Samples obtained from two planting experiments with tropical shrub legumes rich in condensed tannins (CT) (Calliandra calothyrsus, Flemingia macrophylla, Leucaena leucocephala) were analysed in detail for CT concentration, composition and properties, as well as for in vitro degradability. The factors evaluated included cultivation site (fertile mollisol with pH 7.8 and bimodal rainfall vs. infertile oxisol with pH 5.0, high Al saturation and unimodal rainfall) and different levels of mineral fertilisation. In vitro ruminal degradability, CT concentration and properties were affected by cultivation site or fertiliser application or both. In Calliandra and Flemingia, the CT concentration showed a trend (P<0.1) to be higher at the site with the infertile soil. This was consistent with the increased (P<0.001) astringency of CT extracts found in Calliandra when plants were cultivated at this site. In Flemingia, however, astringency was reduced (P<0.001) on the infertile soil. There were certain effects of the cultivation site on the anthocyanidin profile of CT from Calliandra and Flemingia, but these were relatively small and could not explain the differences in astringency observed between sites. Fertilisation did not affect (P>0.05) total CT concentration in Calliandra and Flemingia but clearly increased (P<0.01) astringency of the CT extracts. This suggests that not only CT concentration but also properties of CT may be affected by cultivation site and fertilisation. Leucaena did not grow on the infertile soil and site effects could therefore not be evaluated in this species. Fertilisation did not affect (P>0.05) CT concentration and astringency of CT extracts of Leucaena but reduced (P<0.05) the cyanidin proportion in CT. Cultivation site affected several variables describing fermentation dynamics. Foliage of the legumes grown on the infertile soil exhibited lower (P<0.05) gas production rates. Dry matter degradation was lower (P<0.001) and the proportion of dietary N remaining undegraded after in vitro incubation was higher (P<0.001) when Calliandra was grown on the infertile compared to the fertile soil. The present results indicate that the cultivation site may affect the feeding value of CT-containing legumes. Fertilisation could not compensate the negative effects of the cultivation site on CT properties and ruminal nutrient degradation. Selecting appropriate cultivation sites (with higher soil fertility) may contribute to improve the feeding value of certain tanniniferous tropical shrub legumes.

Results from this greenhouse study indicated that drought resistance of canavalia under low soil fertility conditions is associated with deep rooting ability and vigorous fine root development to explore greater volume of soil. Two Canavalia brasiliensis accessions (CIAT 7969 and CIAT 905) were found outstanding in their total root length production across soil depth under combined stress factors of low soil fertility and drought. Materials and Methods

The biomass production and the nutritive value of three tropical shrub legumes with condensed tannins (CT) (Calliandra calothyrsus, Flemingia macrophylla, Leucaena leucocephala) and two without CT (Cratylia argentea, Desmodium velutinum) as nitrogen-rich feed supplements for ruminants were assessed in two field experiments in Colombia. In one experiment, conducted on two different typical tropical soils (mollisol vs. oxisol; pH, 7 center dot 8 vs. 5 center dot 0; P content, 43 vs. 2 mg kg DM-1), the effects of low and high levels of fertilizer application with P, K, Ca, Mg and S were tested in both the dry and rainy seasons. In a second experiment on the oxisol, the effect of a lower level of application of either P or S fertilizer was assessed. On the oxisol, C. calothyrsus and F. macrophylla had the highest biomass production (93 and 100 g DM plant(-1) in 9 weeks respectively) but the dry season caused extremely low DM yields in all species tested. Leucaena leucocephala did not grow on the oxisol but had the highest biomass production on the mollisol (454 g DM plant(-1) in 9 weeks). On the oxisol, the mineral concentrations of the forage were below the requirements of ruminant livestock. Fertilizer treatment had no clear effects on the nutrient composition of the forage. The non-CT shrub C. argentea had the highest concentrations of most minerals in its forage.

In view of increasing prices for grain based concentrates, alternative feeds are sought to improve competitiveness of smallholder pig production in the tropics. Ensiled forage legumes are seen as an option. The “Rostocker Fermentation Test” allowed a quick evaluation of the ensilability of forage legumes per se and its possible improvement by additives. Vigna unguiculata and/or Canavalia brasiliensis resp. were harvested at 4 different ages (Martens et al., this volume), chopped and minced. 50 g fresh material was weighed in a sterile glass beaker (600 ml) and 200 ml of distilled water were added. Four treatments applied in triplicates: control, control + 2% sucrose on FM base, control + Pediococcus acidilactici preparation (3.0*105cfug-1 FM), control + P. acidilactici + 2% sucrose. The preparations were incubated at 35◦C for 2 days. The pH as indicator for the ongoing acidification was measured at 0, 22, 28, 52 and 58 hours. With Vigna unguiculata, after an initial sour fermentation, the pH of the control rose again during the second day of fermentation until 10 weeks of age. The addition of the lactic acid bacteria preparation slightly improved the fermentation. A significantly better pH development was achieved by adding sugar from 8 weeks onwards. The most promising treatment, independent from the plant age was the combination of the P. acidilactici preparation with sugar, fermenting most rapidly and always leading to a final pH of ≤ 4.0. Best results without additional sugar were achieved at 12 weeks age, the ripening beans probably providing additional carbohydrates (WSC), confirming with findings for lab scale silages (Martens et al., this volume). The pH development of silage cultures with Canavalia without sugar addition was not acceptable throughout the weeks 8–20 hardly going below pH 5.0. The best age for ensiling Canavalia when adding sugar and preferably combine that with lactic acid bacteria inoculation was at 12 and 20 weeks. By this method it was revealed, that in general, the ensilability of Canavalia was worse compared to Vigna, probably due to less WSC and to a higher buffering capacity.

As prices for feed concentrates are rising alternative options for small and medium pig producers in the tropics are searched. Locally grown legumes could contribute to the protein supply. Starchy roots and tubers such as sweet potato (Ipomoea batatas) could add to the energetic value. Vigna unguiculata CIAT9611 and Canavalia brasiliensis CIAT17009 were evaluated at four different ages. Vigna was cut at 6 (pre-florescence), 8 (florescence), 10 (post-florescence) and 12 (pods ripening) weeks of growth; Canavalia was cut at 8, 12, 16 and 20 weeks of growth (no distinct generative stage observed). The forages were wilted to a target dry matter (DM) of 35% and then chopped. On each occasion sweet potato roots were harvested, washed, chopped and dried for several hours. The samples were ensiled in PVC tubes of 1.8 l Volume in triplicates. Three treatments were applied: Vigna or Canavalia only, Vigna or Canavalia mixed with sweet potatoes in the ratio 1:1 on fresh matter base, sweet potato only. The silages were evaluated after storage of approximately 3 months at 25 oC assessing their smell, structure and colour according to the DLG key. The DM content was determined in triplicates at 105 oC and corrected. The pH was measured and judged according to the DLG key 1998 for evaluation of silages based on chemical analysis. Results indicated that when ensiling Vigna or Canavalia alone, the best fermentation results were achieved at 12 weeks age in both cases, with Vigna silage having a more desirable low pH. The mixture with sweet potato improved fermentation. The best Vigna-Ipomoea silage, however, resulted when sweet potato with an unintentionally high DM was used at 12 weeks age of Vigna (silage DM 57%). The best Canavalia-Ipomoea silage was obtained with 16 weeks old Canavalia. Ipomoea only silages with DM contents ranging from 22–40% had the lowest pH values from 3.6–4.2. It can be concluded that the legume harvest age plays a role in the suitability for ensiling and that addition of sweet potato improves legume silages.