An analysis realized at the Institute of Animal Science of Cuba (Ruiz and Febles 2006) reviewed all the information obtained in the country regarding creeping legume persistency in grasslands. It was found that among the aspects influencing on this performance is the amount of rooted points, the sowing methods used, weed presence, as well as the beginning of the exploitation of these species without attaining a satisfactory establishment, among other factors.

Amezquita (1999) indicated that soil preparation, if required, must be directed to correct some physical limitation present in its depth where the roots grow. Once solved the physical problems, those of chemical order must be worked out, associated with amendments and to the application of lacking or deficient nutritive elements for creating a good environment of organic matter in order that beneficial microorganisms can act conveniently.

The objective of this study was to study the effect of the time of starting soil preparation and the number of rows for sowing on the establishment of multiple mixtures of creeping legumes in association with grasses.

Treatments and design. A random block design in factorial arrangement with four replications was applied. Factors were: time of starting soil preparation (February, March and April), number ofrows for sowing (grass: legume; 2:2 and 2:3) and unsown control.

Experimental procedure. The experiments were carried out in a red ferrallitic soil of fast drainage, clayey and deep on limestone (Hernández et al. 1999), equivalent to the subtype cambisol ferralic rodic according to FAO-UNESCO (Duran and Pérez 1994). Research was developed in the experimental area of the Department of Pastures and Forages of the Institute of Animal Science of Cuba, located at Western part of the country at 22º 53' North latitude and 82º 02' of West longitude at an altitude of 80 m.a.s.l.

Soil preparation was by ploughing, two harrow passes and rowing all the area at 0.70 m. Sowing was carried out in June when the rainy period was stabilized. Experimental plots had a net area of 7 x 6 m. Seed were inoculated with the adequate Rhizobium strain for each species. For the legume mixtures, 8 kg/ha of P.G.S. was used. Grass and legume sowing was conducted according the methodology described by Ruiz et al. (2000).

The mixture consisted of the associations of the legumes Centrosema pubescens (centro), Macroptilium atropurpureum (Siratro), Neonotonia wightii (glycine), Calopogonium mucunoides (Calopogonium) and Stylosanthes guyanensis (stylo) with the grass Panicum maximum (guinea grass).

From these measurements, the two formers were transformed according to «x and for the botanical composition, arc sin «x was used. For determining biomass production of legumes and guinea grass, cutting was at 20 and 10 cm height, respectively. All measurements were performed 80 days after the germination of the species sown. An analysis of variance was made and Duncan's (1955) test was applied in the necessary cases.

All treatments received the same number of field labors for soil preparation (ploughing- harrow-harrow).

There was no interaction between factors for the measurements under study (table 1). When the effect of the time of starting soil preparation for sowing, was evaluated with preparation times shorter than three months (table 1), it was found that when preparationstarted in February there were important values for legume and grass yield. The same occurred for other measurements under assessment. The best performance of the number of rows for sowing (grass-legume) at the time of establishment was shown in the relationship 2:3, in measurements as important as higher number of legumes/m² and lower number of weeds/m².

In table 1 is shown that as the numerical value of weed plants increased, the amount of rooted points was affected as it took place in March. Meanwhile, the number of legume plants/m² presented differences and thus, was reflected in its yield when it was lower as it occurred in the treatments of March and April. The above mentioned indicates that the weed factor markedly influenced on the sowing and initial development of the legume-grass plants.

There was interaction for the weeds regarding weight (t DM/ha). It must be highlighted (table 2) that the highest incidence, measured in all treatments at the end of the rainy period, was evident in areas where the association was not sown. This became lower as the preparation time was shorter (April, March and February). This performance can be associated with the fact that the preparation was realized in a phase with some level of rainfall (April) that favors weed germination and thus, their elimination in the process of soil preparation. In that regard, Kolmans and Vásquez (1999) indicated that for a correct land tilling it must be realized in the season where the conditions of humidity, temperature and solar radiation are most favorable.

There were no differences regarding weed presence in sowings realized at the different times for starting soil preparation. Weed incidence is not desirable for starting the exploitation of an association, since they influence negatively on the future stability of the grassland. It must be borne in mind that weed incidence in areas exploited by the animals could provoke differences between them, as an indicator of pasture stability.

In a study carried out by Ruiz and Febles (2008) for determining the performance of different combinations of grass legumes against two times of soil preparation, it was found that sowing of more than one legume presented superior performance for all indicators when compared to the sowing of only one species. Soil preparation by time period longer than three months before sowing does not show marked influence on indicators as important as higher weed presence, lower population values and rooted points of legumes, aspects of great importance for attaining good establishments. The preparation time lower than three months presented more satisfactory values, since lower weed incidence and higher amount of plants and rooted points per area (rooted points) were obtained.

Reyes et al. (1995) stated that the conventional method of soil preparation may be more productive that direct sowing or tilling reduction, but the latter allow an establishment that although slow, its cost is low. These authors recommends using plough + two harrow passes and suggest further studies under different conditions. Similar conclusions were referred by Bernal et al.(2003).

Ruiz et al. (2000) and Padilla et al. (2005) coincide in that better results are attained in pasture establishment when soil preparations are realized during short periods (three months).

In addition, the literature information in the last years is scarce and mainly centered on the use of zero soil tilling in cultures as maize, sorghum and not in pastures and forages. Similar conclusions were cited by Lok et al. (2011) on pointing out the need of studying further soil and pastures indicators that reflect better the stability of legume systems.

It is concluded that the beginning of soil preparation in February favors the harmonic plant and rooted points of creeping legume development. This is displayed in the balance and persistency between the system components.