The rise and fall of Siratro ( Macroptilium atropurpureum ) − what went wrong and some implications for legume breeding , evaluation and management

Siratro (Macroptilium atropurpureum) cv. Siratro was one of the first tropical legumes released for commercial use in the 1960s. It initially showed great promise in experiments and commercial sowings. Early research showed it was unproductive under heavy grazing, but after some 15 years there was increasing concern about its persistence, even under light to moderate grazing pressure. Commercial usage subsequently declined markedly although siratro, usually as cv. Aztec, is still sown to a very limited extent. This paper examines some reasons for this decline and then discusses some implications for research into improving tropical pastures through the use of legumes. In general, early pasture research, such as that on Siratro, failed to recognize that original plants of herbaceous legumes had a limited life span and that, for long-term persistence, new plants had to develop through vegetative or sexual reproduction. However, many studies over a 20-year time span showed that, although Siratro could form new plants, in most cases these replacements were insufficient to maintain an adequate plant density in the long term. Data on stolon density, plant longevity, soil seed banks and seedling survival, under different rainfall regimes and stocking rates, are presented to illustrate this. The major limitation was that soil seed banks were generally inadequate to ensure persistence, especially through a period of drier years, when there would be little or no seed set and possibly the death of all seedlings, which emerged from isolated falls of rain. Autumn spelling of pastures to enhance seed set improved persistence, but not reliably enough to be of widespread practical use. The major implication is that evaluation studies failed to adequately recognize the need for introduced legumes to form new plants after the original ones died. This has implications for future experiments, in terms of: duration; the management regime(s) imposed; the measurements or observations taken; and the need for a more ecological approach in evaluation.


Introduction
Cultivar Siratro was bred from 2 introductions of Macroptilium atropurpureum; a description of the breeding program can be found in Hutton (1962).Siratro was one of the first legumes released for use in sown tropical pastures in Australia.As this species was new to cultivation, nothing was known about the agronomy, ecology or persistence of Siratro as a pasture legume, when it was released in 1960.
It was soon found that Siratro was relatively easy to establish and grew on a wide range of soil types (Jones and Jones 1978).Although it was affected by some pests and diseases, the main limitation, prior to about 1980, was that, in higher rainfall areas receiving in excess of 1,400 mm/yr, it was susceptible to foliar blight caused by Rhizoctonia solani.However, it was later affected by rust caused by the fungus Uromyces appendiculatus.This led to the release of the rust-resistant cv.Aztec, which is otherwise genetically similar to cv.Siratro (Bray and Woodroffe 1995).However, almost all the detailed studies described later have been done on cv.Siratro.Cutting trials in the 1960s showed that yield was depressed by close and frequent cutting (Jones 1974a;1974b) and this depression in yield was soon noticed at high stocking rates on grazed pastures in the early 1970s (Jones and Jones 2003).However, if moderately or lightly grazed, Siratro showed great promise.It sometimes comprised about 30% of pasture yield at the end of the growing season, increased soil nitrogen levels plus the N% of the associated grass, and supported good animal production (Mannetje 1967;Bisset and Marlowe 1974;Mannetje and Jones 1990;RJ Jones 2003;Jones and Jones 2003;Tothill et al. 2008aTothill et al. , 2008b)).Thus, it appeared to have great potential and was well accepted by farmers and graziers; so a special issue of Tropical Grasslands in 1977 (Vol. 11, No. 1), restricted to papers on Siratro, recognized the retirement of Dr Hutton.
However, in the late 1970s, doubts began to surface about the long-term persistence of Siratro in commercial sowings; so QDPI (Queensland Department of Primary Industries) organized a symposium in 1982 to gather data on experiences with Siratro persistence.Divergent views were expressed at the symposium (Brown 1983), with most concerns being raised by QDPI staff, based mainly on experience with commercial sowings.Contributions from CSIRO (Commonwealth Scientific and Industrial Research Organisation) staff, primarily based on results from a large grazing trial at Samford (1,000 mm annual rainfall) and 2 experiments at Narayen (720 mm), were more positive, although they clearly recognized that Siratro was not suited to heavy grazing pressure.Siratro had persisted for 5 years in one of these trials (Tothill et al. 2008a) and for about 10 years in the other 2 (Mannetje and Jones 1990;Jones and Jones 2003); so this positive response was to be expected.However, persistence subsequently declined in all trials and, by the time the trials were terminated after 15−20 years, Siratro was a minor component.Thus, the serious problems with long-term persistence of Siratro reported in commercial pastures eventually occurred in lightly grazed experimental pastures.Despite a marked decline in its popularity with primary producers, about 20 t of siratro seed (presumably both cvv.Siratro and Aztec) were produced annually in the mid-1990s (Smith 1996).Siratro is still used to a limited extent in both long-term pastures and short-term leys, usually as part of www.tropicalgrasslands.info a mixture of legume species (e.g.Clarke 2008;McCamley 2010).
This paper reviews the knowledge of Siratro persistence to give some reasons for the persistence problem and then looks at some wider implications.

Persistence of Siratro
With the early enthusiasm about Siratro and other herbaceous legumes, one basic principle was largely overlooked -herbaceous perennial plants do not persist forever.To maintain a legume of this type in a pasture in the long term, it must form new plants, either from stolons or rhizomes or from seed.Fortunately, some related measurements of Siratro persistence were made in several grazing trials.Almost all of these trials were carried out with fixed stocking rates, usually with continuous grazing.

Life span of individual plants
Siratro plants often have a half-life (time for half the population to die) of about 1.5-3 years (Gutteridge 1985;Jones and Mannetje 1986;Jones and Bunch 1988a;Jones et al. 1993), although this obviously varies with management, site and rainfall.As a typical Siratro sowing rate of 3 kg/ha could equate to 25 seeds being sown per m 2 , 10 or more plants could establish per m 2 .After 2 years there could still be more than 5 plants per m 2 and, as these plants are large, the loss of density might not be reflected in reduced dry matter yield.This could even apply after 4 years with a density of over 2.5 plants per m 2 .
There are data from 4 sites on the effect of stocking rate on the half-life of original plants.At Samford (Jones and Bunch 1988a) and in Thailand (Gutteridge 1985), a higher grazing pressure reduced plant half-life, while in the 2 experiments at Narayen, there was little effect (Jones and Mannetje 1986;Jones et al. 1993).This difference between Samford and Narayen could be partly due to the wider range of stocking rates imposed at the wetter Samford site (1,000 mm rainfall compared with 700 mm), and to the overall lower grazing pressures (weight of animals per weight of herbage) during active summer growth periods at Narayen.Interestingly, in one experiment at Narayen, low soil phosphorus status also had little effect on plant longevity (Jones et al. 1993).
Thus, with time Siratro density must obviously decline to the point where yield is negligible -unless new plants establish.The following sections look at recruit-ment from both stolons and seed, and consider how this is affected by grazing pressure and climate.

New plants from stolons
Formation of new roots from stolons was mentioned in the initial description of Siratro (Hutton 1962).However, while small roots can develop from thin soft stolons on the soil surface, these do not persist.Under some situations, which appear to be restricted to sites receiving in excess of 1,000 mm rainfall and under light stocking (Bisset and Marlowe 1974;Walker 1981;Jones and Bunch 1988a), Siratro can form much firmer woodier stolons, which are often just under the soil surface.Growing points and roots can develop on these stolons.At a moderate stocking rate of 1.7 heifers/ha at Samford, the total length of these stolons was about 1 m per m 2 for about 10 years, but no stolons remained after 16 years.There were more stolons in a pasture lightly stocked at 1.1 heifers/ha, but none under heavy grazing.Further details about stolon length and size, and on the density and size of rooted points, are given by Jones and Bunch (1988a).Stolons were never observed at the drier Narayen site.
Overall, it seems that stolons are not the major pathway for Siratro persistence.

New plants from seed
Data from Samford (Jones and Bunch 1988a) and Narayen (Jones and Mannetje 1986;Jones et al. 1993) show that new plants can develop from soil seed reserves.In a moderately grazed pasture (1.7 heifers/ha) at Samford, seedling replacement maintained good densities of 6-13 plants/m 2 for 7 years after the density of the original plants had fallen below 1 plant/m 2 (Jones and Bunch 1988a;1988b).Long-term persistence seemed assured.In contrast, density declined markedly under a higher grazing pressure, as crowns had a shorter life span and there were progressively fewer new plants.However, during the next 5 years, replacement in the moderately grazed pasture did not match plant death, and Siratro density fell to 1 plant/m 2 .In part, this was due to less favorable rainfall conditions for seed set and to increased grazing pressure in generally dryer years.Siratro rust did not seem to be a major factor, but could have exacerbated the problem of poor persistence; it can reduce leaf and seed yields, but its biggest impact is in monospecific swards or pastures with a high Siratro component (RJ Jones 1982).Other unknown factors could also be involved.

www.tropicalgrasslands.info
Similarly, in 2 experiments at Narayen, although there was certainly some successful recruitment from seedlings, long-term plant density was not maintained (Jones and Mannetje 1986;Jones et al. 1993).
In summary, while there may be some instances where plant density can be adequately maintained after the original plants die out, in many or most instances this does not happen.This decline in successful recruitment must be largely related to fewer seedlings, thus to the size of the soil seed bank, which in turn relates to the amount of seed set.

Seed set and soil seed banks
Measurements of seed reserves in the soil have been made in medium-to long-term trials at Samford (Jones 1979;Jones et al. 1980;Jones and Bunch 1998b) and Narayen (Mannetje and Jones 1986;Mannetje and Jones 1990;Mannetje and Butler 1991;Jones et al. 1993Jones et al. , 2000)), and in spot measurements in some other trials (Tothill and Jones 1977).In summary, seed reserves rarely exceeded 500/m 2 and were usually about 200 or below, even in pastures with a good Siratro content.
Poorer or overgrazed Siratro pastures have much lower or negligible seed levels.In the heavily grazed pasture at Samford, sown in 1968, no Siratro flowers or pods were recorded after 1972, whereas the moderately grazed pasture usually had above 20 flowers or pods/m 2 up to 1982.Not surprisingly, in 1983/85 there were 167 seeds/m 2 in the moderately grazed pasture but only 15 seeds/m 2 in the heavily grazed pasture (Jones and Bunch 1988b).Soil seed numbers were usually significantly reduced by higher stocking rates at Narayen (Jones and Mannetje 1986;Mannetje and Jones 1990;Jones et al. 1993) and were lower at a higher stocking rate in Thailand (Gutteridge 1985).
Although some slight moisture stress can aid initiation of flowering in Siratro, it appears that in dryer areas, such as Narayen, dry conditions in autumn, when seed set occurs, are a major factor contributing to low soil seed numbers.Under dry conditions, aerial stems on Siratro plants can even die back towards the crown and, if stems die, no flowering or seeding occurs.Paradoxically, there was a complete failure of seed set in one year at Samford with sustained wet conditions in autumn (author's unpublished data).There were many pods, but no hard seeds; all seeds were soft or rotten, or prematurely germinated in the pods.
The number of legume seeds measured in good Siratro pastures is about 5-25% of that found in good pastures of other persistent legumes in southeast Queens-land, such as Wynn cassia (Chamaecrista rotundifolia) (Jones 1995;Jones et al. 2000), shrubby stylo (Stylosanthes scabra) (Jones et al. 1993;2000), Bargoo joint vetch (Aeschynomene falcata) (Jones et al. 2000), fine stem stylo (Stylosanthes hippocampoides) (Orr 2008) and white clover (Trifolium repens) (Jones 1982;1984).In a 1993 paper comparing the persistence of Wynn cassia, Verano stylo (Stylosanthes hamata), shrubby stylo and Siratro, there was a suggestion that, for any pasture legume, 'within reasonable ranges, there is a "trade off" between longevity and seed set.Perhaps a key weakness of Siratro is that the seed banks are outside this reasonable range and too low to ensure recruitment, even at moderate stocking rates' (McIvor et al. 1993).

Seedling establishment
Most seed set in pastures cannot subsequently be accounted for.In one experiment at Samford, handcollected Siratro seed was broadcast into 2 sites in closely grazed Siratro-free pastures ideal for seedling emergence.Over the subsequent 6 years, only 21% of the seed was accounted for as seedlings and a further 15% as soil seed (Jones 1981).Similar low recoveries have been recorded with other species in similar experiments.Furthermore, most emerging seedlings, and in many instances all seedlings, will die.This is usually attributed to moisture stress, when the small seedling root system is competing with the established pasture for water (Cook and Ratcliff 1985;Jones and Mannetje 1986;Jones and Bunch 1988b;Mannetje and Jones 1990).Competition from associated grasses can greatly reduce Siratro density when compared with monospecific Siratro swards on clay soils (Keating and Mott 1987;Peake et al. 1990), but reduced half-life, lower seed set and increased seedling mortality could all be involved.Another source of seedling loss in high-yielding and lightly grazed pastures is that there can be almost complete death of emerging Siratro seedlings due to competition for light (Agishi 1974;Jones and Bunch 1988b).

Effects of defoliation
Because of the growth habit of Siratro, with its long twining aerial stems with long internodes, heavy grazing greatly reduces the number of growing points (Clements 1989b) and therefore forage yield.However, as outlined above, the reduction in Siratro persistence under heavy grazing must operate through the effects on plant halflife, seed set, soil seed banks and seedling recruitment, www.tropicalgrasslands.info or, less importantly, on stolon development and persistence.
There are not enough data to make any specific comments about how rotational grazing systems could affect Siratro persistence.At Samford, results from a 4-year experiment suggested that Siratro yield and persistence were slightly better in a rotation with long rest periods, but that this would be obtained only at stocking rates where the pastures were probably overgrazed in terms of animal production (Jones 1979).At Narayen, there were only minor differences in Siratro yield and persistence when continuous grazing was compared with a 4 weeks on:4 weeks off rotation for 4 years (Mannetje and Jones 1990).There is no information about the effect of cell grazing on Siratro persistence.In any rotational system with lengthy periods of stay in each paddock, there is the possibility that the timing of grazing in relation to Siratro flowering and seeding may have a different effect on seed set in different paddocks.

Effects of climate
Siratro has a lower ability to set seed under dry conditions than other herbaceous legumes such as shrubby stylo, Wynn cassia and Bargoo joint vetch.This is illustrated by data from an experiment at Narayen (Jones et al. 2003), supported by data from a wider range of experiments.This is in part due to the growth habit of Siratro, where flowers are borne towards the ends of long aerial stems, and partly from its lower tolerance of moisture stress compared with shrubby stylo.In an experiment at Narayen, where soil seed reserves of Siratro were measured annually, the progressive annual levels were 42,18,285,109,62,32,13 and 11 seeds/m 2 , averaged over all treatments (Jones et al. 1993).Clearly seed reserves increased in only one year and the impact of this one year's seeding on reserves lasted for only 3 years.Tothill et al. (2009) also commented on the effects of a series of years with below average rainfall on Siratro seeding and persistence.In a similar environment to that at Narayen, variation in rainfall had a greater effect on the population dynamics of fine stem stylo than stocking rate (Orr 2008).
Frosting is a feature of the climate in subtropical Queensland.In most years, Siratro flowering and seeding have ended before the onset of severe frosting, so frost is unlikely to have a major effect on size of soil seed banks.Limited evidence also suggests that, in terms of crown survival, Siratro is one of the more frosttolerant tropical legumes (Jones 1969).

Putting the pieces together
As a result of the studies described above, we have a reasonable, but certainly not perfect, understanding of how the different components of Siratro persistenceplant survival, stolons, seed set, soil seed banks and seedling survival -are affected by grazing pressure and rainfall.

How did this understanding of Siratro persistence develop?
This understanding did not develop as a result of measurements made to validate a conceptual model.Rather, as described below, the understanding and model gradually evolved as more measurements were made.The first demographic measurements followed casual observations in 1969 of Siratro seedlings in a pasture established in 1966 (Rees et al. 1976) and from sheer curiosity about their survival.These very limited measurements led to more detailed work on seedling emergence and survival under contrasting stocking rates in an existing experiment at Samford, starting in 1970 (Jones and Bunch 1988a).It soon became obvious that, to understand how these measurements of seedlings fitted into the overall picture, there was a need to measure plant survival, stolons where present, and soil seed banks.These are the key measurements in simple conceptual models of persistence (Jones and Mott 1980;Jones 1985;McIvor et al. 1993) and helped to partly explain the effects of stocking rates on persistence.
These interesting results from Samford then led to some demographic measurements being taken in the later years of 2 existing grazing experiments at the drier Narayen site (Tothill and Tessel 1982;Mannetje and Jones 1990) and to more comprehensive measurements in 2 new Narayen trials that started around 1980 (Jones and Mannetje 1986;Jones et al. 1993).The Narayen measurements gave valuable insights into the role of moisture stress in limiting seed set and seedling recruitment, regardless of stocking rate.It is questionable if this gradual understanding of the mechanisms of Siratro persistence could be achieved under the current research trend to shorter-term and goal-specific experiments.

Implications for Siratro management
The obvious implication from all studies on Siratro is that lighter grazing assists Siratro persistence through increasing seed set and, at times, longevity of crowns, as www.tropicalgrasslands.info well as sometimes increasing stolon development.However, results from commercial experience and research indicated that this alone was not enough to reliably ensure persistence (Brown 1983).The next step was to examine the effects of resting in late summer and autumn on seed set in autumn, the main seeding period.Results have been equivocal.Resting can have a major impact on seed set; for example, in one year a 30-to 60fold increase in seed set was recorded at Narayen with late season resting (Tessel 1983).However, in another study, although resting increased Siratro yield, there was no measureable effect on the soil seed bank (Jones and Jones 2003).
Most of the differences in seeding recorded in these studies could be attributed to differences in rainfall patterns and yield of Siratro.There have also been instances where light grazing or resting of commercial pastures, in conjunction with successive good rainfall years, substantially increased Siratro populations and yield (SJ Cook and CK McDonald personal communications).Other experiments examined the use of rough cultivation of established pastures to enhance Siratro seed germination and to aid seedling survival through reducing competition from established plants (Bishop et al. 1981;Jones and Jones 2003).
The findings from all the related studies are that both rough cultivation and late season resting, especially in years with good rainfall, can sometimes aid Siratro seed set, seedling recruitment and persistence (Hurford 1979;Jones 1979Jones , 1988;;Bishop et al. 1981;Tessel 1983;Jones and Jones 2003;Tothill et al. 2009).However, these practices are not sufficiently reliable to ensure confidence in the persistence of Siratro and hence widespread use in permanent pastures.Furthermore, there will always be some reluctance by many farmers or graziers to lightly graze good Siratro pastures in autumn, when grass quality is declining.This reluctance to graze lightly could even apply to the peak growth period in summer, although cattle usually prefer grass to Siratro during this period (Stobbs 1977).

Implications for further selection or breeding
Unfortunately, the objectives of the second breeding program on Macroptilium atropurpureum during the late 1960s and 1970s (Hutton and Beall 1977) were not related to enhancing the persistence mechanisms outlined above.However, following later characterization of 230 accessions of M. atropurpureum, a group of 45 lines was defined as having more branching and earlier flowering than Siratro (McDonald and Clements 2005).Seventeen of the 230 accessions, including some from this group, were compared for 5 years under farm grazing at 2 sites in southeast Queensland, but no single line maintained significantly higher densities than Siratro or Aztec at both sites (Bray et al. 2000).
If a clearly defined and important breeding target, which would obviously impact on plant survival and seed set, e.g.disease resistance in top growth or roots, can be identified, there is potential for breeding to assist legume persistence.A simple example is the improved persistence of Medicago sativa in southeast Queensland, produced by breeding for resistance to aphids and root diseases (Clements et al. 1984).Although Aztec, a rustresistant cultivar of siratro, has been available for about 15 years, there is no feed-back that rust resistance is appreciably improving siratro persistence.
Disease or pest resistance apart, breeding for improved persistence in a legume, that persists in the long term through seedling recruitment, is very difficult -do you target longer crown life, more seed set, a higher level of hard seed or greater seedling vigor?The target has to be defined and then adequate variation found in plant collections.Furthermore, defining a target, such as higher seed set, is not enough.How much of an increase will you need in this attribute, before it will have an impact in commercial pastures?In addition, there is always the risk that selecting for an increase in one attribute may inadvertently lead to a decline in other desirable attributes of persistence, or even yield.Good advice on breeding for legume persistence has been presented elsewhere (Clements 1989a).

Implications for weediness
In some situations Siratro can be a weed, as it can climb over other plants and set considerable amounts of seed if not defoliated.The characteristics that imbue persistence are similar to those which contribute to weediness.This was not appreciated in the early research on tropical legumes, when Siratro was released, but the potential for weediness must now be considered in any introduction and evaluation program.Extreme conditions of weediness, either minimal or excessive, are relatively easily recognized.White clover, for example, can be a weed in lawns, but world-wide it has been a great asset for increasing animal production and soil fertility with minimal or negligible disadvantage as a weed.However, sometimes the boundary between desirable and undesirable persistence and spread is difficult to recognize and to relate to growth form.For example, glycine (Neonotonia wightii) and Siratro have somewhat similar growth www.tropicalgrasslands.info habits, when grown with grass.Yet, in my experience in coastal southeast Queensland, glycine, with its much greater ability to climb, is a far more serious weed in open or disturbed forests.

Implications for evaluation
This is where the studies on Siratro persistence could have the greatest implications.Initial stages of evaluation of legumes for long-term pastures have often involved lightly defoliated trials lasting only 2-3 years and only yield has been measured.As a result of the factors outlined above, such trials can be potentially useless or at best misleading in assessing long-term persistence, especially if the legumes are grown in pure swards.Preferably, trials of this nature should be conducted under one or more defoliation treatments approximating commercial practice for at least 5 years.This is especially so if the site experiences wide variation in rainfall and, for example, can experience 2 or more consecutive years with well below average rainfall or infrequent periods of sustained heavy rainfall with potential for waterlogging.Longer trials will obviously involve more input of time and labor.However, the time and effort usually spent in cutting and measuring yield in screening trials can be reduced appreciably by using calibrated visual estimates of yield and composition.These are quite adequate to document the higher-yielding lines (Jones 2001).Very useful measurements of legume density, or at least legume frequency (Cameron et al. 1989;Jones 2001), could be made, even if only at the end of the trial.
Given the time-consuming nature of demographic measurements, such as survival of individual plants and soil seed reserves, it is pointless to make them in shortterm assessments of a large number of lines.The important thing is to recognize the limitations of such trials and at least look for flowering, seed set, stolon/rhizome development, seedling recruitment and such like.
Management of evaluation trials should relate to the conditions in which the legumes are likely to be used.In commercial situations, farmers often stock according to average conditions, which means they are overstocked in below average rainfall years.They tend to focus more on the short-term forage needs of their stock than on the long-term survival of the pasture.Hence, evaluation methods need to reflect this.For example, a multi-site study in coastal Queensland evaluated legumes specifically for heavily grazed pastures (Cameron et al. 1989).The legumes were sown into cultivated strips in closely grazed grass swards.Management was controlled in the first year, to allow for legume establishment and seed set.However, in subsequent years, the experiments were just a small section in a pasture which, hopefully, was heavily grazed by the co-operating farmer.The emphasis was on persistence, which was easily measured by species frequency.The results in this experiment matched subsequent commercial experience with the same accessions.In contrast, the grazing or cutting regimes for legumes being evaluated for short-term ley pastures or cut-and-carry systems would be much more lenient.
There may be a case for specific demographic measurements to assist in understanding or extrapolation of the results of later-stage evaluation or management studies.For example, measurements of soil seed reserves of Siratro at the end of a 5-year study of rotational grazing aided prediction of the effects the treatments could have on Siratro persistence, if they had been continued for longer (Jones 1979).In longer-term grazing trials, perhaps with 1-3 accessions subjected to different grazing systems or stocking rates, demographic measurements may help to understand why species did or did not persist and how the results might be extrapolated to commercial conditions (Jones et al. 2000).They could lead to suggestions for improved management for persistence, based on seasonal changes in grazing pressure (Jones 1982(Jones , 1984;;Orr 2008).
Thus, keeping demographic concepts in mind throughout the legume evaluation process may improve predictions of persistence.More comments about what demographic measurements to make, and why, when, and how to make them, can be found elsewhere (Jones and Mott 1980;Jones 1985;Jones and Carter 1989;Hay et al. 2000).

Implications for thinking
During the early years of "pasture improvement", I suggest we may have had a too simplistic view of what we were trying to do.In essence, we were trying to remove an existing plant community, or at least introduce a legume into it, to form a new and more productive community, which would be resilient over time.Perhaps we were too simplistic, thinking more like engineers rather than ecologists.
It was as if we were engineers erecting a new bridge to replace an existing bridge, which had load (animal production) limitations.All you had to do was to supply the seed and fertilizer (concrete and reinforcing), avoid gross mismanagement (exceeding the load limit) and behold the new bridge would be stronger (productive) than the old one, and last for decades.Our approach was www.tropicalgrasslands.info somewhat mechanical rather than biological.Whereas concrete and steel reinforcing usually last for decades, regardless of external forces, herbaceous plants do not.Furthermore, they have widely varying tolerances of external forces such as rainfall, fire, frosting, weed invasion and, especially, grazing pressure and frequency.
Perhaps we should recognize that evaluation of pasture species is, by and large, not science.It is simply a measurement of one or more simple attributes (often yield is the only one), when a range of accessions is subjected to one or more ad hoc defoliation treatments.Even if the defoliation treatments and experimental duration are chosen to reflect particular farm management practices, this does not make evaluation into a sciencealbeit it is a worthwhile thing to do.Science in evaluation is where we are concerned with underlying processes and attempts to answer questions such as the following: What makes some plants more resistant to defoliation than others?Why does one species need more phosphorus than another?How does a species persist in a pasture after the original plants have died?Why does one species spread into adjacent plant communities and another does not?

Some final personal comments
(1) In the 1950s and 1960s, when Siratro was developed and initially evaluated, very little was known about the potential, for good or bad, of incorporating tropical legumes in grazed pastures.At that time, the focus was on identifying legumes with high DM production.In retrospect, the development of Siratro was a good but imperfect start, but now the question is: "Are we learning from past shortcomings and doing things better?" (2) Perhaps one of the surprising features of Siratro is that, given its growth habit, it was so successful and well accepted by graziers and scientists for more than a decade, and is still used to a limited extent.One possible reason for this relates to the climate of Queensland, which incorporates a cool season, usually dry, of 6 months or more.During this time, there is little or no pasture growth, which restricts animal numbers.Given an average or good warm season rainfall, growth is rapid and there is often plenty of forage, so the overall grazing pressure during the growing season is reduced, obviously favoring Siratro.
(3) In view of its ability to improve soil nitrogen status and N concentration in associated grasses (Mannetje and Jones 1990), Siratro may still have a useful role as part of a legume mixture for improving these characteristics in both long-term and ley pastures (McCamley 2010).
(4) Many of the early evaluation experiments were conducted for only 2-3 years, and measured yield only under light defoliation treatments.Unfortunately, they did not continue for long enough, ignored plant attributes related to persistence, and bore little relationship to commercial conditions as defoliation levels were too lenient.
(5) Having a "good" growth habit -prostrate stems that form roots, growing points that are largely protected from grazing, and a capability to set seed under close grazing -does not guarantee that a legume will persist in the long term.Both Lotononis bainesii cv.Miles and Vigna parkeri cv.Shaw have these attributes.However, despite early promise, lotononis has not persisted in the long term and no seed has been commercially produced in Queensland for about 10 years.Vigna parkeri, on the other hand, was rejected in early cutting experiments in favor of taller-growing species such as Desmodium intortum and its value was recognized only when it was seen invading and spreading under grazing at old trial sites (Jones 1984;Cook and Jones 1987).It has persisted well, seed is commercially available, and it is well regarded in the very limited area to which it is adapted.
(6) After having an interest in legume persistence for more than 40 years, I have recently revisited some former experimental sites to observe which legumes are persisting, and have received reports from others who have done likewise.In all cases, I had some experience with the species concerned and had made demographic measurements on many of them.Before looking at or learning about the sites, I formulated views on what the levels of persistence would be.In most instances, these predictions were quite accurate, but occasionally I was completely wrong.The message is: no matter how much experience or knowledge you have -be warned -predicting legume persistence, or botanical change generally, is often somewhat risky.The fewer years of experience, and the lower the demographic understanding, the greater the risk.