Factors affecting biomass production and nutritive value of Calliandra calothyrsus leaf as fodder for ruminants
123*4†2P.K. TUWEI, J.N.N. KANG‟ARA, J.L. STEWART, J. POOLE, F.K.NGUGI 5 AND I. MUELLER-HARVEY
1 Kenya Forestry Research Institute, Regional Research Centre Embu, P.O. Box 27, Embu, Kenya. 2 Kenya Agricultural Research Institute, Regional Research Centre Embu, P.O. Box 27. Embu, Kenya. 3 Oxford Forestry Institute, Department of Plant Sciences, University of Oxford, South Parks Road, Oxford OX1 3RB, U.K. 4 International Centre for Research in Agroforestry, P.O. Box 30677, Nairobi, Kenya 5 Department of Agriculture, University of Reading, P.O. Box 236, Reading RG6 6AT, U.K.
Short title: Biomass and nutritive value of Calliandra calothyrsus
*To whom all correspondence should be addressed.
Email: email@example.com. † Present address: CABI Africa Regional Centre, ICRAF Complex, PO Box 633, Village Market, Nairobi, Kenya.
Calliandra calothyrsus is a tree legume native to Mexico and Central America. The
species has attracted considerable attention for its capacity to produce both fuelwood and foliage for either green manure or fodder. Its high content of proanthocyanidins (condensed tannins) and associated low digestibility has, however, limited its use as a feed for ruminants, and there is also a widespread perception that wilting the leaves further reduces their nutritive value. Nevertheless, there has been increasing uptake of calliandra as fodder in certain regions, notably the Central Highlands of Kenya. The present study, conducted in Embu, Kenya, investigated effects of provenance, wilting, cutting frequency and seasonal variation both in the laboratory (in vitro digestibility, crude protein, neutral
detergent fibre, extractable and bound proanthocyanidins) and in on-station animal production trials with growing lambs and lactating goats. The local Kenyan land race of calliandra (Embu) and a closely-related Guatemalan provenance (Patulul) were found to be significantly different, and superior, to a provenance from Nicaragua (San Ramón) in most of the laboratory traits measured, as well as in animal production and feed efficiency. Cutting frequency had no important effect on quality; and although all quality traits displayed seasonal variation there was little discernible pattern to this variation. Wilting had a much less negative effect than expected, and for lambs fed calliandra as a supplement to a low quality basal feed (maize stover), wilting was actually found to give higher live-weight gain and feed efficiency. Conversely, with a high quality basal diet (Napier grass) wilting enhanced intake but not live-weight gain, so feed efficiency was greater for fresh material. The difference between fresh and wilted leaves was not great enough to justify the current widespread recommendation that calliandra should always be fed fresh.
Utilization of trees and shrubs has long been recognized to be one of the most effective means of improving both the supply and the quality of forage in tropical smallholder livestock systems, especially during the dry season (Gutteridge & Shelton 1994; Robinson 1985). Green fodder from nitrogen-fixing leguminous trees, in particular, contains much higher levels of protein than the poor quality basal feeds (grasses and crop residues such as stover) available during the dry season, and persists for longer owing to the trees‟ deep roots.
The research described in the present paper was conducted at Embu, in the Central Highlands of Kenya, where improved dairy cattle, and increasingly also dairy goats, are raised for milk production, mainly on smallholdings (Paterson et al. 1998). Increasing
human population in this region has led to high pressure on land, with animals managed almost exclusively under intensive zero-grazing systems and fodder produced in small intensively managed plots (Nyaata et al. 2000). Such smallholder dairy production in
Kenya makes a direct contribution to human nutrition, as well as to poverty alleviation through income generation, and increased potential for foreign exchange earnings from export. A common constraint in this region, however, is lack of high quality forage, especially during the dry season because the most common basal feed, Napier grass (Pennisetum purpureum) is depleted during this period (Minae & Nyamai 1988). Crop
residues such as maize stover, bean husks, and banana leaves and pseudostems are used as substitutes for Napier grass, but these are low in digestibility, fermentable energy, crude protein and several essential minerals including calcium (Nangole et al. 1983; Phiri et al.
1992). To provide the protein necessary for high milk yields and good growth rates, farmers in this area therefore try to supplement dairy cattle diets with commercial concentrates, especially during the dry season. However, the increasing price of commercial concentrates (which has not been matched by increases in milk price) has put them out of the reach of many smallholders and this, together with the variable quality of the available concentrates, has led them to seek inexpensive alternative supplements.
Calliandra calothyrsus Meissner („calliandra‟), a tree legume native to Mexico
and Central America, was first introduced to the Central Highlands of Kenya in 1987, and has since been widely promoted and adopted as a supplement to Napier grass (Wambugu et al. 2001). The promotion of calliandra was initially based primarily on its promising agronomic attributes which include, in addition to nitrogen fixation, fast growth and high biomass production of both foliage and wood even during the dry season, tolerance of repeated lopping, and tolerance of soils with low pH and/or high aluminium saturation (Palmer et al. 1994; Chamberlain 2001). More recently, its increasing importance as fodder in many parts of the tropics has focused attention on its nutritive value. Although high in protein (up to 280 g/kg, Kaitho et al. 1993), reports of very high levels of
proanthocyanidins (up to 200 g/kg, Jackson et al. 1996; Stewart et al. 2000) combined
with relatively low dry matter digestibility (Merkel et al. 1999) have raised questions
regarding its value as a feed. Nevertheless, there has already been a high level of calliandra adoption by farmers in the Central Highlands. On-farm trials with dairy animals in the Embu area have suggested that fresh calliandra can substitute for an equivalent amount of concentrate („dairy meal‟), in terms of dry matter, without affecting milk production (Paterson et al. 1999), i.e. that these supplements have a comparable feed
The present study aimed to identify ways both to increase the productivity of calliandra and to optimize its feeding value for ruminants. Several factors were investigated: genetic (provenance) variation, cutting frequency, seasonal variation and post-harvest treatment (feeding the leaves fresh or wilted).
Genetic variation in calliandra has already been studied under a programme of genetic resources research at the Oxford Forestry Institute, OFI (Pottinger 1996). Comprehensive exploration of populations of calliandra in its native range (Central America and Mexico, Macqueen 1992) led to collection of seed from 50 natural
populations (provenances), of which a subset was evaluated in a pantropical trial network co-ordinated by OFI. Most calliandra currently planted outside the native range, including Embu, originates from material introduced from Guatemala to Java in 1937 (Verhoef 1939). This land race material, in which there was no formal selection for quality or yield, was compared with the native populations in many of the trials. Across the network, the overall highest-yielding provenance was from San Ramón, in Nicaragua, but the Indonesian material was also, fortuitously, among the highest yielders (Pottinger & Dunsdon 2001).
Three populations were included in the present study: San Ramón; the local Embu land race; and Patulul provenance from Guatemala, which was also high-yielding in the network trials. Isozyme studies (Chamberlain 1998) have shown Patulul and Embu to be genetically very similar while San Ramón occupies the opposite end of the range of genetic variation within the species. A central question addressed by the present study is not only whether these genetically very distinct provenances also exhibit important variation in quality traits, but also whether they differ in their responses to environmental variation and management.
Provenance variation in nutritive value has been quite widely studied in the laboratory (Duguma & Mollet 1997; Larbi et al. 1998; Premaratne & Perera 1999), mainly
in terms of in vitro digestibility and levels of crude protein (CP), acid and neutral detergent fibre (ADF, NDF) and proanthocyanidins, but always using samples grown under uniform conditions. Previous studies have not, however, included direct provenance effects on animal production.
A major issue in the domestication of calliandra has been the effect of wilting or drying of the leaves on their nutritive value. In Kenya as in many parts of the tropics, fodder trees are used in cut-and-carry stall-feeding systems where, in practice, fodder is usually cut several hours beforehand and is wilted or dry by the time it is fed. Some early studies reported lower intake and dry matter digestibility, in vitro and in situ, for dried or
wilted calliandra than for fresh material (Mahyuddin et al. 1988; Palmer & Schlink 1992),
and this effect seemed to be consistent to a greater or lesser extent across provenances (Palmer & Ibrahim 1996). There is also conflicting evidence, however, suggesting that the feed quality of calliandra may be unaffected, or even improved, by drying, both in vitro
(Stewart et al. 2000) and in terms of voluntary intake and apparent nitrogen digestibility in
vivo (Ahn et al. 1989; Norton & Ahn 1997). In view of these apparent contradictions, there was a need for feeding trials to look directly at the effect on animal performance of wilting calliandra before feeding it.
MATERIALS AND METHODS
There were two distinct components to the research:
(i) An agronomic field trial which was harvested eight times over a 1-year period.
Leaf and wood biomass production were estimated at every harvest, and samples
collected for laboratory analysis to determine effects of provenance, cutting
interval and seasonal variation.
(ii) A series of three feeding trials in which sheep and dairy goats were fed material
from large fodder blocks to measure directly the effects of provenance and leaf
wilting on small ruminant production.
The study was conducted at the Kenya Agricultural Research Institute (KARI) Regional Research Centre, Embu, located in the Central Highlands of Kenya, on the southeastern slopes of Mt. Kenya, at an altitude of 1480 m above sea level. The soils are red Kikuyu loams, mainly comprising humic Nitisols derived from basic volcanic rocks. In the USDA system of classification they fall under humic Palehumult (USDA 1975). They are deep and well weathered with a friable clay texture and moderately high inherent fertility.
Rainfall is moderate with an annual total of 1200-1500 mm (Jaetzold & Schmidt 1983). Rainfall is bimodal with the long rains occurring between mid-March and June and usually providing 650-750 mm. The short rains are from mid-October to December and range from 350 mm to 550 mm. In addition there is a cold misty period in July-August with light and unpredictable rainfall ranging from 50 mm to 250 mm. Mean monthly temperature ranges from 17? to 22?C. Mean monthly rainfall and temperature data at Embu for the period of the trial are given in Fig. 1.
The agronomic trial and the fodder blocks for the feeding trials were both established at this site. Both included three provenances of C. calothyrsus: San Ramón
from Nicaragua, Patulul from Guatemala and the local Embu land race (derived from Indonesian material).
A split plot design with four replicates was used, with two cutting frequencies (6 and 12 weeks) as the main plots and the three calliandra provenances as the subplots. The trees were raised in containers, and planted out in square plots at 1 m ； 1 m spacing (36 trees
per plot). An initial uniformity cut (to 1 m height) was carried out 6 months after planting, followed by eight biomass harvests every 6 weeks for the next year. The 6-week frequency treatments were cut at every harvest, and the 12-week ones at the 2nd, 4th, 6th and 8th harvests. An additional, 9th harvest was used for chemical analysis only (to give compositional data spanning a full year).
Data collection and sample preparation
Biomass was measured directly at each harvest for the 16 central trees in each plot (guard rows also cut). The trees were cut at a height of 1 m and biomass divided into leafy (leaves and green parts of the shoot) and woody biomass (lignified material). Fresh weights were measured directly, and subsamples were dried to constant weight at 103?C for determination of dry matter content. In addition, samples of entire mature leaves were taken from the 16 central trees in each plot, bulked, freeze dried and ground through a 1 mm sieve for analysis.
CP content was determined by the macro Kjeldahl procedure (AOAC 1995). NDF was determined by the method of Van Soest et al. (1991), and in vitro dry matter digestibility
percentage (IVDMD) by the method of Tilley & Terry (1963).
Proanthocyanidin (PA) concentrations in “extractable” and “bound” fractions were
estimated by a modified version of the method of Terrill et al. (1992), with protein- and
fibre-bound fractions combined. Four subsamples of 0.025 g of dried leaf were extracted twice with a mixture of 4 ml of acetone/water (7:3, v/v), containing 1 g/litre ascorbic acid, and 2 ml dichloromethane, the latter to remove lipids and pigments. The aqueous phases from the two extractions were combined and made up to 10 ml with water. PA concentration in this fraction was estimated by the butanol-HCl method (Porter et al. 1986;
Terrill et al. 1992). Of this extract, 0.5 ml was added to 6 ml butanol-HCl (95:5, v/v) and heated for 1 h at 95;C. The absorbance of the red anthocyanidin products was read at 550 nm. PA remaining bound to protein and fibre after the extractions was estimated in the same way, with 3 ml butanol-HCl added directly to the solid residue after residual solvents (dichloromethane, acetone) had been evaporated in a stream of nitrogen.
To convert absorbance values to PA concentrations, crude PA standards were prepared for each provenance separately, using Sephadex LH-20 (Stewart et al. 2000).
These standards were included in each assay at known levels (0.1 mg, 0.2 mg, 0.4 mg and 0.8 mg per tube, all in triplicate). Blanks were prepared separately for each fraction using the plant extracts and the same process but with butanol-HO (95:5, v/v) replacing 2
The relative proportions of cyanidin and delphinidin in the products of the butanol-HCl reaction were estimated by high performance liquid chromatography (HPLC), using a method developed at CIAT, Cali, Colombia (C.E. Lascano, pers. comm.; Stewart et al. 2000). This gives basic information about structural differences between the proanthocyanidins from different treatments, in terms of their constituent subunits, since the assumption is made that the proportions of procyanidin and prodelphinidin in the proanthocyanidins is the same as that of their respective end-products, cyanidin and delphinidin, in the butanol-HCl reaction.
A split-plot analysis of variance (ANOVA) was used to analyse variation in each trait measured (biomass, IVDMD and compositional variables). The main plot was cutting frequency and the sub-plot was provenance. The AREPMEASURES procedure (Genstat 5 1998) was used to analyse changes in treatment effect over time. This carries out an ANOVA on the repeated measurements and checks the assumptions required (i.e. symmetry of the variance-covariance matrix). To adjust for partial failure of the assumptions it calculates a correction factor (Greenhouse-Geisser) for the degrees of freedom in the block.cutting frequency.provenance.time stratum. The correction increases
the rigour of the F-tests in this stratum.
The analysis examined the data in terms of pairs of harvests, i.e. comparing each 12-weekly harvest with the two 6-weekly harvests carried out in the same period. For biomass, the production from the 6-weekly cuts was summed to compare with the biomass from the 12-weekly cut (because cumulative totals are of interest). For all the other variables, mean values from each pair of 6-weekly cuts were compared to the 12-weekly values. This analysis provided information not only about the effects of cutting frequency and provenance, and interactions between them, but also how these may change over time.
Feeding trials 1 & 2 (lamb growth)
Two feeding trials were conducted to determine the effects of provenance and drying on voluntary intake and live-weight gain (LWG) in growing lambs. The experimental treatments were fresh and wilted leaves of each of two provenances: San Ramón and a Patulul/Embu composite. Patulul and Embu material were combined not only because they had been shown by an earlier isozyme study to be genetically very similar (Chamberlain 1998), but also because early results from the agronomic trial had shown minimal differences between these two seed sources. The fodder blocks for the trials were planted 1 year before the start of the trials. The „wilted‟ leaves were cut 24 h before feeding and left to dry; the extent of drying inevitably varied with the weather. An additional treatment, reflecting usual current practice in the area, was also included as a positive control. In this, commercial concentrate („dairy meal‟) was used instead of calliandra.
The two trials were conducted in the wet season (October-December: “short rains”)
and dry season (February-April) respectively. In both experiments, calliandra leaves (including edible, unlignified green stems) or dairy meal were offered ad libitum, by
initially offering the supplement as 0.25 of the diet, and thereafter offering the previous day‟s intake plus 0.1. The main difference between the two trials was the basal diet, which consisted of Napier grass (Pennisetum purpureum Schumach.) in the first experiment
(Trial 1), and maize (Zea mays L.) stover in the second (Trial 2), representing typical local wet and dry season feeding practices. Both the basal feeds were chopped manually into 2-4 cm lengths before feeding. In Trial 1, calliandra was preferred over Napier grass, particularly when wilted, so in these cases voluntary intake of calliandra exceeded the intended 0.25 of total dry matter intake. In Trial 2, intake of stover was so low that the proportion of supplement in the diet rose to over 0.5. The amount of calliandra offered had to be restricted to this level, and as the animals always ate all the calliandra offered, it was therefore not possible to estimate voluntary intake of calliandra in this experiment.
The same forty Corriedale lambs were used in both trials. At the beginning of Trial 1 their live-weight ranged from 14 kg to 20 kg. They were divided into five groups of similar mean live-weight (18 kg), and the five treatments were randomly allocated to these groups.
The animals were individually penned during the trials, with water and mineral licks supplied ad libitum. Each trial included 14 days‟ initial adaptation period followed by 10 weeks of data collection. The amount of feed offered was adjusted weekly, based on the previous week‟s consumption plus a 0.1 surplus. All the basal feed was offered in the
morning (for the Napier grass, within 30 min of cutting), and the supplement was offered in two portions in the morning and afternoon, to ensure freshness in the relevant treatments. Total intake was measured daily by weighing the total refusals. Once a week, the refusals were separated into their components to estimate the proportions of basal feed and supplement consumed by each animal. The lambs were weighed weekly, and average daily LWG estimated by regression.
Trial 2 began after a 4-week rest period during which the lambs were maintained on Rhodes grass (Chloris gayana Kunth.) pasture. The animals were re-assigned to new
groups for Trial 2, with average live-weight 23.5 kg (range: 19.5-27.0 kg). Observations were as for Trial 1, although one treatment, wilted San Ramón, had to be omitted from the second trial owing to shortage of leaf material during the dry season. The animals were therefore divided into four groups of 10 animals each, and these were randomly allocated to the treatments.
In both trials, CP and NDF contents of each component of the diet were determined in triplicate samples. For Napier grass and the two calliandra provenances, samples were taken four times over the course of the trials, whereas the dry stover and dairy meal were assumed to be of constant composition so were each sampled only once (also in triplicate).
Data were analysed using the ANOVA procedure in Genstat (Genstat 5 1998). As well as estimation of overall treatment effects on intake, LWG and feed efficiency, the following orthogonal contrasts were made: dairy meal versus calliandra, fresh versus wilted calliandra, Patulul/Embu versus San Ramón provenance, and the interaction between physical state (fresh/wilted) and provenance (if this interaction is significant, it shows that the provenances are responding differently to wilting). In Trial 2, owing to the absence of the wilted San Ramón treatment, the contrasts were modified to non-orthogonal comparisons: dairy meal versus calliandra, fresh Patulul/Embu versus fresh San Ramón, and fresh Patulul/Embu versus wilted Patulul/Embu.
Feeding trial 3 (goat lactation)
Trial 3 estimated the effects of provenance and drying on intake and milk production in dairy goats. Sixteen lactating goats (local ； German Alpine first generation crosses) were
included in an unbalanced crossover design with five treatments and three periods. The goats were oestrus-synchronized, and all kidded within a 2-week period. The trial started when the kids were aged 2-4 weeks, and was conducted at the end of the “long rains”
(June-August). Each period comprised a 14-day adaptation period followed by 7 days of data collection in which voluntary intake and milk yield were measured. There were 14-day rest periods between periods when the goats were grazed on Rhodes pasture. During the trial periods the animals were individually penned, separately from the kids, who did not have any access to the does during the trial.
The five treatments were the same as in Trial 1, as was the composition of the diet, the amounts of feed offered (previous ad libitum intake plus 0.1), and the daily feeding
regime. Again, water and mineral licks were available ad libitum.
The data were again analysed using the Genstat ANOVA procedure. First, the significance of the carryover effect between periods was tested. This having been found to be non-significant (P=0.559), milk yield, total dry matter intake, and intake of supplement were subjected to analysis of variance with treatment and period as factors. In addition, the same specific, pre-defined comparisons were made as in Trial 1.
The provenance means for leaf and wood biomass production, IVDMD and the four compositional variables (CP, NDF, extractable PA and bound PA) are shown in Table 1. The same table also shows the percentage of cyanidin in the products of the butanol-HCl reaction. The HPLC showed that the anthocyanidins produced from calliandra tannins by the butanol-HCl reaction consist almost exclusively of cyanidin and delphinidin, possibly with traces of pelargonidin. The percentage of prodelphinidin subunits in calliandra PA can therefore be expressed as (100 - % procyanidin).
In Table 1, the values for leaf and wood biomass production show total production over the 48-week period of the experiment under the two cutting regimes, because the trait of interest is cumulative biomass production over consecutive harvests. For the quality traits, however, the values shown are means, also across the whole period of the experiment.
The effects of provenance, cutting frequency, and the interaction between them, are shown for each trait in Table 2. There is a significant difference between provenances in every trait estimated. While San Ramón provenance gave the highest leaf biomass production, Embu and Patulul provenances were superior in terms of the leaf quality traits CP (higher), NDF (lower) and IVDMD (higher). Surprisingly, given that high tannin content generally tends to reduce dry matter digestibility, PA content (both extractable and bound) was lowest in San Ramón provenance. There is also a major structural difference in the PAs from San Ramón compared to those from the other two provenances, with those from San Ramón provenance comprising mainly prodelphinidin tannins, and the other two provenances mainly procyanidin tannins.
Over the year during which biomass was harvested, frequent (6-weekly) harvesting generally gave higher total leaf production than 12-weekly cutting, although the opposite was true for the first pair of harvests (data not shown), and differences between these two treatments was small (Table 1). Wood biomass production, however, was greatly reduced by frequent cutting, with the 12-week cycle producing almost four times as much wood as the 6-week cycle in the course of the year (Table 1), and the latter producing little or no wood at several harvests (Fig. 3). In contrast, the quality traits showed little consistent variation with harvesting regime. The effect of cutting frequency was only (barely) significant for NDF (P=0.022, Table 2). As might be expected, younger
leaves (6-weekly harvest) were generally lower in fibre.
In most traits, provenance effects were unaffected by cutting frequency. The only exception was woody biomass, for which this interaction was highly significant (P<0.001):
over the first two cutting cycles, San Ramón produced much more wood than the other provenances under the 12-week cutting regime, while all three provenances produced similar amounts under 6-weekly cutting.
All the traits measured, except extractable PA, showed significant variation over the course of the year in the analysis of pairs of harvests (“time” in Table 2). Figures 2-9
show seasonal variation in more detail (by using the 6-weekly harvest data only) for all nine harvests. Figure 7 shows that extractable PA did vary greatly from one harvest to the next, but this effect was obscured when values were averaged over pairs of harvests in the
repeated measures analysis. These figures show no consistent pattern of variation among the different traits. Productivity was reduced at Harvests 6 and 7 (July-August) during cool, dry weather (Fig. 1). Leaf quality was also generally reduced at Harvest 6, with high NDF, low CP and IVDMD, and high PA, particularly in bound form, but at Harvest 7 there was an overall improvement of quality despite similar climatic conditions.
Provenance differences generally remained consistent over time (time ；
provenance interactions were all non-significant, except for % procyanidin). The effect of cutting frequency changed over time for some traits but not others: the time ； cutting
frequency interaction was only significant for biomass, NDF and extractable PA. For biomass, the longer (12-week) cutting interval gave higher leaf production initially, but in later harvests 6-weekly cutting was more productive (data not shown). For NDF, although more frequent cutting generally gave lower levels of NDF, this effect was not observed over the first two harvests; and for extractable PA, the effect of cutting frequency on PA level showed no consistent pattern across harvests.
Feeding Trial 1:Napier grass as basal feed for growing lambs
Voluntary intake of Napier grass and supplement (i.e. C. calothyrsus or commercial
concentrate), total intake (all expressed as g dry matter (DM) per kg metabolic body weight), average daily LWG in g (estimated by regression), and feed efficiency (g mean LWG/g mean intake) of the supplement and of the total diet, are shown in Table 3 for the five experimental treatments in Trial 1. One-way analysis of variance (ANOVA) showed highly significant effects of treatment for all these traits (P=0.002 for Napier intake;
P<0.001 for all other traits). Of specific interest in the context of this trial are the differences between Patulul/Embu and San Ramón provenances, between fresh and wilted leaf material of each, and between dairy meal and the different calliandra treatments, as described above. The significance levels of the contrasts between these treatments are shown in Table 3.
Intake of leaves of the two provenances (as the supplementary feed in the diet) did not differ (P=0.826). Feeding San Ramón rather than Patulul/Embu increased the intake of the basal feed (P=0.004) and hence total intake (P=0.009). This difference in intake was
more pronounced for the fresh calliandra than for the wilted material. LWG was much higher with the Patulul/Embu composite (P<0.001), whether fresh or wilted, and this
resulted in higher feed efficiency, particularly in the case of fresh material.
Intake of wilted calliandra leaves was higher for both provenances than that of fresh leaves, and this was reflected in higher total intakes (P<0.001), although intake of
the basal feed was unaffected by the state of the supplement (P=0.108). Wilting, however,
had no effect on animal production. For Patulul/Embu, the combination of lower intake and equivalent animal production for fresh material meant that fresh leaves showed higher feed efficiency. For San Ramón, however, feed efficiencies of fresh and wilted leaves were similar.
The comparisons between dairy meal and the various forms of calliandra consistently showed lower intake of both Napier and supplement, coupled with higher live-weight gain, for lambs fed with dairy meal. The differences in Napier intake were more pronounced with San Ramón than with Patulul/Embu: lambs supplemented with fresh Patulul/Embu consumed similar amounts of Napier to those supplemented with dairy meal. Contrary to expectations (see Introduction) the feed efficiency was much higher for dairy meal in all cases (P<0.001).
Feeding Trial 2: maize stover as basal feed for growing lambs
Intake and LWG data for Trial 2 are given in Table 4. One treatment (wilted San Ramón leaves) was omitted owing to shortage of material. In this trial, the supplement was not given ad libitum because it was always completely consumed in preference to the stover. The proportion in the diet was limited to around 0.5, which although very high by normal
standards, was felt to be necessary to maintain the animals‟ health given their very low voluntary intake of stover. Because of this, only total intake, LWG and feed efficiency were analysed statistically in this trial. The significance levels of specific contrasts are also shown in Table 4.
Unlike that of Trial 1, total intake was unaffected by treatment in this experiment (P=0.347). Despite similar (fixed) intakes of the different supplements, however, the treatments had a highly significant effect on LWG (P<0.001). Wilted Patulul/Embu gave
significantly higher feed efficiency than fresh Patulul/Embu, whilst fresh Patulul/Embu gave significantly higher LWG and feed efficiency than fresh San Ramón (P<0.001 for
all). Fresh San Ramón gave very poor production: only three of the eight lambs showed daily gains over 5 g, and three lost weight over the trial period. In contrast all but one of the lambs fed fresh Patulul/Embu showed a daily gain of at least 20g. Dairy meal showed significantly greater LWG and feed efficiency than the calliandra treatments.
Feeding Trial 3: goat lactation
The intake variables estimated were Napier grass (the basal diet), supplement and total DM. The response variable was milk yield. These data are presented in Table 5. Analysis of variance, with treatment and period as factors, showed a significant treatment effect for Napier intake (P=0.022) and highly significant treatment effects (P<0.001) for the
The orthogonal contrasts again showed that while intake of dairy meal (and corresponding total intake) was significantly lower than intake of calliandra (P<0.001),
milk production was significantly higher with dairy meal than with calliandra (P<0.001).
In this trial, however, provenance had no significant effect on either intake or milk production (P>0.05 for all variables). Wilting had a negative effect on Napier intake (P=0.045) and total intake (P=0.016). However, milk production was unaffected by
Mean CP and NDF contents of the components of the diets in all three trials are presented in Table 6. The differences between provenances are similar to those observed in the agronomic trial, although all the CP values are somewhat lower.
The combination in Patulul and Embu provenances of higher protein content and IVDMD, together with lower NDF content, is consistent with the significantly higher LWG and feed efficiency achieved with this provenance composite in both the sheep production trials. The results of these experiments, taken together, give a sufficiently clear indication of the superior fodder quality of these two provenances, even though no differences were detected in the goat lactation trial. The extremely poor performance of lambs fed San Ramón provenance as a supplement to stover in Trial 2 was particularly striking (although it is unfortunate that only one San Ramón treatment could be included in that trial). In Trial 1, although voluntary intake of San Ramón and Patulul/Embu leaves was similar, supplementation with San Ramón seemed to stimulate consumption of Napier grass, resulting in higher total intake. However, despite this higher intake, LWG was still significantly lower for animals fed San Ramón leaves. In contrast to these results with Napier, fresh San Ramón did not appear to enhance intake of maize stover, the basal feed in Trial 2. The very inferior performance of San Ramón in Trial 2 may reflect the smaller difference between the provenances in digestibility during the short rains (October –
December: Fig. 6), when Trial 1 was conducted, than in February – April, the period of