Forage potential of some native annual Trifolium species in the Ethiopian highlands

Abstract
Materials and methods
Results and discussion
Summary and conclusions
Acknowledgements
References

Levi M.S. Akundabweni
International Institute of Tropical Agriculture
Oyo Road, PMB 5320
Ibadan, Nigeria

Abstract

The intraspecific variability of 34 accessions of Trifolium tembense from the Ethiopian highlands was studied. The effects of planting and harvesting dates and phosphorus application on the dry-matter yield, seed production, leaf tissue mineral contents and forage quality of nine native and three introduced clovers were examined.

Marked genetic variability was found both within and between the T. tembense accessions. The soil characteristics of the site of origin did not appear to have imposed any ecotypic variation. Planting the clovers in March, during the short rainy season, extended the growing season and gave higher yields than planting in June, during the long rains. P fertilization substantially increased both dry-matter and seed yields but had little effect on forage quality.

INTRODUCTION

There are 240 to 300 clover species of the genus Trifolium distributed around the world, and clovers are ubiquitous in natural grasslands in cooler climates (Evans, 1976; Allen and Allen, 1981). Although clovers rank second in forage productivity and feeding value to Medicago sativa in western countries, few Trifolium species are cultivated due to their recent domestication (Davies and Young, 1967).

While the Mediterranean region is the primary centre of diversity of clovers, there are a number of secondary centres of diversity in Africa, including central Ethiopia, Somalia, Tanzania and Kenya (Zohary, 1972; Thulin, 1982).

Forty Trifolium species are reported to be found in sub-Saharan Africa, of which about 25% are endemic to Ethiopia (Gillet, 1952; Thulin, 1982). About two-thirds of the clover species in Ethiopia are annuals and the remainder are perennials; biennials are not found in Ethiopia.

Livestock are important in subsistence agriculture, providing draught power, meat, milk, hides and dung for fuel and fertilizer. However, there are marked constraints to production, the most important being insufficient forage and poor forage quality, particularly during the dry season. During this period animals are mainly grazed on marginal lands and are fed cereal straw. Due to the current lack of importance of planted fodders in the prevailing subsistence farming systems of highland Africa, clovers are little exploited. There are, however, a number of ways in which they can be used to provide more and better-quality feed during the dry season.

Clovers grow rapidly when the growing conditions are favourable and produce a large amount of dry matter. This can be conserved as hay, which can be used to increase the quality of straw-based diets and to overcome seasonal feed shortages.

Some species grow naturally in valley bottoms which are not cultivated due to seasonal waterlogging. Thus they can be managed to increase forage production without competing with food crops for land. Some of the African clovers are adapted to acid and low-P soils (Norris, 1965; Andrew, 1976).

Similarly, clovers can be grown on land that would otherwise lie fallow as part of a crop rotation system. Their ability to fix atmospheric nitrogen will help to improve the fertility of the soil and thus increase the yields of succeeding crops. Large tracts of land are fallowed each year in Ethiopia.

Forage legumes also generally contain a large proportion of crude protein, and are thus a source of high-quality feed for livestock.

Studies were conducted in the Ethiopian highlands to describe, evaluate and screen some annual Trifolium species for their forage potential. The details of the work are published elsewhere (Akundabweni, 1984).

Materials and methods

The experiments were conducted on a low-P (1 ppm) Vertisol with a pH of 6.0 at ILCA headquarters. The site is at 2370 m a.s.1. and receives 1250 mm average annual rainfall. The mean annual temperature is 15.5° C.

Experiment 1

Germplasm of Trifolium tembense was collected in 1982 from along four axes in the Ethiopian highlands:

1. Northeast from Addis Ababa (AA)
2. Southeast from AA
3. Southwest from AA; and
4. West from AA, plus the southeastern region.

The exploration design was based partly on the work of Krajina (1977). Collections were made from a few plants at each site. A soil sample was collected from each collection site and analysed for P and organic-matter contents and for pH.

Thirty-four accessions were selected from the germplasm collection, and seeds were sown in disposable polythene tubes in a screen-house at the beginning of June 1983. The plants were transplanted into the field at Shola, near Addis Ababa, on 26 July 1983. Each accession was space-planted in 4-m rows, with two replicates in a randomised complete block design. Interrow spacing was 40 cm.

Experiment 2

In 1982 an experiment was conducted to examine the dry-matter accumulation of the two annual clovers that are most common in the Ethiopian highlands, T. tembense (ILCA 6278) and T. rueppellianum (ILCA 5791) and two introduced clovers, T. fragiferum (ILCA 6280) and T. resupinatum (ILCA 7022) during the main cropping season.

The experiment was planted on 7 June 1982 in a randomised complete block design with four replicates. There were three plots of each species per replicate, with the plots consisting of five 4-m rows with interrow spacings of 40 cm. The seeds were scarified before planting, and were sown at 9, 8, 5 and 5 kg/ha for T. rueppellianum, T. tembense, T. resupinatum and T. fragiferum, respectively, these rates giving roughly the same number of seeds per hectare. The seeds of the two exotic species (T. resupinatum and T. fragiferum) were inoculated with type B rhizobia.

Dry-matter yields were determined 75, 90, 105, 120, 135 and 150 days after sowing. On each harvest date, a sample was taken from a 0.6 m quadrat in the centre three rows. The plants were cut 2.5 cm above ground level. The herbage was oven dried at 65 C for 72 hours for dry-matter estimation.

In 1983, the experiment was modified:

1. Seven additional clover species/ecotypes were included: T. quartinianum (ILCA 6278), T. quartinianum (ILCA 6777, ecotype 2), T. steudneri (ILCA 6253), T. decorum (ILCA 6303), T. polystachyam (ILCA 6298), T. baccarinli (ILCA 6294) and an introduced cultivar, T. subterraneum cv. Northern (ILCA 7023).

2. The trial was planted on two dates: (a) 15 March, during the short rains, and (b) 15 June, during the long rains, to investigate the effect of early planting on productivity of the clovers.

3. Dry-matter yields were determined 130 and 165 days after sowing, and at the end of the growing season.

The trial was laid out in a split-split-plot design with four replicates, with species as the main-plot treatments, the two planting dates as subplots and harvesting dates as sub-subplots. Plot size was 2 m x 3 m, comprising five rows 40 cm apart. Triple superphosphate was banded below the seed at a rate of 10 kg P/ha. The seed rates of the different species were adjusted to give the same number of plants per unit area as 8 kg of T. tembense seed/ha.

Seed treatment, planting, harvesting and determination of dry-matter yields were carried out as in the 1982 experiment.

Experiment 3

Concurrent with experiment 2 in 1982 a study was made of the effect of P applied at different rates on the dry-matter yields of T. rueppellianum, T. tembense and T. fragiferum.

The trial was planted adjacent to experiment 2 on the same day. The trial was laid out in a randomised complete block design with four replicates. Clover species and P rates (0, 5, 10, 20 and 35 kg P/ha, banded below the seed) were combined factorially. Plot management regimes were similar to those of experiment 2 of 1982.

In 1983 the trial was expanded to include nine species of clover: T. quartinianum, T. tembense, T. decorum, T. steudneri, T. schimperi (ILCA 6290), T. rueppellianum, T. resupinatum, T. subterraneum and T. alexandrinum (ILCA 6810).

The trial was planted on 21 June 1983 in a split-split-plot design, with species as main plots, fertilizer treatments as subplots and harvesting dates as sub-subplots. The subplots were 2 m x 4 m, within which samples were taken from 0.6 m quadrats 75, 90, 105, 120 and 135 days after sowing. Seed rates and treatments and dry-matter yield determination were as previously describes Seed yields was also determined in the 0, 10 and 35 kg P/ha treatments. Samples for seed yield determinations were taken from two 90-cm segments in each guard row in each plot.

Experiment 4

Leaf tissues of plants in experiment 2 were sampled just prior to flowering, according to recommended procedures (Wolf, 1983; Martin and Matocha, 1968). The samples were analysed for P and N contents. Duplicate samples from dried whole plants were also analysed for percentage dry matter content, ash, calcium, neutral detergent fibre (NDF%), acid detergent fibre (ADF%), and lignin, whole-plant P and crude protein (CP) contents according to recommended procedures (Cottenie, 1980).

Results and discussion

Experiment 1

The accessions showed variations in flowering dates and highly significant differences (P<0.01) in branching, height, stem length and stem thickness.

Intraclass correlations of some morphological traits (leaf mark, plant height, leaf width, number of trifoliate leaves) for individual plants (I_iws) were substantially higher than those for accessions (I_s) (Table 1), which indicates that there may be genetic heterogeneity within accessions.

When soil characteristics (P and organic-matter contents and pH) of samples from the germplasm collection sites were subjected to discriminant statistical analysis (data not shown) they were shown to be unique to the axes from which they were collected. However, a similar analysis of the accessions gave imperfect allocation of accessions to the axes from which they were collected. For example, six

Table 1 Variance components for the morphological traits and their intraclass correlations as a measure of variability among T. tembense accessions (experiment No 1)

Trait	Qp2	Q2iwp	Ip	Iiwp
Leaf mark	188.60	431.62	0.30	0.69
Plant height	2.44	13.91	0.15	0.86
Stem length	8.15	7.95	0.46	0.45
Petiole length	1.78	1.40	0.41	0.32
Branching	2.59	4.13	0.38	0.60
Leaf width	0.03	0.04	0.34	0.42
Stem thickness	0.05	0.19	0.18	0.71
Leaf length	0.07	0.13	0.08	0.85
Leaf width	0.01	0.19	0.05	0.94
No of trifoliate leaves	0.00	0.34	0.05	0.80

Note
Q_p² - Population variance (different accessions);
Q²_iwp - Individual plants within populations;
I_p - Population intraclass correlation;
I_iwp - Individuals intraclass correlation iwp accessions (ILCA 8285, 8106, 8258, 8142, 8198 and 8612) that were collected east of the Rift Valley were incorrectly classified into the collection axes lying to the west of the Rift Valley (Table 2) Thus, under the conditions of the trial it was not possible to show ecotypic variation in tembense.

Experiment 2

There were significant (P<0.01) species by harvesting date interactions on the dry-matter yields (Table 3) The dry-matter yields of the two native clovers, T tembense and T. rueppellianum, peaked at 120 days after sowing and then declined, possibly due to leaf loss. In contrast, the introduced species, T. fragiferum and T. resupinatum, continued to grow up to the final harvest 150 days after sowing.

Table 2. Accessions misallocated to other collection axes due to overlapping morphological classification criteria as determined by discriminant statistical analysis and as an index of the absence of uniqueness of a particular population to its site of origin (Experiment 1).

Accession(ILCA No.)	From region	Misallocated into region	Posterior probability for misclassification
8285	45	67	0.578
8106	45	67	0.615
8612	45	81	0.921
8198	45	12	0.538
8142	45	12	0.394
8189	45	67	0.633
8258	45	81	0.458
8000	56	12	0.251
8764	56	67	0.365
8758	56	45	0.411
8001	12	81	0.528
8541	81	56	0.307

Note: Prior probability = 0.200 is less than the posterior probability values indicative of the 'strength' of misclassification (e.g. 0.578 is 57.8% misclassification) among accessions.

It is possible that the early cessation of growth in the native clovers is an adaptive feature, which could be related to soil moisture availability or to photoperiodicity. This was investigated in 1983 using two planting dates.

Results from the 1983 trial show that planting the clovers in March (during the short rains) extended the length of the growing season of the native annual species- by 15 to 87 days and increased the dry-matter yields by 75 to 640%(Table 4). The highest dry-matter yields in the trial were produced by the early plantings of T. quartinianum, T. tembense and T. rueppellianum. The species by planting date by harvesting date interactions (second order) were highly significant (P<0.01). This was due possibly to large differences in yields among species in relation to their degree of adaptation (native vs introduced) and to seasonal environmental effects, e.g. the early-planted crops experienced less waterlogging of the soils in the early stages of growth (300 vs 800 mm rainfall), higher temperatures (23° vs 15° C mean maximum) and more sunshine (8 vs 3 hours of sunshine per day) than the late-planted plots.

Table 3. Mean dry matter yields during the long rains at Shola under natural fertility (Experiment 2), 1982.

Species	Days after sowing
	75	90	105	120	135	150
Controls	kg/ha
T. fragiferum	35	69	116	253	232	395
T. resupinatum	168	299	684	1331	1504	2053
Native
T. tembense	274	237	1276	1626	1035	713
T. rueppellianum	135	353	769	1619	814	873

Experiment 3

In the 1982 P application studies, both the species by fertilizer rate and the species by date of harvest interactions were highly significant (P<0.01). The introduced clover, T. fragiferum, had a much smaller dry-matter yield response to applied P than did the native clovers (Table 5).

Table 4. Dry-matter yields from several clover species grown at two sowing dates in 1983 at ShoIa, near Addis Ababa, Ethiopia.

Species or line	ILCA No.	Sowing date	Days to harvest		End of growth	Days to end of growth
			130	165
			DM yield (t/ha)
T. quartinianum	6301	15/3	3.3	5.0	6.3	233
		15/6	4.9	4.2	3.6	155
T. steudneri	6253	15/3	2.9	4.2	4.8	183
		15/6	2.6	2.5	2.1	136
T. decorum	6303	15/3	2.2	4.8	3.9	220
		15/6	2.9	2.4	1.8	136
T. rueppellianum	5791	15/5	2.7	4.5	5.1	230
		15/6	2.0	1.6	1.5	143
T. tembense	6278	15/3	1.8	3.3	5.8	217
		15/6	2.5	1.9	1.6	136
T. quartinianum	6277	15/3	2.1	2.7	2.6	170
(Ecotype 2)		15/6	1.1	1.1	1.2	155
T. polystachyum	6298	15/3	0.6	2.9	3.7	241
		15/6	0.3	0.1	0.5	160
T. resupinatum	7022	15/3	1.5	2.9	2.4	183
		15/6	0.2	0.1	0.2	190
T. subterraneum	7020	15/3	3.3	0.6	0.4	190
(cv. Northern)	15/6	0.8	0.5	0.7	160
T. baccarine	6294	15/3	1.0	1.6	0.9	198
		15/6	1.3	1.0	0.7	136
T. fragiferum	5280	15/3	0.25	0.15	0.04	356
		15/6	0.04	0.09	0.07	365+

Table 5. Effect of P application on dry-matter yields of three clovers at Shola, 1982 (Experiment 3).

Species	P rates (kg/ha)				Species overall means by W.D.
	0	2	4	10
T. rueppellianum	1305	1491	2094	2904	a
T. tembense	1243	1776	2466	3228	a
T. fragiferum	135	232	260	246	b
P rate overall mean by W-D test	c	bc	ab	a

W-D = Waller-Duncan T-test at K-ratio of 100:1 (P<0.05); Means of the main effects with same letters were not significantly different.

The maximum yields of T. tembense and T. rueppellianum when fertilized with 10 kg P/ha were 4.8 and 4.0 t/ha, respectively, and occurred 120 days after sowing.

There were large dry-matter yield responses to applied P and higher rates of P produced larger yield responses.

The three main effects found in the 1983 P fertilizer trials are shown in Table 6. Regression models were developed for the effect of P on dry-matter yield, based on the 120-day yield data (Table 7).

Optimum rates of P application were found to be in the range of 25 to 30 kg P/ha for the native clovers. At higher rates of P application the increases in dry-matter yield were smaller. In 1984, when elemental P was $1.50/kg and hay was $0.44/kg, these rates were calculated to be the most economical.

The elasticity of response (Dillon, 1968), which indicates the rate of organic-matter synthesis (as an average percentage change of the slope of the response curve per unit of P), was calculated from the quadratic regressions. The elasticity of response value indicated that at 5 and 10 kg P/ha the resultant rate of organic-matter synthesis in native clovers was 60 to 70%. At the optimum fertilizer rates (25 kg P/ha for most of the native species) the ER was only 10%.

Table 6. Mean dry matter yields of several clovers sown at Shola in 1983.

Effects	DM yield(kg/ha)	Waller-Duncan
Species
T. quartinianum	2988	a
T. tembense	2124	b
T. decorum	2102	b
T. steudneri	1727	bc
T. schimperi	1418	cd
T. rueppellianum	1313	d
T. resupinatum	367	e
T. subterraneum	350	e
T. alexandrinum	169	e
Phosphorus kg/ha
0	369	a
5	971	b
10	1519	c
20	1918	d
35	2199	e
Harvest times2
75	496	a
90	943	b
105	1660	b
120	1934	c
135	1944	c

¹Means with the same letters do significantly differ at MSD of 408, 230 and 194 kg DM/ha for species, phosphorus rates and harvest times, respectively.

²Sixth harvest at 150 days omitted due to missing data but DM yields had considerably declined to about 50% of the maximum attained at 120 days.

Table 7. Dry-matter yield response functions due to P rates for various clovers tested at Shola in 1983.

Species	Equation for determining amount of yield	R2
Linear effects	(Y = a + bx)
T. subterraneum	294.6 + 72.5x	0.2408*
T. alexandrinum	69.0 + 48.4x	0.3539*
T. tembense	1363.0 + 613.8x	0.5800*
Quadratic effects	(Y = a + bx + cx2)
T. rueppellianum	201.5 + 947.7x -79.9x2	0.7853*
T. steudneri	755.2 + 1186.6x -98.3x2	0 7321*
T. decorum	765.3 + 1285.7x - 106.9x2	0.7153
T. schimperi	831.2 + 792.5x - 75.7x2	0.6464**
T. quartinianum	904.1 + 2034.3x - 194.1x2	0.7194**

Y = DM yield, x = amount of fertilizer/ha coded as 0,2,4,6 and 7 representing 0,5,10,20 and 35 kg/ha rates in the equation above, respectively, i.e. the coded rates are to be inserted into the above equations in solving for Y. Any other coded rates desired to be inserted like 8,9,10, etc. should be interpreted in multiples of 5 as 40, 45, and 50, kg/ha, respectively.
b = linear regression coefficient
c = quadratic coefficient
*,** = Significant at 0.05 and 0.01 levels, respectively.

These results, and those of the previous trial indicate the ability of the native clovers to respond to P over a wide range of rates of application.

Seed yield was also substantially increased by P application (Table 8). The species differed in their response to P. as indicated by the highly significant (P<0.01) species by fertilizer interaction.

Table 8. Effect of P fertilizer on seed yield for some clovers at Shola in 1983 (Experiment 5).

Species	Phosphate rates (kg/ha)
	0	10	35
	Seed yield
	kg/ha
T. tembense	212	541	1057
T. schimperi	533	817	945
T. quartinianum	478	659	748
T. decorum	305	222	728
T. rueppellianum	117	354	322
T. alexandrinum	12	29	116

¹Exotic clover for comparison. There were significant differences at P<0.01 for among fertilizer rates and among species differences.

Experiment 4

The native clovers had significantly (P<0.05) lower P concentrations in the leaf tissue than T. fragiferum (Table 9), which may indicate that they utilise P more efficiently than the introduced species. However-, this could also have been due to mobilisation of P from the leaf tissues, but it is unlikely that this would have occurred before flowering.

Table 9. Effect of fertilizer and species on concentrations of N and P in leaf tissue.

Main effects	N%	P%
P rate
0	4.1a1	0.213a
2	3.9a	0.233b
4	4.0a	0.253b
10	4.2a	0.253b
Significance level	n.s.	0.01
Species
T. tembense	5.0a	0.211a
T. rueppellianum	4.8a	0.232b
T. fragiferum	2.3b	0.261c
Significance level	<0.01	<0.01
Overall S.E.	0.4	0.02

¹Means followed by the same letter are not significantly different (Waller-Duncan, P<0.05).
S.E. = standard error of the mean.

Phosphorus application had little effect on crude protein content (Table 10) or lignin content (Table 11). However, there were significant (P<0.05) species by age of stand interactions due to the increase in lignin content and decrease in crude protein content as the native clovers matured.

Phosphorus application had no apparent effect on the in vitro digestibility of T. rueppellianum and T. tembense (Table 12).

Table 10. Crude protein content of whole plants as affected by P fertilization and clipping times.

Species	P rate (kg/ha)	Age of stand at clipping (days)			Mean
		120	135	150
	CP %
T. fragiferum	0	21.4	19.2	13.4	18.0
	2	21.8	18.2	13.9	18.0
	4	17.0	15.6	10.6	14.4
	10	13.8	14.6	13.5	14.0
	Mean	18.5	16.9	12.8	16.7
	SE±	1.9	1.1	0.8	1.1
T. rueppellianum	0	18.1	13.4	8.7	13.4
	2	17.8	14.4	9.0	13.7
	4	15.6	11.5	8.1	11.7
	10	16.6	10.4	6.7	11.2
	Mean	17.0	12.4	8.1	12.5
	S.E.±	0.6	0.9	0.5	0.6
T. tembense	0	17.3	10.6	10.3	12.7
	2	14.4	9.6	8.6	10.9
	4	15.8	11.3	10.3	12.5
	10	15.4	10.1	9.4	11.6
	Mean	15.7	10.4	9.6	11.9
	S.E.±	0.6	0.4	0.4
P overall mean	0				14.7
	2				14.2
	4				12.9
	10				12.3
Overall mean		17.1	13.2	10.2	13.5
S.E.±		0.8	1.9	1.4

Table 11. Lignin content of whole plants as affected by P fertilization and clipping times.

Species	P rate (kg/ha)	Age of stand at clipping (days)			Mean	S.E.+
		120	135	150
		% lignin
T. fragiferum	0	4.2	4.2	4.0	4.1	0.1
	2	4.5	3.8	4.8	4.3	0.3
	4	4.1-	3.8	4.0	4.0	0.1
	10	3.9	4.5	4.0	4.1	0.2
	Mean	4.2	4.1	4.2	4.2	0.0
	S.E.±	0.1	0.2	0.2	0.2
T. rueppellianum	0	5.6	7.3	9.2	7.4	1.0
	2	6.8	8.5	9.2	8.2	0.7
	4	6.2	7.9	9.2	7.8	0.9
	10	5.8	7.9	9.9	7.9	1.2
	Mean	6.1	7.9	9.4	7.8	1.0
	S.E.±	0.3	0.2	0.2	0.2
T. tembense	0	6.0	8.4	8.6	7.7	0.8
	2	6.5	6.7	8.8	7.3	0.7
	4	5.7	8.3	8.4	7.5	0.9
	10	5.7	7.8	9.9	7.8	1.2
	Mean	6.0	7.8	8.9	7.6	0.8
	S.E.±	0.2	0.4	0.3	0.1
Overall mean	5.4	6.6	7.5
S.E.±	0.6	1.3	1.6

Table 12. Predicted in vitro dry matter digestibility (DMD%) at full bloom as affected by P fertilization.

	P kg/ha	Plant fraction
		Leaves	Heads	Stems
		DMD1 (%)
T. resupinatum	0	78.5	N/A	77.6
T. rueppellianum	0	78.4	68.6	65.8
T. tembense	0	75.9	69.0	75.2
T. rueppellianum	4	76.6	54.5	66.7
	10	86.6	69.0	72.2
T. tembense	4	74.7	69.0	71.3
	10	76.1	70.5	77.3
Mean and S.E.		77.3+0.8	67.1±2.1	72.3±1.8
Range		74.1-80.6	54.5-70.5	65.8-77.6

¹DMD % = 0.74 (DMS %) + 15.72 (L.J. Lambourne, personal communication) where DMS % = Dry matter solubility (pepsin/cellulase).
N/A There were no seed heads on this plant.

Summary and conclusions

A study of 34 T. tembense germplasm accessions from the Ethiopian highlands indicated that there is a large amount of variation within the species, which would aid plant breeding programmes.

Native clovers showed large increases in dry-matter production at low levels of P application (less than 10 kg P/ha). Higher rates of P application showed that the native clovers could respond to a wide range of P rates (5 to 30 kg

P/ha). The leaf tissues of the native clovers contained less P than an introduced clover (T. fragiferum), indicating the greater efficiency of use of P of the local species. Seed yields were also substantially increased by applying P fertilizer, a factor of potential importance in commercial seed production.

Phosphorus fertilization had little effect on the N. crude protein and lignin contents of the leaf tissues or, as a consequence, on digestibility characteristics.

Early planting, in the short rainy season, resulted in longer growing seasons and higher dry-matter yields than later planting in the long rainy season. The high yields obtained from early planting may justify the risk of the uncertainty of sustained moisture availability during the short rains.

The relatively high yields of the native clovers observed in this study suggest that they are a potentially useful source of forage under proper management within the present farming systems in the Ethiopian highlands.

Acknowledgements

This work is part of a PhD study (Akundabweni, 1984) supported by ILCA and South Dakota State University. The author is grateful to the ILCA Forage Legume Agronomy Group for material and moral support both during and after this work.

References

Akundabweni L M S. 1984. Forage potential of some annual native Trifolium species in the Ethiopian highlands. PhD thesis, South Dakota State University, Brookings, SD. 208 pp.

Allard R W. 1970. Population structure and sampling methods. In: Frankel O H and Bennet E (eds). Genetic resources of tomorrow. F A Davis Co., Philadelphia.

Allen O N and Allen E K. 1981. The Leguminoseae. A source book of characteristics, uses and nodulation. McMillan Publishers, London.

Andrew C S. 1976. Effect of calcium, pH and nitrogen and chemical composition of some tropical and temperate pasture legumes. I. Nodulation and growth. Aust. J. Agric. Res. 27:611-623.

Clausen J and Hiesey W. 1958. Phenotype expression of genotypes in contrasting environments. Record, Scottish Plant Breeding Station, Roslin, Scotland.

Cottenie A. 1980. Soil and plant testing as a basis of fertilizer recommendations. FAO Soil Bull. 38/2. FAO, Rome.

Davies W E and Young N R. 1967. The characteristics of European, Mediterranean and other populations of white clover (T. repens). Euphytica 16:3330-3340.

Dillon J. 1968. Response analysis of crop and livestock production. Oxford Press, London.

Esminger M E and Olentine C J. Jr. 1978. Feeds and nutrition. Esminger Publishing Co., Davis, California.

Evans A. 1976. Clovers. In: Simmonds N W (ed). Evolution of crop plants. Longman, London.

Gillet J B. 1952. The genus Trifolium in southern Arabia and Africa south of the Sahara. Kew Bull. 7:367-404.

Krajina V J. 1977. On the need for an ecosystem approach to forest land management. In: Kimunis J P (ed). Proc. Ecological classification of forest land in Canada and northwestern USA. University of British Columbia, Vancouver, Canada.

Martin W E and Matocha J E. 1973. Plant analysis as an aid in the fertilization of forage crops. In: Walsh L M and Beaten J D (eds). Soil testing and plant analysis. Soil Science Society of America, Madison, Wisconsin.

Norris D O. 1965. Acid production by Rhizobium: A unifying concept. Plant and Soil 22:143-146.

Norris D O and t'Mannetje L. 1974. The symbiotic specialization of African Trifolium species in relation to their agronomic use. East Afr. Agric. For. J. 29:214-235.

Thulin M. 1982. Legumes in Ethiopia: An illustrated guide. Opera Bot. 83.

Wilson J R. 1981. Environmental and nutritional factors affecting herbage quality. In: Hacker J B (ed). Nutritional limits to animal production from pastures. Proceedings of an international symposium, St Lucia, Queensland, Australia.

Wolf B. 1983. A comprehensive system of analysis and its use for diagnosing crop nutrient status. Commun. Soil Sci. Plant Anal. 13:1035-1059.

Zohary M. 1972. Origins and evolution in the genus Trifolium. Bot. Not. 125:501-511.