3.4 Discussion ,
3.4.1 Seed crop development
The sowing date ( 1 3 November) used appeared to be suitable for 'Marc' seed production. Mean air temperatures during establishment ranged from 17 to 3 2°C and moisture supply was high. Although soil temperatures were not measured it is reasonable to assume that soil temperatures fitted into this range or, because the soil was a black cracking clay and therefore had high heat retention, were higher. This suggests that soil temperatures would have been near optimum (26 to 33°C, Njarui et al. , 1 992) for most of the establishment period. Early season growth was rapid with no apparent limitations to growth.
Total water application (rainfall plus irrigation) was between 1 80 and 200 mm per month in the growing period prior to 1 14 days after germination (7 March) and is not thought to have limited vegetative growth. Lower total water application during March ( 103 mm) and April (60 mm) did not appear to stress plants i.e. there was no sign of wilting. However it is possible that the reduced water application may have stressed plants sufficiently so that they became more susceptible to psyllid attack which occurred during this period.
The most important feature of herbage seed crops is usually flowering pattern (Section 2.4.2). Two peaks in mean inflorescence number per plant (96 days and 1 40 days) were observed, the former contributing most of the inflorescences of individual plants. The intensity of the 96 day peak strongly suggests that the inflorescences produced at this time would contribute most to crop yield. Whereas inflorescence number per plant before 1 14 days is the result of normal crop development, it is likely that flowering after this period was affected by other factors. Possible reasons for the sharp decline in mean inflorescence number per plant between 96 and 1 28 days include psyllid damage, moisture stres s or photoperiodic control of inflorescence number. Of these, psyllid damage is the most likely factor considering the changes in vegetative indices, particularly mean leaf number per plant, observed during this period. Moisture supply to
the site was still high during this period and is not thought to have contributed to the decline in inflorescence number.
As there are apparently no photoperiodic controls of flowering in 'Marc' , an early summer sowing seems most appropriate. This better guarantees optimal soil temperatures and establishment rainfall reducing the requirement for irrigation. A different situation may apply for 'Bayarno' and 'Uman' which appear to have short-day requirements for floral initiation and / or induction (Graham et al. , 1 99 1 b). There is the possibility that moisture stress induces more vigorous flowering in ' Marc' as occurs in some indeterminate forage legumes such as Macroptilium atropurpureum (Hopkinson, 1988). If this is the case moisture stress could be used to produce a stronger flowering peak which in turn can enhance yields in single harvest systems. This seems unlikely, at least in early (before 96 days) crop growth because vigorou s flowering was produced under moist soil conditions. It i s more probable that moist soil conditions are required to produce adequate vegetative growth in order to maximise light interception and provide a framework on which reproductive development can occur. The irrigation regime used in this trial u sed the assumption that moisture application was less important once vegetative growth was completed i.e. that
dry
(but not moisture stressed), cool (but frost-free) and settled conditions during late pod development would be conducive to pod development, delay of pod dehiscence and a reduction in diseases. Cool, dry and settled conditions during pod development were provided by the early summer sowing. It is not known whether the virtually complete (practically useful) cessation of flowering was due solely to natural development, psyllid damage or a combination of the two. Partial recovery of flowering indicates that the psyllid was influential. However, the recovery w as only partial which could point to rundown towards the end of the season. Thus it seems likely that both factors played a part.Peaks in mean immature (96 to 1 1 4 and 1 56 days) and mature ( 1 28 and 1 84 days) pods per plant indicated that the inflorescences present at 96 days contributed mostly to seed yield. They also indicate that the periods taken to reach immature and mature pod status are approximately 1 8 days and 32 days respectively. If the second peaks are taken into consideration the periods taken to reach immature and mature pod status is 14 and 42
days. The decrease in late season pod development rate is probably related to decreasing temperatures and / or w ater supply.
Pods appeared to dehisce when they had reached a certain stage of maturity, senesced and dried. Most pods dehisced between 1 28 days and 1 56 days most probably because they represented maturation of umbels involved in peak flowering. Psyllid damage could also have been influential by increasing premature pod abscission. This is supported by the decline in seed weight observed over the season which may also have been caused by pre-mature pod abscission. Although 'Marc ' flowered over an extended period, a large proportion of pods dehisced between 128 days and 1 56 days. Because most seed was presented at one time emphasis is placed on the importance of recognising the correct time of harvest if a single pass harvest system is to be used. This may not be as important in 'Bayamo' and 'Uman' seed crops as pods of these cultivars do not appear to dehisce as readily (Cox pers. comm., 1 995).
In addition to seed yield, seed quality is also important. Mean normal germination was consistently over 80% if hardseeds are assumed to produce normal seedlings. Mean abnormal and dead seeds were consistently around 1 5 % and 2% respectively. Whereas germination characteristics did not change significantly over the season there was a significant decline in seed weight. This indicates that maximisation of seed number per unit area is more important than maximising seed weight (at least for the seed weight range observed) if high yields of viable seed are to be achieved.