Evaluating pasture species for less fertile soils in a subtropical aseasonal low rainfall zone

Grasses, legumes, saltbushes and herbs were evaluated at 6 sites in southern inland Queensland to identify potential pasture and forage plants for use on marginal cropping soils. The region experiences summer heat waves and severe winter frosts. Emphasis was on perennial plants, and native species were included. Seedlings were transplanted into the unfertilized fields in either summer or autumn to suit the growing season of plants, and watered to ensure establishment. Summer-growing grasses were the most successful group, while cool season-growing perennials mostly failed. Summer legumes were disappointing, with Stylosanthes scabra and Indigofera schimperi performing best. Some lines such as I. schimperi and the Eragrostis hybrid cv. Cochise were assessed as potential weeds owing to low animal acceptance. Native Rhynchosia minima grew well at some sites and deserves more study. Cenchrus ciliaris was always easy to establish and produced the highest yields. Persistence of some Digitaria and Bothriochloa species, Eragrostis curvula and Fingerhuthia africana at specific sites was encouraging, but potential weediness needs careful assessment. Standard species were identified to represent the main forage types, such as Austrostipa scabra for cool season-growing grasses, for incorporation into future trials with new genetic materials. The early field testing protocol used should be considered for use elsewhere, if unreliable rainfall poses a high risk of establishment failure from scarce seed.


Introduction
In the Condamine-Maranoa region of southern inland Queensland, Australia (25-29° S), sclerophyllous woodlands and dense tall shrublands dominate the natural vegetation.Native pastures dominated by summeractive grasses are a common forage base for livestock.Their replacement by sown pastures after tree clearing or the infusion of forage legumes seems a good option in some cases (Miller et al. 1988).However, after such disturbance, woody regrowth can thrive and the poisonous pimelea (Pimelea trichostachya) (MLA Australia 2010) can proliferate, especially on abandoned cultivation land.Buffel grass (Cenchrus ciliaris) has been a major exotic pasture success story, but it tends to become a monoculture.
The better structured soils of the region are almost fully utilized for crops, so some less fertile, poorly structured soils are now cropped intermittently in response to market forces.Soil structure on these poorer soils soon breaks down (Douglas 1997), as reported in other countries (Bot and Benites 2005).Significant areas of native pastures have also been seriously degraded, with wiregrasses (Aristida spp.) assuming dominance.Thus, sown pasture species are needed to restore structure to some soils between cropping periods and to augment native pastures, as exists in other subtropical parts of the world, e.g. the Sahel and the Cerrado region in Brazil (Pearson et al. 1995).
Effective rainfall for establishing sown pasture seedlings is unpredictable and most cropping enterprises depend upon stored subsoil moisture to fill the seed of grain crops.Median summer rainfall (October-March) ranges from 300 to 400 mm, median winter rain is 175-200 mm (Rainman 2003), and droughts and floods are common.in winter (Hammer and Rosenthal 1978) and high summer temperatures (>38 °C on many days) routine.
In the mid-1980s, we began to evaluate pasture and forage species that were best adapted agronomically for various roles in that environment and to identify benchmark species against which existing and future genetic material could be compared.The potential roles were: permanent pasture; short-term pasture; legume-augmented native pasture; and soil conservation.We focused on less fertile land that could be cropped intermittently, because that allowed agronomic intervention in a low to moderate cashflow environment.While a range of forage cultivars already existed for fertile soils, e.g.Rhodes grass (Chloris gayana), lucerne (Medicago sativa), leucaena (Leucaena leucocephala), forage sorghums (Sor-ghum spp.) and oats (Avena sativa) (Thompson 1988;Blacket 1992), only buffel grass was suitable for infertile soils.Some plants native to the region were included to ensure that our assessment was not biased by perceived virtues of exotic species (Davis et al. 2011).

Trial site details
We used 5 low-elevation (200-300 masl) field sites on commercial farms, chosen to provide a range of soil types and regional locations; 2 in the Roma district (26.5° S, 149° E; Sites L and N), and 3 near St George (28° S, 149° E; Sites M, U and W).Preliminary screening of accessions and seed increase were also done at Toowoomba (27.5° S, 151.9°E; Site T).Details of the soils and original vegetation at each site are given in Table 1.Site U had grown only one prior wheat crop, while most sites had grown several crops, and Site N had been intermittently cropped for 15 years beforehand.
Soils at the sites varied markedly (Table 1).Two sites (L and W) had a strong duplex profile with a thin, hard-setting, slightly acid loam overlying an impermeable clay subsoil that became saline-sodic at depth.Three other sites (M, T and U) had a gradational increase in clay content with depth, were non-saline and had a general red coloration.At the sixth site (N), the duplex soil was not hard-setting, was alkaline throughout and had carbonate nodules at depth.A strongly bleached A2 horizon at sites L and W was evidence of an impervious subsoil and waterlogging in abnormally wet seasons.Site T was similar to infertile, friable, red cracking clays found south of Roma (Slater and Carroll 1993).

Species evaluated
Since effective rainfall (that which wets the upper 15 cm of soil) can occur at any time, we evaluated both summer-and winter-active species, and included lesstraditional forage plants such as non-legumes and browse shrubs, because their higher mineral contents may counter-balance the low levels in the local, grassdominant pastures.Perennial species were favored, because land can be exposed to erosion for long periods in this environment with annual species.A long assessment period was required to ensure meaningful conclusions were reached.
We chose our test plants on the basis of experience by many pasture agronomists in Queensland and New South Wales, particularly Blumenthal et al. (1985), Day and Silcock (1985), Scattini (1985), Strickland and Greenfield (1988) and Bellotti et al. (1991).Wellregarded species from semi-arid regions overseas were also sourced, e.g.Dorycnium hirsutum and wheatgrasses (Agropyron spp.).The plant accessions tested at field sites are listed in Table 2 along with their homeland and principal traits.Table 2. List of accessions evaluated, their key traits and the sites where not sown (-).An x at a site means that the accession survived there for at least 6-7 years, while t means a tiny amount remained in 1 replicate after 6-7 years.LH Italy CS146 -1 Plant type is described by 2 letters, the first denoting: C=Chenopod, F=C3 grass, L=Legume, P=C4 grass and O=Other; and the second indicating growth form: H=Herbaceous, S=Sward, T=Tussock and W=Woody plant; 2 Nearest Australian town if a native species; 3 Native species; 4 Brackets used if the cultivar name was given later to the accession; 5 Not sown in our trials at this site, but grew well in other trials on adjacent land; 6 Discrimination from similar-looking accessions not certain after many years; 7 Annual species.Details of the origins of each accession can be had by contacting the senior author or by sourcing either the CSIRO Quarterly List of Introductions, starting at CPI11408 in Nbr 16 (1948) through to CPI33946 in Nbr 75 (1963) or thereafter in The Australian Plant Introduction Review (CSIRO:Australia), New Series, Volumes 1 (1964) through to CPI99880 in volume 17 (1985).For cultivars sourced from commercial seed or local ecotypes, that have probably undergone genetic shift since their introduction, their original homeland is named followed by "/ecotype collection locality", such as Zimbabwe/Qld.Plants tested were grouped for presentation of results on their main season of growth, their perenniality and their growth form (Table 3).The seasonal classes are based on the potential season of growth in the subtropical South Queensland environment.
Aseasonal means plants that can grow at any time of the year, when soil moisture is adequate and there are no severe night frosts.Wet winters normally result in mild nights and very few frosts severe enough to freeze green leaf tissue, while in dry winters, morning radiation frosts as cold as -7 °C are common (Hammer and Rosenthal 1978).

Planting details
To minimize field establishment failure, seedlings were transplanted after effective rain and watered intermittently until follow-up rain occurred.Thereafter they competed with local weeds and pests under controlled grazing without the assistance of fertilizer.Some accessions with low seed supplies were increased beforehand at Toowoomba (Site T; 640 masl), and species with many potential accessions available were screened to short-list the agronomically promising lines and to eliminate those which failed badly, such as Russian wildrye (Psathyrostachys juncea), which was extremely susceptible to a stem rust and failed to set seed.In a few cases, nursery plants provided vegetative material for field sowings, when germination difficulties severely limited potential sowing numbers.

Glasshouse procedures
Seedlings and a few grass transplants were established in a glasshouse in Toowoomba.Sieved soil from each site was poured into 15 cm deep expanding blocks of hexagonal paper pots (2.5-4 cm diameter).The soil was wet up with tap water and fertilized with mono-ammonium phosphate (MAP) prior to sowing the seeds.Legume and other non-grass seeds were scarified with sand-paper to remove impervious coatings if necessary, and saltbush fruits were clipped to expose the seed inside and then leached of salt, but grasses were sown untreated as clean seed with any glumes and short awns.When germination rates for grasses were low, plant numbers were augmented with plants struck from nursery ramets, using material from many different parent plants.
After sowing, legume tubes were inoculated with an appropriate rhizobial slurry.Seedlings remained in the glasshouse for 3-5 weeks, until adequate rain had fallen at the trial sites.Sometimes additional MAP was applied to keep larger plants healthy.When excessive seedling numbers occurred, stands in the tubes were thinned, so that grasses tillered and non-grasses suffered limited competitive stress.Just prior to sowing, the tops of large plants were clipped to reduce transpiration load and all were placed outside for a few days to 'harden-up' in natural sunlight and wind.

Field procedures
At field sites, seedlings were sown in rows as spaced plants into a fenced area of wheat stubble.The rows, 3 m apart, were centered along 1 m wide strips sprayed for weed control with glyphosate (1 kg a.i./ha) just before planting.Rows, containing 10-15 randomly located accessions, were blocked into each of the 2 replicates sown at a site.A line of 6 holes at 1 m spacings was dug with a mattock for each replicate of an accession and filled with water.Then the tubed seedlings were planted singly, the soil refilled around them, and each plant rewatered from either a nearby farm dam or the Toowoomba city supply.Until reasonable follow-up rains fell, plants were rewatered about once a week (Table 4).Summer and autumn sowings of appropriate species occurred at each field site (dates given in Table 4).When enough plants failed to establish, accessions were resown after rain the following year at an equivalent time.The initial sowing at each site was of summer-active species, when wheat stubble was still sturdy, but by the autumn sowing, the stubble had often thinned and weeds were established.Domestic stock were excluded by hinge-joint fences and plants were ungrazed until the majority were well established and had flowered and seeded.This time period varied greatly as sites experienced widely differing growing conditions (Table 4).

Grazing management
The owner's stock currently in the paddock (mainly cattle but sometimes sheep and horses) were allowed to graze the plots, for only a few days initially, and stock were completely excluded for some time after the autumn sowing and again for many weeks after resowings in the second year.From about 2 years after the final resowings, plots were left open to grazing for extended periods.As some paddocks beside our plots were recropped, no domestic stock had access for extended periods, usually during winter, but local marsupials and rabbits often more than compensated for this.Grazing pressure steadily increased as the plots aged, partly due to the worsening below average rainfall seasons after 1990.

Data collection
Periodic recordings were made of the number of plants surviving in each plot, their vigor, flowering and seeding status, and general condition.Later, numbers of seedling recruits plus the distance spread from their original plots were recorded.After 5-6 years, the fences were removed and plots subjected to normal grazing.The persistence and spread of surviving plants was monitored for at least 15 years and some lines that showed weed potential were dug out.Sites U and N were replowed by the owners in 1999 and 2003, respectively, providing an opportunity to assess the ability of persistent lines to resist such practices.

Other influences
At Site M, the coarse native wiregrasses amongst the rows were rarely grazed and became quite rank.Hence we occasionally slashed or burnt the whole plot, beginning in November 1988, a management strategy commonly used to control wiregrasses.Site W was exposed to appreciable spray drift (glyphosate + 2,4-D) in autumn 1990, sufficient to defoliate trees beside the plots and brown the foliage of most plants being evaluated.

Seasonal rainfall
Table 4 summarizes the rainfall received around sowing times for each site, while Table 5 and Figure 1 present the seasonal rainfall received at each site during the first 6 years.During the first few years, growing conditions in the cool season were better and in summer were drier than 'normal' (Rainman 2003).After the initial years of reasonable rainfall, a prolonged period of below average rainfall occurred (1991)(1992)(1993)(1994).From 1995 seasons were more favorable, enabling long-term persistence, spread and possible weed potential to be assessed.

Early growth and establishment
The perennial summer-active grasses were the easiest group to establish and their early productivity was generally good.(Strickland et al. 2000).Overall establishment was: quite good at Sites N and U, where only 16 accessions failed within a year; fair at Sites L and W (both hard-setting, strongly duplex soils); and poor at Site M (acidic, infertile), where 48 accessions were lost within the sowing year.
During planting out, healthy root nodules were seen on many legumes, including leucaena, Desmanthus virgatus and lucerne.Locusts were a major problem during the establishment period of the summer-active species and rabbits and marsupials selectively targeted new sowings in dry seasons.Usually a few plants of each accession survived at each site for at least the first growing season and also flowered, while all sown plants of the better-adapted lines survived for years.

Agronomic performance
On overall agronomic performance, 22 out of 128 accessions achieved a rating of ≥3 (out of 5) over the 5 field sites, while 74 had a rating of <2 (poor).Of the highly rated ones, only 2 were native (out of 27 natives), while 18 of the poorly rated lines were native species.Agronomic performance for individual accessions is reported in detail in Supplementary Tables A and B, while Table 2 shows those which persisted well at one or more sites.
In broad terms, the summer-active perennial grasses performed best and the winter-active perennial grasses the worst, despite an aseasonal rainfall pattern and several wet autumns and springs (Figure 1).Of the minor groups tested, e.g.shrubs and saltbushes, none gave an agronomically encouraging performance.Mulga (Acacia aneura TN59) persisted well at Sites L and U, but grew slowly.Although A. angustissima and Indigofera schimperi persisted well, they showed weed potential because of low palatability and root suckers, and were removed from 1995 onwards.Some Eragrostis spp.had low palatability and high seed set, e.g.Cochise and all 3 E. lehmanniana lines, and were dug out systematically over 3-5 years in the mid-1990s.Native species and tropical grasses generally set seed in their first growing season, notable exceptions being Themeda triandra and Cassia sturtii.However, many exotic species failed to set seed in the first year after planting despite fair shoot growth, e.g. A. ischaemum, Dorycnium hirsutum, I. schimperi, Atriplex saltbushes, Merremia aurea, A. angustissima and leucaena.
By plant type category.Grasses.The buffel grasses were clearly superior as a group, but on some soils (Sites L and N) Digitaria and Bothriochloa species showed greater early promise.At Site M, new season growth from buffels was often quite yellow, indicating possible low available soil nitrogen and hence low protein content.The Gayndah buffel type, such as CPI71914, CPI60733 and CPI73390, generally performed better than the Biloela type, like CPI73393 and cv.Biloela.CPI71914 had a leafy, low, tight crown structure and the ability to produce new plants close to parent plants.The hybrid Cenchrus CPI61135 grew well at most sites without gaining a potential 'commercially promising' rating.Eragrostis lines grew and seeded very well, as expected, but long-term persistence and palatability were variable (Table 2 and Supplementary Table B).Eragrostis superba had large seedhead spikelets and was quite leafy but had variable and generally low palatability.The E. curvula lines fell into 2 groups: the conferta form, which has a dense seedhead and thick, stiffly erect culms; and var.curvula, which has long, thin leaves and large, open seedheads that often droop.The former were usually palatable (CPI98914), but none was noticeably more so than the commercial cultivar Consol.
Fingerhuthia africana (a winter-green C4 grass) showed colonizing ability, frost tolerance and persistence on hard-setting solodics, as well as reasonable growth, but was not eaten readily by stock.The Urochloa species tested grew well initially at all sites and the stoloniferous CPI60128 showed good vigor and tolerance of moisture stress and appeared quite palatable.Amongst the other non-commercial species, only Panicum stapfianum and Bothriochloa ewartiana scored 3 or better overall before the drought (Supplementary Table A).
Rhodes grass and silk sorghum grew poorly in the dry summers on the relatively infertile soils chosen, but several Digitaria, Dichanthium and Bothriochloa species performed creditably, e.g.Premier, TBA9 (a bred variant), CPI26832, CPI41192 and CPI59786 (Supplementary Table B).Digitaria abyssinica is regarded as a potential weed (Invasive.org 2006) because of aggressive rhizomes and low palatability.
The exotic Bothriochloa species usually outperformed the exotic Dichanthium species, but native Dichanthium sericeum (Queensland bluegrass) grew well at all sites except M, and the naturalized Medway strain of B. pertusa from central Queensland was outstanding at Sites N, U and W. It shows great potential as a sward-forming grass for use in erosion control and is leafier than most other strains of the species.CPI11408 (B.bladhii ssp.glabra) produces prostrate tillers on winter and spring growth to resist grazing, but its seedheads are later erect and avidly browsed by cattle.This puts seeding in jeopardy and may partly explain why it has not persisted at our sites.Now released as cv.Swann, its origins are unclear, as it is not an Andropogon ischaemum as its passport data indicate.It may not even be from Guyana, but rather a contaminant from a nursery site in Australia.Prostrate tillers were recorded seasonally for other lines such as TN22, TN47 and TBA1024.
Herbaceous legumes.The performance of the summergrowing legumes was not encouraging.On the sandy red soil at Site M, only Stylosanthes and TN15 Rhynchosia minima established well but frosts damaged them severely each winter.Native R. minima and Cullen tenax lines have grown very well at Site L and reasonably at Site N, and their persistence is better than that of most other summer-active legumes.Both sites had reasonable soil fertility and Site N was a non-acidic soil with carbonate at depth.Desmanthus virgatus was not impressive but persisted well initially, and seeded quite freely once it had grown some foliage, but neither line developed into a noticeable stand at any site.The Phaseolus species CPI90752 from Mexico was one of the most persistent perennial legumes but produced only low yields, akin to those of native Glycine tabacina or G. tomentella, which grow at our sites.The atro bean group (Macroptilium atropurpureum) came from a wide range of material from the homeland of the species (Reid 1983) and some had good crown frost tolerance; however, most were susceptible to leaf rust and to persistent summer droughts.None outperformed the commercial cultivar Siratro, which is still not highly regarded in this region (Jones 1998).
Of the annual winter-active legumes, yellow serradella (Ornithopus compressus) was the most consistent performer.Medicago murex grew satisfactorily but seemed to lack adequate hardseededness for persistence in the aseasonal rainfall environment.The 2 lines of naturalized Medicago laciniata tested (TBA7 and TBA13) grew poorly; they persisted and set seed and were difficult to distinguish from other local genotypes of the species.Naturalized woolly burr medic (Medicago minima) built up noticeably at Site U during the trial and was seasonally abundant at Site N.
Others.Sheep's burnet (Sanguisorba minor) did not survive long at the 5 field sites, although it persisted and regenerated from seed for many years at Site T. It was extremely palatable to stock and wilted easily in hot weather, but did not suffer from any obvious pests or diseases.While the other herbaceous plants, TN6 (Evolvulus alsinoides), TN7 (Sida brachypoda) and TN27 (Boerhavia dominii), showed potential as standards for comparisons, they offered no agronomic benefits over the many existing local herbs.Native mulga sida (TN7) and tropical speedwell (TN6) grew reasonably at some sites but produced little edible dry matter.Seed is easy to germinate and TN6 regenerates freely www.tropicalgrasslands.info from seed.TN6 has broader soil tolerance than TN7 and TN27 and its seed was easier to collect and germinate.However, TN27 came from a fertile cracking clay soil and may be different physiologically from the B. dominii found on our trial site earths.
Saltbushes did not thrive, even where very saline subsoils existed and grass competition was minimal, such as at Site W.
By growing season category.Summer-active species.Most summer-growing accessions at Sites U, M and L set seed.Initial persistence was good for most grasses, but legume persistence at Sites M and W was poor, except for a few Stylosanthes plants.This study emphasized the importance of survival after occasional hot, dry summers, either as plants or in a long-lived seedbank.Most summers were drier than 'normal' during our trials and the performance of these grasses augers well, especially as they had negligible disease issues.
Summer-growing legumes performed poorly but improved summer rainfall might enhance their long-term performance, if they set adequate amounts of hard seed to allow regular recruitment of new plants.Most have significant hardseededness (Stylosanthes and Macroptilium) but the latter atro beans were like Siratro and lacked the ability to compete with summer grasses and persist in a grazed pasture (Keating and Mott 1987).The most successful summer legumes were R. minima accessions but they rarely produced much growth.They suffer from legume little leaf virus and have sticky hairs on seed pods that shatter before drying out, making commercial seed production difficult.
Winter-active species.Cutleaf medic (M.laciniata), yellow serradella (O.compressus) and common wheatgrass (E.scaber) flowered in the first growing season and regenerated from seed at Site U. Elymus scaber had few pests or diseases, but seemed to require a fertile soil for adequate growth.Very erect tillering at all times makes it vulnerable to excessive grazing pressure in rangeland grazing systems.Rough corkscrew grass TN10 (Austrostipa scabra) performed creditably and is an obvious choice for a standard against which to compare similar grasses.It is not palatable once dry in summer and its persistence during consecutive dry years is weak, but it does have a very persistent seedbank from which to regenerate, when better cool season rainfall returns.Its sharp seed is most undesirable for wool producers, and its use in field comparisons would have to be well controlled, so as not to result in long-term problems where it was sown.
Aseasonal growers.These include the native Cullen species (formerly Psoralea), mulga, Cassia sturtii and mulga Mitchell grass (Thyridolepis mitchelliana), as well as saltbushes, lucerne, Eragrostis curvula and Fingerhuthia africana.Mulga Mitchell seems unsuited to competing with vigorous grasses in better rainfall environments and, even where it occurs naturally, the Charleville strain (TN22) performed poorly, seemingly unsuited to soils containing any significant amount of clay.As the mulga accession TN59 survived with a very clayey subsoil at Site L, once past the seedling establishment stage, it could be the forage shrub benchmark for semi-arid acid soils, since it grew well at Site U also.However, the species has failed to impress in almost all other semi-arid parts of the world outside Australia, where the soils are generally alkaline, such as the Middle East and North Africa, or are unsuitable in other ways (Gwaze 1989).

Persistence
Long-term persistence of perennial pasture plants is very important to graziers worldwide, as low returns on investments make resowing of pastures unappealing (Murphy 1992).Persistence of the tested lines is summarized in Table 2 based  Fourteen others survived the drought well at all but one site (Table 2; 8 Eragrostis species; 5 perennial grasses; and a native shrub, C. sturtii).These species are adapted to a wide range of soil types, which is advantageous for a successful commercial cultivar.Queensland bluegrass is not in this list, because it is not adapted to acid soils despite its good reputation regionally.In general, the native species (TN codes) showed selective site adaptation and persisted well only at particular sites, despite initially growing well everywhere, e.g.TN10, TN24 and TN27.
A wide range of other species that survived for 6-8 years and recruited seedlings at the less stressful, higher rainfall Toowoomba site, did not survive long at any of the 5 field sites (Table 2).They included legumes, saltbushes, sheep's burnet and the C3 winter-active Eurasian grasses like Agropyron elongatum and Bromus inermis.

The evaluation technique
Establishing a small population of plants in the field from very limited seed supplies proved successful for initial testing.When seedlings or cuttings were grown in the glasshouse, the majority then established in the field with only limited supplementary watering.Planting out after good rain (surface soil wet to >20 cm depth) was most important, but young, pre-flowering plants must be used.If an adequate taproot does not form on nongrasses after transplantation, their drought tolerance may be reduced.This cannot be discounted amongst our results, but we have no direct evidence of it happening.Our method entails more pre-sowing effort but labor costs compared with infrastructure and equipment costs will vary depending on where the study is done and are adaptable to local resource availability.However, in an holistic sense, the major cost is the production, storage and acquisition of exotic seeds.Their death or loss without growth of replacement plants or meaningful data collection is a huge loss.

Locusts, rabbits and kangaroos
Defoliation of small seedlings could seriously compromise the evaluation of an accession.Locusts were a constant threat, particularly in late spring, and could be quite selective, preferring wilted grasses; they were particularly damaging to Rhodes grass seedlings.Winter-active herbaceous species were generally not eaten, because locusts were inactive at that time.Nocturnal rabbits and kangaroos could also heavily overgraze and kill sown plants in small plots, without being seen, and strip planting amongst coarse, unpalatable native species can accentuate this problem.In other countries, small grazers such as jack rabbits (Mexico), antelope (Africa) and capybara (South America) would have a similar effect.Kangaroos appear to prefer Urochloa lines to buffels, which might explain why U. mosambicensis has not been a commercial success in western Queensland.It has become a seasonally prominent roadside species since the 1980s, indicating it has good climatic adaptation.

Species performance compared with existing commercial cultivars
Perennial legumes.This study supports other studies that highlight a serious shortage of perennial herb leg-umes for pastures in semi-arid subtropical Australia (Jones 1998;Strickland et al. 2000;Clem et al. 2001).The stylos tested failed to persist in meaningful numbers, and young seedlings were often yellow and presumably died.Yellow color can indicate poor nodulation, but most Stylosanthes species are non-specific in their cowpea rhizobial symbiosis (Date and Norris 1979).Lack of sufficient frost tolerance may also have been an issue.Recently there has been interest in Caatinga stylo (S. seabrana), which has persisted in other trials (Peck et al. 2012) but was not included in ours.While Siratro and its relatives were very disappointing, their failure did not seem to be due to pests or diseases and was replicated on adjoining trials and in other studies (Clem et al. 2001).
Winter-active grasses.The next most important deficiency in current commercial pasture plants for this region is winter-active grasses.While many Eragrostis species grew well in winter, when soil moisture was available, most were seasonally unpalatable, making them a future weed risk.As E. curvula is a declared weed in nearby New South Wales and regarded as weedy in the Darling Downs and Kingaroy regions east of the Maranoa (DAFF 2013), we are unwilling to subject them to wider testing.
Annuals.Annual medics such as Medicago minima, M. polymorpha and M. laciniata are slowly spreading in this region as their level of hardseededness and their rhizobial symbionts slowly alter genetically and adapt, we consider, to the environment.Their burrs are contaminants in wool and mohair, and their foliage has potential to cause bloat in good spring seasons.Sown cultivars of barrel medic (M.truncatula) grow and persist well on some sandy red earths and we failed to identify anything better adapted to acid soils.Inadequate hardseededness seemed to be a major deficiency of the medic lines that we tested.Yellow serradella grew well for some years, especially on sandier soils, but eventually died out.Whether a bigger area, that was less susceptible to overgrazing by rabbits and kangaroos, would have improved their persistence is not known.
Native species.Few of the native species tested showed promise agronomically.Queensland bluegrass was the best for the more friable, low acidity soils (Supplementary Table B) and it is already recommended for specialized rehabilitation purposes (Huxtable 2003).Desert bluegrass TN47 also showed potential on all soils, but it was more difficult to establish this pasture from seed and www.tropicalgrasslands.info to achieve good increase in the stand from a few plants.Once established, it was very persistent but had only moderate palatability (Supplementary Table B) and would be best suited for long-term pastures, while Queensland bluegrass has a valuable role in short-term, legume-rich ley pastures in farming systems.Kangaroo grass TN35 had weaker plant perenniality but better seedling recruitment than TN47, plus the added historical weakness of susceptibility to overgrazing.
The native legumes tested gave low yields and could not be considered for commercialization.Their seed can be difficult to harvest but germinates easily after scarification (if appropriate).The shrubby legumes mulga and C. sturtii have no apparent commercial forage role in the Maranoa, where there are so many other plants with better growth rates, palatability and digestibility.CPI79501 returned to Australia after giving good results in Israel, so it may still have potential overseas somewhere.
Summer-growing perennial grasses.This group offers the greatest scope for sown pastures in the region, establishing well, growing satisfactorily, seeding and then persisting for some years.Their ultimate value depended mainly on their persistence after the 1992/93 drought, seedling recruitment ability and palatability.Buffel grass hybrid line CPI61135 has a non-hairy seed and persisted well but was not as productive as other buffels, nor was its seedling recruitment as extensive.A similar assessment applies to white Birdwood grass CPI17655 that, though in Australia since 1952, has still not been commercially released.CPI71914 was rated more highly than the similar cv.Gayndah at most sites (Supplementary Table A), but the small advantage does not justify the expense and complications involved in setting up a seed production industry for it.As most people find Gayndah indistinguishable from cv.American, its release would make cultivar integrity very hard to maintain.
Premier digit grass did not recruit seedlings and its high palatability meant that it was often overgrazed in a mixed pasture, so did not persist as well as in higher rainfall areas (McDonald et al. 1998).While stoloniferous Digitaria species grew quite well, they also produced few seedlings and did not persist well, whereas the equivalent stoloniferous Bothriochloa species persisted better and produced more seedlings.Medway Indian couch grass was the best of these (Supplementary Tables A and B), but with its relatively low dry matter yield and vigorous sward production, it is regarded as a weed threat by some people.It is seasonally palatable and resistant to heavy grazing and could have a role stabilizing local waterways and embankments that cannot be slashed.It was not killed by intermittent light cultivation but is susceptible to prolonged drought.As it seeds freely, eradication schemes would require the use of herbicides.
Inverell purple pigeon grass performed poorly away from the heavy clay soils, where it is used commercially.Pioneer Rhodes grass, Bambatsi panic and Hatch creeping bluegrass all showed insufficient persistence to be considered for commercial recommendation on the soil types studied.The climate is too dry and drought-prone for current Rhodes grass cultivars and Bambatsi needs more friable, fertile soils to persist.We recorded no seedlings of Bambatsi anywhere, which is the norm.Hatch also failed to cope here and was rated lower than CPI52193 (Supplementary Table A), which resembles the Bisset cultivar, both of which root down at the nodes much better than Hatch.
These trials have failed to identify new commercial pasture cultivars for arable, poorly structured soils in the subtropical aseasonal rainfall region of southern inland Queensland.However, we have identified pasture species for use as standards in any future assessment program (Table 6).In view of the global diversity of the plants tested, we consider pasture users in many other subtropical regions could benefit from this information.

Table 5 .
Comparison of seasonal conditions at the trial sites over the period 1986-1991 against the long-term data shown in Figure 1.Numbers are the total for all 3-monthly seasons based on 20 seasons (5 years) for Sites L and M and 24 seasons for Sites N, U and W. Seasons (summer, autumn, winter and spring) are described in the Figure 1 caption.
on data collected in 1994 and 1995 from the 5 field sites and in 1991/92 at Site T. Only the following 7 accessions persisted in the long term at all field sites sown, 5 of them being Cenchrus species: CPI17655 C. setiger; CPI71914 C. ciliaris; CPI73393 C. ciliaris; Q10077 C. ciliaris; CPI61135 Cenchrus hybrid; Medway Bothriochloa pertusa and TN47 B. ewartiana.

Table 1 .
Details of the soil and surrounding vegetation at each trial site.

Table 3 .
Numbers of accessions tested in each major forage type group.

Table 4 .
Sowing times, early rainfall received and key site management events during the first 2 years at each site, except the Toowoomba nursery.
Digitaria abyssinica, D. brownii, Themeda triandra, Dactyloctenium spp.andCullen patens were the most difficult species to germinate reliably, and often fewer than 12 plants were sown per site.Lines with the poorest establishment after transplantation (see Supplementary TableA) included a shrub (Atriplex halimus) and the grasses Dactyloctenium spp., Eragrostis truncata and Andropogon ischaemum.Perennial cool seasongrowing lines established well only at Site T (Toowoomba nursery), where the Agropyron species, Elymus scaber and Bromus inermis grew vigorously and spread in a relatively competition-free environment.Wynn cassia (Chamaecrista rotundifolia) grew but failed to set seed at any site, in contrast with its mediocre performance in other studies in the region

Table 6 .
Recommended standard species or cultivars, against which new forage germplasm for infertile soils in southern inland Queensland should be assessed prior to commercial release.Species in parentheses were not assessed in this experiment; -= none can be recommended yet, but an alternative is named; some of the alternatives are naturalized, known or potential weedy species and should be used with care.

Table A .
Summary of overall field performance by each accession on a 1 (very poor) to 5 (very good) scale, as well as for ease of establishment, persistence, spread and yield in the first 6-7 years after sowing.Site Lhard-setting, grey, shallow duplex soil with impervious sodic subsoil; good available P Site Mdeep, infertile, acidic, sandy red earth Site Ngrey, slightly alkaline clay loam with carbonate nodules at depth Site Uslightly acidic, sandy red earth without subsoil constraints Site Whard-setting, grey, shallow duplex soil with impervious sodic subsoil Site Tdeep, friable red loam with fair organic carbon levels www.tropicalgrasslands.info Continued www.tropicalgrasslands.info Continued www.tropicalgrasslands.info Continued www.tropicalgrasslands.info