570 The potential for ploidy level increases and decreases in Crataegus (Rosaceae, Spiraeoideae, tribe Pyreae) Nadia Talent and Timothy A. Dickinson Abstract: Unlike their diploid relatives, some triploid and tetraploid Crataegus frequently produce unreduced megagametophytes. In all cases, pollination is required for successful seed set, but in polyploids, endosperm formation can involve fertilization by either one or both sperm. Apomixis, in which the egg develops parthenogenetically, is widely documented in polyploid Crataegus, and as in many other groups with gametophytic apomeiosis, fertilization of unreduced eggs can also occur. Reciprocal pollinations were made between diploids, triploids, and tetraploids belonging to five taxonomic series in the genus to evaluate opportunities for gene flow between ploidy levels. The ploidy levels of embryo and endosperm in mature seeds, estimated from flow-cytometric DNA measurements, indicate the meiotic or apomeiotic origin of the megagametophyte and whether fertilization has occurred. These experiments demonstrated that although some tetraploids maintain near-obligate apomixis when supplied with pollen from diploids, others produced seeds containing embryos ranging from diploid to hexaploid. Allotriploid embryos were produced when a diploid was provided with pollen from tetraploids. A triploid produced tetraploid embryos when pollinated by a diploid and pentaploid embryos when pollinated by a tetraploid. Gametophytic apomixis in Crataegus thus can be facultative or near-obligate and may be implicated in the formation of interserial hybrids. Key words: Apospory, hawthorn, Maloideae, nonrecurrent apomixis, partial apomixis, polyhaploids. Résumé : Certains Crategus triploı̈des et tétraploı̈des produisent des mégagamétophytes non réduits, contrairement aux diploı̈des apparentés. Dans tous les cas, la pollinisation est nécessaire pour former les graines, mais la formation d’endospermes polyploı̈des peut impliquer la fertilisation par un seul, ou par les deux spermes. L’apomixie, selon laquelle l’œuf se développe de façon parthogénétique, est largement documentée pour les Crataegus polyploı̈des, ainsi que dans plusieurs autres groupes montrant de l’apoméı̈ose gamétophytique, où la fertilisation d’œufs non réduits peut survenir. Afin d’évaluer la possibilité d’un flux de gènes entre les degrés de ploı̈die, les auteurs ont effectué des pollinisations réciproques entre des formes diploı̈des, triploı̈des et tétraploı̈des, appartenant aux cinq séries du genre. Le degré de ploı̈die de l’embryon et de l’endosperme, dans les graines matures, estimé à partir de mesures cytométriques en flux de l’ADN, indique l’origine méı̈otique ou apoméı̈otique du mégagamétophyte, et si une fécondation a eu lieu. Ces expériences démontrent que certains tétraploı̈des maintiennent une apomixie presqu’obligatoire, alors que d’autres produisent des graines contenant des embryons allant de la diploı̈die à l’hexaploı̈die. On obtient des embryons allotriploı̈des lorsqu’un diploı̈de reçoit le pollen d’un tétraploı̈de, alors qu’un triploı̈de produit des embryons tétraploı̈des lorsqu’il est pollinisé par un diploı̈de, et des embryons pentaploı̈des lorsqu’il est pollinisé par un tétraploı̈de. Chez les Crataegus, l’apomixie gamétophytique peut ainsi être facultative ou presqu’obligatoire, et peut être impliquée dans la formation d’hybrides interséries. Mots-clés : aposporie, aubépine, Maloideae, apomixie non-récurrente, apomixie partielle, polyhaploı̈die. [Traduit par la Rédaction] Introduction Morphologically identifiable hybridization is extensive in Received 26 February 2007. Published on the NRC Research Press Web site at canjbot.nrc.ca on 26 July 2007. N. Talent.1,2 Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, ON M5S 3B2, Canada. T.A. Dickinson. Green Plant Herbarium, Department of Natural History, Royal Ontario Museum, 100 Queen’s Park, Toronto, ON M5S 2C6, Canada. 1Corresponding author (e-mail: [email protected]). address: Green Plant Herbarium, Department of Natural History, Royal Ontario Museum, 100 Queen’s Park, Toronto, ON M5S 2C6, Canada. 2Present Can. J. Bot. 85: 570–584 (2007) Crataegus L. (hawthorn) from both Eurasia and North America, and diploids, triploids, and tetraploids are common on both continents. However, gene flow and reproduction in this genus have been difficult to study, and the relationship between hybridization and polyploidy remains unclear (Talent and Dickinson 2005). There is some evidence that the morphologically identifiable hybrids (‘‘wide hybrids’’) of North American Crataegus have historically been largely ephemeral (Phipps 2005). The loss of these taxa is undoubtedly due in large part to changes in land use, but it has also been suggested that the hybrids that briefly flourished after land clearing and abandonment in the 19th and early 20th centuries have disappeared because of low fertility (Phipps 2005). Longley (1924) found numerous triploids among the distinctive forms that had been collected from the wild and doi:10.1139/B07-028 # 2007 NRC Canada Talent and Dickinson were then in cultivation, and the morphology of at least some triploids suggests allopolyploidy (Dickinson 1983; Dickinson and Phipps 1986). Hybrids between native North American diploids have not been observed (Talent and Dickinson 2005), and it is not clear how allotriploids would have arisen. In one case (in section Aestivales), the putative parents are known only as diploids (Phipps 1988b; Talent and Dickinson 2005), but in other cases, diploid–tetraploid hybridization appears to have occurred (in section Crusgalli; Dickinson 1983; Dickinson and Phipps 1986). Apomixis in angiosperms involves clonal seed production through a variety of developmental pathways that are related to the sexual pathway (Koltunow and Grossniklaus 2003). Some triploid and tetraploid Crataegus produce apomeiotic megagametophytes (Muniyamma and Phipps 1979a, 1979b; Dickinson and Phipps 1986; Ptak 1989; Dickinson et al. 1996), and apomictic seeds are produced (Talent and Dickinson 2007), potentially perpetuating these genotypes and increasing the long-term potential for nonclonal reproduction. Pollination is required for seed set, i.e., apomixis is pseudogamous (Dickinson and Phipps 1986; Muniyamma and Phipps, 1979b; Smith and Phipps 1988). The male sterility of some triploid forms (e.g., the two triploids used here; Table 1) suggests long-term apomictic reproduction with resource-based selection for lessened pollen production, in keeping with analytical models of pseudogamous apomixis (Noirot et al. 1997), but repeated formation through hybridization is an alternative scenario. Fertilization of unreduced egg cells in some angiosperms can produce hybrid offspring with higher ploidy levels, socalled BIII hybrids or 2n + n hybrids, and this was observed with open-pollinated Crataegus (Talent and Dickinson 2007). It appears that almost all Crataegus pollen is meiotically reduced (Longley 1924; Muniyamma and Phipps 1979a), and thus unreduced female gametophytes may be an important contributor to ploidy-level changes. The frequency of ploidy-level changes can depend on the genotypes of both parents (Clausen 1961; Nogler 1984), and it is not known whether apparently apomictic tetraploids remain apomictic when pollinated by diploids. A particularly important question is whether triploid apomeiotic Crataegus can produce new tetraploid genotypes if pollinated with the monoploid pollen from diploids. If this occurs, then triploids may have an important role in refreshing the gene pool of tetraploids, some of which appear to be near-obligate apomicts (Talent and Dickinson 2007). Triploids could be an even more important component of the breeding system of Crataegus if they are commonly produced by diploids that receive pollen from apomictic tetraploids. Open-pollinated apomictic triploid and tetraploid Crataegus apparently always require fertilization of the endosperm (Talent and Dickinson 2007), and either one or both sperm can be involved. In sexual diploids, however, the endosperm appears to be strictly triploid (the megagametophyte has the Polygonum-type morphology with a binucleate central cell). Thus, as in some other groups with the aposporous type of apomeiosis (including at least Ranunculus and some Poaceae), the endosperm balance requirement that is common to sexual species in many plant families (Brink and Cooper 1947; Nogler 1984; Katsiosis et al. 1995) is less strict or absent in apomicts. We would therefore expect that crosses be- 571 tween sexual diploids and apomictic polyploids would fail if the maternal parent was diploid and would only succeed with tetraploid apomeiotic mother plants if the unreduced central cell (8x) is compatible with one or both reduced sperm from the diploid (1x or 2x). We carried out pollinations between taxa of different ploidy levels to determine the potential for gene flow between ploidy levels. The pollinations represent the following combinations (maternal parent noted first): diploid
tetraploid, tetraploid
diploid, (male-sterile) triploid
tetraploid, and (male-sterile) triploid
diploid. The ploidy levels of embryo and endosperm, established by flow cytometry (Talent and Dickinson 2007), can reveal whether fertilization of the central cell and the egg cell had occurred and whether the megagametophyte was meiotically unreduced (‘‘apomictic’’) or reduced (‘‘sexual’’). All of the species used for these experiments were studied previously by analyzing seeds from open pollination of individuals in isolated single-species stands or with intrataxon and self-pollinations (Talent and Dickinson 2007). The plants available do not include any likely diploid–autopolyploid pairs, and the material is therefore not ideal for testing the presence of an endosperm-balance requirement that limits interploidy fertility (Brink and Cooper 1947; Lin 1984; Katsiosis et al. 1995). The possibly hybrid triploids in our study are morphologically distinct from their known diploid and tetraploid relatives. Two of the tetraploids (Crataegus crus-galli and Crataegus macracantha) were highly apomictic following intrataxon pollination, but one (Crataegus submollis) produced about 60% of open-pollinated seed with endosperm that apparently developed from reduced megagametophytes (Talent and Dickinson 2007). The species are not taxonomically close (Table 1), but they co-occur in southern Ontario, and their flowering times can partly overlap, particularly at sites with a steep microclimate gradient (this statement is based on personal observations plus data summarized in fig. 1 of Campbell et al. 1991). The pollination experiments were therefore concerned with how these species would interact to produce seed, if pollen should happen to be transferred to plants of a different ploidy level. Materials and methods Plant materials The ploidy levels of all trees used in this study were established previously (Dickinson and Phipps 1986; Talent and Dickinson 2005). The mother trees used for pollination experiments were chosen for their accessibility to Toronto, Ontario; most pollen sources were from warmer sites (more southern localities or from south-facing slopes), thus permitting cross-pollinations with earlier-blooming individuals (Table 1); Crataegus is protogynous. The sequence of bloom of the maternal parents was C. submollis, Crataegus ?grandis, Crataegus succulenta, C. macracantha, Crataegus monogyna, Crataegus punctata, C. crus-galli. The two diploid species (Table 2) are the widespread native species C. punctata (mother trees in Wellington County, pollen parents in City of Toronto) and naturalized C. monogyna (all trees in City of Toronto; the late-blooming mother tree at a cooler site close to Lake Ontario), which is native to much of Eurasia and northern Africa (Phipps et al. # 2007 NRC Canada 572 Can. J. Bot. Vol. 85, 2007 Table 1. Localities in Ontario, Canada, of the Crataegus individuals examined in this study. Ploidy level RM Durham (43.58N, 79.08W), ON27 Section Crataegus C. monogyna Jacq. 2x NT100, 132 Section Crus-galli Series Crus-galli C. crus-galli L. 4x Series Punctatae C. punctata Jacq. C. ?grandis Ashe Section Coccineae Series Macracanthae C. succulenta Link C. macracantha Loudon Series Molles C. submollis Sarg. RM Niagara (43.18N, 79.08W), ON04 NT35, NT72 SC04, 05, 06, 15, NT32, 58, 62, 66 D1696, NT1, 227 3x Wellington County (43.58N, 80.58W), ON43 D1697, MP28, NT114, 116, 117 2x 3x 4x City of Toronto (43.48N, 79.28W), NTON03, 06, 23, 28a, 31, 32, ON40, 41 NT92, 93, 94, 120, 121, 230A, 231 NT29 NT221, 241 NT101, 102, 104, 105, 107, 108, 118, 130, 131, 229, 230 4x NT13, NT26 Note: RM, regional municipality. Entries under each locality are codes for individual study sites, full details of which are given by Talent (2006). Entries for each taxon are codes for individual trees and correspond to vouchers deposited in TRT. The classification is from Phipps et al. (1990). 2003). The triploid mother trees (Table 3) are two native male-sterile taxa; one referred to as C. ?grandis (Dickinson and Phipps 1986; Phipps 1988a) appears to be a hybrid of C. punctata and C. crus-galli; the other triploid is a form of C. succulenta. The three tetraploid species (Table 4) are from three series (two sections) of the genus: C. crus-galli (mother trees in a naturalized stand in Toronto; pollen parents from the Regional Municipality of Niagara in the native range), C. macracantha, and C. submollis. The trees were labelled with aluminum tags, and voucher specimens were deposited at TRT. Both spring and autumn vouchers were filed because these are needed for accurate identification of most Crataegus species. Pollinations and seed collection Pollinations were carried out between morphological species and generally between taxonomic sections (Fig. 1). The closest approximation to a diploid–autotetraploid pollination is the C. punctata – C. crus-galli cross, where the parents are both classified in section Crus-galli, but in different series (Table 1). Pollen was obtained and dried overnight as described previously (Talent and Dickinson 2007) and refrigerated (4 8C) for up to 3 days. Pollinations were done before anthesis and without emasculation because emasculation causes parthenocarpy (Dickinson and Phipps 1986) and is likely to affect pollen-tube behaviour (Greyson and Tepfer 1967; Greyson and Raman 1975; Williams and Knox 1982; Celotti 1995). We suspect that it might also affect rates of sexual and apomictic reproduction. Fruit were collected when red in colour, though not necessarily fully ripe. Fruit that were fully ripe were refrigerated for up to 1 year after harvest, but immature fruit were processed immediately. There is one locule and one style per pistil in Crataegus flowers, but the number of pistils is variable in some species. It is difficult to count the styles while pollinating; therefore, the number of locules pollinated was estimated from the fruit using the average number of locules in a fruit multiplied by the number of flowers pollinated. For pollinations with low fruit yield (fewer than 15 fruit), the number of locules in the original flowers was estimated from the average number of locules in 100 openpollinated fruit from the same tree. Flow cytometry Flow cytometry of a suspension of nuclei from chopped seeds followed previously established procedures for leaf and seed tissue (Talent and Dickinson 2005, 2007), using a Becton Dickinson FacsCalibur (BD Biosciences) with an argon laser for detecting propidium iodide DNA stain. The DNA amount was calculated relative to Pisum sativum using the value of 9.56 pg DNA per Pisum nucleus (Johnston et al. 1999), rather than 8.84 pg (Greilhuber and Ebert 1994). The seeds were analyzed either whole (with embryo and endosperm chopped together) or dissected into separate endosperm and embryo preparations. Endosperm measurements from Crataegus seeds can overlap with the Pisum standard if the samples are mixed, and therefore an external Pisum standard was used. When the seeds were prepared whole, the endosperm measurements were standardized relative to the embryo from the same seed, i.e., the embryo was used as an internal standard for the endosperm signal, and the standard for the embryo measurement (Pisum) was external (Talent and Dickinson 2007). It was previously established (Talent and Dickinson 2007) that maternal contamination via the seed coat is not significant in mature seed. It was # 2007 NRC Canada Talent and Dickinson 573 Table 2. Interploidy pollination results with diploid Crataegus mothers. No. of fruit a Mean no. of seed per fruit Embryo ploidy levels of seeds analyzedb 9 20 20 18 36 36 18 50 42 0 6 1 1 3 1 2 10 (0) 7 (0) — 1.42 1 1 0.66 — 1.5 1.3 1.28 — 8 1 3x (1) 3x (2) — 2x (1), 3x (2) 13 9 C. punctata
C. macracantha NT92 2000 NT101 20 NT93 2000 NT101 20 NT94 2000 NT101 40 NT230A 2003 NT105 53 NT231 2003 NT105 101 5 3 4 1 2 1.4 1.33 1 — 1 7 4 4 — 3x (2) C. punctata
C. submollis NT231 2003 NT13 39 2 C. monogyna
C. crus-galli NT117 2003 NT72 50 23 NT117 NT117 43 34 6 8 C. monogyna
C. macracantha NT117 2002 NT104 24 NT117 2003 NT105 106 11 19 Mother tree Year Pollen source C. punctata
C. crus-galli NT92 2000 NT32 NT93 2000 NT32 NT94 2000 NT32 NT93 2002 NT72 NT120 2002 NT35 NT120 2002 NT72 NT121 2002 NT72 NT230A 2003 NT72 NT231 2003 NT72 2004 2004 NT72 NT35 C. monogyna
C. submollis NT117 2003 NT13 No. of flowers 50 0 1 2x (1), 3x (1) 0.87b No embryo (1), 3x (19), 4x ? (1) 3x(6) No embryo (2), 3x (8) 1 1 — 0.63b 11c No embryo (1), 3x (11) — — a Parentheses indicate the number of fruit remaining after samples for morphological analysis were removed. Until 2002, seed set only was counted. Parentheses indicate the number of seeds analyzed by flow cytometry. c The tree was vandalized before the fruit ripened, and the seeds were in poor condition when found. b also established that a low level of endopolyploidy throughout the tissues of the seed did not interfere with resolving the separate embryo and endosperm signals. The seeds come from species that apparently vary slightly in their basic DNA amounts, but ploidy levels of 2x, 3x, 4x, and 5x are distinguishable with 0.82 pg ± 10% per genome copy (Talent and Dickinson (2005), adjusted using Johnston et al. (1999)), although there is an overlap in the estimated pentaploid and hexaploid ranges (3.69–4.51 pg and 4.43– 5.41 pg, respectively). Thus, almost all embryo ploidy levels could be readily distinguished. We used the standard deviation of the fluorescence calculated according to the method contributed by Warren Lamboy (Dickson et al. 1992) that takes into account the distributions of both the Crataegus signal and the Pisum standard. If the ratio of the standard deviation to the mean (2/) was greater than 15% for any measurement (embryo or endosperm), then that measurement was excluded (one seed from C. macracantha
C. monogyna was therefore excluded). The estimated midpoint DNA amount for the ploidy level of the embryo was used to adjust the endosperm measurement relative to the embryo (Talent and Dickinson 2007): Endospermadjusted ¼ ðEndospermmeasured
DNAmidpoint Þ=Embryomeasured with DNA midpoint estimates of 1.64 pg for diploids, 2.46 pg for triploids, 3.28 pg for tetraploids, 4.10 pg for pentaploids, and 4.92 pg for hexaploids. Chi-square tests of the seed set from pollinations were performed using Haber’s correction for continuity (Zar 1999). The cumulative distributions of two data sets were compared using a Kolmogorov–Smirnov comparison (Zar 1999), calculated using the statistical package of the R language and environment (The R Development Core Team 2005). Results Seed set All but four of the 16 intertaxon interploidy pollinations produced seed (Fig. 1), the exceptions being the reciprocal # 2007 NRC Canada 574 Can. J. Bot. Vol. 85, 2007 Table 3. Interploidy pollination results with triploid Crataegus mothers. Mother tree Year Pollen source No. of flowers No. of fruit Mean no. of seed per fruit Embryo ploidy levels of seeds analyzeda 0 — — 0 — — Triploid
diploid C. ?grandis
C. monogyna NT29 2004 MP28 101 C. ?grandis
C. punctata NT29 2004 NT227 87 C. succulenta
C. monogyna NT221 2004 MP28 73 NT241 2004 MP28 69 10 20 0.67 0.80 3x (3), 4x (3) 3, 3x (7), 3–4x (1), 4x (5) Triploid
tetraploid C. succulenta
C. submollis NT221 2004 NT26 NT241 2004 NT26 32 40 1.00 0.98 22, 3x (5), 5x (5) 33, 3x (1), 5x (5) 92 66 a Parentheses indicate the number of seeds analyzed by flow cytometry. crosses of tetraploid C. submollis with diploid C. monogyna and triploid C. ?grandis pollinated from two diploids. Pollen quality does not explain the failures; for example, pollen from diploid C. monogyna produced zero seed on triploid C. ?grandis (101 flowers, estimated total locules = 247), but later pollination using the same pollen on triploid C. succulenta produced significantly greater seed set (142 flowers, estimated total locules = 417; 22 seed; X2 = 16.67 with Haber correction for continuity, df = 1, P < 0.001). The seed set apparently relates to the identity of the parents and not simply to their ploidy levels. Diploid mother plants Pollinations on C. monogyna were successful from both C. crus-galli and C. macracantha (Table 2; 27.56% and 23.08% of locules filled, respectively), but a small experiment (50 flowers = 50 pistils) with pollen from C. submollis yielded no seed (which is a significantly different success rate from the pooled data for the other pollen from tetraploids; X2 = 11.58 with Haber correction for continuity, df = 1, P < 0.001). Of the seeds that were tested by flow cytometry, nine (seven
C. crus-galli, two
C. macracantha) were dissected and separately analyzed as embryo and endosperm, with the result that the embryo signal was triploid (2.052–2.508 pg DNA) and the endosperm signal was tetraploid (2.736–3.344 pg DNA). The remaining seeds were prepared whole; of these 30 gave both triploid and tetraploid signals, another 12 gave a triploid signal only, and five gave a tetraploid signal only. The volume of the embryo is relatively large, and thus the seeds without a tetraploid signal appear to be preparation failures in which the endosperm signal was lost. Of the seeds with only a tetraploid signal, four were flat seeds with no visible embryo, and the cytometric signal apparently belonged to the endosperm. The fifth of these seeds was brown throughout with a plump brown embryo, and it is not clear whether this embryo was tetraploid or did not yield a signal. The seed set on diploid C. punctata pollinated from tetraploids was low (Table 2). Cytometry of three of the seeds yielded both embryo and endosperm signals. One seed (with 3x embryo and 4x endosperm) probably indicates fertilization by the pollen from the tetraploid, or alternatively by un- reduced self-pollen. Another seed suggests self-pollination or pollen contamination from a diploid (2x embryo, 3x endosperm). The third seed (with 2x embryo and 6x endosperm, pollinated from C. submollis) suggests apomixis: a meiotically unreduced megagametophyte, a parthenogenetic embryo, and endosperm derived from a binucleate central cell fertilized by one sperm from the tetraploid. Triploid mother plants Hand pollinations of C. ?grandis using pollen from two diploids produced no seed (Table 3; Fig. 1), but the other triploid male-sterile species, C. succulenta, produced seed with pollen from a tetraploid and a diploid (Figs. 1, 2). An unusual seed had an aneuploid embryo accompanied by an endosperm that seems to derive from a meiotically reduced megagametophyte, but the majority of the seeds were consistent with meiotically unreduced megagametophytes with triploid nuclei. The embryo ploidy levels indicate that both parthenogenesis (the 3 + 0 case in Fig. 2) and fertilization of the egg cell (the 3 + 1 and 3 + 2 cases in Fig. 2) occurred. Two exceptionally high endosperm signals (Fig. 2) appear to indicate a megagametophyte with unusual morphology: a form with a trinucleate central cell that has been observed in Crataegus and in other apomictic Rosaceae tribe Pyreae (Muniyamma and Phipps 1984a; Jankun and Kovanda 1986; Campbell et al. 1987). The other endosperm ploidy levels were consistent with a binucleate central cell with reduced nuclei fertilized by either one or two (reduced) sperm, as previously observed in apomictic Crataegus (Talent and Dickinson 2007). Tetraploid mother plants The two highly apomictic species C. crus-galli and C. macracantha, pollinated without emasculation using pollen from the diploids (Table 4; Figs. 1, 3, 4), showed only tetraploid embryos in the 70 seeds that were tested. Three seeds had hexaploid endosperm, consistent with sexual megagametophytes. The majority of the seeds had higher endosperm measurements close to 10x, 12x, 14x, or 16x (Fig. 4). The 10x and 12x ploidy levels are consistent with meiotically unreduced binucleate central cells combined # 2007 NRC Canada Talent and Dickinson 575 Table 4. Interploidy pollination results with tetraploid Crataegus mothers. Mother tree Year Pollen source No. of flowers No. of fruit C. submollis
C. monogyna NT26 2003 D1697 45 0 Mean no. of seed per fruit Embryo ploidy levels of seeds analyzeda — — C. submollis
C. punctata NT26 2003 D1696 63 28 1.03 C. crus-galli
C. SC05 2000 NT58 2001 NT62 2001 SC06 2001 NT62 2003 NT66 2003 monogyna NT100 NT132 NT132 NT132 NT116 D1697 20 20 20 20 26 35 9 4 2 3 1 1 0.33 0 1 0.67 1 — C. crus-galli
C. SC04 2000 SC15 2000 NT58 2001 NT62 2001 SC06 2001 SC06 2002 NT62 2003 NT66 2003 NT66 2003 SC06 2003 punctata NT94 NT93 NT93 NT93 NT93 NT120 D1696 D1696 NT231 D1696 20 20 40 40 40 32 62 97 41 91 5 6 12 9 8 12 12 7 9 12 0.6 0.58 0.79 0.44 0.69 0.5 0.25 0.71 0.22 0.58 3 3 9 4 5 4x 4x 4x 4x 4x C. macracantha
C. monogyna NT107 2000 NT100 40 NT108 2000 NT100 21 NT102 2002 NT114 30 NT104 2002 NT114 30 NT118 2003 NT116 50 NT131 2003 NT116 10 NT229 2003 NT116 20 NT230 2003 NT116 20 15 2 18 17 11 1 9 (4)a 12 (6)a 0.87 0.50 0.86 0.54 0.82 1 0 0 14 1 4x (15) 1, 4x (6) 2, 4x (9) 4x (1) — — C. macracantha
C. punctata NT101 2000 NT1 10 NT103 2000 NT1 10 NT130 2001 NT93 20 NT131 2001 NT94 20 NT133 2001 NT93 21 NT130 2003 D1696 70 NT133 2003 D1696 70 NT229 2003 D1696 40 NT230 2003 D1696 20 1 3 3 9 4 21 6 16 (8)a 10 1 1 1 0.55 0.75 0.52 0.33 0.25 — 1 1 5 4 2 4x (11) 4x (2) 4x (2) — 2x (2), 2–3x (1), 3x (12), 4x (4), 5x (1), 6x (1) 4 — 2 2 4x (1) — (6) (3) (5) (2) (7) a Until 2002, seed set only was counted. Parentheses indicate the number of seeds analyzed by flow cytometry. with one reduced self-sperm (10x) or with two reduced selfsperm or one unreduced self-sperm (12x). As with the triploid mother plant, the exceptions (ploidy levels higher than 12x) suggest the involvement of more than two nuclei from the megagametophyte. The endosperm ploidy levels from these interploidy pollinations are not the same as those from previous intrataxon pollinations of tetraploid C. crus-galli and C. macracantha (Talent and Dickinson 2007): the proportion of 10x measurements is higher (in a Kolmogorov– Smirnov test of the cumulative distributions of the DNA measurements, excluding one seed with a hexaploid embryo, n1 = 53, range1 = 4.64–12.09 pg, median1 = 7.97 pg, n2 = 150, range2 = 4.18–12.60 pg, median2 = 9.25 pg, D = 0.3194, P < 0.0004719). The 10x endosperm is consistent with fertilization of the central cell either by one sperm from self-pollination or by two monoploid sperm from the diploid, but curiously, no 9x endosperm from fertilization by a single sperm from the diploid is apparent. The pollination of tetraploid C. submollis from diploid C. punctata was one of the more successful crosses. The ploidy levels of the embryos vary greatly (Fig. 5): most were triploid, some were the tetraploids that we would ex# 2007 NRC Canada 576 Can. J. Bot. Vol. 85, 2007 Fig. 1. Crossing design for pollination experiments. Cross-pollinations were made between plants of different ploidy levels (2x, 3x, and 4x), and predominantly between different taxonomic sections. Known hybrids (Eggleston 1908, 1923; Palmer 1956; Wells and Phipps 1989) are indicated by an oval outline. Arrows represent the direction of pollen transfer, and numbers adjacent to the arrow heads represent the seed set, e.g., 47/986 means that an estimated 986 pistils (locules) produced 47 seed at the rate of one seed per locule. The flowers were not emasculated; the two triploids are male-sterile but the other species produce copious pollen. pect from either apomixis or self-pollination, others were pentaploid, hexaploid, or diploid, and one was apparently aneuploid. The endosperm ploidy levels also indicate that about 70% of the seeds that matured were derived from meiotically reduced megagametophytes: the diploid and triploid embryos occur with pentaploid endosperm, confirming that the megagametophyte was meiotically reduced (Fig. 5). The endosperm of seeds with 4x embryos indicates both reduced and unreduced megagametophytes: one endosperm is 6x, two are 9x or 10x, and one is 12x or 13x. The seed with 6x endosperm likely represents a reduced megagametophyte fertilized by diploid pollen. Diploid pollen might arise from some meiotic nonreduction in the pollen from C. punctata, but because these flowers were not emasculated, selfpollination with reduced self-pollen is also possible. The other three seeds with 4x embryos are consistent with apomictic megagametophytes, as was seen with C. macracantha and C. crus-galli. The endosperm measurement close to 9x might have been fertilized with sperm from the diploid (4 + 4 +1 = 9), but the 9x and 10x DNA ranges cannot be reliably distinguished with the cytometric technique used here (Talent and Dickinson 2005), and conse- quently it is possible that this seed had a 10x endosperm. The seeds with 5x and 6x embryos are consistent with apomeiosis (4x eggs) and fertilization (1x or 2x pollen), and the 5x embryo would therefore occur with a 9x endosperm (4 + 4 + 1), although the cytometric measurement appears to be closer to 10x than to 9x. The expected error in the cytometric measurement (Talent and Dickinson 2005) is great enough, however, that 9x could be the correct ploidy level for the endosperm of this seed. Discussion Taxonomic implications The seed set of these interploidy, intertaxon pollinations in Crataegus (Fig. 1) has been as good as or better than that achieved by other workers with intrataxon pollinations (Dickinson 1983; Wells 1985; Macklin 2001). Twelve of our 16 interspecific pollinations produced some seed, and the ploidy levels of at least some of the embryos indicated hybridization, except where the mother plant was a highly apomictic tetraploid. Differences in flowering time provide one explanation of # 2007 NRC Canada Talent and Dickinson 577 Fig. 2. Embryo and endosperm ploidy levels from male-sterile triploid C. succulenta pollinated from diploid C. monogyna and tetraploid C. submollis. Pollinations on male-sterile triploid C. succulenta involved pollen from both a diploid and a tetraploid (Table 3). With the exception of one seed that appears to be aneuploid 3x–4x, the DNA measurements of the embryos were euploid and have been adjusted to the estimated midpoint for that ploidy level, with the corresponding endosperm signal adjusted accordingly (see Materials and methods for details). Where multiple data points overlap, the number of points is indicated numerically. Along the axes the inferred ploidy levels of the constituent gamete nuclei are indicated, with female gametes listed first in parentheses (e.g., (3+3) +1+1 indicates two triploid central-cell nuclei and two monoploid sperm). Open circles, pollen from the diploid was used; solid circles, pollen from the tetraploid was used. 11 13 (3 + 3 + 3) + 2 + 2 9 (3 + 3 + 3) + 2 11 (3 + 3) + 2 + 2 10 2 1 8 9 7 9 8 2 (3 + 3) + 2 OR (3 + 3) + 1 + 1 8 (3 + 3) + 1 7 8 6 5 Ploidy level of endosperm 12 (3) + 1 (3) + 2 6 (3) + 0 Endosperm DNA amount (picograms) 10 5 4 2.5 3 3.5 4 4.5 5 5.5 6 Ploidy level of embryo why Crataegus species found growing together do not (apparently) hybridize (Palmer 1932, 1943; Campbell et al. 1991; Phipps 2005). Our results suggest at least three more reasons. Interspecific pollinations may fail to result in hybrid seeds because ovules are removed from play, firstly by selfing in self-compatible pollen-fertile polyploids and secondly by parthenogenesis in pseudogamous apomicts. Thirdly, our results suggest that there may also be some postpollination incompatibility, at least when ploidy levels differ between parents in intra- and inter-series crosses. Diploid C. punctata is probably a close relative of triploid C. ?grandis, but its pollen produced no seed on the triploid. We have not yet discovered a diploid pollen source that can produce seed on triploid C. ?grandis, but trees at the same locality that were treated with pollen from the tetraploid Crataegus conspecta Sarg. produced a seed in 22% of treated locules (30 flowers, 95% confidence interval 4.8%– 42.3%; Dickinson and Phipps 1986). Thus the most effective pollen source so far discovered for C. ?grandis of section Crus-galli is a tetraploid from section Coccineae. Seed set on triploid C. succulenta also occurred with distantly related pollen parents, one tetraploid and one diploid. On the other hand, the small experiment with reciprocal pollinations of tetraploid C. submollis and diploid C. monogyna (total 95 flowers, 245 pistils) was unsuccessful and might indicate intersectional incompatibility. Hybrids between different taxonomic series of Crataegus have been known in horticulture for centuries, and other putative hybrids have been identified in nature (Christensen 1992; Phipps et al. 2003). These hybrids include diploids, triploids, and tetraploids (Talent and Dickinson 2005), but cross-pollinations of North American species have been primarily designed to test diploid–diploid hybridization (e.g., Love and Feigen 1978; Wells and Phipps 1989; Celotti 1995). There is, however, a statement by Brown (1910) that suggests that seed set possibly occurs after a variety of cross-pollinations: ‘‘During the spring of 1908, I pollinated a few Crataegus monogyna (English hawthorn) flowers with pollen from C. brainerdi, a native [North American] species. They set fruit which matured. During the flowering season of 1909, Mr William Moore and the writer made cross pollinations between the majority of the native species of the local flora. Most of these cross pollinations were effective — fruit set and matured, being entirely normal apparently. (These experiments are still in progress, the details of which will be published later.)’’ The results of Brown and Moore’s work do not appear to have been published, and it is unclear whether the description of the fruit as ‘‘entirely normal apparently’’ included checking for seed (the experimenters might have preferred to try to germinate the fruits or the woody pyrenes without opening them). Our experience is that C. monogyna is likely to set fruit that contain seed, but other Crataegus species are highly parthenocarpic (for example, the C. succulenta plants used for our experiments are male-sterile and isolated from other hawthorns, and a sample of 162 fruit (472 pistils) contained no seed). We can therefore only say that the high seed set of the interploidy intertaxon pollinations reported here comes from an assortment of taxonomically separate ‘‘species’’ and further work will be needed to test whether wide hybridization between ploidy levels occurs generally throughout the genus. The seeds of C. submollis show much higher rates of meiotic reduction than in the other tetraploid Crataegus, but it is not clear that sexual hybridization with C. punctata occurs in nature, although pollen transfer may be possible, at least at the study site, which has a temperature gradient because of its proximity to Lake Ontario. The flowering times of the two species did not overlap at an inland study site in Durham (Hoy 1992). If this individual is representative of its species, then the fitness reduction due to a high rate of nontetraploid (triploid and some diploid, pentaploid, and hexaploid) offspring could be a factor that has selected for the almost complete separation of flowering times between it and the diploid species that commonly occurs in proximity. Sexual species and near-obligate apomictic forms are quite common in the genus Crataegus, and the taxonomy is complicated by segregates that are thought to vary in chromosome number and reproductive behaviour. However, embryological investigation has shown partial (facultative) apomixis in some tetraploid Crataegus individuals (Muniyamma and Phipps 1985), and we have now confirmed that both meiotic and apomeiotic embryo sacs can produce significant quantities of seed in C. submollis. This species therefore offers a valuable research opportunity to test whether some taxa that have been, and should continue to be, recognized as species are made up entirely of partially # 2007 NRC Canada 578 Can. J. Bot. Vol. 85, 2007 Fig. 3. Sources of tetraploid seed from pollinations of unemasculated flowers. The cross-pollinations shown in Fig. 1 produced seed with embryos of various ploidy levels. Tetraploid individuals constitute a majority of apomictic Crataegus, and the pollinations that produced seed with tetraploid embryos are highlighted. sexual forms (as suggested by Muniyamma and Phipps 1985). Endosperm balance constraints In plants with the Polygonum-type megagametophyte, like Crataegus, we would expect to find that the endosperm is less viable when it contains maternal and paternal genome copies in a ratio other than 2:1 (Brink and Cooper 1947; Nogler 1984; Katsiosis et al. 1995). The 2:1 ratio was violated in intrataxon pollinations of apomictic tetraploid Crataegus (Talent and Dickinson 2007) where a majority of the seed has 10x endosperm (a 4:1 ratio). Similar endosperm formation has been noted in some but not all other aposporous apomicts. In the indeterminate gametophyte (ig) mutant of maize (Lin 1984), very high endosperm ploidy levels show a reduced endosperm-balance requirement. Quarin (1999) proposed a model based on this for aposporous Paspalum (Poaceae) where endosperm with high ploidy levels (maternal input 6x or more) loses the endosperm balance constraint, and thus apomixis is successful only in polyploids. This explanation, however, does not explain the readiness of diploid Crataegus to accept pollen from tetraploids. A complete lack of an endosperm balance requirement is a more satisfactory explanation, and it has not yet been possible to cross-pollinate a diploid and its autotetraploid to test this directly (the parent species of C. ?grandis are not a diploid–autotetraploid pair and may not even be very closely related; Talent and Dickinson 2005). There is very little evidence that Rosaceae generally have an endosperm balance requirement (we are not aware of any studies that have directly investigated the question), but there are good indications that such a requirement exists in Rubus (Dowrick 1966; Jennings et al. 1967; Topham 1970). An alternative explanation is that the endosperm balance number of polyploid Crataegus is reduced, as in polyploids of some other plant families (Hawkes and Jackson 1992). This explanation, however, does not explain the variability in the apomicts, where either one or two sperm can fertilize the endosperm (Talent and Dickinson 2007). Curiously, although either one or two sperm from a tetraploid or two sperm from a diploid can fertilize the endosperm in apomeiotic megagametophytes of tetraploids, the only evidence that a single sperm from a diploid is sufficient for this purpose (two seeds; Fig. 5) comes from C. submollis, which is partly sexual. This may indicate that endosperm formation in strongly apomictic species is not # 2007 NRC Canada Talent and Dickinson Fig. 5. Embryo and endosperm ploidy levels from (unemasculated) tetraploid C. submollis pollinated from diploid C. punctata. A tetraploid C. submollis mother tree was pollinated from diploid C. punctata (Table 4) without emasculation. With the exception of one seed that appears to be aneuploid 2x–3x, the DNA amounts of the embryos were euploid and have been adjusted to the estimated midpoint for that ploidy level, with the corresponding endosperm signal adjusted accordingly (see Materials and methods for details). Two diploid and 12 triploid embryos are indicated; all other points represent a single seed. Along the axes, the ploidy levels of the inferred constituent gamete nuclei are indicated as in Fig. 2. Open circles, monoploid pollen from the diploid was inferred to be responsible for fertilization; solid circles, diploid pollen (possibly self-pollen) was inferred to be responsible; open square, an aneuploid embryo with hexaploid endosperm suggestive of a meiotic megagametophyte. 11 13 10 (4 + 4) + 4 12 8 (4 + 4) + 2 10 (4 + 4) + 1 9 7 8 6 7 5 4 12 2 (2 + 2) + 2 6 (2 + 2) + 1 5 Ploidy level of endosperm 11 9 (2) + 1 (4) + 0 OR (2) + 2 (4) + 1 (4)+ 2 4 3 (2) + 0 Endosperm DNA amount (picograms) Fig. 4. Endosperm ploidy levels from highly apomictic polyploids pollinated (without emasculation) from diploids. Crataegus crusgalli and C. macracantha, previously shown to be highly apomictic with open and self-pollination (Talent and Dickinson 2007), were pollinated from diploids (Tables 1 and 4) without emasculation. The DNA amount of the embryo has been adjusted to the estimated midpoint for tetraploids (3.04 pg), with the corresponding endosperm signal adjusted accordingly (see Materials and methods for details). One seed, not shown, from C. crus-galli tree NT62 pollinated from C. monogyna tree NT116 had hexaploid endosperm. (a) C. crus-galli trees NT62, NT66, and SC06 pollinated from C. punctata trees D1696, NT120, and NT231; (b) C. macracantha trees NT102, NT104, NT118, and NT131 pollinated from C. monogyna trees NT114 and NT116; (c) C. macracantha trees NT130, NT133, and NT229 pollinated from C. punctata tree D1696. 579 2 3 4 5 6 3 2 1 7 Ploidy level of embryo completely flexible: a weak endosperm balance requirement remains, and most nonaploid endosperm in tetraploid mother plants does not mature. Possible apomixis in diploid C. punctata One of the seeds from diploid C. punctata pollinated from tetraploid C. submollis had a diploid embryo and hexaploid endosperm, which would be consistent with a meiotically unreduced megagametophyte in which the central cell was fertilized by sperm from the tetraploid. Alternatively, these measurements could indicate endoreduplication of a triploid endosperm. It may therefore be significant that a possible aposporous initial from C. punctata was photographed by Dickinson (1983), although in an earlier test (Talent and Dickinson 2007), this species and other diploids (84 openpollinated seeds from six diploid species, of which 45 seeds were from C. punctata) appeared to produce only sexually derived mature seed (with diploid embryos and triploid endosperm). Several authors have argued that apomixis is likely to arise de novo in polyploids, particularly in polyploid hybrids (Stebbins 1941; Camp 1942; Stebbins 1950, 1980; Carman # 2007 NRC Canada 580 2001). However, apomeiotic development can occur through a variety of processes. The maternal archesporium of Rosaceae is commonly multicellular (Davis 1966), but the boundary between it and the nucellus is difficult to define cytologically (Jankun and Kovanda 1988). Apospory (in the nucellus) is commonly noted in Rosaceae and Crataegus, but diplospory of the Taraxacum type (which involves only the megaspore mother cell or a cell in the equivalent position; Nogler 1984) was seen in a triploid individual of Crataegus (Muniyamma and Phipps 1984b). The cell lineages that give rise to aposporous initials in Crataegus have not yet been studied. This suggests the possibility that apospory and diplospory could be initiated as developmental variations of the same basic process in diploids. Dickinson (1983) has hypothesized that a mutation in polyploid Crataegus completes the later stages of apomixis by permitting parthenogenesis (i.e., development of an apomictic embryo from the unreduced egg cell). Thus, it would be of great interest if apomeiosis can lead to a mature seed in a diploid; if it could, then this would indicate that some factor other than the lack of a mutation for parthenogenesis has prevented the general success of apomixis in diploids. Our data suggest that C. punctata warrants further investigation for rare apomixis, and pollination from C. submollis may be an enabling condition for detecting mature apomictic seed in that species. Nonrecurrent apomixis and fertilization Gametophytic apomixis in angiosperms involves a meiotically unreduced megagametophyte and the formation of a seed with an embryo that is a clone of the mother plant (Nogler 1984). The contrasting ‘‘nonrecurrent apomixis’’ was defined by Maheshwari (1950) to apply to diploid mother plants and included only the development of embryos from cells of reduced (haploid) gametophytes, i.e., parthenogenesis and apogamy that yield haploid embryos. The definition has been extended by Mazzucato (1996) to include polyhaploidy (in which megagametophytes are reduced and parthenogenesis of the egg cell yields a nonmonoploid product, such as a dihaploid embryo from a tetraploid individual) and nonreduction coupled with fertilization (BIII hybridity), and we have seen that both these processes occurred in polyploid Crataegus. Dihaploid plants (this term is used sensu Bender (1963) and subsequent authors, not for doubled monoploids, to which the same term is sometimes applied) are produced in low numbers by parthenogenesis in tetraploids of many genera but can be a significant proportion of the seedlings when particular pollen parents are used, with pollen irradiation, with interspecies pollinations, and with interploidy pollinations (Nogler 1984). It may therefore be significant that the pollination of tetraploid C. submollis that produced diploid embryos involved pollen from another section of the genus, as well as a different ploidy level. Two mechanisms of diploid formation from tetraploid mother plants can be difficult to distinguish: parthenogenesis of an unfertilized egg cell (dihaploidy) and chromosome elimination following fertilization (Solntseva et al. 1998). The two diploid embryo DNA measurements from C. submollis pollinated from C. punctata could therefore be parthenogenetic dihaploids or might have derived through chromosome loss from a trip- Can. J. Bot. Vol. 85, 2007 loid embryo. In either case, the pentaploid endosperm indicates that a meiotically reduced megagametophyte and monoploid sperm were involved. An important question in apomixis research is whether apomeiosis and parthenogenesis are under separate, but possibly linked, genetic control. Polyhaploids, involving parthenogenesis without apomeiosis, were observed only in the partly sexual tetraploid and, if they occurred in the triploid, would probably have been lethal. Apomeiosis without parthenogenesis (BIII hybrid embryos) occurred at a low rate in all the tetraploids, including those that are partly sexual and those that are highly apomictic (Talent and Dickinson 2007). BIII hybridization was particularly evident in the triploids (Talent and Dickinson 2007; Fig. 2) and yielded tetraploid and pentaploid embryos. We cannot say that genes for apomeiosis and parthenogenesis are uncoupled in any of these Crataegus, but only that parthenogenesis does not necessarily occur in apomeiotic megagametophytes, particularly in the triploid C. succulenta. A principal cause of ploidy-level changes in these experiments did not involve any sort of apomixis, but rather compatible crosses on diploid mother plants with pollen from tetraploids (but see Appendix A for discussion of the fertility of some such crosses). The probable importance of triploids In general, sexual tetraploids arising among a population of ancestral diploids are incapable of hybridizing with the diploids, and their establishment is unlikely (the minority cytotype mating disadvantage; Levin 1975; Husband 2000). However, several factors can reduce this disadvantage, including self-compatible pollen, which occurs in neotetraploid Rosaceae (Crane and Lewis 1942; Lewis 1943; Entani et al. 1999; Takayama and Isogai 2005) and has been demonstrated in tetraploid Crataegus (Dickinson and Phipps 1986; Smith and Phipps 1988; Dickinson et al. 1996; Macklin 2001). Gene flow from diploids to tetraploids via triploids could be important to the success of tetraploids if there is no triploid-block effect, as we have demonstrated is a close approximation to the situation in Crataegus. The importance of gene flow via triploids has been demonstrated in Chamerion angustifolium (L.) Holub (Onagraceae; Burton and Husband 2001; Husband 2004), a plant that lacks a triploid-block effect. However, tetraploid Crataegus enjoys further advantages not available to Chamerion: both triploid and tetraploid Crataegus are long-lived perennials with a capacity for both vegetative and apomictic reproduction, and tetraploids can apparently be produced at quite a high rate from some triploid–diploid pollinations. Gene flow via triploids has not yet been fully demonstrated in Crataegus, but our pollination experiments with wide hybrids suggest considerable potential that warrants investigation in natural settings. Selection for mutations that confer apomixis will be particularly strong in triploids that are otherwise largely sterile (Camp 1942). Acquisition of such mutations, and the resulting reproductive success of the triploids with those mutations, has likely contributed to the development of apomixis in Crataegus. Such triploids would have the potential to give rise to tetraploids when their apomeiotic female gametes are fertilized by reduced sperm from diploids. Those tetraploids would inherit the apomixis mutations from their # 2007 NRC Canada Talent and Dickinson triploid parents without necessarily losing the capacity for meiosis. The first tetraploids might not have been as strongly apomictic as C. crus-galli, C. macracantha, and C. douglasii sensu lato are today (Talent and Dickinson 2007). Simple genetic models, sometimes backed by simulation studies (e.g., Charlesworth 1980; Stebbins 1980; Marshall and Brown 1981; Mogie 1992; Noirot 1993), have been used to argue that apomixis will spread to fixation except for a necessary residual capacity for some sexual reproduction to avoid the dangers of Muller’s ratchet (Muller 1932). The tetraploids studied so far, like other studied apomicts (Savidan 2000), retain a low level of sexuality that might in itself be sufficient. On the other hand, new alleles descending from sexual diploids to triploids and then to near-obligate apomictic tetraploids might be sufficient to offset the deleterious mutations that accumulate. However, gene flow from diploids to tetraploids has yet to be documented. The genetic relationships between near-obligate apomictic tetraploid Crataegus and their diploid and triploid neighbours thus repay further study. Self-compatible pollen and apomictic polyploids The question of whether apomixis occurs in diploid angiosperms has received considerable attention (Stebbins 1941; Fagerlind 1944; Stebbins 1950, 1980; Nogler 1984; Savidan 2000), and a recent opinion is that diploids do not reproduce by apomixis with ‘‘a few academic exceptions, some of them possibly controversial’’ (Savidan 2000). However, apomictic diploids of Ranunculus (which might not have sufficient vigour to compete in a natural setting) can form via parthenogenesis in meiotically reduced megagametophytes of partly apomictic tetraploids (dihaploidy; Nogler 1982). Jankun and Kovanda (1988) made embryological observations of a Sorbus hybrid (Sorbus
eximia Kovanda in Rosaceae tribe Pyreae, like Crataegus), and found apomeiosis (both apospory and diplospory) in both diploid and tetraploid cytotypes. They suggested two possibilities: that the diploids arose as dihaploids from the tetraploids, or alternatively that the diploid was a hybrid that inherited apomixis genes from one of its parents. It therefore appears that the most interesting question about the genus Crataegus is not how apomixis became established, but rather why it is scarcely evident in diploids. One possibility is that an allele or combination of alleles essential to a high rate of apomixis cannot be transmitted via monoploid gametes, as was found in Ranunculus and Tripsacum (Nogler 1982; Grimanelli et al. 1998). An alternative explanation, however, concerns the self-compatibility seen in many apomictic tetraploid Crataegus (Dickinson and Phipps 1986; Smith and Phipps 1988; Dickinson et al. 1996; Macklin 2001), which require pollination to form the endosperm of the seed (Talent and Dickinson 2007). The competitive interaction model of self-incompatibility in Rosaceae holds that heteroallelic pollen bypasses the selfincompatibility system and that most neopolyploids are heteroallelic and therefore produce some self-compatible pollen (Entani et al. 1999; Ridout et al. 2005; Takayama and Isogai 2005). Models for Poaceae (Noirot et al. 1997) indicate that self-compatible pollen is essential for the maintenance of apomictic populations that, like Crataegus, require pollination. It would therefore be of great interest to know whether 581 the exceptional diploid apomicts in S. eximia (Jankun and Kovanda 1988) were self-compatible (the endosperm was fertilized). Now that we have the means to rapidly assess the breeding system of individual Crataegus from cytometry of the DNA in seed tissues, we may be able to discover whether diploid apomictic populations have been successful in this large genus and whether such diploid populations are self-compatible. Acknowledgements We thank two anonymous reviewers for suggestions about the manuscript; Sean Graham for pointing out the recent ambiguity of the word ‘‘dihaploid’’; the University of Toronto Faculty of Medicine, Department of Immunology, for the use of their equipment; Cheryl Smith for flow cytometry advice; Bruce Hall for advice about growing Pisum sativum; Barbara Thomson of the Statistical Consulting Service of the University of Toronto for advice about statistics; Graeme Hirst for comments on an earlier version of the manuscript; Richard Dickinson, James E. Eckenwalder, Eugenia Lo, and Melissa Purich for help with general fieldwork; Richard Dickinson, Eric Harris, Sophie Nguyen, Jeannine SaintJacques, Joan Talent, and Dao Tran for help with pollinations; Arunika Gunawardena for photographic assistance; and, crucially, Richard Dickinson, James E. Eckenwalder, and Lena Spoke for locating the diploid trees growing in places where pollination experiments were possible. Access to trees was kindly provided by Heide Bateman and Kim Delaney and by the Toronto and Region Conservation Authority via Natalie Iwanycki. Funding was provided by the Ontario Graduate Scholarship Program, the Ontario Graduate Scholarships in Science and Technology Program, the Natural Sciences and Engineering Research Council of Canada (grant A3430), and the Royal Ontario Museum. References Bender, K. 1963. Über die Erzeugung und Enstehung dihaploider Pflanzen bei Solanum tuberosum. Z. Pflanzenzücht. 50: 141–166. Boom, B.K. 1959. Benaming, geschiedenis en kenmerken van een anntal houtachtige planten. Jaarboek — Nederlandchse Dendrologische Vereeniging, 21: 85–178. Brink, R.A., and Cooper, D.C. 1947. The endosperm in seed development (concluded). Bot. Rev. 13: 479–541. Brown, H.B. 1910. The genus Crataegus, with some theories concerning the origin of its species. Bull. Torrey Bot. Club, 37: 251–260. doi:10.2307/2479407. Burton, T.L., and Husband, B.C. 2001. Fecundity and offspring ploidy in matings among diploid, triploid and tetraploid Chamerion angustifolium (Onagraceae): consequences for tetraploid establishment. Heredity, 87: 573–582. doi:10.1046/j.1365-2540. 2001.00955.x. PMID:11869348. Camp, W.H. 1942. The Crataegus problem. Castanea, 7: 51–55. Campbell, C.S., Greene, C.W., and Bergquist, S.E. 1987. Apomixis and sexuality in three species of Amelanchier, shadbush (Rosaceae, Maloideae). Am. J. Bot. 74: 321–328. doi:10.2307/2443808. Campbell, C.S., Greene, G.W., and Dickinson, T.A. 1991. Reproductive biology in subfam. Maloideae (Rosaceae). Syst. Bot. 16: 333– 349. doi:10.2307/2419284. Carman, J.G. 2001. The gene effect: genome collisions and apomixis. In The flowering of apomixis: from mechanisms to genetic engineering. Edited by Y. Savidan, J.G. Carman, and T. Dresselhaus. CIMMYT Publications, Houston, Tex. pp. 95–110. # 2007 NRC Canada 582 Celotti, N. 1995. The pollen tube pathway and obturator in hawthorn sexual reproduction. B.Sc. Honours thesis, Department of Biology, Queen’s University, Kingston, Ont. Charlesworth, B. 1980. The cost of sex in relation to mating system. J. Theor. Biol. 84: 655–671. PMID:7431946. Christensen, K.I. 1992. Revision of Crataegus sect. Crataegus and nothosect. Crataeguineae (Rosaceae–Maloideae) in the Old World. The American Society of Plant Taxonomists, Ann Arbor, Mich. Clausen, J. 1961. Introgression facilitated by apomixis in polyploid Poas. Euphytica, 10: 87–94. doi:10.1007/BF00037208. Crane, M.B., and Lewis, D. 1942. Genetical studies in pears. III. Incompatibility and sterility. J. Genet. 43: 31–43. Davis, G.L. 1966. Systematic embryology of the angiosperms. John Wiley & Sons, Inc., New York. Dickinson, T.A. 1983. Crataegus crus-galli L. sensu lato in southern Ontario: phenotypic variation and variability in relation to reproductive behavior. Ph.D. thesis, Department of Plant Sciences, University of Western Ontario, London, Ont. Dickinson, T.A., and Phipps, J.B. 1986. Studies in Crataegus (Rosaceae: Maloideae). XIV. The breeding system of Crataegus crus-galli sensu lato in Ontario. Am. J. Bot. 73: 116–130. doi:10.2307/2444284. Dickinson, T.A., Belaoussoff, S., Love, R.M., and Muniyamma, M. 1996. North American black-fruited hawthorns. I. Variation in floral construction, breeding system correlates, and their possible evolutionary significance in Crataegus sect. Douglasii Loudon. Folia Geobot. Phytotaxon. 31: 355–371. Dickson, E.E., Arumuganathan, K., Kresovich, S., and Doyle, J.J. 1992. Nuclear DNA content variation within the Rosaceae. Am. J. Bot. 79: 1081–1086. doi:10.2307/2444917. Dowrick, G.J. 1966. Breeding systems in tetraploid Rubus species. Genet. Res. 7: 245–253. Eggleston, W.W. 1908. The Crataegi of the northeastern United States and adjacent Canada. Rhodora, 10: 73–84. Eggleston, W.W. 1923. Crataegus L. In Report of the State Botanist for 1921. Edited by H.D. House. New York State Museum Bulletin. pp. 63–65. Entani, T., Takayama, S., Iwano, M., Shiba, M., Che, F.-S., and Isogai, A. 1999. Relationship between polyploidy and pollen self incompatibility phenotype in Petunia hybrida Vilm. Biosci. Biotechnol. Biochem. 63: 1882–1888. doi:10.1271/bbb.63.1882. PMID:10635553. Fagerlind, F. 1944. Der Zusammenhang zwischen Perennität, Apomixis und Polyploidie. Hereditas, 30: 179–200. Greilhuber, J., and Ebert, I. 1994. Genome size variation in Pisum sativum. Genome, 37: 646–655. Greyson, R.I., and Tepfer, S.S. 1967. Emasculation effects on the stamen filament of Nigella hispanica and their partial reversal by gibberellic acid. Am. J. Bot. 54: 971–976. doi:10.2307/ 2440720. Greyson, R.I., and Raman, K. 1975. Differential sensitivity of ‘‘double’’ and ‘‘single’’ flowers of Nigella damascena (Ranunculaceae) to emasculation and GA3. Am. J. Bot. 62: 531–536. doi:10.2307/2441962. Grimanelli, D., LeBlanc, O., Espinosa, E., Perotti, E., González de León, D., and Savidan, Y. 1998. Non-Mendelian transmission of apomixis in maize–Tripsacum hybrids caused by a transmission ratio distortion. Heredity, 80: 40–47. doi:10.1046/j.1365-2540. 1998.00264.x. PMID:9474775. Hawkes, J.G., and Jackson, M.T. 1992. Taxonomic and evolutionary implications of the endosperm balance number hypothesis in potatoes. Theor. Appl. Genet. 84: 180–185. Hoy, D. 1992. A test of the hypothesis that leaf and flower emergence phenology correlates with vessel diameter and density within the Can. J. Bot. Vol. 85, 2007 genus Crataegus. Department of Botany, University of Toronto, Toronto, Ont. Unpublished report. Husband, B.C. 2000. Constraints on polyploid evolution: a test of the minority cytotype exclusion principle. Proc. R. Soc. Lond. B. Biol. Sci. 267: 217–223. Husband, B.C. 2004. The role of triploid hybrids in the evolutionary dynamics of mixed-ploidy populations. Biol. J. Linn. Soc. 82: 537–546. doi:10.1111/j.1095-8312.2004.00339.x. Jankun, A., and Kovanda, M. 1986. Apomixis in Sorbus sudetica (embryological studies in Sorbus 1). Preslia (Prague), 58: 7–19. Jankun, A., and Kovanda, M. 1988. Apomixis at the diploid level in Sorbus eximia (embryological studies in Sorbus 3). Preslia (Prague), 60: 193–213. Jennings, D.L., Craig, D.L., and Topham, P.B. 1967. The role of the male parent in the reproduction of Rubus. Heredity, 22: 43–55. Johnston, J.S., Bennett, M.D., Rayburn, A.L., Galbraith, D.W., and Price, H.J. 1999. Reference standards for determination of DNA content of plant nuclei. Am. J. Bot. 86: 609–613. doi:10.2307/ 2656569. PMID:10330063. Katsiosis, A., Hanneman, R.E., and Forsberg, R.A. 1995. Endosperm balance number and the polar-nuclei activation hypotheses for endosperm development in interspecific crosses of the Solanaceae and Gramineae, respectively. Theor. Appl. Genet. 91: 848–855. Koltunow, A.M., and Grossniklaus, U. 2003. Apomixis: a developmental perspective. Annu. Rev. Plant Biol. 54: 547–574. doi:10. 1146/annurev.arplant.54.110901.160842. PMID:14503003. Levin, D.A. 1975. Minority cytotype exclusion in local plant populations. Taxon, 24: 35–43. doi:10.2307/1218997. Lewis, D. 1943. The incompatibility sieve for producing polyploids. J. Genet. 45: 261–264. Lin, B.-Y. 1984. Ploidy barrier to endosperm development in maize. Genetics, 107: 103–115. PMID:17246209. Longley, A.E. 1924. Cytological studies in the genus Crataegus. Am. J. Bot. 11: 295–317. doi:10.2307/2435389. Love, R., and Feigen, M. 1978. Interspecific hybridization between native and naturalized Crataegus (Rosaceae) in western Oregon. Madroño, 25: 211–217. Macklin, J.A. 2001. Systematics of Crataegus series Coccineae (Rosaceae). Ph.D. thesis, Department of Plant Sciences, University of Western Ontario, London, Ont. Maheshwari, P. 1950. An introduction to the embryology of the angiosperms. McGraw-Hill, New York. Marshall, D.R., and Brown, A.H.D. 1981. The evolution of apomixis. Heredity, 47: 1–15. Mazzucato, A. 1996. Which gene(s) are we looking for? Apomixis Newsletter, 9. Mogie, M. 1992. The evolution of asexual reproduction in plants. Chapman and Hall, London, UK. Muller, H.J. 1932. Some genetic aspects of sex. Am. Nat. 66: 118–138. doi:10.1086/280418. Muniyamma, M., and Phipps, J.B. 1979a. Meiosis and polyploidy in Ontario species of Crataegus in relation to their systematics. Can. J. Genet. Cytol. 21: 231–241. Muniyamma, M., and Phipps, J.B. 1979b. Cytological proof of apomixis in Crataegus (Rosaceae). Am. J. Bot. 66: 149–155. doi:10. 2307/2442517. Muniyamma, M., and Phipps, J.B. 1984a. Studies in Crataegus. XI. Further cytological evidence for the occurrence of apomixis in North American hawthorns. Can. J. Bot. 62: 2316–2324. Muniyamma, M., and Phipps, J.B. 1984b. Studies in Crataegus. X. A note on the occurrence of diplospory in Crataegus dissona Sarg. (Maloideae, Rosaceae). Can. J. Genet. Cytol. 26: 249–252. Muniyamma, M., and Phipps, J.B. 1985. Studies in Crataegus. XII. # 2007 NRC Canada Talent and Dickinson Cytological evidence for sexuality in some diploid and tetraploid species of North American hawthorns. Can. J. Bot. 63: 1319–1324. Nogler, G.A. 1982. How to obtain diploid apomictic Ranunculus auricomus plants not found in the wild state. Bot. Helv. 92: 13–22. Nogler, G.A. 1984. Gametophytic apomixis. In Embryology of angiosperms. Edited by B.M. Johri. Springer, Berlin, Germany. pp. 475–518. Noirot, M. 1993. Allelic ratios and sterility in the agamic complex of the Maximae (Panicoideae): evolutionary role of the residual sexuality. J. Evol. Biol. 6: 95–101. doi:10.1046/j.1420-9101. 1993.6010095.x. Noirot, M., Couvet, D., and Hamon, S. 1997. Main role of selfpollination rate on reproductive allocations in pseudogamous apomicts. Theor. Appl. Genet. 95: 479–483. doi:10.1007/ s001220050586. Palmer, E.J. 1932. The Crataegus problem. J. Arnold Arbor. Harv. Univ. 13: 342–362. Palmer, E.J. 1943. The species problem in Crataegus. Chron. Bot. 7: 373–375. Palmer, E.J. 1956. Crataegus in Ohio with description of one new species. Ohio J. Sci. 56: 205–216. Phipps, J.B. 1988a. Typification of Crataegus crus-galli L. (Rosaceae) and nomenclatural notes on some related species of hawthorn. Bot. J. Linn. Soc. 96: 359–369. Phipps, J.B. 1988b. Crataegus (Maloideae, Rosaceae) of the southeastern United States. I. Introduction and series Aestivales. J. Arnold Arbor. Harv. Univ. 69: 401–431. Phipps, J.B. 2005. A review of hybridization in North American hawthorns — another look at ‘‘the Crataegus problem’’. Ann. Mo. Bot. Gard. 92: 113–126. Phipps, J.B., O’Kennon, R.J., and Lance, R.W. 2003. Hawthorns and medlars. Royal Horticultural Society, Cambridge, UK. Phipps, J.B., Robertson, K.R., Smith, P.G., and Rohrer, J.R. 1990. A checklist of the subfamily Maloideae (Rosaceae). Can. J. Bot. 68: 2209–2269. Ptak, K. 1989. Cyto-embryological investigations on the Polish representatives of the genus Crataegus L. II. Embryology of the triploid species. Acta Biol. Cracov. Ser. Bot. 31: 97–112. Quarin, C.L. 1999. Effect of pollen source and pollen ploidy on endosperm formation and seed set in pseudogamous apomictic Paspalum notatum. Sex. Plant Reprod. 11: 331–335. doi:10. 1007/s004970050160. Ridout, M.S., Xu, X.-M., and Tobutt, K.R. 2005. Single-locus gametophytic incompatibility in autotetraploids. J. Hered. 96: 430–433. doi:10.1093/jhered/esi063. PMID:15829726. Savidan, Y.H. 2000. Apomixis: genetics and breeding. Plant Breed. Rev. 18: 13–86. Smith, P.G., and Phipps, J.B. 1988. Studies in Crataegus (Rosaceae, Maloideae). XIX. Breeding behaviour in Ontario Crataegus series Rotundifoliae. Can. J. Bot. 66: 1914–1923. Solntseva, M.P., Pendinen, G.I., and Chernov, V.E. 1998. Haploids in crosses Hordeum marinum (4x)
H. vulgare. Apomixis Newsletter, 10. Stebbins, G.L. 1941. Apomixis in the angiosperms. Bot. Rev. 7: 507–542. Stebbins, G.L. 1950. Variation and evolution in plants. Columbia University Press, New York. Stebbins, G.L. 1980. Polyploidy in plants: unsolved problems and prospects. In Polyploidy: biological relevance. Edited by W.H. Lewis. Plenum Press, New York. pp. 495–520. Takayama, S., and Isogai, A. 2005. Self incompatibility in plants. Annu. Rev. Plant Biol. 56: 467–489. doi:10.1146/annurev. arplant.56.032604.144249. PMID:15862104. Talent, N. 2006. Gametophytic apomixis, hybridization, and poly- 583 ploidy in Crataegus (Rosaceae). Ph.D. thesis, Department of Botany, University of Toronto, Toronto, Ont. Talent, N., and Dickinson, T.A. 2005. Polyploidy in Crataegus and Mespilus (Rosaceae, Maloideae): evolutionary inferences from flow cytometry of nuclear DNA amounts. Can. J. Bot. 83: 1268–1304. doi:10.1139/b05-088. Talent, N., and Dickinson, T.A. 2007. Endosperm formation in aposporous Crataegus (Rosaceae, Spiraeoideae, tribe Pyreae): parallels to Ranunculaceae and Poaceae. New Phytol. 173: 231–249. doi:10.1111/j.1469-8137.2006.01918.x. PMID:17204072. The R Development Core Team (Editor). 2005. An introduction to R, version 2.2.1. Topham, P.B. 1970. The histology of seed development following crosses between diploid and autotetraploid raspberries (Rubus idaeus L.). Ann. Bot. (Lond.), 34: 137–145. Wells, T.C. 1985. An analysis of interserial hybrid crosses in Crataegus (Rosaceae). M.Sc. thesis, Department of Plant Sciences, University of Western Ontario, London, Ont. Wells, T.C., and Phipps, J.B. 1989. Studies in Crataegus (Rosaceae: Maloideae). XX. Interserial hybridization between Crataegus monogyna (series Oxyacanthae) and Crataegus punctata (series Punctatae) in southern Ontario. Can. J. Bot. 67: 2465–2472. Williams, E.G., and Knox, R.B. 1982. Quantitative analysis of pollen tube growth in Lycopersicon peruvianum. J. Palynol. 18: 65–74. Zar, J.H. 1999. Biostatistical analysis. Prentice Hall, Upper Saddle River, N.J. Appendix A. Seed morphology and triploid hybrids The Crataegus monogyna
C. crus-galli cross that produced seed in this experiment is not known, as far as we are aware, from nature or from horticulture. That the seed coats had an unusual morphology might therefore indicate that the seeds will not germinate (Fig. A1b), unlike the seeds from C. monogyna
C. macracantha that had coats that were smooth like those of the open-pollinated seeds (Fig. A1a). Thus, we suspect that inter-sectional incompatibility can occur even with healthy embryos. Although hybrids between C. monogyna and C. macracantha have not been recorded as such to our knowledge, there are horticultural hybrids known as C.
mordenensis that have a very similar ancestry, which suggests that C. monogyna – C. macracantha hybrids might be as viable as their healthy-looking seed suggests (Fig. A1a). The original description of the C.
mordenensis hybrid (cultivar ‘Toba’; Boom 1959) says in the Latin description that hand-pollinating flowers of C. oxyacantha ‘Paul’s Scarlet’ with pollen from C. succulenta was involved. Phipps et al. (2003) consider the name C. succulenta sensu lato to include C. macracantha, and this group includes diploids, triploids, and tetraploids (Talent and Dickinson 2005). They also update the nomenclature of the female parent of C.
mordenensis to C.
media, the diploid hybrid (Talent and Dickinson 2005) of C. monogyna and C. laevigata. Thus, C.
mordenensis is a hybrid that would be closely related to the triploid C. monogyna
C. macracantha seeds obtained in our experiments. The story of the horticultural experiment that produced C.
mordenensis is rather confused, however. The English summary in Boom’s paper (1959) contradicts both the Latin description cited above and the Dutch comments in the same # 2007 NRC Canada 584 Can. J. Bot. Vol. 85, 2007 Fig. A1. Healthy- and unhealthy-looking triploid seed from different crosses. (a) Although the 14 seeds from C. monogyna (tree NT117) pollinated from C. crus-galli (tree NT35) had healthy-looking triploid embryos, the seed coats had a pale, bubbled surface. (b) On the left (labeled 1): seeds with triploid embryos from C. monogyna (tree NT117) pollinated from C. macracantha (tree NT105; the 12 seeds were extracted in 2003 and photographed October 2004) had normal seed coats. On the right (labeled 2): normal open-pollinated seed with diploid embryos from C. monogyna (tree NT117; collected in 2003 and extracted in October 2004). Scale bar = 2 mm. paper. The English refers to ‘‘crossing flowers of C. succulenta with pollen of C. oxyacantha ‘Paul’s Scarlet’’’. Phipps et al. (2003) also suggest that the C.
media parent of C.
mordenensis was probably the cultivar ‘Rubra plena’ rather than ‘Paul’s Scarlet’. In our experience, C.
media ‘Rubra plena’ has no stamens, but Boom’s English summary has some plausibility because any staminate flowers that did occur would provide an attractive horticultural opportunity. Thus, it is not clear whether the triploid seeds obtained in our experiment are a match for C.
mordenensis. There are two different C.
mordenensis cultivars. The original cultivar ‘Toba’ has very double pink flowers somewhat like those of C.
media ‘Rubra plena’ and a strongly lobed leaf shape intermediate between those of the parents. ‘Snowbird’ was derived from a seed of the nearly seed- sterile ‘Toba’ (Phipps et al. 2003). ‘Snowbird’ has white flowers that are more like those of the wild parent, and a leaf that is more like that of C. macracantha. Flow cytometry showed that C.
mordenensis ‘Snowbird’ is triploid like the seeds obtained from the C. monogyna – C. macracantha pollinations, but ‘Toba’ is diploid — measurements that mirror the number of parental genome copies suggested by the morphology. These hybrids, therefore, might be evidence that Camp (1942) was correct and triploid Crataegus can be subsequently derived from semi-sterile diploid hybrids. Alternative scenarios are suggested by the new results from seeds, however; for example, that ‘Toba’ was the result of hybridization with a diploid C. succulenta and that the paternal parent of ‘Snowbird’ was a nearby white-flowered tetraploid C. succulenta or C. macracantha. # 2007 NRC Canada