Increasing the profitability of traditional, planted rubber agroforests at the Tapajós river, Brazilian Amazon

The planted rubber agroforests of the Tapajós region in eastern Pará State, Brazilian Amazon, are a traditional land use practice of intermediate intensity between extractivism in natural forest and plantation agriculture, and thus of potential interest especially for inhabited conservation areas and Extractive Reserves. We measured rubber production of five groves of 23–40 years of age on representative soil types and landscape positions of the region during one tapping season and tested options for improving rubber production in a series of on-farm experiments. The groves contained 650–700 rubber trees with a dbh >10 cm ha −1, plus variable numbers of smaller rubber trees. Rubber trees contributed 72–97% to the total basal area of the groves. Average dry rubber production per grove varied between 30 and 75 g per tree per week, with an average of 50 g. Income from rubber tapping at the present level of productivity was unlikely to exceed one Brazilian minimum salary for most farmers. Stimulation of the latex flux with a conventional chemical stimulant, ethephon, increased rubber production per week by up to 38%, but the response was highly variable between groves. The attempt to substitute the chemical stimulant by a locally produced, biological product was unsuccessful. There seemed to be considerable potential for improving tree health while increasing rubber production per unit time through the use of improved tapping practices, especially for old trees. From the two tapping knives and methods that are used in the region, the Amazonian method gave significantly higher yields per tree than the southeast Asian method; this adds to better protection of the trees from panel diseases and therefore greater longevity of the groves. Stem diameter, crown form and crown position of the rubber trees were all significantly related to rubber production, suggesting that too early tapping (which reduces growth) and excessively dense stands should be avoided; however, the predictive power of these variables was poor, and this may explain why farmers are reluctant to thin their stands and believe that mature rubber trees are not negatively affected by a secondary forest environment. Even in groves with low average productivity some very high-yielding trees were found, with the most productive ones attaining 250–300 g of dry rubber per tree per week. This points to a considerable value of these traditional agroforests as genetic reservoirs, and to opportunities to estab