Pollen viability and stigma receptivity
The viability of the pollen mixture used in 2003 was determined to be high using both the DAB (92·2 % ± 4·0 s.d.) and MTT (94·0 % ± 7·5 s.d.) techniques. It was assumed that the pollen mixture used in 2002 had similarly high viability. Freshly opened flowers had the largest proportion of high stigma receptivity as measured by oxygen generating activity, and had the highest mean receptivity grade (Fig. 2). Peroxidase activity was measured in all stages of flower development.
Significant differences were observed among the pollination treatments for percentage fruit set in 2002 (F = 22·43, d.f. = 5, P ). The zero-stigma pollination treatment had almost no fruit set and differed from the remaining treatments, and the one-stigma pollination treatment differed significantly from only the two-stigma pollination treatment (Fig. 3). In 2003, the same trend was observed with significant differences observed among the pollination treatments (F = 12·15, d.f. = 5, P Fig. 3). No differences were detected among the remaining treatments (Tukey's HSD test, P = 0·05).
Seed production and distribution
In 2002, significant differences in mean seeds per fruit were found among the pollination treatments (F = 2·86, d.f. = 4, P = 0·026), but not the trees (F = 1·71, d.f. = 2, P = 0·185). The data from the trees were subsequently pooled and analysed with Tukey's HSD test (P = 0·05). Significant differences were observed between the one-stigma and three-stigma pollination treatments, and between the one-stigma and five-stigma pollination treatments, but not among the other treatments (Fig. 4). In 2003, the same trend was observed with significant differences found among the pollination treatments (F = 5·15, d.f. = 4, P = 0·001), but not among the trees (F = 0·38, d.f. = 3, P = 0·77). The 1-stigma pollination treatment differed from all other treatments (Tukey's HSD test, P = 0·05) (Fig. 4).
In both years, the mean number of seeds per apple exceeded those expected from an imperfectly syncarpous arrangement in the one-, two- and three-stigma pollination treatments, but was less than expected for the five-stigma treatment (Fig. 4). Similarly, the total seeds produced was also significantly higher than expected for imperfect syncarpy for the one-, two- and three-stigma pollination treatments, but less for the five-stigma pollination treatment (2002: 2 = 202·3, d.f. = 4; 2003: 2 = 121·5, d.f. = 4; data pooled across trees in each year) (Fig. 5). The percentage of fruits with seed-bearing carpels from the different pollination treatments also did not correspond to those expected from an imperfectly syncarpous arrangement, as the majority of fruit from all treatments had five carpels bearing seeds (Table 1). No mature fruits had more than two seedless carpels.
The external fused portion of the styles, as measured from the hypanthium to the area of stylar separation, of Summerland McIntosh flowers is approx. 3 mm in length (Fig. 6). Internal microscopic examination indicated that the compitum is much smaller, and is confined between the upper part of the ovaries and the lower part of the connate styles (Fig. 6).