Integrating cattle into the slash-and-burn cycle on smallholdings in the Eastern Amazon, using grass-capoeira or grass-legume pastures

In the humid tropics of northern Brazil, where slash-and-burn cropping is the prevailing smallholder land use system, pastures ecologically degrade after 7–10 years of use, mostly due to invading vegetation, decline of soil fertility (N, P) and insect pests. Degraded pasture areas cannot be easily restored or returned to cropping, so large areas are abandoned. To avoid this degradation process, two alternative pastures were tested to incorporate cattle into the slash-and-burn cropping cycle. It was thought that the inclusion of secondary woody fallow vegetation, locally called capoeira, or of legumes in pastures restores the soil for a subsequent cropping phase while allowing an acceptable level of animal performance. This hypothesis was tested in a researcher-managed on-farm experiment in Igarapé-Açu (Bragantina region, Pará). A grass-legume pasture (GLP), combining the grass Brachiaria humidicola with three multi-purpose legumes, was compared with a pasture, where controlled regrowth of capoeira was tolerated (GCP), and tested against two controls in the form of a conventional B. humidicola pasture (GP) and undisturbed regrowth of capoeira (UC). The pastures were grazed in a rotational system for 2 years at 1.5 livestock units (LU) ha−1 in the first year and at 1.2 LU ha−1 in the following year. The results showed that the GCP kept the full regenerative potential of the capoeira, showing no significant loss of phytodiversity (GCP 67 species 100 m−2; UC 72 species 100 m−2) and low dissimilarity in species composition measured by Euclidian distance (UC/GCP = 51; UC/UC = 52; GCP/GCP = 33). On the GLP, the legumes Arachis pintoi and Cratylia argentea almost disappeared due to the grazing pressure, while Chamaecrista rotundifolia var. grandiflora was little palatable and invaded the plots by seed propagation. None of the legumes produced a meaningful quantity of nodules, and no difference in total soil N was found between GLP and the other pasture treatments. With regard to animal production, both alternative pastures were less productive than GP over 21 months of grazing (GLP = 384 kg liveweight ha−1, GCP = 474 kg ha−1, GP = 659 kg ha−1), which was attributed to overstocking in the second year. GCP was considered to be an interesting option for smallholdings, provided the balance between the vigorous regrowth of capoeira and forage grass cover can be maintained. In contrast, on GLP a more palatable Ch. rotundifolia accession would be desirable and the establishment of all legumes needs to be improved. It was concluded that GCP would support a subsequent cropping phase better than GLP. However, on both pastures stocking rates would have to be frequently adjusted a management practice that is of secondary consideration in a crop/livestock system in which cattle are kept as cash-generating assets.