Predicting plant yields using soil testing for phosphorus: Effects of error in the estimation of the maximum yield on interseasonal comparisons of yield prediction

Results from long-term field experiments in south-western Australia are presented in the form of relationships between yield, expressed as a percentage of the maximum yield, and soil test for phosphorus (P) values. Maximum yields were not always indicated by well defined yield plateaux. Different methods have been used to estimate the maximum yield value which is used to calculate yield as a percentage of the maximum yield so as to remove interseasonal variation. For all of these methods and for the same site, the same P fertilizer (superphosphate), and the same plant species, the relationship between yield and soil test P differed for different years. Consequently fertilizer recommendations based on the assumption that this relationship is constant are likely to be incorrect. We therefore question the validity of the common practice in soil testing programmes of using percentage yield values to remove interseasonal variation.     

The long-term residual value of rock phosphate and superphosphate fertilizers for various plant species under field conditions

In three, long-term field experiments on different lateritic soils in south-western Australia, the effectiveness of superphosphate and rock phosphate fertilizers applied 10 years (one experiment) or 4 years previously was measured relative to the effectiveness of freshly-applied superphosphate (relative effectiveness or RE) using several different plant species. For the species comparisons, RE values were estimated using the initial slope of the relationship between yield and the level of P applied. In addition, RE values were also determined for different levels of application to test whether RE values for previously-applied fertilizer changed with increasing level of application. Soil samples were collected 3–5 months before sowing for a soil test for phosphate (P) and the soil test values were related to plant yields measured later that year. At each site, the RE value of previously-appliedrock phosphate was calculated using initial slopes and was mostly consistently low and was similar (0.04–0.18) for all plant species. The exceptions were that the RE value about doubled for barley in one experiment and for another experiment the effectiveness of calcined (heated) C-grade ore (Calciphos) was about 2–3 times that of the untreated (i.e. unheated) fertilizer. In most cases, the RE value of previously appliedsuperphosphate at each site was similar (0.23–0.34) regardless of plant species. The exceptions were that the RE value was about double for barley in one experiment and about half for triticale in another. Rock phosphates applied 4 or 10 years previously were between about one twentieth to one quarter as effective as freshly applied superphosphate. Superphosphate applied 4 or 10 years previously was between about a quarter to one third as effective as freshly-applied superphosphate. At each site, the yield of each species was closely related to the P content of plant tissue and the relationship was independent of the fertilizer type or when the fertilizer was applied. At each site and for each plant species, the RE value of the previously-applied rock phosphate was estimated for different levels of application and generally decreased with increasing level of application, whereas the RE value for previously-applied superphosphate mostly remained approximately constant. At each site, the relationship between yield and soil test values (i.e. soil test for P calibrations) differed depending on the fertilizer type and the plant species.     

A comparison of five soil phosphorus tests for five crop species for soil previously fertilized with superphosphate and rock phosphate

The P_i, Colwell, Bray 1, calcium acetate lactate (CAL) and Truog phosphorus (P) soil test reagents were assessed in two field experiments on lateritic soils in Western Australia that had been fertilized four years previously (1984) with triple superphosphate, North Carolina rock phosphate, Queensland rock phosphate, and in one experiment, Calciphos. Soil samples to measure soil P test were collected February 1987. Soil P test was related to seed (grain) yields measured later in 1987. Different crop species were grown on different sections of the same plot at each site. The species were lupins (Lupinus angustifolius), barley (Hordeum vulgare) and oats (Avena sativa) at one site, and lupins, oats, triticale (×Triticosecale) and rapeseed (Brassica napus) at the other site. For each reagent, the soil P test calibration, which is the relationship between yield, expressed as a percentage of the maximum yield, and soil P test, generally differed for different plant species and for different fertilizer types. Variations in soil P test required to produce half the maximum yield of each species at each site was least for the CAL reagent followed by the Colwell reagent.     

Summary of research on soil testing for rock phosphate fertilizers in Western Australia

The relationship between plant yield and values of soils tests for phosphorus (P) was studied in long-term field experiments in south-western Australia for soil previously fertilized with rock phosphate and superphosphate. The rock phosphates studied were: Queensland (Duchess) apatite rock phosphate; reactive apatite rock phosphate from North Carolina; and rock phosphate from Christmas Island (as either C-grade ore or Calciphos). The P fertilizers were applied once only at the start of each experiment, and in subsequent years, soil samples were collected in January-March to measure soil test values. These were compared with plant yields measured later on in that year. The Colwell alkaline bicarbonate soil test was used in all years in all experiments. Olsen, Bray, lactate and Troug tests were used in some years in some experiments. For all soil tests the relationships between yield and soil test values was generally different for rock phosphate and superphosphate. For a given source of P, none of the different soil test reagents was significantly superior for predicting plant yields. The relationship between yield and soil test value was also generally different for different plant species. At one site cultivation was included as a treatment and the relationship varied depending on the cultivation treatment of the topsoil before sowing oats (Avena sativa). The relationship between yield and soil test also differed between years.     

Comparison of the Pi,Colwell, Bray 1, calcium acetate lactate (CAL) and Truog soil phosphorus test for predicting growth of oats,barley, triticale and clover in the field in lateritic soils fertilised with superphosphate and rock phosphate

The P_i soil test for phosphorus (P), which uses an iron oxide impregnated paper, was evaluated in three field experiment on lateritic soils in south-western Australia fertilised with triple superphosphate, North Carolina rock phosphate, Queensland rock phosphate, and in one experiment, Calciphos. Soil samples were collected February to March from 1985 to 1988. The P_i, Colwell, Bray 1, calcium acetate lactate (CAL) and Truog soil P tests were used. Soil test P values were related to yields of triticale (×Triticosecale) or oats (Avena sativa), barley (Hordeum vulgare) or dry herbage yields of subterranean clover (Trifolium subterraneum). The Colwell soil test, which is commonly used in Australia, and the Pi soil test were almost equally predictive, but showed considerable error in prediction of yield. For each soil test and plant species the relationship between yield and soil test P differed with fertilizer type and year. For combined data for all sites, fertilizers and years, the CAL soil test was the most predictive and the Truog soil test was least predictive of plant yield.     

Residual value of superphosphate measured using yields of different pasture legume species and bicarbonate — extractable soil phosphorus

The residual value of superphosphate was measured in three glasshouse pot experiments using three different lateritic soils (pH CaCl₂: 4.8–5.3) from south-western Australia. The residual value was estimated relative to levels of freshly-applied superphosphate using yield of dried tops and bicarbonatesoluble P extracted from the soil (soil test values). Up to five successive crops were grown. In each experiment, four different pasture legume species fertilized with mineral nitrogen were grown in rotation with a cereal species. The legume species includedMedicago polymorpha, M. murex, Trifolium subterraneum, Ornithopus compressus, O. perpusillus andO. pinnatus. The cereal species includedTriticum aestivum, ×Triticosecale, andHordeum vulgare. The comparative phosphorus (P) requirement of the different pasture legumes was estimated from the amount of P required to produce 50 or 90% of the maximum yield measured for each species at each harvest. Soil samples for the soil test were collected just before sowing each crop, and were related to the plant yields of that crop.Relative to freshly-applied superphosphate, the residual value of superphosphate measured using plant yield was similar for all pasture legume species, and decreased markedly, by about 50 to 80% between the first and second crop, and by a further 5 to 30% for subsequent crops. The decrease in residual value estimated using soil test values was less marked. For freshly-applied superphosphate, and for the same plant species, the relationship between yield and the level of P applied differed for different crops.There was no consistent, systematic trend for the comparative P requirement of the different legume species within and between crops of the three experiments and soils.For all crops, the relationship between yield of dried tops and P concentration in dried tissue generally differed for the different legume species, indicating the different species usually have different internal efficiency of P use curves. However, for each experiment, when the same cereal species was grown in all the pots, the relationship between yield and P concentration in tissue was similar for previously- and freshly-applied superphosphate, regardless of the pasture legume species grown in previous crops.The relationship between yield and soil test values usually differed, within each crop, for different plant species and for previously- and freshly-applied superphosphate. For the same plant species, the relationship also differed between different crops.     

Evaluation of the Bray 1, calcium acetate lactate (CAL), Truog and Colwell soil tests as predictors of triticale grain production on soil fertilized with superphosphate and rock phosphate

In a field experiment in Western Australia, six different levels of three different phosphorus (P) fertilizers (triple superphosphate, TSP; Queensland (Duchess) rock phosphate, QRP; North Carolina rock phosphate, NCRP) were applied at the start of the experiment in 1984. Grain yield of triticale (×Triticosecale) was measured from 1984 to 1988. In February-March of each year from 1985 to 1988, soil samples were collected to measure soil extractable P (soil test values) using four reagents (Bray 1, calcium acetate lactate (CAL), Truog and Colwell). Soil test values were related to triticale grain yields, determined either as absolute yield or percentage of the maximum yield, produced later on in each year. The relationship differed with fertilizer type, reagent and year. All four soil test reagents were equally predictive of yield. It is concluded that these soil P tests provide crude predictions of plant yield regardless of the reagent used.     

Cultivation reduces fertilizer residual effectiveness and affects soil testing for available phosphorus

The agronomic effectiveness of superphosphate and two rock phosphates that had been applied once only to the soil surface 8 to 12 years previously was measured in a field experiment with oats on a lateritic soil in south-western Australia. The soil was either undisturbed or cultivated with a rotary hoe before sowing. The rock phosphates were Christmas Island C-grade ore (C-ore, a calcium ironaluminium rock phosphate), and C-ore calcined (heated) at about 500°C (Calciphos).Cultivation reduced the effectiveness for all three fertilizers by 20 to 50%. The effectiveness of phosphorus (P) applied as superphosphate decreased with increasing period from time of application whereas the effectiveness of the rock phosphates increased but they were always much less effective than superphosphate.The relationship between grain yield and P concentration of plant tissue (i.e. the internal efficiency of P use curve) was similar regardless of fertilizer type, year of application of fertilizer, and whether or not the soil was cultivated. Thus differences in fertilizer residual effectiveness were solely due to the amount of P taken up by the plants.Values of bicarbonate-soluble P (i.e. soil test for P values) for superphosphate treated soil were reduced by about 20 to 25% when the fertilizer was incorporated into the soil whereas for the rock phosphate treated soils the values were little affected by cultivation. The relationship between yield and soil test for P values varied depending on cultivation treatment and fertilizer.We conclude that cultivation decreases the effectiveness of residual fertilizer P and that cultivation and fertilizer type influence the accuracy of yield prediction from soil test values.     

Tissue testing for phosphorus in dried tops and seed: Effect of fertilizer type,plant species,and time of sampling within a year and in different years

The apparent relationship between yield and phosphorus (P) concentration in dried whole tops or seed was studied in field experiments. Yield and P concentration in tissue were measured at the time of sampling for several harvests within each year to estimate the current P status of the plants. For each sampling time, there were close relationships between yield and P concentration in tissue that gave a good indication of the current P status of the plants. For the same site and plant species, the relationship did not differ for different fertilizer types (superphosphate or rock phosphate) or method of applying the fertilizer. However, the relationship did differ for different harvests within and between years, and differed for different plant species.Tissue test for P values are sometimes used to predict future plant yields. In the experiments, the P concentrations measured in tissue harvested earlier during the growth of the annual plants were related to subsequent plant yields measured later on within each year. The relationship between future plant yields and tissue test for P values differed for different P fertilizer types and for different sampling times, both within and between years. It also differed for different plant species.     

The current and residual value of superphosphate for lupins grown in rotation with oats and wheat on a deep sandy soil

In a field experiment on a deep pale-yellow sand in a 600 mm per annum rainfall Mediterranean environment of south-western Australia, six levels of phosphorus (P) as superphosphate (O up to 546 kg P ha^–1) were applied once only, to the soil surface, before sowing lupins (Lupinus angustifolius). The lupins were grown in a continuous arable cropping rotation with, in successive years, oats (Avena sativa), wheat (Triticum aestivum), lupins. Five such rotations were started in the experiment from 1985 to 1989. The experiment continued until the end of 1990.The relationship between lupin seed (grain) yields and the level of P applied was measured in the year of P application for five successive years (1985 to 1989). The relationship had the same general form but it varied between years, largely due to different maximum yields (yield plateaux) in each year.The residual value of superphosphate applied three years previously was measured for lupins on two occasions (1988 and 1989) relative to superphosphate applied in the current year. The residual values was different in the two years. The superphosphate applied three years previously was about 30% as effective as freshly applied superphosphate in 1988, and 12% as effective in 1989.At each harvest, the relationship between grain yield and the P concentration in the grain differed for different species. However, for each species at each harvest, the relationship was similar regardless of when the P was applied in the previous years. Thus each species had the same internal efficiency of P use curve, and yields varied only with P concentration in tissue.Bicarbonate-extractable soil P was determined on soil samples taken in mid-July of 1989 and 1990. These soil test values were related to grain yields at harvest. The relationship between yield and soil test values had the same general form but varied for different species within years and for each species between years. It also varied for each species within years depending on the year the P was applied.     

Residual value of superphosphate for wheat and lupin grain production on a uniform yellow sandplain soil

In a field experiment on a sandplain soil in a low rainfall (326 mm per annum) Mediterranean environment of south-western Australia, the effectiveness of superphosphate applied in 1986 was measured in three subsequent years relative to freshly-applied superphosphate each year, using grain (seed) yields of wheat (Triticum aestivum) and lupins (Lupinus angustifolius). The wheat and lupins were grown in rotation and both crops were grown each year starting in 1986. Bicarbonate-soluble phosphorus was determined on soil samples taken in mid June from where the P treatment was applied in 1986 only. These soil test values were related to the grain yields produced that year.For each level of superphosphate applied in 1986, soil test values decreased with increasing time from application. The relationship between grain yield and soil test values had the same general form within each year for both plant species, but varied between years.For both species, the effectiveness of superphosphate decreased by about 70–80% between the year of application and the first and second years after application, and by a further approximate 10% in the third year. The relationship between grain yield and the level of superphosphate applied became sigmoidal by 1989.     

Reactive rock phosphate fertilizers and soil testing for phosphorus: The effect of particle size of the rock phosphate

North Carolina rock phosphate (NCRP) (highly carbonate—substituted apatite) was ground to produce three samples with different particle size distributions. The effectiveness of these fertilizers was compared with the effectiveness of superphosphate in a field experiment and three glasshouse experiments using lateritic soils from south-western Australia. Non-reactive Queensland rock phosphate (low carbonate-substituted apatite from the Duchess deposit) was also used in the pot experiments. Bicarbonate-soluble phosphorus extracted from the soil is widely used in Western Australia to predict plant yields from previously-applied fertilizer dressings. For both field and pot experiments bicarbonate-extractable phosphorus (soil test value) was measured and related to subsequent plant yields.As calculated from the initial slope of the relationship between yield and the level of P applied, finely powdered NCRP was about 5–32% as effective as freshly-applied superphosphate in the year of application and also for two years after application in the field experiment, and for two successive crops in the three pot experiments. For both field and pot experiments, finely powdered NCRP, was at best, 1.5–2.0 times as effective as granular NCRP. Relative to freshly-applied superphosphate, the effectiveness of rock phosphates usually decreased with increasing level of application.For each of the crops in the field experiment, the relationships between yield and phosphorus content of plants (i.e. internal efficiency curves) were similar for all fertilizers. Thus the low effectiveness of the rock phosphates relative to superphosphate was solely due to much less phosphorus being taken up by plants. By contrast, in the pot experiments internal efficiency curves differed for different fertilizers. This is attributed to differences in the rate of phosphorus uptake by plant roots during the early stages of plant growth.For both field and pot experiments, soil test calibrations (the relationship between yield and soil test value) differed for rock phosphates and superphosphate. For superphosphate, soil test calibrations also differed for the three different years after the initial application of this fertilizer in the field experiment. For the second crop in the pot experiment, soil test calibrations differed for superphosphate applied at different times (before the first and the second crop). These results point out the difficulty of applying soil testing procedures to soils that have experienced different histories of fertilizer application.     

Evaluation of two rock phosphates and a calcined rock phosphate as maintenance fertilizers for crop — pasture rotations in Western Australia

Two long-term (11 and 12 y) field experiments in south-western Australia are described that measured the relative effectiveness of three rock phosphate fertilizers (C-grade ore, Calciphos and Queensland (Duchess) rock phosphate), single, double and triple superphosphate. The experiments were on established subterranean clover (Trifolium subterraneum) — based pasture that had received large, yearly, applications of single superphosphate for many years before the experiments began so that in the first year the nil phosphorus (P) treatment produced 80 to 90% of the maximum yield. The experiments were conducted using a rotation of one year cereal crop (oats,Avena sativa at one site, and barley,Hordeum vulgare, at the other): 2 y pasture, a typical rotation on farms in the region. Five levels of each P fertilizer were applied every third year with the crop. Grain yield of cereals, P content of grain, pasture yield, and bicarbonate-soluble P extracted from the soil (available P) were used to estimate fertilizer effectiveness values.The three superphosphate fertilizers had identical values of fertilizer effectiveness. Superphosphate was always the most effective fertilizer for producing grain. The rock phosphate fertilizers were one-seventh to one-half as effective per kg P as superphosphate when assessed on the yield or P content (P concentration × yield) of grain within each cropping year. Bicarbonate-extractable soil P values demonstrated that superphosphate was two to fifteen times as effective as the rock phosphate fertilizers. The relationship between grain yield and P content in grain (i.e. the internal efficiency of P use curve) was similar for the different P fertilizers. Thus for all P fertilizers yield was not limited by other factors as it varied solely in response to the P content, which in turn presumably depended on the P supply from the fertilizers.The relative agronomic effectiveness of rock phosphates is greater for marginally P deficient soils than for highly P deficient soils but rock phosphate remains less effective than superphosphate. We conclude that the rock phosphates studied should not be substituted for superphosphate as maintenance fertilizers for soils in Western Australia that are marginally deficient in P. This result is consistent with the results of many field experiments on highly P deficient soils in south-western Australia. These have shown that a wide variety of rock phosphate fertilizers are much less effective than superphosphate in both the short and long term.     

Effect of fertilizer type,sampling depth,and years on Colwell soil test phosphorus for phosphorus leaching soils

The relationships between (i) soil test phosphorus (P) (Colwell sodium bicarbonate procedure) and the level of P applied (from 0 to 1000 kg total P ha^–1) (relationship 1), and (ii) yield and soil-test P (relationship 2, the soil P test calibration), were measured in two field experiments on very sandy, P-leaching soils in the high rainfall (> 800 mm annual average) areas of south-western Australia. The soils were humic sandy podzols, or haplohumods, comprising 97% sand (20 to 2000 m). The experiments started in April 1984 and were terminated at the end of 1990. Soil-test P, measured on soil samples collected to 5, 10 and 25 cm depth each January in the years after P application, was related to yields of dried clover (Trifolium subterraneum) herbage measured later in each year. The four P fertilizers studied were single superphosphate, coastal superphosphate (made by adding, just before granulation, extra rock phosphate together with elemental sulphur while manufacturing single superphosphate), apatite rock phosphate, and Calciphos.Relationship (1) was adequately described by a linear equation (R² > 0.80, most being > 0.90). The slope coefficient estimates the extractability of P from the soil by the Colwell procedure, and is called extractability. Relationship (2) was adequately described by the Mitscherlich equation (R² > 0.75, most being > 0.90). For relationship (2), use of percentage of the maximum (relative) yield eliminated differences due to different maximum yields and yield responses (maximum yield minus the yield for the nil-P treatment). Soil test P ranged from about 4 to 150 g Pg^–1 soil. Soil test P and extractability were generally higher for samples of the top 5 cm of the soil than the top 25 cm, and were largest for single superphosphate and lowest for apatite rock phosphate. Both extractability (relationship (1)) and the curvature coefficient of the Mitscherlich equation (relationship (2)), differed for different P fertilizers and different soil sample depths. The curvature coefficient also differed for different yield assessments (harvests) in the same or different years. Different soil P test calibrations were required for different P fertilizers, soil sample depths and harvest in the same or different years. It is concluded that soil P testing provides a crude estimate of the current P status of P-leaching soils in Western Australia.     

Response of intensively grazed ryegrass dairy pastures to fertiliser phosphorus and potassium

Intensively grazed, rain-fed dairy pastures on the predominantly sandy soils in the high rainfall (>800 mm annual average) Mediterranean-type climate of south-western Australia comprise >90% ryegrass (annual ryegrass, Lolium rigidum Gaud. and Italian ryegrass, L. multiflorum Lam.). To maximise pasture use for milk production, the pastures are rotationally grazed by starting grazing when ryegrass plants have 3 leaves per tiller, and fertiliser nitrogen (N) and sulfur (S), in the ratio of 3–4 N and 1S, need to be applied after each grazing for profitable pasture dry matter (DM) production. In addition, farmers usually also apply low levels of phosphorus (P) and potassium (K) fertiliser to these pastures after each grazing, despite Colwell soil test P usually being well above critical values for pasture production, and fertilizer K being only required for clover in the traditional clover (Trifolium subterraneum L.) ryegrass pastures of the region. In field experiments undertaken May 2006–June 2010 on intensively grazed ryegrass dairy pastures in the region, no significant ryegrass DM responses to applied fertiliser P or K were obtained, regardless of level or method of P or K application. When no P was applied, soil test P declined gradually, by between 4.4 and 7.1 mg/kg per year, and remained above the critical value for the soils at 2 sites, but declined below the critical value for soil at a third site. Critical soil test P is located near the maximum yield plateau in the flat part of the relationship between yield and soil test P, particularly when, as appropriate for dairy production, the critical value is for 95% of the maximum pasture DM yield. Consequently, when no P is applied and soil test P decreases, significant pasture DM yield decreases will only occur when soil test P approaches the steeper part of the relationship, which can take some time. In addition, as occurs on farms, faeces deposited by cows while grazing supplied P to pasture even when no fertiliser P was applied. Soil K testing proved unreliable for indicating the need for fertiliser K applications to pasture in the next growing season because many soil samples collected within and between urine patches contained elevated levels of K deposited by cows while grazing. We conclude fertiliser P should only be applied to intensively grazed ryegrass dairy pastures when soil testing indicates it is required. Further research is required to assess if plant K testing is an alternative, but urine patches may also pose a problem for plant testing.     

A comparison of seven soil P tests for plant species with different external P requirements grown on soils containing rock phosphate and superphosphate residues

Seven soil tests for phosphate (P) (Bray 1, Bray 2, Truog, ammonium oxalate, Colwell, iron oxide-strip (Pi) and resin-strip soil tests) were evaluated for predicting the yield of plant species which have very different external P requirements. Two acid, sandy soils that had been fertilized six years previously with superphosphate and three rock phosphates were used. A glasshouse pot experiment with lettuce, wheat and maize was used to calibrate the soil tests.For some soil P tests, different calibrations relating yield to soil P test values were required for each plant species, P fertilizer and soil combination. The Bray 2 and Truog soil P tests were the worst predictors of yield for both soils and all plant species. The Pi and ammonium oxalate tests were the most predictive tests for one soil when data for all fertilizers were considered. The Bray 1 and Colwell soil P tests were the most predictive for the other soil. The resin-strip P test was poorly predictive of yield of lettuce and wheat for both the soils. The accuracy in prediction of yield on the basis of P test value decreased in the sequence maize > wheat > lettuce. This rank is opposite to the increasing external P requirements of these species.     

Comparison of single and coastal superphosphate for subterranean clover on phosphorus leaching soils

Coastal superphosphate, a partially acidulated rock phosphate (PARP), is being considered as an alternative fertilizer to single superphosphate for pastures in high rainfall (> 800 mm annual average) areas of south-western Australia. The effectiveness of single and coastal superphosphate, as P fertilizers, was measured in two field experiments using dry herbage yield of subterranean clover (Trifolium subterraneum). The experiments were started in April 1990 and were terminated at the end of 1993. In the years after P applications, soil samples were collected each January to measure Colwell soil-test P, which was related to plant yields measured later on that year, to provide soil P test calibrations.Relative to freshly-applied single superphosphate, the effectiveness of freshly-applied coastal superphosphate and the residues of previously-applied single and coastal superphosphate were less effective in some years (from 3% as effective to equally effective), and up to 100% more effective in other years. This large range in effectiveness values in different years is attributed to different climatic conditions. Soil P test calibrations were different for soils treated with single or coastal superphosphate. The calibrations were also different for different yield assessments (harvests) in the same year, and in different years. Consequently soil P testing can only provide a very crude estimate of the current P status of the soils.     

The maximum yield potential of subterranean clover swards achieved from superphosphate applications is influenced by the maturity and the plant density of the swards

In two field experiments, one conducted in 1987 and the other in 1988, the effect of maturity grading (days from emergence of seedlings to appearance of first flowers) of four clover cultivars (Trifolium subterraneum cvv. Northam, Dalkeith, Junee and Karridale) on the relationship between yield (herbage and seed) and the level of superphosphate applied was measured for dense (170 kg seed ha^–1), single-strain, undefoliated swards. In another two field experiments, one conducted in 1987 and the other in 1988, the effect of the density of clover plants, produced by sowing 1, 10, 100 and 1,000 kg seed ha^–1 ofT. subterraneum cv. Karridale in 1987 and cv. Junee in 1988, on the relationship between yield (herbage and seed) and the level of superphosphate applied was measured for single-strain, undefoliated swards. In all experiments, phosphorus concentration in dried herbage or seed (tissue test for P) were related to plant yields.For herbage production, the maturity of the clover cultivar largely affected the maximum yields (i.e. yield plateaux) achieved for the relationship between yield and the level of phosphorus (P) applied. The P requirements of the different cultivars were similar. For seed production, however, the different cultivars achieved different maximum yields and the P requirement of the clover cultivars were different. The later-maturing cultivars would have experienced greater water stress whilst producing seed which may have affected the P requirements of the different cultivars for seed production.For herbage production in the two plant density experiments, as the density of clover plants in the swards was increased, then the maximum herbage yields for the relationship between yield and the level of P applied increased markedly. The P requirements of the different density swards was in most cases not greatly affected. By contrast, the maximum seed yields tended to decrease drastically with increasing plant density. The 10 kg seed ha^–1 sowing rate produced the largest maximum seed yields. Swards with larger plant densities produced greater herbage yields which presumably increased water stress during seed production thus reducing the maximum seed yields for the relationship between yield and the level of P applied.When tissue test for P values were related to absolute yields, foreach harvest, different relationship between yield and tissue test for P values were required for different clover cultivars or for swards with different plant densities. However, expressing yield as a percentage of the maximum yield for the relationship between yield and tissue test for P values reduced differences foreach harvest due to different clover cultivars or different plant densities. But regardless of whether absolute yield or percentage of the maximum yield were used, different relationships between yield and tissue test for P values were required fordifferent harvest, both in the same or different years.     

The effect of water supply on the response of subterranean clover,annual medic and wheat to superphosphate applications

The effect of water supply on the response of subterranean clover (Trifolium subterraneum), annual medic (Medicago polymorpha) and wheat (Triticum aestivum) to levels of phosphorus (P) applied to the soil (soil P) was studied in four glasshouse experiments. P was applied as powdered superphosphate. In one experiment, the effect on plant yield of P concentration in the sown seed (seed P) was also studied. There were two water treatments: the soil was returned to field capacity, by watering to weight, either daily (adequate water, W1) or weekly (water stress, W2). In three experiments: (i) P concentration or content (P concentration × yield) in plant tissue was related to plant yield, and (ii) soil samples were collected before sowing to measure bicarbonate-extractable P (soil test P) which was related to subsequent plant yields.Compared with W1, water stress consistently reduced yields of dried tops and the maximum yield plateau for the relationship between yield and the level of P applied, by up to 25 to 60% in both cases. Compared with W1, the effectiveness of superphosphate for producing dried tops decreased for W2 by 11 to 45%, for both freshly-applied and incubated superphosphate. Consequently in the field, water supply, which varies with seasonal conditions, may effect plant yield responses to freshly — and previously — applied P fertilizer.Seed P increased yields, for W1, by 40% for low soil P and 20% for high soil P; corresponding values for W2 were 20 and 12%. Consequently proportional increases due to seed P were smaller for the water-stressed treatment.The relationship between yield and P concentration or content (internal efficiency of P use) differed for W1 and W2, so that the same P concentration or content in tissue was related to different yields. Estimating the P status of plants from tissue P values evidently depends on water supply, which in the field, differs in different years depending on seasonal conditions.The relationship between yield and soil test P differed for W1 and W2. Predicting yields from soil test P can only provide a guide, because plant yields depend on both P and water supply, which in the field may vary depending on seasonal conditions.     

Comparative responses of annual pasture legume species to superphosphate applications in medium and high rainfall areas of Western Australia

The comparative phosphorus (P) requirement of different annual pasture legume species was measured in seven field experiments in south-western Australia. Superphosphate was applied once only, at the start of each experiment. The duration of the experiments was from one to three years. The amount of P required to produce 90% of the maximum yield of each legume was used to estimate the comparative P requirements of the legumes at each harvest. Ornithopus spp. (O. compressus, O. perpusillus andO. pinnatus) required less P thanTrifolium subterraneum, the most widely sown pasture legume in Western Australia. The P requirements ofMedicago polymorpha varied with soil type when compared to that ofT. subterraneum. M. polymorpha required less P on a soil with a neutral pH value, but had a similar P requirement on a more acidic soil.M. murex, generally required more P thanT. subterraneum. In some experiments, the comparative P requirement of the different legumes varied for different harvests.At each harvest in each experiment, the relationship between yield and P concentration in tissue (internal efficiency curves) usually differed for different legumes. Presumably different legumes take up P from the soil at different rates within each harvest, and utilize the absorbed P differently to produce herbage and seed. The exceptions were that similar internal efficiency curves were measured forO. compressus andT. subterraneum in one experiment, and three cultivars ofO. compressus in another experiment.