The gynoecia of most taxa within the Maloideae are typically described as syncarpous, but Rohrer et al. (1991, 1994) indicate that much variation exists in the level of connation among the carpels, including at least three genera which are strictly apocarpous. Connation among carpels in the Maloideae is at two levels (Rohrer et al., 1994). At the level of the ovaries, connation is normally congenital (Endress, 1994). Most maloid genera have carpels that are fully fused to each other at this level, but variable degrees of fusion exist in some genera, including Pyrus (Rohrer et al., 1994), of which some species have been described as apocarpous (Sterling, 1965b). In the Maloideae, connation at the level of the ovaries is external (Rohrer et al., 1994) which probably limits the degree of inter-carpel communication in some genera, leading to the assumption of imperfect syncarpy in Malus (Carr and Carr, 1961; Cresti et al., 1980; Anvari and Stösser, 1981; Pratt, 1988; Weberling, 1989), a condition which does exist in other genera (Gorchov and Estabrook, 1987; Gorchov, 1988). However, small openings may be present at the centre of the core (Rohrer et al., 1994), which may allow some degree of inter-carpel communication. The second level of connation among carpels occurs in the styles (Rohrer et al., 1994), which develop from the apical portions of the carpel primordial (Evans, 1999). Some genera (e.g. Pyrus) develop styles which appear completely separate throughout their length (Rohrer et al., 1994; Aldasoro et al., 1998). Both congenital and post-genital fusion can lead to the formation of a compitum within connate areas of the carpels, including the styles (P. K. Endress, pers. comm.). However, Endress and Igersheim (2000) indicate that the degree of post-genital fusion is probably dependent on its time of commencement, being more intensive in forms which fuse earlier in development.
Several studies have presented microscopic examination of carpel structure (Sterling, 1964, 1965a–c, 1966; Rohrer et al., 1991, 1994; Evans, 1999) and, at least partially, pollen tube pathways in the Maloideae (Stott, 1972; Cresti et al., 1980; Gorchov, 1988; Embree and Foster, 1999; Kaufmane and Rumpunen, 2002; Broothaerts et al., 2004). However, no studies have traced the transmitting tissue within each carpel in its entirety, from stigma to micropyle. For instance, the microscopic/histochemical study by Cresti et al. (1980) of Starkrimson apple clearly shows five separate areas of transmitting tissue within a transverse section of the gynoecium just below the point of stylar union (indicated here in Fig. 6), supporting their claim of imperfect syncarpy. Unfortunately the authors did not continue to examine tissue below the point of stylar union, as it is below this level that the majority of perfectly syncarpous gynoecia have a compitum (Endress, 1994). Endress (1994) reports that in flowers with free stigmatic lobes (as in Malus), only 0·5 mm of joined transmitting tissue is required to evenly distribute pollen tubes among carpels when not all stigmas are pollinated.
The present findings support the notion that not all stigmas have to be pollinated to obtain uniform pollen tube distribution and full fertilization in Summerland McIntosh. Seed yield in the present study evidences the presence of a compitum, hence perfect syncarpy, in the apple flower which allowed pollen tubes growing down individual styles to cross into any of the five carpels. Seed number was higher than expected from an imperfectly syncarpic gynoecium when four or fewer stigmas were pollinated (Figs 4 and 5). It is believed that neither pollen viability nor incompatibility contributed to the lower than expected seed yield observed for the five-stigma pollination treatment in both years. This was probably a result of unrealized seed development in some fruit, as ovules do not always develop into seeds in the Maloideae (Rohrer et al., 1991). Further evidence of perfect syncarpy was provided by the even distribution of seeds among the carpels within fruits from all treatments receiving pollination (Table 1). High levels of fertilization and seed development among the carpels resulted in >40 % fruit set in both years (Fig. 3) when at least one stigma was pollinated. Only flowers which were not pollinated failed to set fruit.
Other studies have indicated that several apple cultivars may have gynoecia that are perfectly syncarpic (Beaumont, 1927; Visser and Verhaegh, 1987, and references therein). Perfect syncarpy is considered a derived condition within the angiosperms (Endress, 2001), and several evolutionary advantages over apocarpous and imperfectly syncarpous gynoecia have been recognized (Williams et al., 1993; Endress, 1982, 1994). Among these are greater seed set through more regular pollen tube distribution, economy in flower construction, and increased gametophyte selection among pollen grains in a unified transmitting tract (Endress, 1982, 1994).
Additional advantages may be gained with apically subdivided stigmatic surfaces for pollen capture (Howpage et al., 1998). The flowers of Malus and most other Maloideae attract a wide range of pollinators, and are not specialized for a single group (Cambell et al., 1991). For Malus, which has at least 44 bee visitors in Nova Scotia (Sheffield et al., 2003), pollination efficacy among apoidean visitors varies considerably (Boyle and Philogène, 1983; Boyle-Makowski, 1987; Free, 1993, and references therein; Goodell and Thompson, 1997; Vicens and Bosch, 2000). However, subdivided stigmatic surfaces promote pollen capture from a variety of positions, and accommodate differing foraging behaviours, albeit deposition may be uneven among stigmas. Perfect syncarpy via the compitum allows pollen tubes to be evenly delivered to ovules, despite unequal deposition during pollination. As a result, total pollen deposition by different bee visitors of Malus may not be as indicative of potential fruit quality as previously thought, and smaller or less effective floral visitors may be contributing significantly to fruit production if at least one stigmatic surface is adequately pollinated. More important factors in determining seed set and fruit production in some cultivars of apple are pollen viability, pollen compatibility and pollen dispersal. The importance of these factors, particularly with respect to orchard design, was recently investigated by Kron et al. (2001a, b).
Visser and Verhaegh (1987) indicate that imperfect syncarpy may occur in some apple cultivars. Horticultural practices and breeding programmes have developed over 2000 apple cultivars worldwide (Morgan and Richards, 2002), many of which show variability in fruit form (Rohrer et al., 1991; Morgan and Richards, 2002). Differences in floral form have also been reported among many cultivars (Stott, 1972; Ferree et al., 2001). For instance, Stott (1972) reports differences in the proportion of stylar fusion among many apple cultivars. Some of the differences in floral form among cultivars can be great enough to cause variation in pollinator floral handling behaviour (Schneider et al., 2002), in some instances to a level that pollinator effectiveness (i.e. stigma contact) declines, as reported with the ‘sideworking’ behaviour of honey bees on ‘Delicious’ apples (Robinson, 1979; Degrandi-Hoffman et al., 1985). It is possible that intensive cultivar development has created or caused the loss of the compitum in some apple cultivars, either through variable degrees of carpel separation at the level of the ovaries or through reduced stylar fusion. Gynoecial arrangements in the non-cultivated forms of Malus (55 species reported in Phipps et al., 1990), including the ancestor of the domesticated forms, M. sieversii (Ledeb.) Roem. (Morgan and Richards, 2002), are presently unknown. Clearly, gynoecial structure in apple cultivars warrants further study as the presence or absence of a compitum directly influences fruit quality when pollination is incomplete.
This study comprises part of the PhD research undertaken by C.S.S. which was funded in part by the Agri-Focus 2000 Technology Development Program (Nova Scotia Department of Agriculture and Fisheries) and Agri-Futures-Nova Scotia Association (Agriculture and Agri-Food Canada) through the Nova Scotia Fruit Growers' Association. This paper is contribution number 2288 of the Atlantic Food and Horticulture Research Centre. We thank Susan Rigby, for experimental assistance during this project and for her illustration, and the following for field assistance: Kim Jansen, Meg Hainstock, Michelle Larson, Stephanie Moreau, Derek Maske and Darrin Moran. For reviewing the manuscript, we thank Dr Klaus Jensen, Atlantic Food and Horticulture Research Centre–Agriculture and Agri-Food Canada, Kentville, Nova Scotia, and two anonymous reviewers. We also thank Dr Peter K. Endress, Institute of Systematic Botany, University of Zurich, Switzerland for clarification of floral morphology and terminology, and Dr Rodger C. Evans, Acadia University Biology Department, Wolfville, Nova Scotia for his comments and for providing a valuable piece of literature.