Narrow-leafed lupin is a rain-fed crop in southern Australia whose yields are limited by the amount and distribution of rainfall. Drought terminates the growing season, and the timing and intensity of development of this (terminal) drought is a likely cause of much yield variability. We investigated this yield variability by manipulating terminal drought with trickle irrigation. Despite comparatively high crop dry mass of nearly 9 t/ha in the absence of irrigation, seed yield was only 1·2 t/ha with a harvest index of only 14%. Delaying the terminal drought and thus prolonging the period for crop ripening raised crop dry mass by 18-42%, with large increases in harvest index and seed yields, which rose by 45-75% and 95-135%, respectively. Pod set started 110 days after sowing (DAS) but appreciable pod filling did not start until 140 DAS, by which time, in the unirrigated treatment, only about 15% of the plant-available soil water remained, leaf diffusive conductance had fallen by 75%, and leaf water potential was -1·7 MPa. Leaf senescence and abscission had already begun; in the control treatment only 25% of the green area remained at 150 DAS, when net vegetative growth ceased and appreciable seed filling began. During irrigation, plant-available soil water was maintained at 35-70% of that at field capacity, with correspondingly better plant water relations. Pod and seed filling started at the same times as in the unirrigated control, and vegetative growth ceased at the same time. However, green area declined more slowly and reproductive growth continued for longer and at a faster rate. Yields were highly correlated with the number of productive pods (and seeds) at maturity, which in turn was associated with pod survival rather than pod set. Irrigation increased the number of surviving pods and seeds, both of which sometimes aborted at comparatively late stages of filling. Average seed weights were stable across treatments, except where there was a period of drought before irrigation, in which case seeds were heavier, compensating for prior loss of pods. It is argued that a better ideotype for the Mediterranean environment of southern Australia would switch to reproductive growth earlier, before severe water deficit develops, and with less overlap between vegetative and reproductive growth.

Studies under controlled environment conditions indicate that transient high temperatures (34-38˚C) during grain filling can significantly reduce weight per seed in narrow-leafed lupin (Lupinus angustifolius L.). This study has shown that on average, lupin pods reach temperatures about 3-5˚C higher than the maximum daily air temperature during seed filling under field cropping conditions. These differences do not appear to be markedly influenced by the amount of radiation intercepted by the canopy, stage of pod development, or position of the pods in the canopy, but fluctuate more as a result of differences in radiation intensity, wind speed, and water availability. Trickle irrigation reduced the difference between pod and air temperature by about 2˚C. Lupin species with larger pods (L. cosentinii and L. atlanticus) reached higher maximum daily temperatures than those of L. angustifolius. Long-term meteorological data indicate that air temperatures during seed filling of lupins in Western Australia are likely to exceed 30˚ C and will occasionally exceed 35˚ C. In lupin-growing areas of Western Australia, pod temperatures exceeding 33-35˚ C can be expected about 1 year in every 3, and more rarely (about 1 in 10 years), pod temperatures exceeding 38-40˚ C can be expected. These transient high temperatures are likely to increase significantly the year to year variation in yields of lupin grain.

Highly variable yields are a weakness of narrow-leafed lupins. Yield variability could be caused by many factors, including hot days during seed filling. This paper investigates the effects of 2 hot days at various stages of seed filling in L. angustifoliusL. cv. Merrit. Exposing adequately watered plants to a total of 6 h at 34, 36, or 38˚C, compared with 20˚C, over 2 consecutive days reduced weight per seed by 4, 8, or 12% at maturity, respectively. The 38˚C treatment, applied when seeds averaged 4% of their final weight, also caused significant seed abortion. High temperatures reduced weight per seed at all stages of seed growth, except when seeds were <5-12 mg dry weight (3 and 6% of final seed dry weight, Expts 1 and 2, respectively). The reductions in weight per seed were not associated with reduced assimilate supply because: (a) neither photosynthesis nor leaf longevity were reduced by heat treatment; (b) competing inflorescences and branches were not allowed to develop; (c) the plants produced very large seeds for this cultivar (174-190 mg); and (d) leaves remained green well after the pods had matured. Seed N concentration decreased and fat concentration increased by small, although statistically significant, amounts in response to heat treatment at the last stage of seed development tested (57% of final weight per seed when treated) but not at earlier stages. This study indicates that hot days with pod temperatures as low as 34-36˚C during seed development can cause reductions in weight per seed, and hence yields, in narrow-leafed lupin crops.

The relationship between flowering time and daylength and temperature is described for L. angustifolius using multiple linear regression. The main cultivar was Gungurru but cvv. Danja, Yorrel and the L. albus cv. Kiev Mutant were also studied. Regression analyses were performed on time to flowering observations for lupins grown with serial sowings at up to 12 sites over up to 5 years in Western Australia (there were 102 separate observations of time to flowering for Gungurru). Time to flowering in the L. angustifolius cultivars was best explained by a model incorporating terms for average temperature and daylength between sowing and flowering. Models of this form were not satisfactory for L. albus, probably because of vernalization requirements which the L. angustifolius cultivars do not have. Using data from the experiment with the widest range of sowings, 94.6% of the variation in time to flowering was explained by the above model and an additional 3.5% was explained by including an interaction term. The rate of progress through all stages of development to flowering, except for the period between appearance of the last leaf and flowering, was sensitive to temperature. The rate of leaf appearance responded to both temperature and daylength and the rate of progress through the period between appearance of the last leaf and flowering was sensitive only to daylength.