Soil Management for the Establishment of the Forage Legume Arachis pintoi as a Mean to Improve Soil Fertility of Native Pastures of Mexico

lective


Introduction
Pasture (rangelands) degradation in the humid tropics of Latin America is a fact that dates back several decades, and to date not only has not been resolved, but tends to worsen according to the unfavorable economic situation of livestock in the region [1].
In Mexico, according to a report [2], 75% of the degradation is caused by deforestation (25.8%), overgrazing (24.6%) and changing land use (agricultural and urban-industrial, 25.5%). The report adds that, in the north as well as in the southern of Mexico, livestock have overgrazed pastures and supports excesives stocking rates, causing a radical change in the floristic composition of rangelands and reduced permeability of the soil, increasing runoff and causes accelerated erosion thereof.
In this paper we addressed several land management practices for the establishment of the forage legume Arachis pintoi (CIAT accesions 17434, 18744 and 18748) as a means to improve soil fertility on native pastures of Mexico. Arachis pintoi was selected because it is a forage species that has enormous potential to improve the vegetation cover of the grazing areas in the Mexican tropics, and its contribution of nutrients to the soil, improving the fertility of this. All experiments were conducted in the northern of Veracruz state, Mexico, in a hot and humid climate, where soils are classified mainly as Ultisols or Oxisols. Some of the experiences were developed in native pastures and or in citrus plantations because this is a very important crop in this region.
In the case of coverage, the trend of the data indicated the existence of an asymptotic response, so exponential models were fitted to a maximum, logistic and sigmoid, using the routine "Regression Wizard" program SigmaPlot [8]. The model that final showed the best fit to the data coverage with rational values, was the three-parameter sigmoid, which is described below: where "Y" is the coverage in percentage, at a "X" time, given in weeks; "a" is the maximum coverage value predicted by the model; "e" is the base of natural logarithms; "X0" is the time to "Y" reaches 50% of the value of "a"; and "b" is a constant of proportionality indicating the slope of the "S" on the right side (the higher the value, the greater slope), ie how fast it reaches the value of "a".

Second experiment
In this case, the orchard was located in the municipality of Misantla, Veracruz, and consisted of an orange plantation with coffee plants from 14 and 8 years old, respectively. The arrangement of citrus planting was 6 x 6 m, with four coffee plants around each orange tree. A week before the start of the experiment, the native vegetation was controlled with mechanical slashing and application of glyphosate (2 L/ha). The establishment of Arachis pintoi CIAT 17434 was evaluated for three methods of site preparation: disking, chiseling and hoeing, with two A. pintoi plant arrangements: plants at 35 and 50 cm within furrows separated by 75 cm; and two fertilizer levels: with and without P+K+Mg. P was used as triple superphosphate (50 kg/ha P 2 O 5 ), for K, potassium chloride (50 kg/ha of K 2 O), and Mg, magnesium sulfate (20 kg/ha) applied every 30 days post -planting. Plant material consisting of stolons of 20-25 cm was used, placing of 3-4 stolons per plant site.
A randomized blocks design was used, in an split-split plot arrangement, being fertilization treatment the main plot and subplot planting method, sub-divided into two planting densities. This resulted in 12 treatments with four replicates each. The total area was 2268 m 2 , and the experimental unit was 144 m 2 .
We measured the percentage of coverage, number of plants/m 2 and plant height (cm, five plants per replication), at 4, 8, 12, 16 and 20 weeks post planting. Coverage data, number of plants and plant height were subjected to analysis of variance, and means were compared using the Tukey test from the SAS statistical package [9]. The soil was analyzed at the beginning of the experiment and 16 months later to determine changes in organic matter, soil acidity, as well as levels of nitrogen, phosphorus and potassium. Economic estimates were made to determine costs of establishment, maintenance and return on investment, compared to traditional management of weed control in citrus plantations. Were considered: the cost of slashing of the land, legume plant material and its planting labor, fertilization (P, K, Mg), land preparation, with disking, hoeing; and herbicide application.

First experiment
Number of plants. Table 1 shows the average number of plants (and its standard error) for each accession. For the first and the second accesion an increase from week 4 to 12 was registered, while for the third accesion, the average remained constant during the period evaluated; achieving at 12 weeks an overall average of 5.1 plants/m 2 . The analysis of variance did not detect any statistically significant difference within each accession, considering the weeks of sampling. Plant height. Except for the evaluation at 4 weeks, the range was kept between 10 and 20 cm, and the latter value was more frequent in ecotypes 18744 and 18748.

CIAT accesion
Coverage. In the three ecotypes the model was highly significant (P <0.0001), and in all cases the model parameters were different from zero at the same level of probability. R 2 values were greater than 0.8 (Table 2). ¥ "Y" is the percentage of ground covered by the plant, "a" is the maximum coverage, "X0" is the time in weeks to reach half of "a", 'X' is the time in weeks, since planting date; and "b" is a constant of proportionality. Coverage. Table 4 shows the percentages of coverage, achieved five months after establishment. In all treatments the highest values were achieved with the higher plant densities and fertilization treatment, except for planting treatments with hoeing. Treatments involving the disking had values far above the other ones, regardless of the plant density and/or fertilization applied. Analyses of variance performed within each site preparation, indicated statistically significant differences (P ≤ 0.05) considering the variables plant density and fe application or not of fertilizer.
Changes in the soil. In relation to changes in soil properties, increases were recorded for the content of nitrogen, phosphorus and potassium, although there was a decrease in organic matter content (Table 5).
Economic considerations. Economic estimates indicated that establishment costs per hectare (in U.S. dollars) for the year in which the experiment was performed, varied according to the evaluated treatments, being lower for those without fertilization (US $ 294, 410 and 396) in compared with those receiving fertilizer (US $ 356, 472 and 473) for treatments with disking, weeding and hoeing, respectively. Moreover, the expenses incurred to control weeds in one hectare included the purchase of a commercial herbicide (glyphosate), an adherent and implementation of both. It imported US $ 222.  Table 2.

First experiment
The number of plants for the three ecotypes at 12 weeks, was on average lower compared with those found by [10] in one of three experiments with Arachis pintoi in native pastures of that region. He observed that Ap 17434 presented at that time more than 10 plants/m 2 , using plant material grown in field where vegetation was controlled with a machete and herbicide, with or without burning the dead material and fertilized or not with P. The smaller number of plants found here could be due in part to the month after planting (May) was relatively dry (<50 mm), consequently affecting plant emergence. In the mentioned experiment [10] the growth period immediately to planting date had higher humidity (> 150 mm).  Moreover, it appears that the ecotypes evaluated here shown in the early stages of establishment a tendency of erect growth. It has been indicated [10] a range from 14.4 to 21.0 cm at 12 weeks post planting date, regardless of the treatments.
With respect to coverage, the R 2 values showed good predictive power for the environmental conditions during the study. No one model predicted a maximum coverage of 100%, because the measurement time was only 24 weeks.
In Colombia [4] evaluated in citrus plantations the same ecotypes, and found 8 months after that Ap 17434 was much lower coverage (32%) compared with the other ones (73% on average). On native pastures [10], found in another experiment with Ap 17434 that its establishment was even slower, since the accession planted with no-tillage or reduced tillage, with or without fertilization (P, K, Mg, Ca, Zn, Cu and B), needed 20 to 21 weeks to achieve 50% coverage.
The lower rate of coverage by the accession 17434 was also confirmed [11], on the experiment developed in this same region comparing four species of forage legumes (Desmodium ovalifolium, Neonotonia wightii, Pueraria phaseoloides and Stizolobium deerigianum) associated to a citrus plantation. This slowness in the establishment was also reported in Costa Rica [12] to associate in banana plantations.   Table 5. Changes in the soil with the use of Arachis pintoi 16 months after planting.

Soil Fertility
Moreover, the ecotypes established by seed showed a higher rate of coverage, however, however, these differences in the velocity of establishing tend to disappear as time passes.

Second experiment
The coverage obtained with the disking treatment with plants every 35 cm along the furrow, and with or without fertilization are considered acceptable and are superior to those reported for Ap 18748 in coffee plantations of Nicaragua for high plant densities using vegetative material (strips of 3.3 m wide, with furrows 50 cm) and three weedings in the first 90 days [13]. This author reported that at 158 days post seeding, the legume exceeded 60% of ground cover. In Brazil [14], assessed A. pintoi at plant densities of 8 to 16 plants/linear m, reaching a 50% coverage to 84 and 68 days post seeding, respectively; whether the separation between furrows was 25 or 50 cm. The above percentages indicate superior performance under these conditions that found here, which is explained by the higher plant density used.
Respect to changes detected in the soil properties, the increase in the concentration of N could be attributed to a transfer to the soil of the element present in the leaves of Arachis pintoi by the effect of decomposition thereof. In this regard, [15] estimated litter decomposition of grasses and legumes, among whom was A. pintoi. They found that the decomposition of organic matter and nitrogen in leaves of this legume, along with that of Stylosanthes capitata, decomposed faster than the other species studied, although the amounts released of P, K, Ca and Mg were similar among grasses and legumes. Other researchers [13], working with Ap 18748 or Desmodium ovalifolium CIAT 350 associated with coffee plants, found no differences for any legume in N, P and K soil, three years after establishment; unlike [6], who in Australia, in banana plantation with or without Arachis pintoi after 5.5 years found significant increases in the association, in terms of organic matter (3.94 vs. 3.71%), N (0.42 vs. 0.39%); the K, Ca, Mg and Na increased to at 52, 26, 43 and 23%, respectively.
By comparing these costs with the traditional management of weed control in citrus orchards, we found that the costs for these plantations were around US $ 222 per year. Economic estimates in coffee plantations in Nicaragua [13], mentioned that the relative costs (%) in the establishment and maintenance of the associations were higher in the first two years, compared with the traditional control of weeds, but at that time the use of herbicides was lower between 30-50%. Establishment costs in the three experiments [see 10] fell in the range of US $ 282 to 623 (the exchange rate in 2001) in terms of inputs applied. Although costs for the establishment of Arachis pintoi is higher, this is recovered in about a year and a half or two, with the advantage of having a highly competitive species for weed control and its long persistence in the land, plus inputs of nutrients to the soil as an additional benefit.

Conclusions
Arachis pintoi is a promising legume to associate as a cover crop with citrus plantations and other crops of high commercial value, such as bananas, pineapple, coffee and papaya. In the case of the first experiment, Ap ecotypes CIAT 18744 and 18748 represent for citrus plantations area of Veracruz, a better option compared to Ap CIAT 17434, due to the slowness of this accesion to cover the ground. Regarding the second experiment, the disking treatment, proved to be the best treatment for the establishment of the legume, but the costs of establishment will vary depending on the inputs applied, but a long-term coverage will absorb these costs converting this costs in an effective alternative.

Establishment of Arachis pintoi in native pastures of Mexico
Research results from the hot humid areas of México and from other parts of Latin America showed that the forage legume Arachis pintoi CIAT 17434 has the ability to be associated with grasses, because it has shows better persistence than other legumes and also has high nutritive value and palatability [16,[17][18][19][20]. A. pintoi establishment techniques range from a complete soil tillage and planting with seed to zero tillage and planting with vegetative material (stolons) into an existing pasture [17]. The objective of this study was to evaluate the agronomic performance of different techniques of establishing A. pintoi CIAT 17434, as well as the accessions CIAT 18744 and 18748, into existing native pastures in the humid tropics of the coastal plains of the Gulf of México. The climate is hot and humid, with rains all year round. Mean yearly rainfall was 1,917±356 mm from 1980 to 1997. Monthly rainfall is highly variable being September (322 mm) and October (248 mm) the rainiest months while March (85 mm) is the driest. The coldest and hottest months are January (18.9 °C) and June (27.8 °C). Minimum daily temperatures from November to February (winter) are around the critical range of 8-10 °C, below which the growth of C 4 tropical grasses is severely reduced [21][22][23]. These combinations of rainfall and temperature lead to a seasonal DM production pattern, a common situation in the tropics of Latin America: A high growth rate on the rainy season followed by poor growth during the winter and dry seasons.

Site characteristics
The experiments were conducted in different years. Temperatures were typical of each season, but the current maxima were below, and the current minima above the long term (1980-1997) mean ( Figure 3a). Total rainfall during experiment 1, December 1991 to September 1992, was 39% above average ( Figure 3b). Rainfall in the experimental planting seasons was 339 mm in winter (November 29, 1991 to February 14, 1992), 637 mm in the dry season (March 2 to May 18 of 1992) and 1,352 mm in the rainy season (July 2 to September 17 of 1992). Rainfall was 19% above average during experiment 2 in 1993, but rains in 1996 were 43% below average for experiment 3 ( Figure 3b). The soils are acid Ultisols (Durustults), with a range in pH from 4.1 to 5.2, and an impermeable hardpan between 0 and 25 cm in depth, that result in a inadequate drainage during the rainy and winter seasons. The soil texture is clay-loam with low levels of P (< 3 ppm), S (< 30 ppm), Ca (< 3 meq/100 g) y K (< 0.2 meq/100 g). Both cation exchange capacity and aluminum saturation increase with depth, but the latter do not reach toxic levels for pasture plants [24].

Experiment 1. Reduced and zero tillage, with or without fertilisation
The study was conducted to test the combined effects of tillage type: reduced and zero, and fertilisation with (kg/ha): P 22; S 25; K 18, Mg 20; Ca 100; Zn 3; Cu 2 and B 1, or no fertilisation, in a four treatment combination: T1, reduced tillage and fertilisation; T2, reduced tillage without fertilisation; T3, zero tillage and fertilisation; and T4, zero tillage without fertilisation. Reduced tillage consisted of four passes of a disk harrow, while zero tillage only required the elimination of pasture vegetation by machete to ground level.
The experimental area was 2,000 m 2 (50 m x 40 m split in two plots of 1,000 m 2 -25 m x 40 m). These plots were divided in two sub plots of 500 m 2 (25 m x 20 m), of which one sub plot was fertilised. Three 2,000 m 2 -experimental areas were used: one per each climatic season (winter, dry and rainy season).
Arachis pintoi was planted on sub-plots of 500 m 2 on 29 November, 1991 (winter season), 2 March, 1992 (dry season) and 2 July, 1992 (rainy season). Three to four stolons, approximately 15 cm in length and with five nodes per stolon, were planted per planting position. On the reduced tillage treatments the distance between rows and planting positions were 1.0 m and 0.5 m, respectively. Planting was done on 3 m wide strips, which alternated with 3 m intact native pasture strips. Three rows of the legume were planted per strip and 3 strips were contained in a subplot, being the sampling quadrat size 3.0 m x 1.5 m. On the zero tillage treatment, distance between rows and positions was 2 m and 0.5 m, respectively, with the subplot containing nine sampling rows also and a sampling quadrat dimensions of 6 m x 3 m. Even though this planting arrangement was confounded with tillage treatments, it gave a similar number of planting positions per sub-plot and two sampling hills/m 2 in each sampling quadrat, regardless of type of tillage. Fertiliser was broadcast 30 days after planting.

Experiment 2. Type of control of native pasture growth, with or without P fertiliser
This experiment tested the combined effect of the type of pasture vegetation control: herbicide (glyphosate) or slashing (by machete) with or without burning of dead vegetation, and with or without localised P-fertilisation which resulted in eight treatment combinations. The choice of treatments attempted to reduce competition to A. pintoi from existing native pasture vegetation, enhance legume establishment and early growth, following the approach described by [25] for the establishment of legumes into existing Speargrass (Heteropogon contortus) native pastures, in Australia.
Slashing was done by machete and burning was carried out between 1-5 days after slashing. A 2% aqueous solution of glyphosate (480 g of isopropyl amine salt of glyphosate/l) was applied on a 0.25 m wide strip 15 days before planting; burning was done 15 days after herbicide application.
The planting legume was done between June 28 and July 3. Application of herbicide and herbicide plus burning, and slashing or slashing plus burning, were applied 15-16 days and 3-5 days earlier, respectively. Vegetative material, 0.25 m length stolons with eight nodes, was used for planting. This material was inoculated just prior to planting with a specific

Experiment 3. Establisment of Arachis pintoi accessions using seed pods
This experiment compared the establishment of three A. pintoi accessions using seed pods: CIAT 17434 (cv. Pinto peanut or Amarillo), 18744 and 18748. Seed germination was assessed in the laboratory at room temperature; using 125 seeds per accession. Petri dishes, bottomlined with filter paper, were used and were watered twice daily. The seeding rate was equivalent to 10 kg of germinable seed pods per hectare, based on quadruplicate germination tests. The experimental plots (10 m x 5 m; ten 5 m length rows/plot) were established within a grazing experiment where milk production from native pastures and native pastures associated with A. pintoi was to be compared. Three replicates were established in one paddock and three in another. Each replicate had three plots, with an accession each. Plots were excluded from grazing for the 12 weeks of the establishment period. A 2% aqueous solution of glyphosate was applied on a 0.30 m wide strip 15 days before planting to eliminate competition from existing vegetation. Distance between rows and planting positions was 1.0 m and 0.5 m, respectively. Seed pods were placed in a 5 cm deep hole made with pointed wooden stick, and lightly covered with soil by the planter's foot. Three replicates were planted on August 2 and three on September 3, 1996. Fertiliser was not applied.

Measurements and statistical analyses
The response variables were: 1) plant number (PN, plants/m 2 ) by counting; 2) plant height (PH, cm), on each plant within the sampling quadrat, measured with a ruler from the soil surface to the uppermost part of the plant; and 3) soil covered by the legume or cover (COV, % of quadrat area covered by the legume) measured with the aid of a 1 m 2 quadrat, divided into 25 squares, which was placed over the row. These measurements were done on weeks 4, 8 and 12 after planting [26]. In experiment 1, PH was not measured, but COV was measured again at 24 weeks after planting.
In experiment 1, there were no field replications, since it was perceived that treatments applied in larger areas would have a closer resemblance to that of farmers' fields. Also, if several sampling quadrats were used within each treatment plot, this would yield information as useful as that obtained from randomised complete block designs. In experiments 2 and 3, the design was a randomised complete block design with 3 blocks as replicates. The treatment arrangement was a split-plot in experiment 2, where the main plot was the combination of type of pasture vegetation control (slashing and herbicide), while the combinations of burning (with and without) and P application (with and without) were the sub-plots; additionally the effect of time after planting was considered a sub-sub-plot. The treatment arrangement of the third experiment was a split plot, in which the main factor was the combination of month of planting by accession and time after planting the sub-plot. Here, number of plants was expressed as "plants/50 m 2 ", in order to be clearer and avoid fractions of plant/m 2 . Analyses of variance were done with linear additive models in accordance to the experimental design [27]. The natural log transformation of the response variable was used if its response to time was exponential. If necessary, linear or exponential relationships provided rates of increase with time in the measured variables. Also, means comparisons using Tukey's test were done when was necessary.

Experiment 1. Reduced and zero tillage, with or without fertilisation
The main effect of treatment on plant number (PN) was highly significant (P<0.01) in all seasons. The linear effect of week after planting was highly significant (P<0.01) on PN in the winter season of 1991-92 and the rainy season of 1992, but it was not significant (P>0.05) in the dry season of 1992 (Table 6). There was no significant treatment x week interaction on PN in any season. The main effects of treatment and week after planting and its interaction were highly significant (P<0.01) on COV, except for the interaction in the rainy season. Weeks to reach 50% cover were 21, for T2 (winter season) and T4 (dry season); and 20, for T1 and T4 in the rainy season (Table 7).

Experiment 2. Control of native pasture growth, with or without P fertiliser
The effect of time after planting was highly significant (P<0.01) upon all response variables. Height values increased with time, but to a different degree on each main plot combination. The increase in plant height (PH) with time was much larger than the increases with time shown by the other two response variables. The standard deviations were high in all cases and increased with time also (Figure 4). The coefficients of variation remained relatively uniform through time: 28% to 31% for plant number (PN), 29% to 35% for plant height, and 75% to 83% for cover (COV).
When herbicide was applied, the burned plots produced taller plants than the non-burned ones (P= 0.01), but the contrary happened on slashed plots (P<0.05) ( Table 8).
P fertilisation did not increase (P>0.05) legume cover in any vegetation control by burning combination. Slashing without burning and without fertiliser, the treatment with the least external inputs, had significantly (P<0.05) less legume cover than the herbicide plus burning plus fertilisation treatment, the treatment requiring the most external inputs (Table 9).

Experiment 3. Establishment of three A. pintoi accessions using seed pods
The averages of percentage of seed germination at 7 days on the laboratory were of 44 Table 9. Combined effect of vegetation control x burning x P-fertilization treatments upon A. pintoi CIAT 17434 mean cover (COV, %, ± standard error).

Discussion
In experiment 1, reduced tillage gave better results than zero tillage during the winter season, but the opposite occurred in the dry season. As soil moisture and temperature conditions increased in the rainy season, the difference between reduced and zero tillage not disappeared and was significant. Other trials conducted in the same region have indicated the advantage of reduced tillage over zero tillage to establish vegetatively A. pintoi [28]. The literature shows a general agreement among researchers in that some sort of soil disturbance is necessary to assure establishment [29,30]. Means followed by the same letter are not statistically different at P≤0.01.

Month
NS= Non-significant. It has been suggested [25] that seedlings facing more root competition from existing vegetation responded to fertilisation, whereas those without competition had a lesser or nil response.
In the winter season planting of experiment 1, fertilisation failed to stimulate COV of slashed plots, those supposedly with a larger competition from existing pasture. In the dry season planting, fertilisation was detrimental to COV in the slashed plots, in contrast to [25]; finally, in the rainy season the effect of fertilisation was negligible. The second experiment showed a positive effect of fertilisation on COV only when herbicide was applied and the dried vegetation was burned. When plots were slashed, but not burned, the effect of fertilisation on COV was positive. Nevertheless, when the slashed plots were burned, the fertilisation effect on COV was negative.
As suggested by the inconsistent results of our trials and those of the literature, fertilisation appears not to be of great importance for the establishment of A. pintoi, when vegetative material is used. The lack of P response on Arachis species has been reported by other researchers. In experiment 2, single superphosphate was used, and perhaps the use of this source could explain, partially, the lack of response. Also, the very low P levels on soils at CEIEGT (0.6 to 1.2 µg g -1 soil on 0-30 cm depth), could limit N mineralization [31], resulting in a poor legume performance.
In experiment 2, burning was directed to reduce competition from existing grasses, since the way A. pintoi vegetative material was planted assured a close contact with the soil. However, burning, as well as fertilisation, did not show a clear positive trend either on COV or on PHT.
When only herbicide was applied in bands in experiment 2, pasture canopy height was not reduced, leading to reduced PH of A. pintoi. On the other hand, when the herbicide treated vegetation was burned, PH of A. pintoi was not impeded. Non-burned plots gave slightly taller A. pintoi plants than those burned. A. pintoi CIAT 18744 flowers less and produces a denser stolon mat than the other two accessions and it also has a vigorous initial growth, covering the soil more rapidly than the CIAT 17434 accession [32][33]. For this reason, a better behaviour during establishment, particularly with respect to COV and PN was expected from this cultivar. Nevertheless, in experiment 3, COV performance at the end of establishment was similar to that of CIAT 17434 (8.5% vs. 8.7%) and only slightly better than CIAT 18748 (7.5%). Then, the 3 accessions behaved similarly during establishment. Rates of plant emergence are considered to be good, as A. pintoi is a legume that can have a strong dormancy [34]. However, emergence (from 125 seeds originally planted/plot) of new branched plants/plot was not so bad, considering that these values ranged from 70% to 90% for three accessions. Therefore, there was low coverage but high number of new branched plants. This situation is common for A. pintoi, which is characterized by its slow establishment, as has been reported [6,[35][36]. Zero tillage failed to stimulate a rapid establishment of A. pintoi in these trials, the reproductive mechanisms of this species ensure that eventually it will establish and encroach within the pasture. Our experience with this legume is that eventually it ends up to be the dominant species when associated with native pasture, Stargrass, or to both. A good strategy would be to establish A. pintoi in strips with reduced tillage at high density. This will result in a rapid establishment of a mixed sward in a minimum of time.

Conclusions
Neither fertilisation nor burning were successful in enhancing A. pintoi establishment; slashing did not improve establishment either. On the contrary, herbicides were effective and improved establishment over slashing. The best alternative to introduce A. pintoi into a native pasture is by reduced soil tillage in strips using, within the strips, 8 kg of pure live seed pods/ha; or 0.70 m between rows and 0.35 m between planting positions for vegetative material.

Establishment of Arachis pintoi CIAT 17434 and Pueraria phaseoloides CIAT 9900 using minimum tillage in Veracruz, Mexico
In the watershed Gulf of Mexico region, there is a highly seasonal pasture production due to climate variability. The main genera are components of Paspalum, Panicum and Cynodon (Gramineae), and in smaller proportions Centrosema and Desmodium [37]. Among the legumes evaluated in that area, A. pintoi CIAT 1434 and Pueraria phaseoloides CIAT 9900 outstanding for their performance and good adaptation [38].
The cost of establishing pastures in native savanna vegetation is high when following traditional methods. Given this, it is justified to evaluate planting systems cheaper, to promote the adoption of new forages and their use to recover degraded pasture [39]. Therefore, this trial is performed to supporting evidence to assess the effect of various types of tillage and application of phosphorus on the establishment of A. pintoi CIAT 17434 and Pueraria phaseoloides CIAT 9900 in native pastures.

Characteristics of the experimental site
The research was conducted at the Centre for Teaching, Research and Extension in Tropical Animal Husbandry of the Faculty of Veterinary Medicine, of the National University of Mexico (UNAM), located in north-central region State of Veracruz, 20 ° 4 'north longitude 97° 3' W and a height of 105 meters above sea level. The climate is hot and humid with rain all year, type Af (m) with average daily temperature of 23.4 ° C and average annual total precipitation of 1840 mm (1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989). The soil texture ranges from sandy loam to sandy clay. The area has a hard horizon with low permeability that occurs between 5 and 25 cm deep. The soils are acidic (pH 4.1 to 5.2), and are classified as Ultisols.
We used an area of 6.000 m 2 of degraded native pasture grazed by cattle. The treatments were the type of weeding (slashing, S; and herbicide, H) and the burning (B) application or not (with + B and without -B), to temporarily control the growth of existing vegetation (larger plots), and thus prove its effectiveness to allow the establishment of the legumes Arachis pintoi CIA 17434 (Ap) and Pueraria phaseoloides CIAT 9900 (Pp). Additionally we evaluated the application of phosphate fertilizer or not (+P addition; no-P as single superphosphate). The factorial combination between legumes and fertilizer was the subplot.
Treatments were applied between 28 May and 3 June 1993. The slashing (S) was a machete to the whole plot. In S + B, the burning was applied between one and five days after slashing. The application of herbicide (H) was done using a backpack sprayer, applied in bands 50 cm wide, spaced 1 m apart from the center of each. The dose was 0.96 kg (2 l) of a nonselective systemic herbicide (Glyphosate). The product was dissolved in 200 l of water and applied 15 days before seeding. In H + B, herbicide application was the same way as above, burning 15 days after application, only the bands where the herbicide is applied.
Ap vegetative material, was inoculated with the specific Rhizobium, by means of a suspension prepared with nodulated roots, washed and crushed to release Rhizobium bacteria, then adding cold water and molasses, placing the suspension in a refrigerator, performing all procedure in the shade. Each kg of root was added 1.5 kg of molasses (as adherent) and 7.5 l of cold water.
Legumes were planted between 3 and 5 days after applying treatments S or S + B, and between 15 and 16 days after applying treatments H or H + B. Ap was planted with stolons of 20 cm long, inoculated with the suspension of Rhizobium already described. By planting, we used a metal digging stick, to make a biased hole of 15 cm lenght and 5 cm depth. Three stolons were placed by hole and soil was compacted with foot to ensure contact with the ground. The distance between plants and rows was 0.5 and 1 m, respectively.
For planting of kudzu (Pp) botanical seed was used, previously scarified with sulfuric acid to 98% for 10 minutes. This ensured the seed germination in three days post seeding. The effectiveness of this treatment has been verified by other researchers [40].
Planting density was 2 kg/ha of pure and viable seed. After scarified, the seed was inoculated and seeded similarly as Ap placing about 8 seeds per site, but was not covered preventing soil compaction. The distance between plants and rows was 0.5 and 1 m, respectively. Single superphosphate (333 kg/ha = 30 kg P/ha) was applied at planting time in 5 cm band from the seed or plant material.
We used two sampling sites per plot at random. Firstly, two rows of each plot were chosen, and then the sampling site within each row. The recommended [26] variables were, number of plants, plant height (cm) and coverage (%) at 4, 8 and 12 weeks post seeding. A randomized complete block design was used, with three replications and a split-plot arrangement, with the factorial combination between weeding and burning as main plots, and the combination of the two legumes with or without fertilization as subplots. We considered the costs for materials and labor costs per treatment.

Results and discussion
Climate. -The climatic parameters of precipitation and temperature were recorded from May to September 1993. The monthly average temperature was similar for the periods, ranging from 25. The small number of Pp plants is also attributed to the seed rot because of soil waterlogging. It has been mentioned [41] that heavy rainfall limits the development of Kudzu (Pp). Also, surprisingly, the number of plants decreased as time passes (P <0.01): There were 1.27, 1.18 and 1.0 plants/m 2 for first, second and third samples. Effects such as slashing, burning, and their interaction were not significant (P> 0.05), which coincides with other experiment [42]. These authors, who established three species of legumes (Centrosema pubescens, Macroptilium atropurpureum and clitoria ternatea) using total soil preparation, harrowing, plowing and burning, with no significant difference found (P> 0.05) among the different methods,and concluded that burning favored the establishment of legumes.
Plant height. -Analysis of variance showed highly significant differences (P <0.01) between the species: Ap with 18.3 cm and 9.5 cm with Pp ( Figure 6). This difference is attributed to Ap was seeded with plant material starting its growth as seedling, which gave to Ap an advantage over Pp that was sown with seed. The interaction slashing X burning was highly significant (P <0.01). The application of H+B resulted in greater plant height with 16.0 cm, followed by S-B with 15.6 cm, being H-B method the lowest height with 11.1 cm. In the case of H+B, the plant height was attributed to no competition between the legume and native grasses; also, burning causes release of soil nutrients that legumes can absorb quickly, making their establishment more effectively.
Burning, releases mineral nutrients immobilized in plant tissues, and others are transformed into simple soluble salts, readily available to the plant [43]. In the treatment of S-B, the largest plant height was mainly due to competition for sunlight by the grass. Competition for sunlight between Aeschynomene, seeded with the grass Hemarthria altissi-ma resulted in greater height during legume establishment [44]. Here, the combination of H-B produced lesser height.
Sampling at 4, 8 and 12 weeks showed highly significant differences (P <0.01) with 7.10, 12.23 and 22.39 cm, respectively (Figure 7). This increase in plant height in time was an expected effect. The interactions species x weeding x fertilization, and weeding x species x sampling were significant (P <0.05), while weeding x fertilization x species x sampling were highly significant (P <0.01). Most probably is that the latter would have been highly significant because it contained the first two.
These results coincide with those of an experiment in Cuba [45], who evaluated different methods of control vegetation during the establishment of Leucana leucocephala, and reported that the best method was the application of systemic herbicide, achieving a plant height of 162.5 m and plant coverage of 96% to 5 months post seeding, concluding that the best promoter of the successful establishment of this legume was the control of vegetation.
Coverage. -The analysis of variance showed highly significant differences (P <0.01) between species: Ap showed a coverage of 2.6% and 0.5% Kudzu. We also found highly significant difference (P <0.01) between samples, with 0.7, 1.2 and 2.8% at 4, 8 and 12 weeks post seeding. The species x sampling interaction was highly significant (P <0.01). Ap was the best species, averaging 1.2, 2.0 and 4.6% while Pp averaged 0.2, 0.5 and 0.9% for the same samples ( Figure 8). The interaction slashing x fertilization x burning was also significant (P <0.05), resulting in the best combination of the H+B+P with 2.5% coverage, followed by S-B-P with 1.9%. Burning + fertilization promoted a good establishment of legumes. The combinations in which was planted after herbicide application, showed no significant differences for the variable coverage.
The burning x slashing x sampling interaction was also significant (P 0 <0.05). In the third sampling, treatment H+B+P was the best combination of coverage averaging 4.2%, followed by H-B-P with 2.4%. The other combinations were not significantly different from each other. The combination S+B+P coverage reached 3.7%, which is the highest value, which shows that the burning had positive influence on legume development, although interacted differently with the type of weeding and fertilizing.
The elimination of competition below and above ground, by applying H+B+P promotes the successful establishment of legumes. The lack of competition, plus the application of P, al-lowed to establish successfully the legume Siratro (Macroptilium atropurpureum) on the native grass [25].
The higher cost of treatment to establish Ap, was the S+B+P, or S+B-P (USD $ 195.00/ha), whereas the application of H-B-P was less expensive to establish Pp (USD $ 86.00/ha). Herbicide application was more economical compared to the slashing treatment. (Table 11).

Conclusions
The banded herbicide application without application of fertilizer is the best method for introducing vegetatively Ap in native grass pastures in north-central region of Veracruz State, Mexico.

Establishment of Arachis pintoi CIAT 17434 by two tillage methods in a native pasture of Veracruz, Mexico
In the humid tropics of Mexico, native pastures are affected by climatic variations from one season to another that make it difficult, to obtain stable yields of forage during the year.
Also, financial constraints of most producers in the tropics must be considered when trying to introduce forage species [40]; so it is justified, evaluate and implement systems-on planting native vegetation, different from the traditional, in order to encourage the adoption of new and improved grass species, the lower potential economic costs.
In order to improve the botanical composition of native pasture in north-central region of Veracruz, Mexico, was evaluated two methods of establishment to incorporate the forage legume Arachis pintoi CIAT 17434.

Location
Centre The study was conducted during the three seasons representative of this region: winter or "North" from November to February; drought: March to June, rain or summer: July to October. Weather conditions for the experimental period by time are presented in Figure 9.
Experimental design and treatments. We used a completely randomized design with factorial arrangement of 2 x 2: Conventional or minimum tillage, and fertilization or not, within each period and 12 observations (no repetitions) per treatment (T), resulting in: Land preparation. In T1 and T2, were allocated strips of 3 m x 20 m, alternating with native grass, where the vegetation was slashed with desvaradora, followed by 4 to 5 passes of harrow and plowed with a hoe. On the strips, the distance between rows and plants within them was 80 and 50 cm, respectively. The legume is seeded with a seed depth of 15 cm.
Minimum tillage. For T3 and T4, there was a land clearing with machete, were traced rows of 20 m long, spaced every two meters, and the rows were holes (seed points) every 50 cm, diameter and depth of 20 and 15 cm, respectively.
Planting dates were in Nov 29/1991, March 2/1992 and jul 15/1992, with plant material, placing 3-4 stems of 15 cm long, with only three or four leaves in the air. After 30 days, treatments were applied "with" and "without fertilization. These works were carried out in each season and in the corresponding area. Weed control was made with a hoe, in the first three months, for each treatment and time.
Variables. Data were collected at 4, 8, and 12 weeks post-seeding for number of plants, and 4, 8, 12 and 24 weeks for coverage. The useful area was 9 m 2 (T1 and T2) and 18 m 2 (T3 and T4). The first variable was the number of facilities within the useful area and in the second, the proportion was estimated visually apparent that the legume covered the area. The data were analyzed separately for each planting season, using ANOVA, and Tukey's test was used to compare means [9]. Regression coefficients were estimated to number of plants (linear) and coverage (exponential) to observe trends.   Figure 10A shows the increase in coverage during the establishment period, for each treatment at 4, 8, 12 and 24 weeks. There is a considerable increase for all treatments from week 12. The maximum coverage at 24 weeks is presented in conventional tillage treatments.

Winter season
The rate of coverage of the ground, expressed as the average time in weeks for the plants to cover 10% of area, is presented in Table 14.  In conventional tillage, were less plants than at minimum tillage treatments. Theere are not significance for age effect, neither its interaction with soil treatment Table 13.

Coverage
The best coverage (>25%) was at at 3 and 4 treatments (P≤0.05). Figure 10B shows the soil coverage at each evaluation frequency. An outsatndinh behaviour was observed for T1 after 8 weeks, achievinig 80% coverage to 24 weeks.
The age effect and its interaction with treatments were statistically significant. The shortest time to cover 10% of soil was during dry and rainy seasons at T4 (Table 14).

Coverage
During the rainy season, soil coverage was similar among treatments ( Figure 10C). The overall average was 34.5% with a coefficient of variation of 26.1%. For the rate of ground coverage, the lowest average time was observed in T4 with 2.0 weeks to cover 10% of the area (Table 14).

Number of plants in each season
The average of plants/m2 was largest during the rainy season (19.4 plants) followed by winter season (14.2 plants), and dry season (8,7 plants). The analysis of variance and regression coefficients for number of plants/season are shown in Table 15.

Soil coverage and age of plants
The best percentages of soil coverage by A. pintoi occurred at 24 weeks, highlighting the rainy season planting date. Table 16 shows the analysis of variance and regression coefficients between soil coverag and age of plants on each season.

Discussion
Although the two ways to establish Arachis pintoi tested here are not the only ones, the results with conventional tillage are attractive, in the frequencies tested. In this regard, the method [46], using a planting implement, designed for them, allowed that two months after planting shown good development. Here, at 24 weeks, the ground cover in all treatments was above 80%, while the total coverage (100%) in treatments with conventional tillage was achieved approximately eight months post-planting.
Should be noted that the availability of plant material is an advantage in the evaluation of the species, as well as attempts to disseminate the same among low-income producers, because of the ease of material handling.
The null effect of fertilization on the establishment of Arachis pintoi found here, was also observed in Colombia [47], who applied seed and fertilizer pellets to a degraded pasture of Brachiaria. The fertilizers were the same as those applied here, except Zn, Cu and B, although in much smaller quantities. This lack of effect could be explained based on the relatively short period of observation (12 and 24 weeks for number of plants and coverage, respectively) per day, such as to indicate their presence nutrients to the crop, especially in the case of P, which is referred to their low mobility in soil.