Major changes to the global environment and their impact upon statistical requirements are presented. The need for extensive yet accurate datasets is stressed and the difficulties of achieving appropriate sampling strategies are described. Examples are presented of environmental studies in use in diverse field such as global warming, sustainability, agroforestry and intercropping, and local community studies. The history of the use of environmental statistics is highlighted as is the need for modern methods to handle large, multivariate datasets likely to be subject to both spatial and temporal variability.

Four components of the water budget for a growing season, namely storage, drainage, transpiration and direct evaporation from the soil surface, were estimated using a suite of techniques. The only data requirements were rainfall, neutron probe measurements of soil water content and microlysimeter measurements of evaporation from the soil. Data from four growing seasons at Kiboko, Kenya between 1990 and 1992 were used to provide examples of the estimations. Drainage was significant (about 10% of rainfall) in one season only; in the other seasons, total evaporation comprised at least 95% of the seasonal rainfall.

Drainage was determined using a relationship between unsaturated hydraulic conductivity and soil water content that was determined during the early part of the rainy season when water was penetrating to depth. This analysis made it possible to identify a critical water content at the base of the soil profile, above which there would be significant drainage. However, there are large errors associated with estimation of drainage if significant drainage occurs.

Estimates of direct evaporation from the soil surface were used as the basis of distinguishing transpiration from total evaporation. Microlysimetry was used to develop a model of evaporation from these sandy soils, which was based on the assumption that the evaporation from the soil surface following heavy rainfall is a unique function of time from rainfall, and little influenced by the presence of a sparse crop. This method showed that direct evaporation from the soil accounted for between 70 and 85% of total evaporation in seasons when total evaporation estimates ranged from 150 to 325 mm.

Plants of rice (Oryza sativa)cultivar IR36 were grown under flooded and free-draining conditions at a constant air temperature of 28°C with an 11.5 h d−1 photoperiod in both a growth cabinet and a glasshouse. The nutrient solution within flooded pots was maintained at 20°, 24° or 28°C, or was not controlled. Rate of floral development was enhanced by warmer nutrient solution temperatures at the same air temperature; times to panicle emergence were reduced by an average of 17 d with increase in the nutrient solution temperature from 20° to 28°C. In comparable treatments, plants grown in the growth cabinet reached panicle emergence later than those grown in the glasshouse, possibly because mean air temperature was on average 0.8°C cooler in the growth cabinet. An independent model, derived from a previous investigation, provided a good prediction of the times from sowing to panicle emergence for plants grown in both the glasshouse and the growth cabinet when based on the mean of the temperatures of the nutrient solution and the air, showing that it is important to determine water/root temperature as well as air temperature when investigating the effect of environment on development of the rice plant.

The effect of mixtures of cultivars on yield and risk distribution in four maize cultivars grown at four different population levels was studied in semi-arid environments in Ethiopia. Mixtures yielded between 2 and 29% more than the pure stands, but late maturing pure stands produced more biomass than mixtures. Mixtures of cultivars with similar flowering periods yielded 60% more than the pure stands in dry growing seasons, but only 30% more when there was more rain. Yield gain was associated mainly with varietal synchrony of anthesis and silking. Mixtures composed of cultivars with different flowering times were less at risk from yield loss in the case of end-of-season drought and yielded 100% more than their late flowering component in pure stand. But in long growing seasons, with an early dry spell, pure stands of the late cultivar outyielded the mixtures whereas pure stands of the early flowering cultivars yielded less than the mixtures, except in years with a serious end-of-season drought. The results indicate that it would be profitable to grow mixtures in semi-arid regions if cultivars with similar height and synchronized flowering time were grown at populations of between 65 000 and 90 000 plants ha−1.

Wheat was planted at different altitudes in the Ethiopian highlands. Increased altitude led to a lower neutral detergent fibre (NDF) content and a higher in vitro organic matter digestibility (IVOMD) of the leaf blades, leaf sheaths and stems. The varieties tested did not differ in NDF content, however, because of the improved NDF digestibility of all three straw fractions. The semi-dwarf varieties had a higher IVOMD than the standard tall wheats. The local durum wheat variety showed a much higher sodium content and a more favourable Na:K ratio. Undersowing with an equal mixture of Trifolium ruepellianum (Fres.) and Trifolium steudneri (Schwf.) led to a small reduction in straw yield but increased the crude protein content of the crop residues from 2.3 to 7.1% and the IVOMD from 44 to 51% as compared to the sole wheat stand.

Defoliation affected inflorescence production by oil palms, causing a decrease in sex ratio (ratio of female inflorescences to total) and an increase in abortion rate. Clones differed in response to defoliation. One of seven clones showed no increase in abortion rate; in one clone there was a tendency for female inflorescences to abort in greater numbers than males. The change in sex ratio occurred much sooner after defoliation in some clones than in others, and comparison of flowering records with dissections done at the time of defoliation indicated that inflorescence sex was affected at more than one stage in development. This has important consequences for attempts to predict yields.

Field experiments are commonly designed in blocks and there are sound reasons for using them, for example, control of unwanted environmental differences and administrative convenience. If used, they should be chosen to correspond to perceived differences in the site or to simplify farm operations and not merely to conform to statistical desiderata. Thus, it is not essential that each must contain one plot for each treatment, though there are advantages if they do. Some of the consequences of using other block sizes are examined, it being borne in mind that modern computer packages will perform most of the tiresome arithmetic. The effectiveness of blocks is considered and it is noted that they sometimes do harm rather than good. The analysis of variance is explained in terms of strata as used in many modern computer programs and is extended to include the recovery of information and resolvability. Recommendations are made as to randomization.

The implications of a novel system of seed production for tossa jute (Corchorus olitorius), involving late sowing and intercropping with vegetables, were studied in 1989 and 1990. The greatest jute seed yield was obtained from jute-radish intercropping, where radish yielded 12–13 t ha−1 roots and gave the highest average gross margin, although jute-amaranthus intercropping performed marginally better in 1989. This system of husbandry yielded 3–4 t ha−1 of jute stick for fuelwood, which is of great value to rural households, and proved to be a viable means of producing jute seed, vegetables and fuel.

A computer crop simulation model of the growth and development of maize was validated using data sets obtained from field experiments run at various sites in the mid-altitude maize zone of central Malawi between 1989 and 1992. The model was used to provide information concerning management options such as the timing and quantity of nitrogen fertilizer applications and to quantify the weather-related risks of maize production in the region. It was linked to a Geographic Information System to provide information at a regional level that could ultimately be of value to policy makers and research and extension personnel.

The possible use of tropical forage legumes in pastures based on Brachiaria decumbens and Andropogon gayanus pastures was tested on farms in the humid tropics region of Pucallpa, Perú. The pastures were established by farmers using manual labour. They were associated with maize in some cases and were normally grown without fertilizers. The pastures were established in areas where fallow regrowth was felled and burnt. These pastures, with grass-alone controls, were incorporated by farmers into their normal paddock rotation and were grazed by dual-purpose cows. Despite the variable terrain and soil conditions, all the pastures established and persisted well. Over a four year period under grazing, the legume contribution to the forage averaged 21% and ranged between 1 and 66%. Desmodium ovalifolium was the dominant legume in paddocks that were not burnt, whereas in pastures regularly burnt to control weeds it tended to disappear and Stylosanthes guianensis was the main legume present. It is concluded that where farmers ensure the maintenance of adequate levels of forage, grass-legume mixtures are a viable and persistent option even if the pasture is occasionally burnt.

Microcomputer-based simulation models are increasingly being recommended as multipurpose tools for agricultural research. Use of a model should be conditioned by an evaluation of its performance and understanding of its limitations. This paper evaluates the responses of the process-oriented growth model for dry bean (Phaseolus vulgaris), BEANGRO V1.01, with an emphasis on the factors related to cultivar differences for production in tropical environments. Simulations of seed yield from beans grown under conditions of a known water deficit showed good agreement with observed data. The qualitative response to plant population resembled that of a field trial, and the model showed the expected linear relation between days to maturity and seed yield. Overall, the results suggest that BEANGRO has utility for certain types of agronomic studies, but that improvements are possible, particularly with respect to prediction of phenology.