indicate the need to dry seed to 8–10 percent moisture or less and store it in sealed containers at temperatures of about 15–18°C and 40 percent relative humidity. More recently it has been reported that the highest percentage of normal seedlings of Centrosema pubescens, Pueraria phaseoloides and Stylosanthes guianensis were obtained from seeds stored at -12°C/70% RH, while the optimal conditions for Calopogonium mucunoides were -1°C/60% RH and for Desmodium ovalifolium 5°C/40% RH (Bass, 1983).

The main investment in the tropics should be for seed drying and sealed storage, with air conditioning or cooling facilities as a secondary consideration. An idea of the rapidity of seed viability decline is seen in the data of Win et al., (1975) who noted that the viability of C. pubescens stored at 33°C and 90 percent relative humidity decreased from 66 percent to about 2 percent after only 4 months. In Vanuatu, Evans et al. (1992) noted that recorded signal grass germinations declined from 60 percent on arrival in January to 30 percent in April in cool, non-airconditioned storage. Even under air-conditioning in Vanuatu grass seed should not be stored for more than three months, otherwise the quality of pasture establishment will decline.

Advice on the harvesting, cleaning and storage of home grown tropical pasture seeds has been given by Javier and Mendoza (1976). An example of the minimum standards of seed germination and purity expected in Queensland, Australia for a range of tropical pasture species is given in Table 69. In Thailand the Department of Livestock Development has established minimum quality standards for grass and legume seeds produced in the Northeast (Boonpakdi and Leeratanachai, 1989) and with the recent expansion of Thailand's beef and dairy industries there has been a significant increase in the seed production of ruzi grass (B. ruziziensis) Verano stylo (Stylosanthes hamata cv. Verano) and a range of other legumes mostly on small farms (Anon., 1994a).

Chin (1989) stresses that high quality seeds are a pre-requisite in crop production. It pays to invest in high quality seeds which represent only a small percentage (10– 20 percent) of the total outlay and failure to do so could be very expensive in terms of poor establishment. Anyone wishing to test seed viability with a germination test should follow the procedures outlined in the international rules for seed testing (International Seed Testing Association, 1985) or refer to Chin (1989). When importing seed it is important to ensure that the seed has been treated and contains no live insects or weed contamination.

Cuttings should be used on the same day or within 24 hours of cutting. Rootstocks, pieces and stakes can be kept for 2 or 3 days depending upon the species. Napier grass canes, for example, can be kept for some weeks if stored carefully in the shade, but it is also best to plant rootstocks, pieces and stakes within 24 hours of cutting. All planting material should be kept in the shade before use.

Table 69. - Standards regulations for seed quality

From Humphreys and Riveros (1986) after Queensland DPI (1984)

4.6.4 Seeding rate and plant spacing

The amount of seed required for each species is governed by a number of factors:

1. Seed quality (purity and germination). With poor seed larger quantities will be required. Many failures to establish tropical pastures are directly attributable to poor quality seed (Jones and Jones, 1971). Seed rates are therefore based on high purity and germination percentages. Details of seeding rates for different grass and legume species are given by Humphreys (1974) (see Tables 70 and 71). Although rates vary from about 1 to 8 kg, a rate of about 3–4 kg of good quality seed (PGS, pure germinating seed) for both grasses and legumes is the general recommendation for good seed beds. As seed is expensive the main aim is to achieve good germination and a satisfactory pasture stand with the minimum quantity. However, considering the high cost of establishing pastures it is unwise to skimp on sowing rates in order to ‘reduce’ costs or to plant a larger area (Teitzel and Middleton, 1978).

2. Climate. A higher seeding rate can be used in higher rainfall areas to provide a quicker cover thus suppressing weed growth (Skerman, 1977).

3. Seed-bed type. Heavier seeding rates should be used on roughly prepared seed-beds (Teitzel and Middleton, 1978). One example is where 9 kg ha^-1 of Macroptilium atropurpureum seed was recommended on roughly prepared seed-beds, 4.5 kg ha^-1 on partially prepared seed-beds and only 1 kg ha^-1 on well-prepared seed-beds (Skerman, 1977).

4. Potential weed problem. Where heavy weed growth is expected because of previous cropping history or type of soil preparation the seeding rate should be increased accordingly. However, to ensure a completely successful establishment the resident vegetation must be destroyed before sowing. It is important to remember that tropical pastures are sown at about 100–700 seeds per m² (Silcock, 1980; Humphreys, 1979), whereas natural seed loads already in pasture soils may be 5,000–50,000 per m² (Johnston et al., 1979; Jones and Evans, 1977).

5. Method of sowing. Where seed is broadcast the seed rate should be increased up to 50 percent (Skerman, 1977).

Many different spacings are used for cuttings, pieces and stakes. Because bunch grasses do not send out stolons, pieces should be planted close together at about 1 m² and later be allowed to set seed. For rapid establishment of planted cuttings (either by hand or machine) 1 m² is also recommended (see Figure 121), although with species like B. mutica or I. aristatum broadcasting followed by discing may be used. In general, the wider the spacing of cuttings or pieces (and the lower the amount of seed) the slower the establishment of a complete pasture cover. At CIAT (Anon. 1981c) one ploughing and one rotovation to a depth of 15 cm followed by planting 30 cm long stolons of B. humidicola and B. decumbens at 1 m × 1 m spacing gave the best establishment in terms of establishment percentage, fresh weight yield and percentage cover (Aminah et al., 1989).

Figure 121. - Stolons of Signal grass (planted at 1 × 1 m) spreading rapidly to cover the ground between planting sites.

For bush legumes like leucaena, spacing will depend on use. The following is recommended: for grazing, double rows (1–2 m apart) between each two rows of coconuts; for cut forage, closer spacing is recommended for forage napier grass, e.g., in Colombia stem pieces of Pennisetum purpureum and Axonopus scoparius spaced at 0.5 × 0.5 m gave higher forage yields than at 2.0 × 2.0 m (Lotero et al., 1967; 1969).

Table 70. - Seed rates for the main tropical grass species

¹ - Vegetative propagation
Source: O'Reilly (1975), Steel et al., (1980).

Table 71. - Seed rates and inoculum requirements for the main tropical legume species

*1 - v.p. = vegetative propagation
Source: O'Reilly (1975), Stobbs (1976)

In Nigeria, Okeagu and Agishi (1990) showed that dry matter yields of rooted sprigs of signal grass (B. decumbens) at 20 × 20 cm spacing were significantly higher than spacings of 40 × 20 cm and 60 × 20 cm in the first two years of an experiment but in the third year forage production from the three row spacings was similar (see Table 72). In Sri Lanka, Jayawardana (1985) recommended that cuttings should be spaced at 30 × 30 cm, with fodder type cuttings at 60 × 60 cm.

Table 72. - The effect of row spacing on DM yield (t ha^-1) of signal grass (Okeagu and Agishi, 1990)

Research suggests that although sowing rate (see Table 73) and seed size have an early effect on seedling size and pasture yields, there is no significant long term effect because of the importance of the management factor (Cook and Stillman, 1981; Jones 1975; Silcock, 1980; Ludlow and Wilson, 1972; Tow, 1967). However, in order to ensure good legume establishment it is best to increase the legume rather than the grass sowing rate, or even to sow the legume before grass cuttings are established in order to reduce competition (Middleton, 1973a; Roberts, 1974).

Table 73. - Effect of sowing rate on Siratro dry-matter yield (kg ha^-1) (Middleton, 1973)

4.6.5 Legume seed preparation

Because most legume and some grass species have a high percentage of ‘hard’ seeds with impermeable seed coats, preplanting treatment (seed scarification) is recommended to increase germination percentages (see Table 74). (For some grass seeds dormancy is more of a problem - Hopkinson, 1993). McIvor and Gardener (1987) examined the level of hardseededness in some commercial seed samples and while there was a wide range for all species, samples varying from 0 to 98 percent hardseed, median values were: S. guianensis 40 percent; S. hamata 55 percent; S. scabra 63 percent and S. humilis 72 percent. Methods of overcoming hard seededness include mechanical abrasion, acid treatment, dry heat and hot water treatment (Butler, 1983; Gray, 1968; Jones, 1969; McIvor and Gardener, 1987; Mott and McKeon, 1982; Mott et al., 1982; Olvera and West, 1985; Whiteman, 1980). Among the major grasses with impermeable seed coats is B. decumbens (Grof, 1968). When very small quantities of seed are involved they can be rubbed with moderate pressure between a folded sheet of glass or sand paper or by treatment in a small hand or electric scarifier or by rubbing the seed with a flat piece of timber on a rough cement floor (MacFarlane, 1992). When large quantities are involved, a revolving drum scarifier, commercial rice huller or even a concrete mixer can be used. In the process seed coats are scratched on the rough sides as the barrel turns, or on gravel which is added with the seed. Also used are fine mesh seed filled bags soaked in hot water (80°C) for 2 to 5 minutes, followed by washing in cold water, or for 2–5 seconds in water at 100°C. The latter treatments are perhaps more feasible as they can be made without the use of a thermometer (Oakes, 1984), however, as demonstrated by McIvor and Gardener (1987) the optimal period varies with species. For S. guianensis and S. scabra this was between 2 and <20 seconds respectively (with both being very sensitive to longer exposures, resulting in a high percentage of dead seeds) while seeds of S. hamata survived immersion in boiling water for 20 minutes. There also appears to be variation between seed lots which may be caused by differences in seed moisture content, so testing of a small sample of seed is essential prior to treatment. In Tanzania a sulphuric acid treatment method has been used (Grundy, 1959; Skerman, 1977), and in Colombia the effect of chemical scarification with sulphuric acid and storage (22°C and 80 percent humidity, and cold-room storage at 18°C and 50 percent humidity) on the germination of Centrosema spp. was determined (Burbano, 1990). Mott and McKeon (1982) described the use of a rotating drum machine in which Stylosanthes hamata seed was heated for 15–20 seconds at 155°C. Butler (1983) subjected S. guianensis (cv. Schofield, Cook, Endeavour and Oxley) seed to dry heat in small drying ovens with the six-hour treatment at 80°C giving the best results. Hopkinson and Paton (1993) demonstrated that scarification of Stylosanthes scabra by hammer milling was preferable to heat treatment. In Brazil, Cameiro et al. (1990) studied the effect of mechanical scarification on the germination of glycine seeds (using a scarifier at different rpm for different time periods with two types of sand paper and different cylinder/seed volume ratios. Scarified seed should be sown fairly soon after treatment, as the seed tends to lose viability in storage more rapidly than unscarified seed.

For some grasses, particularly Signal and Sabi grass, a period of 9–12 months is required after seed harvest before there is a high percentage of viable seed.

Inoculation of legumes prior to sowing is strongly recommended, particularly when introducing new species into new areas, to ensure that species are nodulated by the most effective Rhizobium strains (Norris, 1967). Peat based cultures are commercially available for seed treatment. Legumes can be broadly classified into 3 groups on the basis of their requirement for a specific rhizobium to obtain effective nodulation (Humphreys, 1974):

• specific (a very specific strain required), e.g. L. leucocephala
• moderately specific e.g. C. pubescens;
• promiscuous (this group will form an effective symbiosis with a range of rhizobium in the cowpea group) e.g. M. atropurpureum.

The cowpea type of Rhizobium produces an alkaline reaction in the growth medium whereas specific types produce a strong acid reaction. If a test shows that bacteria from a newly introduced legume are acid producers, it is a strong indication that the legume will probably respond to lime additions on acid soils (Skerman, 1977).

Details of inoculum requirements are given in Table 71. Inoculants must be stored in a refrigerator and expiry dates observed closely. Storage life can be up to one year. Skim milk or a 10 percent sugar solution can be used to stick the bacteria to the seed. After applying the peat cultures in slurry form to the seed and thoroughly mixing seed and slurry, seeds should be dried away from direct sunlight in a cool, dry place and the seed should be planted within one week of scarification/inoculation. MacFarlane et al. (1992) describe the steps in inoculating 2 kg of seed:

1. mix two teaspoons of sugar in a cup of water and then mix with the seed to make it sticky, so that the inoculum will adhere to the seed;

2. put the correct amount of inoculum into a container (2 heaped teaspoons for 2 kg of seed, in this case);

3. mix water or coconut juice with the inoculum until it is like mud;

4. tip the mixture onto the legume seed and mix thoroughly until the seed is coated in the mixture;

5. store the seed (mixing and storage should be done in the shade as direct sunlight will destroy the inoculum).

When no inoculants are available, the soil (already infected with the appropriate bacteria, where the particular legume is already growing and nodulating) can be mixed with the soil in the new area.

4.6.6 Seed treatment against insects

Preplanting treatment of seeds with insecticides may be necessary to prevent damage by ants, bean fly etc. (Campbell, 1966). Pelleting some legume seed with lime or rock phosphate may also reduce ant damage as well as protecting the inoculum on the seed from acid fertilizers and dry conditions at planting (Jones, 1965; Russell et al., 1967; Skerman, 1977; Whiteman, 1980).

Table 74. - Effect of scarification methods on Leucaena leucocephala seed germination (Gray, 1962)

¹ - Water temperature and immersion time
² - Immersed in commercial sulphuric acid for 20 minutes.

4.7 Depth of sowing/planting

With most small seeded grasses and legumes, surface broadcast followed by harrowing or planting depth of about 5 mm is suitable (Silcock, 1980), while larger seeds can be sown at about 1.5–2.0 cm. Sowing at 3 cm or deeper often leads to failure (Humphreys, 1974; Teitzel and Middleton, 1978: Whiteman et al., 1974), although Cooksley (1982) found that 5–6 cm was the optimum depth when sowing leucaena into moist, well cultivated soil, while with rainfall after sowing the optimum depth was reduced to 3 cm. Newman and Moser (1988) found that seedling emergence percentage generally decreased with increasing planting depth (from 1.5 to 6.0 cm). Cuttings and pieces are best planted at least 10–15 cm deep. Failure to cover the seed with soil or to plant the grass at a good depth may cause moisture stress as the soil dries out from the surface with resultant non-germination of seed and withering of plant material. Stakes should be pushed well into the soil to a depth of at least 15–20 cm and leucaena bare stem seedlings may be planted even deeper depending on the length of the tap root.

4.8 Sowing/planting time

The optimal period is at the start of the wet season taking advantage of a long wet spell for establishment. In areas where rainfall distribution is bimodal, two major planting seasons may be possible depending on length and reliability of rainfall (Reynolds, 1983). Planting can be done in the wet season whenever the soil is sufficiently moist and when a wet spell is likely to persist for some time. MacFarlane et al. (1992) suggest that “the best time to plant cuttings is when soils are saturated, even when it is still raining”.

4.9 Method of establishment

Most legumes are usually established from seed but where labour costs are low and planting material is available, then most grasses and some of the legumes can be established under coconuts from cuttings, root stocks, pieces and stakes. One of the advantages of seed is that large areas can be planted very quickly however, because of the cost of seed most smallholders are more likely to use vegetative establishment techniques and slowly expand their areas of improved pasture by planting additional areas (relative to the amount of family labour available each wet season) each year.

Typical seed mixtures for commercial plantations under old coconuts in Vanuatu on good soils (MacFarlane et al., 1992) would be:

The advantage of using a number of legumes is that each may dominate the pasture at different times throughout the year when others may temporarily die back.

For smallholders in heavily shaded areas under coconuts on good soils the emphasis would be on cuttings of:

Buffalo grass
Arachis repens
Desmodium ovalifolium
Arachis pintoi

Seeds - Various types of seed drilling machinery are available (Skerman, 1977; Teitzel and Middleton, 1978), but in many countries the resources available will limit the choice to either broadcasting by hand (from a seed sack hung in front of the sower), or by using a hand seed spinner. Small quantities of seed can be more evenly distributed if a ‘spreader’ such as dry sand, soil, sawdust or rice hulls is mixed with the seed. When the sower has little experience it is important that the seed is divided into a number of portions and/or that the area is sown several times avoiding seed concentration on only one part of the area (Steel et al., 1980).

For direct drilling of seeds into an existing sward (and application of herbicide) special (zero-till) machines have been developed (see Figure 122). In Australia, Davidson (1985) found that band establishment of Siratro and green panic into spear grass pastures with direct drilling and band herbicide application along the row gave much better results than broadcasting. In Malaysia, Ahmad et al. (1981) found that spot placement of legume seeds was much better than broadcasting. Cameron (1989) reported on the successful establishment of Aeschynomene americana, Brachiaria decumbens, Calopogonium mucunoides, Centrosema pubescens and Macroptilium lathyroides into existing Digitaria decumbens/Paspalum plicatulum swards in Australia by spraying the grass swards with Roundup (360 g glyphosate l^-1) at 10 ml litre^-1 and sowing seed into the sprayed sward on the same day.

Figure 122. - A zero-tiller at work on Elbee Ranch, Efate, Vanuatu (Photo D. MacFarlane).

Cutting and pieces -

1. Planting machines can be used, ranging from modified sugarcane, tobacco or cabbage planters to simple machines consisting of a tool bar with two or three tines and press wheels surmounted by a partly enclosed platform for carrying planting material (see Figure 123a). With the machine illustrated in Figure 123b, the tractor proceeds at a walking pace, the planting material is put by hand into furrows formed by the tines and covered by soil pushed over the furrow by the feet of the planters walking behind the machine.

2. Cuttings of some species like B. mutica (only recommended for less shaded sites) and I. aristatum can be scattered on bare soil and disced in by tractor and disc harrow. Steel et al. (1980) suggested that if possible the area should be rolled after discing.

3. A team of planters (planting by hand - see Figure 116b) are paid either by contract or at piece rates. A practical method is for planters to work in lines where each person is 1 or 2 m from the next, using a hand how, mattock or tapered spade with cuttings carried in a frond basket. A variant would be for two people to work together on the strip between each two rows of coconuts: one would make the holes with a hand digging tool while the other would carry and plant the cuttings. Legume seed could be sown later with a hand drill by one person. When legume pieces (e.g. hetero) are used, these could be interplanted with the grass in alternate rows or, if grown together in a nursery, planted simultaneously. In Vanuatu, a planting iron is used which performs well under very wet conditions provided care is taken not to pull the cuttings out of the ground when removing the iron from the ground (MacFarlane et al., 1992). Another successful technique used is to plant legumes such as hetero, Arachis pintoi or Arachis repens into food gardens using available labour and then later to establish grass species from cuttings so that pasture follows food gardens in the rotation and gradually larger grazing paddocks are established (MacFarlane, 1993).

   Crowder and Chedda (1982) suggested that six men can transplant 1 ha (with vegetative planting material) in 1 day. In Vanuatu, Evans et al. (1992) indicate that in an efficient operation 15 person days are required to plant 1 ha of Koronivia grass; and Mullen (1993) suggests that hetero can be planted at 2 × 2 m spacing at the rate of 5 man days ha^-1. Planting of bare rooted leucaena stems (including preparation) in rows under coconuts at populations of 4,500 plants ha^-1 takes 10–12 man days ha^-1.

4. Where a chisel plough has been used, cuttings can be hand planted along the furrows and legume seed sown as in paragraph iii) above.

5. Where heavy grazing has churned the soil, plant as recommended above (from i) to iv)); where strips have been sprayed planting can be done by hand as described in paragraph iii) or in Section 4.10.

Stakes - Hand plant in rows or blocks according to whether for grazing or cut forage.

In general, level of control of the existing vegetation and proximity of the cuttings, when planted in moist soil, will favour a quicker establishment of the new sward. Where soil is moist and free of weeds, germination of seed and seedling emergence is likely to be rapid. With good soil fertility the seedlings can establish and grow and sward consolidation will begin. In Vanuatu (Mullen, 1993) where buffalo grass or hetero is established under coconuts and where the paddock cannot be destocked, newly planted cuttings are protected by partially covering them with cow dung (or by planting into cow dung). Cattle will not eat around dung patches for at least 3 months (MacFarlane et al., 1992).

Figure 123a. - Divided rootstocks of guinea grass being planted using a ‘planting’ machine with press wheels.

Figure 123b. - Rapid hand establishment of B. brizantha using a simple ‘planting’ machine.

The effect of seed-bed type on the establishment of legumes and grasses in N. Queensland was examined by Thompson et al., (1983), while the effect of seed-bed preparation and sowing time on the establishment of perennial Stylosanthes species has been reported by McIvor, (1983). In Brazil, germination and establishment of Galactia striata and C. pubescens were highest in those treatments where soil covered the seed and where the soil was compacted (Novelly et al., 1985). In Malaysia, the method which had no tillage gave the poorest results in establishing Brachiaria sp. (Wong and Sharuddin, 1981).

4.10 Oversowing and sod seeding

Low cost pasture improvement has been achieved in many areas by sowing or planting a legume into existing native tropical grasslands. Usually existing grasses are temporarily set back by heavy grazing, burning, chemical spraying or discing so that a legume can be oversown/planted or sod seeded with a special heavy-tine seeding implement (Breakwell and Jenkins, 1953; Cook, 1980; Gutteridge and Whiteman, 1977a; Luck and Douglas, 1966; Middleton, 1973b; Murtagh, 1963; Norman, 1961; Olsen et al., 1981; Sheafer and Swanson, 1982; Skerman, 1977; Welty et al., 1981). The feeding value of the pasture is thus improved. This may be used where, for example, Imperata cylindrica or Axonopus compressus occur in coconut areas.

Evans et al. (1992) and MacFarlane (1993) describe in detail the methods used to establish legumes in grass dominant pastures in Vanuatu. They stress the need to create conditions that will favour the germination and establishment of the legume seedlings which means that competition from the established root system of the existing grass pasture must be reduced to ensure good soil/seed (or vegetative material) contact for the introduced species. This can be done as follows:

1. Heavily graze the pasture to a height of 5–10 cm to reduce the leaf canopy and the accumulated layer of litter. This is essential so that subsequent seeding operations achieve adequate soil contact. Vegetative material hand planted into a heavily grazed existing pasture (e.g. Koronivia and hetero into carpet grass, or Signal and hetero into T-grass) is then faced with less competition for light in the early stages of establishment.

2. Legumes can be successfully established in such pastures either by sowing into disced strips or by sod-seeding into rows using a disc seeder, preferably a triple disc seeder and spraying along the rows when planting the seed with a non-residual, non-selective herbicide such as glyphosate. Where 2 m wide strips are disc harrowed every 5 m into heavily grazed pasture (see Figure 124) the strips are planted to a suite of legumes which later spread out into the unploughed areas (Mullen, 1993).

They conclude that while both methods have led to successful pasture establishment legume seedling establishment from disc strips appears to be superior to zero-tilled treatments, for the same seed sowing rate ha^-1 and is potentially more applicable than zero-tilling because disc harrows are readily available. Zero-tilling with the Mason Deere maxi-strike planter or disc stripping legumes into 15 year old Signal increased legume content in Signal pastures from <5 percent to at least 30 percent in one year with light grazing resuming six weeks after oversowing and full grazing resuming after five months. (Recently in New Zealand, Lother et al. (1993) and Woodman (1993) demonstrated that an experimental strip-seeder drill had considerable potential compared with the conventional triple disc drill in terms of cost-effective pasture establishment in difficult environments). In Queensland, Australia Partridge et al. (1993) described trials with a new Connor Shea Napier band-seeder. Similar increases in legume content have been recorded in IRHO para grass pastures where animal production has increased by at least 25 percent ha^-1 as a result. The usual cost of legume oversowing in strips 1.5–3.0 m wide every 5.0 m ranged from 8,000 to 10,000 VT ha^-1 (US$ 67–83) with 3 litres of glyphosate per sprayed ha necessary for adequate control of vigorous grasses such as Signal or para. Zero-tilling into Koronivia requires 4–5 l of glyphosate per sprayed ha for good control. Figure 125 demonstrates the effect of disc stripping a legume mixture into Signal grass at Elbee Ranch, Rentabao on Efate, Vanuatu.

Figure 124. - Strips disc harrowed in a Signal grass pasture to enable legumes to be established (Photo D. MacFarlane).

Ahmad and Wahab (1982) described several experiments carried out at the MARDI Research Station in Serdang, Malaysia to evaluate three low cost techniques of establishing Stylosanthes guianensis cv. Schofield into existing native pastures of Paspalum conjugatum and Iallang (Imperata cylindrica). The mowing treatments and sowing seed into soil disturbed with the use of hoe or tines were superior to the non mowing and broadcast treatments and mowing plus burning was superior on Iallang plots. After one year of establishment the percentage of stylo in the hoe and tined (mowed and non-mowed) treatments was more than 60 percent whereas in the broadcast mowed and non-mowed treatments stylo percentage was 41 and 23 percent respectively. Bellotti and Blair (1989a, 1989b and 1989c) demonstrated that grasses can be successfully established by using direct drill sowing methods provided existing vegetation is adequately suppressed by a pre-sowing application of herbicide. The promotion of rapid early seedling growth through cultivation or herbicides considerably enhanced survival of the sown perennial grasses. Figure 122 shows a machine used for sod seeding (in Vanuatu) and band-spraying of herbicide.

4.11 Fertilizer needs and application

Due to the different soil, climatic and management conditions found in the various coconut growing areas, precise recommendations for fertilizer use at establishment are not possible. However, it is important that the fertilizer requirements for both coconuts and intercrops (pastures) are considered. Competition between the pasture species and the palms for plant nutrients should be eliminated otherwise coconut yields will suffer. Although each coconut growing area will have its own fertilizer recommendations for coconuts, a useful general recommendation is to apply 10-5-20-12 or 12-9-22 fertilizer at a rate of 2.5 to 3 kg tree^-1 year^-1 around the base of each palm within a 2 m radius of the trunk (Guzman and Allo, 1975).

(a) Signal grass	(b) Legumes spreading from the disc strips in the Signal grass

Figure 125. - A comparison of Signal grass between the disc strips and where legumes were sown in the disc stripped area.

Methods for the assessment and correction of soil fertility in relation to tropical pastures have been described by Whiteman (1980). As pasture dry matter yields and protein content are often closely related to nitrogen availability, and since it is deficient in many tropical areas, it is very important that pastures established under coconuts have sufficient nitrogen. The high cost of N fertilizer (Whiteman, 1980) and the relative advantage of legume nitrogen over N fertilizer in terms of energy efficiency (see Table 59), has prompted considerable research, especially in Australia, in the use of legumes in tropical pastures. The effects of using relatively low cost phosphorus fertilizers on the establishment of grass-legume pastures with a high legume content have been widely reported (Whiteman, 1980), while the importance of phosphorus for N fixation is shown in Table 75.

Table 75. - Phosphate effects on nitrogen content and yield of S. humilis (Whiteman et al., 1974 after Shaw et al., 1966).

As potash is the major requirement for coconut palms (see Table 76), K fertilizer is likely to be required by pastures, particularly in old coconut plantations.

Table 76. - Nutrients removed from 1 ha of coconuts (Guzman and Allo, 1975 after Georgi and Teik, 1932)

For hybrid coconuts (copra, husk, shell and water) with a yield of 4 t ha^-1 Teoh et al. (1986) calculated total nutrient removal from the soil as N 47 P 8 K 106 Ca 4 Mg 9 kg ha^-1.

The amount of nutrients returned to the soil by the droppings of grazing animals cannot maintain soil fertility. It has been found that 60 percent utilization of a fresh herbage yield of 10 tons ha^-1 year^-1 could deplete the soil of 72 kg N, 18 kg P₂O₅ and 78 kg K₂O. In spite of the amount supplied by an animal there would be a yearly deficit of 51 kg N, 1 kg P₂O₅ and 55 kg K₂O. The situation may be remedied by using inorganic fertilizers (Anon., 1982d).

However, Liyanage et al. (1989) found that nutrients returned in the form of dung and urine reduced the cost of inorganic fertilizers by 69 percent. Some 73 kg palm^-1 of dung and 30.4 l of urine provided 0.812 kg N, 0.219 kg P₂O₅ and 0.802 kg K₂O palm^-1 year^-1.

Relationships between level of fertilizer application, intensity of grazing and yield of forage and coconuts were investigated in experiments carried out in Sri Lanka (Santhirasegaram, 1966a, 1967a; Ferdinandez, 1968, 1969, 1970a). Fertilizer levels used have been summarized by Plucknett (1979). Good forage and nut yields were achieved on B. brizantha pastures with 56 kg N, 18 kg P and 125 kg K ha^-1 year^-1. Appadurai (1968) recommended for Cori grass in Sri Lanka the following amounts: 50 kg N, 25–30 kg P and 60–65 kg K ha^-1 split into two applications at the start of each monsoon season. Liyanage (1986) has recommended Urea 25 kg ha^-1 Saphos phosphate 25 kg ha^-1 and Muriate of Potash 25 kg ha^-1 every three months for pasture grasses, with double the amount of Urea for fodder grasses. In Western Samoa, during the establishing of grass-legume pastures, Reynolds (1981) used a fertilizer application of 250 kg ha^-1 of 30 percent potassic superphosphate (7% P, 14% K) split into two equal applications, while Rika et al., (1981) used triple superphosphate at a rate of 108 kg ha^-1 in establishing grass-legume pastures under coconuts in Bali, Indonesia. In some areas sulphur and trace elements (Mo) have also been used (Watson and Whiteman, 1981a). In the Philippines the normal fertilizer recommended for pastures grown under coconuts is 60–100 kg N and 30–60 kg P₂O₅ ha^-1 year^-1 split into three or four broadcast applications (Anon., 1982d; Felizardo, 1979); fertilizer is applied separately to the palms themselves. Manidool (1983) particularly stressed the importance of magnesium and potassium where pastures are grown under coconuts.

In Seychelles, where many of the soils are coralline, Walker (1992) recommended the following fertilizer mixture when establishing grass-legume pastures: 200 kg ha^-1 of Nitrophoska, 200 kg ha^-1 of triple superphosphate and 100 kg ha^-1 of sulphate of potash. In the absence of Nitrophoska 50 kg ha^-1 of urea, 250 kg ha^-1 of triple superphosphate and 150 kg ha^-1 of sulphate of potash. Maintenance fertilizers should be applied three times per year at one sixth the above rates at each application. Iron deficiency (and also Cu, Zn, Co and Mn deficiencies) could also be a problem. In zero grazing systems (i.e. cut-and-carry) large amounts of nutrients are taken from the area so soil nutrients are quickly depleted. In addition to FYM being applied in the rows at 20 tonnes ha^-1 the following fertilizer rates are suggested by Walker (1992) for establishment and optimum production: 200 kg ha^-1 of Nitrophoska (i.e. 100g/10 metres of row) 200 kg ha^-1 of triple superphosphate and 100 kg ha^-1 of sulphate of potash. If Nitrophoska is not available use urea and amounts as for grass-legume pastures.

In Vanuatu, Evans et al. (1992), while acknowledging that there are many areas of relatively fertile soils, also identify areas where deficiencies of potassium occur (especially on coralline soils) and deficiencies of copper and possibly other trace elements on Efate and Santo. Soil phosphorus levels also vary between different soils and are too low for good legume growth on Erromango, North Tanna, Montmartre and parts of Efate, and interior areas of the Santo plateau. They stress that in the Pacific low soil P is usually the main soil fertility factor contributing to legume loss in a pasture. Although they recommend the band application of fertilizer at planting on soils of marginal P or K status, especially in the case of legume oversowing into existing pastures where 65 kg K ha^-1 and 30 kg P ha^-1 considerably improves the persistence and productivity of some oversown legumes, the overall approach has been to research and recommend groups of legumes (and grasses) adapted to a range of soil pH and available phosphorus conditions rather than to seek to ameliorate soil fertility as a general practice for beef production. For example, green panic, guinea grass, glycine and Siratro provide maximum production on the most fertile soils, whereas Signal grass, Koronovia and the legumes mimosa, hetero, Seca and Cook stylo grow well on lower fertility soils. Koronovia grass in fact tolerates low fertility best of all the native and introduced grasses.

While actual fertilizer rates should be based on local assessments, early legume establishment will benefit from an application (at planting) of 100 kg ha^-1 of superphosphate, or potassic superphosphate on low potash areas, either broadcast, or for oversown legumes, banded as recommended by Evans et al. (1992).

4.12 Cost of establishment

Because costs will vary considerably depending on the country, the region, the nature of the vegetation to be cleared, whether family/community or hired labour is to be used, whether machinery is to be used and chemicals, whether livestock are available to assist with vegetation clearance, and the cost of seed etc., detailed costings are not provided here. However, some information for Vanuatu is given below (and in Section 4.13) and for information on pasture establishment in Western Samoa refer to Section 8.3.11.

For Vanuatu costs for the first year for a typical smallholder pasture improvement project are given in Table 77. MacFarlane et al. (1992) emphasize that if the costs of pasture improvement are greater than 50,000 VT ha^-1 (US$ 416.7 ha^-1) it will be difficult to make a reasonable profit from the farm.

Table 77. - Inputs and average costs for clearing regrowth bush and improved pasture establishment over the first year on an average smallholding in Vanuatu (MacFarlane et al., 1992)

The cost of rehabilitation of weed infested native pastures will depend on the methods used and whether family labour is available. In most cases costs were less than 18,00–20,000 VT ha^-1 (US$ 150–166.7 ha^-1).

4.13 Specific establishment techniques used for pastures under coconuts in Vanuatu

Some of the most recent work on pasture establishment under coconuts has been reported in a series of Technical Bulletins by Evans and MacFarlane (1990), Evans et al. (1990), Evans et al. (1991) and Evans et al. (1992).

The major problem in most coconut plantations (see Figure 126) is that of replacing total weed infestations of Anona muricata, Cassia tora, Codiaeum variegatum, Hibiscus tileaceus, Lantana camara, Psuedoelephantopus spicatus, Psidum guajava, Solanum torvum, Stachytarpheta urticifolia and other bush regrowth with sown pastures, or of replacing native pastures with improved pastures or of establishing legumes into legume deficient improved pastures.

For plantations with access to machinery the same principles which apply to rehabilitating open weed infested areas also apply to areas under coconuts, except that costs are approximately 10 percent higher due to reduced efficiency of working machinery and the need for weeding around coconuts. The most important factor is to establish an aggressive and productive pasture which will compete with weeds and result in a sustainable pasture.

Figure 126. - An overgrown coconut area.

Some examples of on-farm rehabilitation under coconuts with owner operated machinery (excluding depreciation):

• Indigofera suffruticosa and Hibiscus tileaceus under coconuts; the area was windrowed, levelled with a heavy bar and sown with glycine to smother the regrowth; the area was further treated to one (high) slashing in the first year and was left ungrazed for 18 months; cost 25,000 VT ha^-1 (US$ 208)².

  ² Exchange rate 120 VT US$ 1 (April 1993).

• Lantana camara under coconuts; fire was put through the area in the dry season (without damaging the coconuts) followed by a heavy duty slash, three disc harrowings and a heavy chain harrowing; it was then seeded with Sabi grass and legumes; weeds were controlled with spot spraying or hand weeding; cost in year 1 was 20,000 VT ha^-1 (US$ 167).

• Cassia tora under coconuts; the area was windrowed with a small bulldozer, slashed and then zero-tilled with a Signal grass/legume mixture; cost 15,000 VT ha^-1 (US$ 125). An alternative treatment involved slashing followed by burning, two discings, harrowing and broadcasting of Signal or Sabi grass seed with a legume mixture; this was followed by chain/roller; cost 15,000 VT ha^-1 (US$ 125).

• A native pasture of carpet grass was oversown by zero-tilling Signal grass and legumes into the coconut interrows; cost was approximately 8,000 VT ha^-1 (US$ 67).

In Western Samoa where mintweed (Hyptis capitata/pectinata) is a considerable problem (and in some areas dense stands reach heights of 1–2 metres prohibiting the harvesting of nuts) control methods have focused on use of herbicides (Butoxone at 8 ml l^-1 water) or biological control through the use of smothering legumes (Evans, 1992) prior to good pasture establishment. In a grazing situation one farmer has found that the application of 100 kg ha^-1 (approx.) of superphosphate to Batiki pastures has been very effective in suppressing reinfestation (Stephen Lee personal communication). Smothering legumes were also used to control weeds like Cassia tora in Vanuatu - see Figure 127.

Evans et al. (1992) stress that successful rehabilitation with little or no use of machinery involves the careful timing of operations, e.g., deferring grazing for long enough to allow climbing legumes such as Siratro, glycine, centro, dolichos lablab or puero to smother Solanum torvum, Sida acuta, Sida rhombifolia, Pseudoelephantopus spicatus or Stachytarpheta urticifolia is an essential first step in the rehabilitation process which avoids the manual removal of the problem weed.