Collection of pollen for hand pollination and determination of pollen viability
In both years of study (2002 and 2003), branches with several flower clusters from different apple Malus x domestica Borkh. (Rosaceae, Maloideae) cultivars, maintained as breeding stock at the Atlantic Food and Horticulture Research Centre, Kentville, Nova Scotia (45°05'N, 64°28'W), were collected and placed into a greenhouse compartment with the temperature maintained continuously above 18°C and ambient light conditions. Branch ends were snipped diagonally and quickly placed into 4-L jars of water. At the balloon stage of flowering, petals were peeled away and the swollen anthers were removed from the filaments by rubbing the open flower on wire mesh, following the procedures of Galletta (1983). Anthers were collected in a glass Petri dish and allowed to dehisce for 24–48 h. Dehisced anthers were placed into a glass vial and crushed to release more pollen. The vial was sealed, placed in a jar with anhydrous calcium sulfate (DrieriteTM) and kept in a cool, dry place until used (within 1–5 d).
In 2003, viability of a pollen sub-sample was determined using the MTT [3-(4,5 dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide; Sigma M 2128] (Rodriguez-Riano and Dafni, 2000) and DAB (Sigma FastTM 3,3'-diaminobenzidine tablets; Sigma D-4168) techniques outlined by Dafni (2001). A control of non-viable pollen was prepared for viability analysis simultaneous to the treatment; pollen was killed by spreading a small amount of the pollen mixture into a 70 % ethanol droplet on a glass microscope slide, which was then heated with a flame. This procedure was repeated twice. Control and treatment pollen samples were placed on glass microscope slides and the reagents MTT or DAB were applied following Dafni (2001). The droplets were mixed and spread out to minimize pollen clumping and to facilitate drying, and the slides were placed on a slide warmer to dry. This procedure was repeated twice. Randomly selected fields of view were examined with a compound microscope at x100 magnification. The total of viable and non-viable pollen grains was counted in each field of view for a total count of at least 300 pollen grains. Non-viable pollen grains, which stay light coloured, were distinguished from viable ones, which turn a dark colour (violet-purple-brown in MTT; brown-purple-red in DAB), indicating the presence of dehydrogenases or peroxidases, respectively (Dafni, 2001). Percentage viability was calculated for both reagents.
Stigma receptivity period for hand pollination studies
Stigma receptivity of ‘Summerland McIntosh’ flowers at different stages of development was investigated in 2002 using a method which rated oxygen bubble generation on stigmatic surfaces. Receptive stigmatic surfaces produce peroxidase enzymes (Zeisler, 1938; Galen and Kevan, 1980; Dafni, 1992) which break down hydrogen peroxide, generating oxygen bubbles (Galen and Plowright, 1987; Dafni, 1992). Flowers at various stages of development were collected in plastic bags and immediately placed into a cooler containing ice packs. A ‘blind’ design was used to evaluate the level of oxygen generation where the evaluator had no knowledge of stage of flowering. It was assumed that the rate of bubble production from the stigmatic surface in hydrogen peroxide was directly related to its level of receptivity. A grading scale of 1–5 was developed to rank receptivity (1 = no bubble production; 5 = very rapid bubble production). The flowers were randomly selected for evaluation. The styles were then removed from each flower just above the point of emergence from the hypanthium and passed to the evaluator who placed the stigmatic ends into a depression slide containing fresh 3 % hydrogen peroxide (after Galen and Kevan, 1980). The evaluator graded and recorded the level of bubble production for each flower.
Stigma excision and pollination experiments
A 3-ha orchard at the Agriculture and Agri-Food Canada research farm in Canard, Kings Co., Nova Scotia (45°07'N, 64°28'W) was used for the experiments. The orchard consisted of alternating north-to-south double rows of 18-year-old ‘Summerland McIntosh’ (semi-dwarf ‘MM 106’ root stock, pruned to modified central leader) and 8-year-old ‘Royal Court Cortland’ (semi-dwarf ‘MM 106’ root stock) apple trees, with 6-m spacing between rows, and 4·5-m tree spacing within rows.
A randomized complete block (CB) design was used. The blocks were three (2002) and five (2003) randomly selected ‘Summerland McIntosh’ trees. On each tree, six limbs were selected based on flowering potential. Six levels of pollination treatment (i.e. none, one, two, three, four or five stigmas remain) were randomly assigned to each of the six limbs on each tree. Previously opened, damaged, unhealthy and under-developed flowers were removed. All remaining flowers at, or slightly beyond, the balloon stage had their styles snipped and cauterized (using forceps heated with a mini-torch) approx. 3–5 mm below the stigmatic surface to leave the desired stigma number. The remaining stigmas were then hand pollinated using the eraser end of a pencil coated with the previously collected pollen mixture. This procedure was repeated on two consecutive days to ensure pollination of open flowers. The total number of flowers receiving the pollination treatments remaining on each branch was recorded. Branches were then covered with approx. 45 x 50 cm Delnet® apertured film bags (DelStar Technologies, Inc., Middletown, DE) for 3–4 weeks to minimize insect feeding damage.
Percentage fruit set was determined for each of the branches on all trees by counting the developing fruit remaining 4 weeks after petal fall. Percentage fruit set data were subject to arcsine transformation (Zar, 1999) for normalization and to achieve homoscedascicity of variance prior to balanced analysis of variance (ANOVA) (Minitab, 2000). At harvest in mid-September, all fruits were collected and placed in cold storage at approx. 4 °C until fruit quality was determined. Fruits were cut in half just above the equatorial axis, and the number of plump seeds per carpel and their distribution within the fruit was determined. The number of seeds per fruit was analysed with ANOVA (generalized linear model procedures) (Minitab, 2000). The number of seeds produced per pollination treatment was also compared with those expected using chi-square analysis for contingency tables. Expected values were determined by first multiplying the maximum number of seeds expected per fruit for each of the k pollination treatments (i.e. two seeds per remaining stigma) by the number of fruit which developed for each treatment (i), to obtain the value Si. These values were then subject to the following transformation:
is the total seeds produced in all the fruit for all treatments. This procedure was done to balance the observed and expected totals in the contingency table for analysis while maintaining the relative proportions of expected seeds for each treatment. All tests were conducted at the 5 % level of probability.
Gynoecium structure and the pollen-tube pathway
Twenty flowers were collected and fixed in 3 : 1 ethanol : acetic acid for 20–24 h (Kearns and Inouye, 1993). Flowers were then rinsed in 50 % ethanol for 0·5 h, and then stored in 70 % ethanol at 4 °C. Flowers were dissected longitudinally to determine floral form, specifically the length of various sections of the gynoecium, and transversally to determine the location and structure of the pollen transmitting tissue within the styles. Measurements were made using a binocular microscope with an ocular micrometer.