Search

The analysis of series of crop variety trials has a long history with the earliest approaches being based on ANOVA methods. Kempton (1984) discussed the inadequacies of this approach, summarized the alternatives available at that time and noted that all of these approaches could be classified as multiplicative models. Recently, mixed model approaches have become popular for the analysis of series of variety trials. There are numerous reasons for their use, including the ease with which incomplete data (not all varieties in all trials) can be handled and the ability to appropriately model within-trial error variation. Currently, the most common mixed model approaches for series of variety trials are mixed model versions of the methods summarized by Kempton (1984). In the present paper a general formulation that encompasses all of these methods is described, then individual methods are considered in detail.

Summary Following estimation of effects from a linear mixed model, it is often useful to form predicted values for certain factor/variate combinations. The process has been well defined for linear models, but the introduction of random effects into the model means that a decision has to be made about the inclusion or exclusion of random model terms from the predictions. This paper discusses the interpretation of predictions formed including or excluding random terms. Four datasets are used to illustrate circumstances where different prediction strategies may be appropriate: in an orthogonal design, an unbalanced nested structure, a model with cubic smoothing spline terms and for kriging after spatial analysis. The examples also show the need for different weighting schemes that recognize nesting and aliasing during prediction, and the necessity of being able to detect inestimable predictions.

In response to concerns over excessive discarding from Australian recreational round traps (with four funnel entrances) used to target giant mud crabs, Scylla serrata, an experiment was done to assess the independent and cumulative utility of paired, bottom-located horizontal escape gaps (46×120 mm) and increasing mesh size (from 51 to 101 mm). Compared to conventional traps comprising 51-mm mesh throughout, those with the same mesh size and escape gaps caught significantly fewer (by 95%) undersize (

The recommendation of new plant varieties for commercial use requires reliable and accurate predictions of the average yield of each variety across a range of target environments and knowledge of important interactions with the environment. This information is obtained from series of plant variety trials, also known as multi-environment trials (MET). Cullis, Gogel, Verbyla, and Thompson (1998) presented a spatial mixed model approach for the analysis of MET data. In this paper we extend the analysis to include multiplicative models for the variety effects in each environment. The multiplicative model corresponds to that used in the multivariate technique of factor analysis. It allows a separate genetic variance for each environment and provides a parsimonious and interpretable model for the genetic covariances between environments. The model can be regarded as a random effects analogue of AMMI (additive main effects and multiplicative interactions). We illustrate the method using a large set of MET data from a South Australian barley breeding program.

In undertaking cultivar trials, the variability of the response of the cultivars to the different environments in which they are grown introduces the possibility of release errors and non-release errors in the decisions made on the basis of the trial results. In this article a model is developed that accounts for the economic costs of those errors as well as the costs of operating the trials, and enables the features of the optimal cultivar testing program to be identified. The model is illustrated by application to wheat cultivar trials in central and southern NSW.

A fully efficient approach for the analysis of multi-environment early stage variety trials is considered that accommodates a general spatial covariance structure for the errors of each trial. The analysis simultaneously produces best linear unbiased predictors of the genotype and genotype by environment interaction effects and residual maximum likelihood estimates of the spatial parameters and variance components. Two motivating examples are presented and analyzed, and the results suggest that the previous approximate analyses can seriously affect estimation of the genetic merit of breeding lines, particularly for models with more complex variance structures.

Plant improvement programs involve the evaluation of a large number of genotypes (varieties) in a series of designed experiments known as multi-environment trials (MET). The combined analysis of MET data is a complex statistical problem which requires extensions to the standard linear mixed model. The analysis must accommodate spatial correlation structures for the plot errors from each trial and appropriate genetic covariance structures. ASReml (Gilmour, Cullis, Welham & Thompson, 1998) provides a broad range of variance structures for both the errors and the random effects in a linear mixed model. The gains in statistical efficiency resulting from the use of more complex but more realistic variance structures are large. With ASReml they can be achieved at very little extra cost since the algorithm and use of sparse matrix methods ensures timely analyses. In this paper the computational strategy of ASReml will be described and some of the scope of the program will be demonstrated in the analysis of a MET data set.

The retrospective analysis of a large database on wheat variety testing in New South Wales (NSW) is considered. This analysis involved three key steps. Initially error variance heterogeneity is modelled, indicating significant differences in error variance due to trial location, year of trialling, sowing date and trial mean yield. The implication of this modelling for the estimaion of variance components is discussed.

Selection for malting quality traits is a major breeding objective for barley breeding programs. With molecular markers linked to loci affecting these traits, this selection can be undertaken at an earlier stage of the breeding program than is possible using conventional tests. Quantitative trait loci (QTLs) associated with malting quality traits were mapped in 2 populations derived from parents with elite malting quality. Progeny from an Arapiles/Franklin population grown in 4 environments and an Alexis/Sloop population grown in 5 environments were tested for grain protein percentage, α-amylase activity, diastatic power, hot water extract, wort viscosity, wort β-glucan, β-glucanase, and free α-amino acids. QTL analysis was performed using a one-stage approach, which allowed for modelling of spatial variation in the field, and in each phase of the malting quality analysis in the laboratory. QTLs for malting quality traits were detected on all chromosomes and for both populations. Few of these QTLs were significant in all of the environments, indicating that QTL × environment interactions were important. There were many coincident QTLs for traits that are expected to be related such as diastatic power and α-amylase activity, wort β-glucan and wort viscosity and for some traits that are not expected to be related such as hot water extract and malt viscosity.

‘Managing Acid Soils Through Efficient Rotations (MASTER)’ is a long-term pasture–crop rotation experiment commenced in 1992. One of the objectives was to demonstrate the extent of crop, pasture, and animal responses to lime application on a typical acidic soil in the 500–800 mm rainfall zone of south-eastern Australia. Two types of pastures (perennial v. annual pastures) with or without lime application were established in 1992. Fifteen- to eighteen-month-old Merino hoggets were used as test animals and were changed annually. This paper reports the results of sheep responses to liming from the 4 continuous pasture treatments over 6 years from 1992 to 1997. The stocking rate was the same on all plots within a treatment during each rotation period, but was varied between treatments based on the pasture availability and sheep body condition. The most important findings from this study are that the limed treatments carried 29% and 27% more stock (up to 4 DSE/ha) than the unlimed treatments for perennial and annual pastures, respectively. As a result, the limed perennial pastures produced 27% more liveweight gain (62 kg/ha.year) and 28% more greasy wool (13 kg/ha.year) than unlimed perennial pastures, whereas the limed annual pastures produced 34% more liveweight gain (77 kg/ha.year) and 24% more greasy wool (11 kg/ha.year) than unlimed annual pastures. The significant responses to lime in liveweight and wool production were detected from the second growing season after the pastures were established. The increased sheep productivity on the limed treatment was due to a combination of increased pasture production and improved pasture quality. Perennial pastures showed a slight advantage in wool production, but not in liveweight gain. However, the seasonal variation of liveweight was greater on annual pastures than on perennial pastures. The larger variation in liveweight change could lead to more adverse effects on wool quality especially at high grazing pressures. Grazing management can be used to manipulate pasture and animal productivity to increase profits from lime use.

Extensibility and dough strength are key traits for varietal selection in most wheat-breeding programs. As there are several techniques for measuring these traits there is interest in examining the agreement between methods in terms of genotypic (varietal) rankings. We investigated this issue using 2 different extension methods; namely, small-scale (modified Kieffer) and large-scale (Brabender Extensograph) methods. Data were obtained from a doubled-haploid population (190 lines) from a Chara (high extensibility, excellent dough strength) × WW2449 (low extensibility, poor dough strength) cross that was grown in a field trial at the Wagga Wagga Agricultural Institute (WWAI) in 2000. Six extensional rheological traits were measured and compared according to a multivariate mixed statistical model. The estimated genetic correlation matrix for 4 of the 6 extensibility traits (R_Max, area, height, and resistance at 5 cm extension) revealed that for these dough strength related parameters, both methods were measuring equivalent traits. Comparisons of the extensibility traits length and extensibility at Rmax demonstrated that, although substantial amounts of the variance are controlled by the same glutenin loci, the traits differed in the allocation of variance across the loci, and the sources and magnitude of non-genetic variance. The data verified that small-scale testing is a robust and efficient alternative to large-scale testing for both commercial breeding and research.

A doubled haploid population of 120 individuals was produced from the parents Chebec, an Australian 2-row barley of feed quality with resistance to the cereal cyst nematode, and Harrington, a 2-rowed, Canadian variety of premium malting quality. This paper describes 18 field and laboratory experiments conducted with the population and summarises the traits mapped and analysed. The genomic location of 25 traits and genes is described and marker–trait associations for 5 traits (malt extract, diastatic power, resistance to cereal cyst nematode, early flowering, resistance to pre-harvest sprouting) important to Australian efforts to improve malting barley varieties have been used in practical breeding programs. Detailed maps for these populations are shown in this paper, while a consensus map incorporating these maps and further experiments on the populations are described elsewhere in this issue.

Research in Western Australia and South Australia indicated that fertiliser phosphorus (P) banded below the seed of narrow leaf lupin (Lupinus angustifolius L.) at sowing was a more effective method of applying P�fertiliser than the usual placement of P with the seed. This technology has not been investigated in southern New South Wales where lupins have been known to be unresponsive to fertiliser P.We conducted 4 field experiments to examine the effect on lupin yield of applying 6 rates of P (0, 5, 10, 15, 20 and 40 kg/ha) either by placement with or below the seed. To further test responsiveness to P, an additional set of treatments was used; applying P at 40 kg/ha before sowing and then placing additional P below the seed at the 6�rates of application. The grain yield of lupin was increased by P application at all sites, despite the medium to high P�status of 3 of the 4 sites used in these experiments. However, the technique of banding P fertiliser below the seed depth rather than placing it in direct seed contact had only a small advantage in grain yield responsiveness to applied fertiliser P (P = 0.09). Fitted response curves indicated that when P was applied at 15 kg/ha, grain yield increased by 60 kg/ha at one site and 30 kg/ha at the other 3 sites, if P was deep-placed rather than applied in seed contact. This advantage of deep placement of P fertiliser was much smaller than has been reported in Western Australia.Placement of P below the seed of lupin when sown on the red earth and red-brown earth soils of southern New South Wales slightly enhanced the availability of fertiliser P. This applied even when sowing was quite shallow (2–3�cm), provided recommended rates of P fertiliser were used at conventional row spacing (17 cm). Separation of seed and fertiliser to avoid reduced germination may be an advantage when using double row spacing and higher P�application rates.

Two permanent pastures (annual pasture v. perennial pasture) were established in 1992 as part of the long-term field experiment, MASTER — Managing Acid Soils Through Efficient Rotations. The primary objective of the experiment was to develop an agricultural system that is economically viable and environmentally sustainable on the highly acidic soils in south-eastern Australia. This paper reports on the effects of lime on the botanical composition changes of annual and perennial pastures over 9 years. In general, lime increased the proportion of the desirable species, such as phalaris (Phalaris aquatica) in perennial pasture and subterranean clover (Trifolium subterraneum) in annual pastures, and decreased the proportion of the undesirable species, such as Vulpia spp., in both annual and perennial pastures, ultimately improving the quality of feed-on-offer to animals. As a result, the limed pastures carried 24% more sheep than the unlimed pastures, while maintaining individual animal performance similar for both limed and unlimed pastures. The phalaris-based perennial pasture was more stable in terms of maintaining the sown species than the annual pasture. Lime improved the persistence of phalaris and the longevity of the phalaris-based pasture should be at least 10 years. Lime changed the direction of plant succession of annual pastures. Without lime, Vulpia spp. gradually became more dominant while ryegrass and subterranean clover became less dominant in annual pastures. With lime, barley grass (Hordeum leporinum) gradually invaded the sward at the expense of ryegrass, thus reducing the benefits of lime, but this effect was less for the perennial pastures than for annual pastures. Liming perennial pastures should be more beneficial than liming annual pastures because of the beneficial effects on pasture composition. In addition, previously published work reported that liming perennial pastures improved sustainability through better use of water and nitrogen.

A new approach for multi-environment quantitative trait locus (QTL) analysis based on an appropriate genetic model is presented. To accommodate a multi-environment analysis, the size of a QTL effect is assumed to be a random effect. The approach results in a multiplicative mixed model for QTL × environment interaction of the factor analytic type. The full genetic model may also include a factor analytic model for the residual genotype × environment interaction, whereas the environmental model for the non-genetic variation involves local, global, and extraneous variation. The approach is used to determine QTLs for yield in the Arapiles × Franklin doubled haploid population of the National Barley Molecular Marker Program. Analysis leads to the determination of 8 QTLs. Many of these QTLs are associated with other traits.

It is well known that the response to selection for grain yield is improved with the use of appropriate experimental designs and statistical analyses. The issues are more complex for quality traits since the data are obtained from a 2-phase process in which samples are collected from the field then processed in the laboratory. This paper presents a method of analysis for quality trait data that allows for variation arising from both the field and laboratory phases. Initially, an analysis suitable for standard varietal selection is presented. This is extended to include molecular genetic marker information for the purpose of detecting quantitative trait loci. The technique is illustrated using two doubled haploid wheat (Triticum aestivum L.) populations in which the trait of interest is milling yield.

A long-term trial, known as ‘managing acid soils through efficient rotations’ (MASTER), commenced in 1992 to develop and demonstrate a cropping system that is economically viable on the highly acid soils of the traditional permanent pasture region in south-eastern Australia, so that their fertility is sustained or improved. There were 2 permanent pasture systems and 2 pasture–crop rotations, each with and without lime. This paper reports the effect of lime on crop production over the first cycle (6 years). On annual pasture–crop rotations, lime significantly increased the dry matter production at anthesis and grain yields of wheat (cv. Dollarbird) compared with the unlimed treatments. Averaged across years from 1992 to 1997 (excluding the severe drought year 1994), wheat crops produced 1.6 t/ha more grain on the limed treatments than on the unlimed treatments (3.6 v. 2.0 t/ha). On perennial pasture–crop rotations, the lime effects varied with crops grown at each phase and year. For example, despite being tolerant of acidity, oats (cv. Yarran) responded to lime in 1996. Likewise, triticale (cv. Abacus) responded to lime in 1997. Wheat (cv. Dollarbird) that is moderately tolerant to acidity responded to lime in phase 6 from 1992 to 1997 excluding 1994 (3.5 v. 1.7 t/ha). Acid-tolerant wheat varieties, triticale, and narrow-leaf lupins are considered the most viable crops for the soil and climatic conditions encountered in this high rainfall (5000—800 mm per annum) area of south-eastern Australia.

Potassium (K) deficiency of wheat and pasture species was found at a site in the south-western slopes of New South Wales. The soil was a subnatric yellow sodosol. Subterranean clover (Trifolium subterraneum) was found to be less competitive for K in the soil compared with its associated grasses. Higher soil K concentrations were required to achieve the same subterranean clover K concentration in the grass–legume mixtures than in a subterranean clover monoculture. For wheat (Triticum aestivum) production, a soil exchangeable K (Kex ) below 0.25 cmol(+)/kg appeared to be deficient for the limed treatments, but there was no obvious critical value for either limed or unlimed treatments. The critical K exvalues for the grass –legume mixtures could not be simply specified because the values were affected by competition between species growing in swards of variable botanical composition. An annual rate of 20 kg K/ha for the pasture–crop rotations (50/50%) and 29 kg K/ha for the permanent pastures was estimated to be sufficient to replenish the K losses from product removal and animal excreta transferred to campsites at this trial site.

Acidic soils constrain wheat yield in some parts of central and southern New South Wales. This paper describes research designed to evaluate the usefulness of aluminium (Al) tolerance, manganese (Mn) tolerance and the interaction of lime use and Al tolerance in improving wheat yields. Closely related pairs of wheat lines with tolerance and sensitivity of Al and Mn were bred. Carazinho (a Brazilian wheat cultivar) was used as a source of Al tolerance in a backcrossing, recurrent selection program to introduce tolerance into an Egret background (a locally adapted Australian cultivar). Aluminium tolerance was determined using the haematoxylin root tip test and Mn tolerance was determined using a subirrigated gravel bed system. Eight pairs divergent in tolerance of Al were evaluated for their yield on 3 acidic soil types in 5 field experiments. Grain yield increased in Egret-derived lines when Al tolerance from Carazinho was introduced. Yield from the Al-sensitive genotypes (averaged over the 8 pairs) compared with the Al-tolerant genotypes was 0.43 to 0.98 t/ha and 0.88 to 1.38 t/ha respectively on an acid earthy sand in central western New South Wales (Binnaway) in 2 seasons, and 1.08 to 1.96 t/ha and 1.29 to 1.88 t/ha on an acid podsolic soil in southern New South Wales (Borambola). On a moderately acidic red earth site (pH Ca 4.8) at Wagga Wagga, no such advantage accrued to the Al-tolerant group with the average yield for the sensitive and tolerant pairs being 5.00 and 4.78 t/ha, respectively. Manganese tolerance was assessed in only 1 of these experiments (Borambola) using 6 pairs of lines tolerant of Al but with contrasting Mn tolerance. No advantage of Mn tolerance was apparent at this site. At Binnaway the tolerant and sensitive lines responded to lime application with the tolerant lines yielding on average about 0.42 t/ha of grain more than the sensitive lines even when 5 t/ha of lime was applied. At Borambola the tolerant lines yielded 0.59 t/ha of grain more than the sensitive lines when no lime was applied. With lime application this difference in performance disappeared and Al-tolerant and sensitive lines yielded equally. At the Wagga Wagga site, the addition of lime did not affect the yield of the sensitive lines, while the tolerant lines showed a yield depression of 0.32 t/ha of grain with lime application. Three different relationships between lime application and Al tolerance were observed. While 2 relationships fit with our previous understanding, grain yield depression resulting from the combined use of Al tolerance and lime at our least acidic site remains poorly understood.

Summary. Farmers in southern New South Wales began lime application in the 1980s. Many have now limed most of the acidic soils on their properties, and are considering reliming. This is an important economic consideration as lime is a costly input. We accurately located an old lime experiment established in 1982 and applied and incorporated lime in 1992, to give factorial combinations of 5 rates of lime applied in 1982 and 5 rates applied in 1992. The plots were soil sampled and cropped to wheat (cv. Janz) in both 1992 and 1993. The rate of soil pHCa decline in the 0–10 cm soil from 1983 to 1993 following lime application in 1982 was dependent on the pHCa increase achieved 1 year after lime application (1983). The rates of decrease varied from 0.10 pHCa units/year, after 5000 kg/ha was applied, to 0.02 pHCa units/year following application of 500 kg/ha of lime. Evidence in 1992 and 1993 suggested that the pHCa effect of lime applied in 1982 had moved down the soil profile below 10 cm. Wheat yield in 1992 responded to lime applied in 1982 but not to lime applied in 1992. In the 1993 season, the 1982 and 1992 applied lime gave significant yield increases. The response in grain yield in 1993 to 1992 applied lime was greatest where no lime, or low rates of lime, had been applied in 1982. Grain yield in 1993 was described as a function of pHCa in the 0–10 cm and 10–20 cm layers in that season. Maximum yield of the aluminium sensitive cultivar Janz was obtained where the pHCa was about 5.5 in both layers. Where the soil pHCa was 5.0 in the 0–10 cm soil layer, grain yield increased with increasing pHCa in the 10–20 cm layer from 3.2 t/ha at pHCa 4.1 to 3.9 t/ha at pHCa 5.3. Reliming at 2000 kg/ha increased grain yield in 1993 by about the same amount as an initial application of the same lime rate. We suggest that the residual benefit in grain yield was due in part to movement of the lime effect to the subsurface soil. It appears that maximum yields may only be achieved with the amendment of the subsurface soil by a series of lime applications over several decades or by the combined use of shallow incorporated lime and plant tolerance of soil acidity.

Summary. Studies were conducted on a natural pasture dominated by Aristida ramosa (wiregrass) in northern New South Wales. In each of the 4 studies, treatments were designed to reduce wiregrass presence by grazing, using stocking rates up to 12.5–15.0 dry sheep equivalents (dse/ha) in summer and autumn (1983–88), slashing (1984–86), or burning (2 studies, 1985–88; 1986–88) in spring before heavily grazing in summer–autumn. Effects of these treatments on A. ramosa dry matter yield and basal cover, Danthonia linkii (wallaby grass) basal cover, sheep liveweight change and fleeceweight were measured for the different time periods. Wool quality was determined from mid-side samples taken in 1983–85. All treatments reduced A. ramosa dry matter yield and basal cover and increased D. linkii basal cover, compared with control plots continuously grazed at 5 dse/ha. From 1983 to 1988, A. ramosa dry matter yields in the control plots increased from about 1000 to 3500 kg/ha and its basal cover increased from 7 to 13%. In contrast, A. ramosa dry matter yields in heavily grazed, and spring burnt and heavily grazed treatments were

The analysis of a series of crop variety trials often proceeds using a mixed model in which the data are the combined means from individual trials. The residual variation for this model consists of genotype by environment (G.E) interactions and within-trial error variation. The latter is regarded as known from the analyses of individual trials, and any associated heterogeneity can be accounted for in the overall mixed model by the use of weights. The G-E interactions may also have nonconstant variance and, since the variances themselves are often of interest, we propose that heterogeneity arising from these sources be accommodated by modelling the G.E variances as a log-linear function of explanatory variables. We present a residual maximum likelihood estimation method and develop a diagnostic technique for detecting dependence. The approach is demonstrated using a large unbalanced set of crop variety testing data. The methodology is easily generalized to residual variance modelling in any mixed model application.

We identify three major components of spatial variation in plot errors from field experiments and extend the two-dimensional spatial procedures of Cullis and Gleeson (1991) to account for them. The components are nonstationary, large-scale (global) variation across the field, stationary variation within the trial (natural variation or local trend), and extraneous variation that is often induced by experimental procedures and is predominantly aligned with rows and columns. We present a strategy for identifying a model for the plot errors that uses a trellis plot of residuals, a perspective plot of the sample variogram and, where possible, likelihood ratio tests to identify which components are present. We demonstrate the strategy using two illustrative examples. We conclude that although there is no one model that adequately fits all field experiments, the separable autoregressive model is dominant. However, there is often additional identifiable variation present.

In 2 field experiments, varying lime rates were applied to acidic soils and incorporated to 10 or 20 cm depth by rotary hoeing. In Expt 2, soil was also limed in 10-cm layers to 40 cm depth, and the profile was reconstructed. The aim was to remove acidity as a constraint to plant growth. In both experiments, 3 cereal cultivars (1 barley and 2 wheat) of varying aluminium (Al) tolerance were sown. Grain yield was monitored in 4 seasons (Expt 1) or 5 seasons (Expt 2). Incorporation of lime to 20 cm rather than to 10 cm improved yields of the cereals in 2 of 4 seasons in Expt 1, and improved yields markedly in a drier season in Expt 2. In Expt 2, cereal yield was demonstrated to be a function of soil pHCa in both the 0-10 and 10-20 cm layers. There was no advantage of amendment to 40 cm depth compared with 20 cm depth at this site. At both sites, the 10-20 cm soil (corresponding to the upper A2 layer) was very acidic, and the incorporation of lime to 20 cm removed this barrier. The effect of amendment or pHCa increase in the 10-20 cm layer is clearest and most consistent for barley, the most Al-sensitive species, and least consistent for the tolerant wheat cultivars. It is probably not realistic in commercial agriculture to incorporate lime to 20 cm depth. The alternative strategies are to wait for the effects of amendment of the 0-10 cm layer to move down the soil profile, or to amend part of the soil below 10 cm depth by lime injection. Until the subsurface soil is amended, the best strategy is to combine surface liming (0-10 cm) with the use of Al-tolerant cultivars.

Seasonal variation in the response of crops and pastures to limestone application has been observed on acidic soils in south-eastern Australia. Our hypothesis was that temporal variation in soil acidity related factors may contribute to this variable response. Soils from 4 annual pasture sites were sampled at least monthly for 3 years during 1988–1990 to monitor changes in pH(CaCl2) and in concentrations of exchangeable aluminium (Al) and manganese (Mn). The sites received no fertiliser or cultivation and therefore allowed for the estimation of natural temporal variation. Temporal variation in soil pH during a year ranged from 0 to 0·45 pH units depending on the site, soil depth, and the weather conditions. The larger changes in soil pH were associated with more extreme climatic conditions than normal, e.g. following the break of season after a hot, dry summer (autumn 1988) or during periods of above-average rainfall in autumn and early spring (1990). Temporal variation in pH was less than the spatial variability at the sites but greater than the long-term net acidification rate reported for the region. Temporal variation in the concentration of exchangeable Al ranged from 0 to 0·4 cmol(+)/kg during a year and varied primarily with the inverse of pH. Variations in the concentration of exchangeable Mn ranged from 0·05 to 0·35 cmol(+)/kg during a year. The concentration of exchangeable Mn increased over summer to an extent dependent on the drying of the soil. At the 2 sites with duplex profiles, maxima in the concentration of exchangeable Mn also occurred in spring, particularly in the warm wet spring of 1990. Soil tests for soil acidity therefore represent guides to probable risks of toxicity, as pH and the concentrations of exchangeable Al and Mn may change between soil sampling, sowing, and the period of crop or annual pasture growth. Such variations will alter the responsiveness of crops and pastures to lime.

A strategy of using an average information matrix is shown to be computationally convenient and efficient for estimating variance components by restricted maximum likelihood (REML) in the mixed linear model. Three applications are described. The motivation for the algorithm was the estimation of variance components in the analysis of wheat variety means from 1,071 experiments representing 10 years and 60 locations in New South Wales. We also apply the algorithm to the analysis of designed experiments by incomplete block analysis and spatial analysis of field experiments.

Growth of cattle was measured between 1974-75 and 1980-81 on a white clover (Trifolium repens) and carpet grass (Axonopus affinis) pasture receiving superphosphate applied at 0, 12, 24, and48 kg P/ha annually. Hereford weaner steers grazed the pastures at 1.67, 2.5, and 3.3 steers/ha. Linear responses to superphosphate and stocking rate were significant (P

Sultana vines, either own-rooted or grafted to Dog Ridge or Ramsey rootstocks, were trained to 4 trellis systems: 0.3-m narrow T, 0.9-m wide T, arch cane, swing-arm. The swing-arm trellis generally gave the highest yields over the 9 years of the trial. Average annual yields were 19.5, 22.0, 15.6, and 26.2 kg/vine (1442 vines/ha), respectively, for the 4 trellis systems. Vines grafted to Ramsey rootstock had higher overall yields than own-rooted vines and vines grafted to Dog Ridge rootstock (24.0 v. 20.8 v. 17.8 kg/vine). The yield advantage conferred by Ramsey rootstock was greatest in years 1-3 of the trial. There was no significant trellis x rootstock interaction. Trellis effects on yield were primarily due to effects on bunch number, while rootstock had a greater effect on berry and bunch weight. Grafted vines were less fruitful than own-rooted vines. There was no significant difference in quality of fruit dried in a dehydrator, but when fruit was trellis-dried, quality was higher for the swing-arm than for the wide T-trellis, with the other trellises intermediate. Own-rooted vines produced better quality trellis-dried fruit than did vines grafted to Ramsey rootstock. The swing-arm trellis was not well suited to commercial production, but it produced good yields and had other desirable characteristics that are incorporated in some new trellis systems being developed in major dried vine fruit production areas in Australia.

The potential dry matter production in autumn, winter, and spring of 15 annual and 7 perennial pasture legumes was assessed in ungrazed plots at Tamworth on the Northern Slopes of New South Wales. Seed was sown into a prepared seedbed, and dry matter yield was estimated on 24 occasions from August 1983 to November 1987. The presence of green material at each sampling time. and the actual and total dry matters, were examined. Among the annual legumes, the highest proportion of plots presenting dry matter at all sampling times, and the highest total yields, were found for Trifolium hirtum cv. Hykon (rose clover), T. subterraneum var. subterraneum cvv. Seaton Park and Woogenellup, T. subterraneum var. brachycalycinum cv. Clare (subterranean clovers), and Medicago aculeata (CPI No. 19416). The highest proportion of bare plots and the lowest dry matter yields were found for local ecotypes of the naturalised legumes T. glomeratum, M. minima and M. polymorpha, Astragalus hamosus cv. Ioman (milk vetch), and T. subterraneum var. subterraneum cv. Nungarin. After October (day 300) in 1983-87 the proportions of plots with >10 kg/ha of dry matter present as green material were Clare 85%; Hykon 78%; Woogenellup 74%; M. aculeata, Seaton Park, and Vicia villosa var. dasycalpa cv. Namoi (woolly pod vetch) 67%; M. scutellata cv. Sava (snail medic) 56%; and M. truncatula cv. Sephi (barrel medic) 52%. Total annual yields of Nungarin were often lower than those of the other subterranean clovers, with Clare having the highest yield in each year. Cumulative yields of Namoi declined markedly after 1985, while those of T. lairtum cv. Hykon increased. Little or no linear relationship was found between maturity grading and the relative yield of the annual legumes, although 4 of the 6 highest yielding legumes were late maturing. Among the perennials, yields of M. sativa cv. Pioneer 581 (lucerne) were above average in 1984-87. Lucerne cv. Pioneer 581 produced more dry matter over 5 years than Onobrychis viciifolia cv. Eski (sainfoin), but about the same amount as O. viciifolia cvv. Othello and Remont. Sainfoin production was high in the first 2 years of the experiment, but lower towards the end. The poor performance of T. repens cvv. Haifa and Huia (white clover) was attributed to high temperatures and low effective rainfall in summer. From these data, 5 annuals (Seaton Park, Woogenellup, Clare, Hykon, and M. aculeata) and lucerne show the most potential for sowing as ley pasture in the low elevation, summer rainfall environments of the Northern Slopes of New South Wales.

The effect of light on in vitro germination of urediniospores of Tranzschelia discolor was studied over time at different intensitities (up to 400 8E m-2 s-l) within the temperature range 5�C to 20�C. A model was also developed from the data to predict germination at different combinations of light, temperature and times of leaf wetness. Light retarded the germination process, and its effect increased in direct proportion to intensity. At 20�C, for example, the time taken to exceed 80% germination increased from 2 h in the dark to 9 h at 200 8E m-2 s-l. The model showed that there was an interaction between light and temperature, with the effect of light becoming more pronounced as the temperature declined below 20�C. Germination percentages of the order of 90% were, however, recorded within 24 h at all combinations of light intensity and temperature studied. Light also influenced germ tube growth, causing a reduction in the rate of growth. As in germination, its effect increased with increasing light intensity. At 20�C, the average germ tube length at 9 h was 541 8m in the dark, compared with 227 8m at 200 8E m-2 s-1 and 148 8m at 400 8E m-2 s-l. A similar effect was observed at 5�C, where the average germ tube length at 24 h was 274 8m in the dark compared with 157 8m at 200 8E m-2 s-l. The effects of light on the germination and germ tube growth of urediniospores under field conditions are discussed.

Five salinity treatments, ranging between 0.37 and 3.47 dS m-1, were applied through a trickle irrigation system to own-rooted sultana grapevines for six years. The changes in soil salinity levels and the relationship between soil salinity and yield were studied, and a simplified salt balance model was developed to calculate leaching fractions. Soil salinity was strongly influenced by soil texture as well as by salt treatment, because leaching fractions were lower in heavier soils; they averaged 23% in the lightest soils and 10% in the heaviest. Leaching fractions also increased with salt treatment, from 7% in the 0.37 dS m-1 treatment to 24% in the 3.47 dS m-1 treatment. This was probably because water use by salinized vines was lower. Yield was correlated with mean soil salinity, ECe, but the relationship was not as good as with plant salinity levels. The fitted model accounted for between 52 and 62% of the variance. It was concluded that soil salinity levels at the end of winter should be maintained below 1.0 dS m-1 in order to keep yield losses below 10%. For own-rooted sultana grapevines in Sunraysia, this requires a leaching fraction of about 8%. Rootzone depth and root density were lower in the heavier soils, and were decreased by salt treatment. The deleterious effects of salt treatment on clay dispersion and soil hydraulic conductivity were also greater in the heavier soils. Soil properties must therefore be considered when predicting the effects of saline water on crop productivity, especially in the long term.

The effects of the salinity of irrigation water on the productivity of mature grapevines was studied. We injected NaCl solutions into River Murray water to generate five salinity levels ranging from 0.37 to 3.47 dS m-l, with sodium adsorption ratios of between 2 and 37. These treatments were applied through a trickle irrigation system for six years to own-rooted sultana grapevines growing in a vineyard at Dareton, N.S.W. The effect of irrigation water salinity was most severe on vines growing in the most heavily textured soil, and increased with time. The yield (y) response was well described by the generalized logistic function: y = D[1+(ECi/ECih)�]-1 where y is yield, ECi is salinity of irrigation water, D is the theoretical yield at ECi=0, Ecih is the half-effect ECi and a: is the shape parameter. Soil texture affected the response via an effect on ECih and, to a lesser extent, on �. These parameters were related to % (silt & clay) in the final model, which was fitted for each of years 2 to 6 and accounted for 76.2% of the variance in yield. The yield response on the lightest soil most closely resembled the Maas-Hoffman 'bent-stick' model, being relatively flat at low levels of ECi and becoming steeper at moderate salinity levels. Yield losses for heavy soils were much greater than predicted by the Maas-Hoffman model. Bunch number was the primary determinant of yield, while a lower berry weight also contributed to the reduction in yield in salinised vines. Effects on dried fruit quality were apparent only when yield was severely reduced.

Effects of salinity on the mineral composition, growth and physiology of field-grown sultana grapevines were studied by adding NaCl solution to River Murray water. Five salt treatments, ranging between 0.37 and 3.47 dS m-l, were applied for six seasons. Petiole levels of Cl and Na were better predictors of yield than were lamina levels. There was no evidence of safe threshold salinity levels, so grapevine growers should aim to keep tissue salinity levels as low as possible, certainly below 1.5% and 0.5% for petiole Cl and Na respectively. Yield was affected not only by the salinity of the tissue in the current season, but also by tissue salinity in the preceding seasons. A model was developed which included previous tissue salinity measurements. Salt treatment reduced all growth parameters measured-pruning weight, shoot length, cane number and leaf and petiole weight. Decreases in stomatal conductance and photosynthesis were measured in the field in leaves of salinised vines, but leaf water potential was not affected. These decreases in photosynthetic rate are the probable cause of the severe yield and growth reductions in salinised vines.

The one-dimensional spatial analysis procedure proposed by Gleeson and Cullis (1987, Biometrics 43, 277-288) is extended to two dimensions using the subclass of separable lattice processes to model the errors. Residual maximum likelihood estimation of the models is described and diagnostics for testing model adequacy are derived. Results from the analysis of 24 sets of uniformity data indicate the frequent need for a two-dimensional analysis even when the plot shape is highly rectangular. These results also indicate the potential gain from using a two-dimensional spatial analysis rather than a row + column analysis. An example is presented of the analysis of a field experiment on tobacco.

The effects of paraquat and glyphosate on the nutritive value of dry residues of vulpia [Vulpia bromoides (L.) S. F. Gray], and its subsequent regeneration the following year were examined at Wagga Wagga during the spring, summer and winter of 1986/87 and 1987/88. Paraquat (100 and 200 g a.i./ha) and glyphosate (135 and 270 g a.i./ha) were applied as spraytopping treatments at heading, anthesis and early grain filling stages of vulpia. For both herbicides, time of application was critical to the level of regeneration obtained. Glyphosate gave 84 and 83% control when applied at heading and anthesis respectively, delaying application until early grain filling reduced the level of vulpia control to 28%. Paraquat gave 81% control when applied at anthesis, while delaying application until early grain filling, or earlier application at heading, gave 59% control. Crude protein (CP) and water-soluble carbohydrate (WSC) concentrations, and in vitro organic matter digestibilities (OMD) were measured in vulpia residues collected for 16 weeks after herbicide application. Paraquat increased CP levels most when applied at heading (from 4-8 to 7.2% in 1986, and from 4.9 to 6.5% in 1987). Glyphosate increased CP levels most when applied at heading in 1986 (from 4.8 to 5.4%), but at anthesis in 1987 (from 4.9 to 6.5%). Glyphosate increased WSC most when applied at heading (from 5.7 to 10.6% in 1986, and from 3.5 to 6.3% in 19871, while paraquat reduced WSC in both years. Application of glyphosate at heading was the only treatment to increase OMD (from 50.5 to 54.7%).

The influence of 5 rootstocks (Parramatta sweet orange, trifoliate orange, Troyer citrange, Appleby Smooth Seville and Somersby Smooth Seville) on the productivity and fruit quality of Marsh and Davis grapefruit was studied over 15 years in the Lower Murray region of New South Wales. Annual yields of mature trees and cumulative yield were highest on Troyer citrange; cumulative yield was lowest on trifoliate orange. Canopy volume was largest on Troyer citrange and smallest on trifoliate orange. Production efficiency (fruit yield per unit of tree size) was significantly greater on trifoliate orange than all other rootstocks. Canopy volume was 44% smaller on trifoliate orange than on Troyer citrange, but cumulative yield was only 26% lower. Because of their small size and high production efficiency, trees on trifoliate orange would be suitable for high density plantings. Trifoliate orange produced the best quality fruit. Fruit from trees on trifoliate orange had the highest total soluble solids (TSS) and TSS :acid ratio. Sweet orange and Appleby Smooth Seville produced the least acceptable fruit due to low TSS :acid ratios. Fruit size was largest on Troyer citrange and trifoliate orange. Trees on trifoliate orange accumulated excessive chloride (Cl-) during periods of high irrigation water salinity; however, no adverse effects on growth or long-term productivity were apparent. Leaf C1- concentrations for trifoliate orange varied from 0.42 to 1.10%. Leaf C1- concentrations remained below 0.30% for the other rootstocks. There were no significant effects of scion cultivar on tree size, yield or production efficiency, and differences in fruit quality were slight. Fruit size was larger with Marsh than Davis.

The effects of soil pH and seasonal conditions on the responses of barley (Hordeum vulgare L.) and wheat (Triticum aestivum L.) to chlorsulfuron were investigated at 3 sites in southern New South Wales in 1986, 1987 and 1988. The sites varied in soil pH (4.3-6.2) and annual rainfall (360-560 mm). In addition to the variation in soil pH between sites, 2 surface pH levels were obtained at each site by the addition of lime. The effect of post-emergence applications of 7.5, 15.0, 22.5, 30.0 and 37.5 g a.i./ha chlorsulfuron on the yield of weed-free barley and wheat varied with season, site and the addition of lime. The yield reduction was greatest in 1986, and the extent of the reduction was always greater in barley than wheat. In 1986, a recommended rate of chlorsulfuron (15 g a.i./ha) significantly (P

Five field trials to screen a range of barley germplasm for tolerance to saline soil conditions were conducted on irrigation farms in southern New South Wales, in areas affected by secondary salinisation from shallow watertables. Three trials were located on heavy grey clay soils and 2 on red-brown earth soils. An electromagnetic soil conductivity meter (EM-38) was used to quantify the salinity of individual field plots. Cultivars were compared in terms of their grain yield response to soil salinity. Yields were significantly reduced by soil salinity at all sites except 1 on red-brown earth. Both genetic and site differences in salinity response were identified. The reduction in yield per unit increase in electrical conductivity of the saturated paste (EC,), averaged across sites, varied from 4.7% for Forrest to 6.6% for Schooner. However, the yield reduction per unit increase in EC,, averaged across cultivars, varied from 4.1% in a red-brown earth to 6.4% in heavy clays.

In repeated measures experiments how treatment contrasts change over time is often of prime interest and modelling these contrasts is the aim of any analysis. We present an approach to the analysis of repeated measures data in which both the mean and the covariance matrix are modelled parametrically. We use the correlogram and semivariogram for identification of the covariance structure and linear models for specifying the change over time of the treatment contrasts. Incomplete data are handled in the approach and estimation is based on residual maximum likelihood. Examples which motivated the work are presented to illustrate the applicability of the method

SUMMARYUse of a one-dimensional neighbour method of analysis in 1019 variety trials of a range of crops conducted by plant breeders in four states of Australia in 1985–87 resulted in an average reduction of 42% in variances of varietal yield differences compared with conventional randomized complete block analysis. Of these trials, 219 were designed as square, rectangular or generalized lattices and the average reduction in variances of varietal yields with incomplete block analysis and recovery of interblock information was 33%. The results emphasized that plots should be wide enough to avoid interplot competition, and that neighbour analysis is of most benefit in trials with short plots or when the field layout has many plots in a row.

From a set of 12 factors thought to be related to eating quality, only percentage soluble solids and penetrometer estimates of flesh firmness were useful in deriving a predictive equation. A practical grade standard might be 10% minimum solids with upper and lower limits for flesh firmness of 2.0 and 1.0 kg force.

SummaryTwo scries of simulation experiments were used to investigate the accuracy of treatment and variance estimation with a neighbour analysis of field trials proposed by Gleeson & Cullis (1987). The first series examined the accuracy of residual maximum likelihood (REML) estimation of seven theoretical error models applicable to field trials. REML estimation provided accurate estimates of the variance parameters, but the Ftest of treatments was slightly biased upward (to +2·4%) for first differences models and slightly biased downwards (to –1·4%) for second differences models. The second series of simulations, based on 19 uniformity data sets, illustrated that treatment effects were consistently estimated more accurately by the REML neighbour (RN) analysis of Gleeson & Cullis (1987) than by incomplete block (IB) analysis with recovery of interblock information. The relative gain in accuracy of RN over IB depends on the amount of systematic variation or ‘trend’ in the trial, and ranged from 6 to 18% with an average of 12% for a range of trend and error variances commonly encountered in field trials. The predicted average standard errors of pairwise treatment differences from the RN analysis were in close agreement with their empirical estimates, indicating that the predicted average S.E.D. is approximately valid.

Tubers, cv. Sebago, previously treated or untreated with Rindite were stored under different light and temperature regimes in 2 seasons for up to 8 weeks to produce physiological conditions ranging from dormancy in the dark to vigorous sprouting in light. No treatment prior to inoculation withErwinia chrysanthemi (Sabet) Victoria, Arboleda & Munoz reduced the incidence of severity of soft rot compared with tubers stored under conditions approximating those in the field. Rindite treatment to break dormancy increased incidence and severity of rotting in tubers previously stored at 5°C but had no significant effect on ambient stored tubers. Exposure to light had no consistent significant effect although there was a tendency to reduce the incidence, but not the severity, of rotting. Reducing sugar levels and moisture content were not correlated with soft rot either on a treatment or individual tuber basis. The data showed no predictive correlation between these parameters and the incidence or severity ofE. chrysanthemi soft rot.

Summary The vec operator arranges the columns of a matrix one below the other. When the matrix is symmetric such elements are not distinct but an extraction of only the distinct elements on or below the diagonal forms the operation denoted by vech. For other types of patterned matrices a ‘patterned vech’ operator is defined. The transformations from vech to vec are not uniquely defined. Here we examine properties of linear transformations which overcome the lack of uniqueness and develop properties of such linear transformations.

An enzootic, Australian, atypical strain of Aeromonas salmonicida isolated from diseased goldfish, Carassius auratus (L.), was inoculated into Atlantic salmon, Salmo salar L., brown trout, S. trutta L., rainbow trout, S. gairdneri Richardson, and brook trout, Salvelinus fontinalis (Mitchill), fingerlings by intraperitoneal injection (i.p.) and by bath challenge, the latter with and without prior abrasion of skin. The 10-day LD50 (i.p.) was estimated to be 7·4 × 10-3 colony forming units (cfu) for Atlantic salmon, 3·0 × 10-2 cfu for brown trout, 3·7 × 102 cfu for brook trout and 6·4 × 103 cfu for rainbow trout. Brown, rainbow and brook trout succumbed to bath challenges with between 105–106 cfu/ml, developing ulcers of the skin and septicaemia. The organism was trasmitted from inoculated fish to five of 195 within-tank control fish via water and established a carrier state in one of 14 Atlantic salmon. It was concluded that the organism poses a significant threat to the salmonid farming industry and wild salmonid fisheries in Australia.

Key message Factor analytic mixed models for national crop variety testing programs have the potential to improve industry productivity through appropriate modelling and reporting to growers of variety by environment interaction. Abstract Crop variety testing programs are conducted in many countries world-wide. Within each program, data are combined across locations and seasons, and analysed in order to provide information to assist growers in choosing the best varieties for their conditions. Despite major advances in the statistical analysis of multi-environment trial data, such methodology has not been adopted within national variety testing programs. The most commonly used approach involves a variance component model that includes variety and environment main effects, and variety by environment (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V\times E$$\end{document}V×E) interaction effects. The variety predictions obtained from such an analysis, and subsequently reported to growers, are typically on a long-term regional basis. In Australia, the variance component model has been found to be inadequate in terms of modelling \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V\times E$$\end{document}V×E interaction, and the reporting of information at a regional level often masks important local \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V\times E$$\end{document}V×E interaction. In contrast, the factor analytic mixed model approach that is widely used in Australian plant breeding programs, has regularly been found to provide a parsimonious and informative model for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V\times E$$\end{document}V×E effects, and accurate predictions. In this paper we develop an approach for the analysis of crop variety evaluation data that is based on a factor analytic mixed model. The information obtained from such an analysis may well be superior, but will only enhance industry productivity if mechanisms exist for successful technology transfer. With this in mind, we offer a suggested reporting format that is user-friendly and contains far greater local information for individual growers than is currently the case.