Introduction
Offspring traits in plants come from the mother and father, but the mother has more influence on these traits, especially seed dormancy/germination, than the father. Thus, the mother contributes to the traits of her offspring via nuclear and cytoplasmic genetic effects; maternal environmental (including epigenetic) effects; provisioning seeds with nutrient resources, mRNA transcripts, protein, abscisic acid (ABA), which has a positive influence on seed development and can inhibit viviparous germination, and nitrate, which can lower the level of ABA in the seed and act as a signalling molecule in decreasing/overcoming seed dormancy; regulating transfer and distribution of nutrient resources from endosperm to embryo; controlling nutrient storage in and growth of the perisperm, a maternal sporophytic (2n) tissue for embryo growth in seeds of some angiosperms; providing two-thirds of the genetic contribution to the triploid endosperm; and supplying 100% of materials for development of seed coat and diaspore structures external to the seed (Fig. 1). In short, then, the mother plant plays the major role in determining, via diaspore dormancy and dispersal characteristics, maternal architecture (including growth form and height) and seasonal effects (phenology), when and where the seed germinates (Roach and Wulff, Reference Roach and Wulff1987; Donohue, Reference Donohue1999, Reference Donohue2009; Raz et al., Reference Raz, Bergervoet and Koornneef2001; Finkelstein et al., Reference Finkelstein, Gampala and Rock2002; Frey et al., Reference Frey, Godin, Bonnet, Sotta and Marion-Poll2004; Alboresi et al., Reference Alboresi, Gestin, Leydecker, Bedu, Meyer and Truong2005; Kucera et al., Reference Kucera, Cohn and Leubner-Metzger2005; Kanno et al., Reference Kanno, Jikumaru, Hanada, Nambara, Abrams, Kimiya and Seo2010; Matakiadis et al., Reference Matakiadis, Alboresi, Jikumaru, Tatematsu, Pichon, Renou, Kamiya, Nambara and Truong2009; Herman and Sultan, Reference Herman and Sultan2011; Thomson et al., Reference Thomson, Moles, Auld and Kingsford2011; Costa et al., Reference Costa, Yuan, Rouster, Wyatt, Dickinson and Gutierrez-Marcos2012; Jacobs and Lesmeister, Reference Jacobs and Lesmeister2012; Rubio de Casas et al., Reference Rubio de Casas, Willis, Donohue, Clobert, Baguette, Benton and Bullock2012; Tamme et al., Reference Tamme, Götzenberer, Zobel, Bullock, Hooftman, Kaasik and Pärtel2014; Vu et al., Reference Vu, Chang, Moriuchi and Friesen2015; Piskurewicz et al., Reference Piskurewicz, Iwasaki, Susaki, Megies, Kinoshita and Lopez-Molina2016; Yan and Chen, Reference Yan and Chen2017; Augspurger et al., Reference Augspurger, Franson and Cushman2017; Povilus et al., Reference Povilus, Diggle and Friedman2018).
Conceptual model of the various ways in which maternal and paternal plants can contribute to the phenotype of their progeny. The maternal phenotype portion of the diagram is modified from Roach and Wulff (Reference Roach and Wulff1987), and paternal phenotype portion is based on information from various sources. Not shown are the various interactions among the maternal environment (Em), maternal nuclear genotype (Gm), offspring environment (Eo) and offspring nuclear genotype (Go), such as Gm × Em in determining maternal phenotype (Pm) and Gm × Eo, Em × Eo and Gm × Em × Eo in determining offspring phenotype (Po) (e.g. see Rossiter, Reference Rossiter1996). DNA methyl (5′meC), DNA methylation; epig. modif., epigenetic modification; mt, mitochondrion; pl, plastid; PSGE, parent (maternal or paternal) specific gene expression; 2m, 2x (where x = base chromosome number), contribution of maternal parent to endosperm; 1p, 1x, contribution of paternal parent to endosperm. The contributions of Pm to Po via physiology and structure can be found in the Introduction.
Furthermore, evidence is accumulating that the store of active hydrolytic enzymes, anti-microbial substances and nutrient elements in the dry, dead cells of the maternally derived seed coat, pericarp and floral bracts, which can be released upon rehydration, play a role in seed persistence in the seed bank, seed germination and seedling establishment and vigour (Godwin et al., Reference Godwin, Raviv and Grafi2017; Raviv et al., Reference Raviv, Aghajanyan, Granot, Makeover, Frenkel, Gutterman and Grafi2017a,Reference Raviv, Granot, Chilifa-Caspi and Grafib). Finally, in seeds of weedy holoparasitic species of Orobanche the perisperm is involved in reception of the simulant from the host plant that is required for the parasite to germinate (Plakhine et al., Reference Plakhine, Tadmore, Ziadne and Joel2012).
Influence of the father on the traits of his offspring may occur via nuclear (genotype) and non-nuclear genetic material [i.e. mostly plastids transmitted to the progeny by pollen at a low frequency (Corriveau and Coleman, Reference Corriveau and Coleman1988; Azhagiri and Maliga, Reference Azhagiri and Maliga2007)] and paternal environmental (including epigenetic) effects (Schmid and Dolt, Reference Schmid and Dolt1994; Lacey, Reference Lacey1996; Galloway, Reference Galloway2001a,b; Herman and Sultan, Reference Herman and Sultan2011) (Fig. 1). In the gynodioecious species [coexistence of male-sterile (female) and hermaphrodite plants in the same natural interbreeding population] Silene vulgaris and S. acaulis, parental leakage of mitochondrial DNA can result in biparental or strictly paternal inheritance of the mitochondrial genome (Welch et al., Reference Welch, Darnell and McCauley2006; McCauley and Olson, Reference McCauley and Olson2008; Pearl et al., Reference Pearl, Welch and McCauley2009). In most angiosperms, the contribution of the father to the embryo of sexually produced offspring via non-nuclear genetic material is generally minimal compared with that of the mother. In gymnosperms, plastid inheritance is paternal in the conifers, whereas mitochondrial inheritance is paternal in some (taxonomic) families and maternal in others, including the largest family Pinaceae (Mogensen, Reference Mogensen1996).
In their highly-cited review of ‘Maternal effects in plants’, Roach and Wulff (Reference Roach and Wulff1987) included a very short section on paternal effects. However, although the authors refer to several studies on paternal cytoplasmic inheritance, other non-nuclear per se paternal environmental effects on offspring performance are not mentioned. The primary purpose of this paper is to review the literature on the influence of the paternal parent on germination of F1 seeds. We also briefly discuss the influence of the pollen parent on seed size and development.
Influence of paternal parent on seed germination
Multiple paternity in plants, which encompasses the number of pollen donors and the non-random ability of the donors to sire offspring (Snow and Spira, Reference Snow and Spira1991a,Reference Snow and Spirab), is thought to be common and to be caused by (1) sequential visits by several pollinators, each carrying pollen from a different single male; (2) deposition on the stigma of a mixed pollen load by a single pollinator; and (3) deposition on the stigma of a mixed pollen load by several pollinators; and to have fitness consequences (Ellstrand, Reference Ellstrand1984; Marshall and Ellstrand, Reference Marshall and Ellstrand1985, Reference Marshall and Ellstrand1986; Brown et al., Reference Brown, Grant and Pullen1986; Ellstrand and Marshall, Reference Ellstrand and Marshall1986; Marshall, Reference Marshall1988, Reference Marshall1990, Reference Marshall1991; Marshall and Whittaker, Reference Marshall and Whittaker1989; Karron and Marshall, Reference Karron and Marshall1990; Dudash and Ritland, Reference Dudash and Ritland1991; Ibarra-Perez et al., Reference Ibarra-Perez, Ellstrand and Waines1996; Campbell, Reference Campbell1998; Bernasconi, Reference Bernasconi2003; Mitchell et al., Reference Mitchell, Karron, Holmquist and Bell2005; Karron et al., Reference Karron, Mitchell and Bell2006; Teixeira and Bernasconi, Reference Teixeira and Bernasconi2007; Llaurens et al., Reference Llaurens, Castric, Austerlitz and Vekemaus2008). Using microsatellite DNA markers, Reusch (Reference Reusch2000) also demonstrated multiple paternity in Zostera marina, a marine angiosperm with hydrophilous (subaqueous) pollination.
The rank order of one, two, three and five pollen parents for seed germination of the F1 offspring of the tropical herb Costus allenii was three (76.6%) > two (72.9%) > one (67.0%) > five (63.0%) and not statistically significant; standard deviations ranged from 11.5 to 16.9% (Schemske and Paulter, Reference Schemske and Pautler1984). Mean germination percentage across pollen donors was higher (but not significantly so) for seeds of Vaccinium corymbosum flowers pollinated by three pollen donors than for those of flowers pollinated by a single donor (Vander Kloet and Tosh, Reference Vander Kloet and Tosh1984). In a study of single vs mixed donor crosses in Chamaecrista fasciculata, an annual legume with physical dormancy, germination (ca 38–47% for single-donor progeny and ca 34–37% for three-donor progeny) did not differ significantly among selfed, near-crossed and far-crossed seeds (not scarified, overwintered outside) from single-donor and three-donor pollen loads. For all crosses (single- and three-donor pollen donors), the range of relative performance (RP) for inbreeding depression (see Appendix) was narrow, i.e. –0.04 to 0.09 (Sork and Schemske, Reference Sork and Schemske1992). In general, multiple pollinations did not increase germination percentage in seeds of Vaccinium elliotii (Wenslaff and Lyrene, Reference Wenslaff and Lyrene2001).
Seed germination percentage of the rare endemic Cochlearia bavarica did not increase with number of pollen donors, i.e. one, three and nine from the same population and nine from a neighbouring population. However, pollen recipient (maternal parent) had a significant effect on germination percentages (Paschke et al., Reference Paschke, Abs and Schmid2002). In another study, C. bavarica germination percentage did not differ statistically between seeds of one pollen donor hand-cross pollinated plants (47.4%) and those of open-pollinated plants (40.8%), ‘…which likely involves less similar or even several pollen donors’ (Fischer et al., Reference Fischer, Hock and Paschke2003). However, assuming several pollen donors for open-pollinated plants, RP was –0.14 (see Appendix). In which case, seeds from a single-pollen donor outperformed those from multiple-pollen donors (see Appendix).
The results of comparisons by Himes and Wyatt (Reference Himes and Wyatt2005) of self-sterile and self-fertile multiple and (a) single pollen donor(s) on seed germination percentage of Asclepias exaltata were as follows: (1) self-sterile single donor > self-sterile multiple donors; (2) self-sterile single donor = self-fertile single donor; (3) self-sterile single donor = self-fertile multiple donors; (4) self-fertile single donor > self-sterile multiple donors; and (5) self-fertile single donor = self-fertile multiple donors. Thus, germination percentages of seeds sired (fathered) by single donors were equal to or greater than those of seeds sired by multiple donors. In weedy Raphanus sativus, pattern of seed germination (days to emergence of each seedling) differed (P < 0.0001) among maternal plants but not among the three pollen donors; altogether 97.1% of 450 seeds planted germinated (Marshall et al., Reference Marshall, Reynolds, Abrahamson, Simpson, Barnes, Medeiros, Walsh, Oliveras and Avritt2007). However, pollen donor did have a significant effect on some aspects of fitness, i.e. growth and reproduction (Marshall et al., Reference Marshall, Reynolds, Abrahamson, Simpson, Barnes, Medeiros, Walsh, Oliveras and Avritt2007; see also Marshall and Whittaker, Reference Marshall and Whittaker1989; Karron and Marshall, Reference Karron and Marshall1990). Maternal, but not paternal, parent had a significant effect on germination percentage of the wind-pollinated tree Betula pendula (Pasonen et al., Reference Pasonen, Pulkkinen and Käpylä2001).
In a study of the effect of pollen load size and donor diversity on Mirabilis jalapa, days to seedling emergence did not differ significantly among five pollination treatments: (1) large load/multiple donors; (2) large load/single donor; (3) small load/multiple donors; (4) small load/single donor; and (5) single outcross pollen grain. Neither was there a maternal effect on days to emergence (Niesenbaum, Reference Niesenbaum1999). Thus, there was no effect of donor diversity on days to seedling emergence. Also, there was no effect of donor diversity on seed mass. The number of donors did not have a significant effect on proportion of seeds of Collinsia heterophylla that germinated. However, proportion of seeds that germinated was significantly affected by pollen recipient × paternity diversity, showing that seeds of some maternal plants germinated better when sired by multiple donors (Lankinen and Madjidian, Reference Lankinen and Madjidian2011). Neither germination percentage of seeds of the weedy annual mustard Raphanus raphanistrum (Snow, Reference Snow1990) nor those of the bignoniaceous woody vine Campsis radicans (Bertin, Reference Bertin1986) differed significantly between progeny of single- and multiple-donor pollinations. Furthermore, neither germination nor number or mass of seeds from fruits sired by one, three and five pollen donors differed significantly in C. radicans (Bertin, Reference Bertin1986). However, the best pollen donors of this species gave rise to statistically more and heavier seeds than the donor mixtures, and although percentage germination was higher for seeds sired by the best donors (46 vs 41%) the difference was not significant.
Seeds of Swertia perennis from open-pollinated (i.e. natural, not supplemented by hand pollination, multiple-pollen donors assumed) flowers germinated to 40.83%, whereas those from flowers whose stigmas were saturated with hand-crossed pollen from one donor germinated to 32.24% (P < 0.05) (Lienert and Fischer, Reference Lienert and Fischer2004). Thus, the RP was 0.21, showing that seeds from multiple-pollen donors outperformed those from a single-donor (see Appendix). The authors suggested that the difference might be due to pollen quality, i.e. lower pollen quality representing the single donor than that of open pollen, which they suggested was ‘… most likely from several donors’. In a pollen competition experiment with Dalechampia scandens, there was no evidence for paternal (or maternal) effects on seed germination. Furthermore, paternal effects on seed maturation time, seed mass and seedling vigour at 1 month were limited and not statistically significant (Pélabon et al., Reference Pélabon, Hennet, Bolstad, Albertsen, Opedal, Ekrem and Armbruster2016).
Emergence (germination) of all seedlings of Asclepias speciosa from crosses by three donors in a pollen competition experiment (84.8%) was significantly higher than that of seedlings in the single donor experiment (74.8%) (Bookman, Reference Bookman1984). The author stated that, ‘Donors which are superior competitors, therefore, father more seedlings with a higher percentage of emergence than seedlings fathered by all donors’. However, when other results of her study also were considered, Bookman suggested that the differences in seedling emergence were not due to higher pollen vigour of the competitors that fathered the seedlings with the highest emergence percentages but to sperm quality or fertilizing ability.
In Crepis tectorum subsp. pumila, an alvar species on the Baltic island of Öland (SE Sweden), pollen donor had a significant (P < 0.05) effect on seed mass and width and a marginally significant (P = 0.053) effect on seed length, and seeds that germinated had a significantly greater mass, length and width than those that did not germinate (Andersson, Reference Andersson1990). Thus, the author concluded that genetic variance among the pollen donors was responsible for these differences. However, Mazer and Gorchov (Reference Mazer and Gorchov1996) questioned this claim. They suggested that maternal effects, extranuclear genes, environmentally induced gene expression acting differently among the pollen donors or gametophytic selection within different pollen donors before pollen maturation could rule out additive genetic variation as being responsible for these differences. Recently, Marshall and Evans (Reference Marshall and Evans2016) presented strong evidence that the ability of pollen donor families of Raphanus sativus to sire seeds in mixed pollinations under competition (i.e. with other pollen donors) is heritable and can respond to selection.
Pollen donor had a significant effect on seed germination rate (speed) but not on germination success (%) in Iris hexagona. The variation in germination rate of seeds of this species among pollen donors indicated ‘… a potential for a paternal contribution to seed quality’. Pollen donor × parental population was significant for both germination rate and germination success (Van Zandt and Mopper, Reference Van Zandt and Mopper2004). In a study of cryptic self-fertility in Campsis radicans by Bertin et al. (Reference Bertin, Barnes and Guttmann1989), days to germination of seeds sired by (1) pollen recipient/pollen donor 12 (maternal and paternal parent selfed) and pollen donor 1 (outcross paternal parent), and (2) pollen recipient/pollen donor 12 and pollen donor 3A were: self – 24.2, cross – 25.3; and self – 36.6, cross – 33.7, respectively. Self and cross seeds sired by 12/1 did not differ significantly in days to germination, whereas self seeds sired by 12/3A germinated in significantly fewer days than outcrossed seeds. Bertin (Reference Bertin1990) tested the effect of different pollen ratios of two donors (paternal parents) on germination percentage of seeds produced on two pollen recipients (maternal parents) of C. radicans. For pollen recipient 8, germination percentage of seeds from four pollen combination ratios (amount) of two pollen donors were [(2:3A) = (2:1)] < [(1:10) = (10.3)]. For pollen recipient 10, germination percentage of seeds from two pollen combination ratios were (2:3) < (2:1). Thus, not only pollen donor identity per se but also the ratio of the amount of pollen from different donors in a multiple-donor pollen load may affect seed germination.
In nature, plants can be limited in the number of seeds produced due to low pollinator activity. Experimentally, pollen limitation is shown to occur when plants that are open-pollinated plus hand-cross pollen supplemented (Ps, treatment) produce more seeds than those that are open-pollinated only (Po, control), i.e. no pollen supplementation (Burd, Reference Burd1994; Larson and Barrett, Reference Larson and Barrett2000; Ashman et al., Reference Ashman, Knight, Steets, Amarasekare, Burd, Campbell, Dudash, Johnston, Mazer, Mitchell, Morgan and Wilson2004; Knight et al., Reference Knight, Steets, Vamosi, Mazer, Burd, Campbell, Dudash, Johnston, Mitchell and Ashman2005). Only a small percentage of the numerous studies on pollen limitation have measured the germination responses of seeds from treatment vs control, and some of them showed that pollen supplementation increased seed germination percentage. Of 30 cases (18 species) reviewed by Baskin and Baskin (Reference Baskin and Baskin2018) that compared germination of seeds produced by Ps and Po plants, Ps > Po in 12, Ps = Po in 11 and Ps < Po in 7. Thus, based on relative performance (see Appendix), in 40% of the cases pollen supplementation enhanced germination percentages/rate (speed). In these 12 cases, pollen supplementation may have been equivalent to an increase in number/quality of pollen donors.
For seeds of Eschscholzia californica, dormancy in interpopulation crosses ‘… seems to have been maternally inherited, although in a few cases (828-4, 873-2) the pollen parent may have had some effect’ (Cook, Reference Cook1962). In a study of the quantitative genetics of life history and fitness components of Raphanus raphanistrum, Mazer (Reference Mazer1987a) found no effect of pollen parent on seed germination date or on seed mass. Twelve of 15 paternal genotypes of Raphanus sativus exhibited higher rates of germination (1/days to germinate) at high than at low or medium planting densities (Mazer and Schick, Reference Mazer and Schick1991a). In another study by these authors on R. sativus (Mazer and Schick, Reference Mazer and Schick1991b), paternal genotype had a significant effect on germination rate in medium planting density plots, but not in low and high planting density plots.
A study of the dark germination of reciprocal hybrid 6-mo afterripened seeds from light-requiring (R) and indifferent (I) tobacco (Nicotiana tobacum) selections (Kasperbauer,1968) showed that both parents contributed to light sensitivity of the seeds, but the contribution of the seed parent was greater [or slightly greater sensu Karssen et al. (Reference Karssen, Brinkhorst-van der Swan, Breekland and Koornneef1983)] than that of the pollen parent, i.e. germination percentage of R [seed (maternal) parent] × I [pollen (paternal) parent] in darkness was < (I × R). Overall, (R × R) < (R × I) < (I × R) < (I × I) (uninterrupted darkness) and (R × R) = (R × I) = (I × R) = (I × I) (illuminated). Interestingly, in a much earlier study than that of Kasperbauer, Honing (Reference Honing1930) reported that the light requirement for germination of tobacco seeds was influenced by both parents, with the maternal parent predominating. The results of a diallel cross between five sugar beet (Beta vulgaris) plants showed that genotype of the maternal parent controlled germination to a large extent (Battle and Whittington, Reference Battle and Whittington1971). Reciprocal hybrid seeds of lettuce (Lactuca sativa) showed paternal control over seed dormancy. Genotype MQS was more dormant than 466, and 466 (seed parent) × MQS (pollen parent) was more dormant than MQS × 466 (Rideau et al., Reference Rideau, Monin, Dommergues and Cornu1976). Seed dormancy in the annual weedy mustard Sinapis arvensis had both a maternal and an embryo component, but there was an ‘… overriding importance of the maternal genotype in seed dormancy …’ (Garbutt and Witcombe, Reference Garbutt and Witcombe1986). Reciprocal crosses of the most dormant (Th7) and least dormant (M30) lines of Petunia hybrida showed paternal control over seed dormancy (Girard, Reference Girard1990). The order of seed dormancy of parents and reciprocal hybrids was Th7 > (M30 × Th7) > (Th7 × M30) > M30 for primary dormancy and for the ability of the seeds to enter secondary dormancy. Thus, dormancy is predominantly under paternal control.
In a diallel cross using five parental plants (A→E) of Lupinus texensis, maternal and paternal effects on germination percentage were similar in A, C, D and E. For parental plant B, however, seeds of the paternal outcross geminated to a considerably higher percentage than those of the maternal outcross and self, which were similar. Both maternal and paternal parents and their interaction significantly affected seed mass, and seeds that germinated had significantly more mass than those that did not germinate (Helenurm and Schaal, Reference Helenurm and Schaal1996). The paternal parent had significant effects on germination time of the progeny in two of three sets of reciprocal diallel crosses in the monocarpic perennial species Aster kantoensis, whose seeds are non-dormant at maturity (Kagawa et al., Reference Kagawa, Tani and Kachi2011).
For the grass Anthoxanthum odoratum, paternal genotype did not have an overall effect on germination percentage, but in two of six randomized blocks in the experimental design it did have a significant effect on germination percentage. Also, germination percentage differed significantly for seeds sired by different fathers in four of the six blocks for maternal genotype D2 (Schmitt and Antonovics, Reference Schmitt and Antonovics1986). For Lychnis flos-cuculi, Biere (Reference Biere1991a,b) found significant differences for germination among paternal offspring within families of female progeny and significant variation in time to emergence of progeny of maternal and of paternal parents with different genotypes. Male parents had a significant influence on seed germination in within-population crosses of Lobelia cardinalis (Schlichting and Devlin, Reference Schlichting and Devlin1992). Germination of seeds from crosses between the weedy Silene vulgaris and the narrow Swedish alvar endemic S. uniflora var. petraea was strictly determined by the pollen parent. Thus, hybrid seeds germinated to the same percentage and rate as outcrossed (non-hybrid) seeds of the paternal species rather than to a percentage and rate intermediate between those of the parental species (Andersson et al., Reference Andersson, Månsby and Prentice2008). The authors hypothesized ‘… that the germination behavior of Silene seeds is affected by nonnuclear (cytoplasmic) factors inherited from the male parent or that nuclear genes from the maternal parent are “silenced” during germination’.
Days to emergence in seeds of Campanula americana did not differ between high and low pollen loads, but there were significant maternal and paternal effects on this trait. Seed mass had no effect on days to emerge (Richardson and Stephenson, Reference Richardson and Stephenson1992). Pollen parent had a significant effect on germination percentage in Purshia tridentata seed progeny in response to 2 weeks of chilling (cold stratification). Among-year variation in maturation environment was not significant, indicating no paternal environmental effect (Meyer and Pendleton, Reference Meyer and Pendleton2000). Paternal genotype of Eucalyptus globulus had a significant effect on six measures of proportion and rate (speed) of seed germination (Rix et al., Reference Rix, Gracie, Potts, Brown, Spurr and Gore2012). In red (R, Morus rubra) and white (W, M. alba) mulberry, progeny of W mothers had the highest cumulative fitness (Burgess and Husband, Reference Burgess and Husband2004). Offspring (seeds) of W mothers germinated to a significantly higher percentage than those of R and hybrid (R × W) mothers, which did not differ. However, the paternal parent did not have an effect on cumulative fitness or germination, and paternal × maternal interactions were not significant. Thus, the authors attributed the strong influence of the mother and lack of influence of the father on fitness to non-nuclear (maternal) effects.
Sire (paternal parent) had a significant effect on percentage and time to germinate and on dormancy in seeds of Nemophila menziesii (Platenkamp and Shaw, Reference Platenkamp and Shaw1993). However, in a follow-up study on this species, Byers et al. (Reference Byers, Platenkamp and Shaw1997) found that the paternal effect on time to germinate (and on seed mass) was weak and inconsistent. In both studies, the dam (maternal parent) had a greater effect on time to germinate than the sire. In Brassica campestris, percentage germination of F1 (high pollen load HF1) was greater than that of F1 (low pollen, LF1), but there were no second generation effects, i.e. no difference in percentage germination of HF1 and LF1 progeny using the same amount of pollen. However, there was a paternal influence on plant fitness, including germination (Palmer and Zimmerman, Reference Palmer and Zimmerman1994). Lassere et al. (Reference Lassere, Carroll and Mulcahy1996) found no effect of pollen competition on seedling emergence time in Silene latifolia. However, paternal parents had a significant effect on days required for seedling emergence.
In Campanula rapunculoides, strength of self-incompatibility (weak, intermediate, strong) of neither the father nor the mother had a significant effect on seed germination (Good-Avila and Stephenson, Reference Good-Avila and Stephenson2003). Germination percentage of seeds of Silene latifolia was strongly influenced by the population of origin of the female parent but not by that of the male parent (Jolivet and Bernasconi, Reference Jolivet and Bernasconi2007). For S. latifolia, number of pollen donors (one vs two) had no effect on percentage or rate (speed) of germination, whereas paternal family had a significant effect on both measures of germination (Teixeira et al., Reference Teixeira, Foerster and Bernasconi2009). Many studies have shown the effects of the environment of the maternal plant during seed formation/maturation on germination, but very little attention has been given to paternal environmental effects, which can be prezygotic only (Fig. 1). However, although paternal environmental effects are considered to be minimal they do occur. Raphanus raphanistrum seeds sired by pollen of low- vs high-nutrient stressed plants did not differ in number of days to germination (Young and Stanton, Reference Young and Stanton1990). In a growth chamber experiment on Plantago lanceolata, paternal (prezygotic) temperature influenced seed germination more than maternal (prezygotic or postzygotic) temperature (Lacey, Reference Lacey1996). However, in a follow-up combined growth chamber-field experiment on this species there was essentially no prezygotic paternal or maternal temperature effect on germination, whereas postzygotic (maternal) temperature strongly influenced germination (Lacey and Herr, Reference Lacey and Herr2000). Seeds of male parents of Solidago altissima grown in soil germinated faster than seeds of male parents grown in sand (Schmid and Dolt, Reference Schmid and Dolt1994). The authors suggested that this positive effect on germination, as well as that on post-germination growth, was probably due to differences in pollen quality than in pollen quantity. In studies by Galloway (Reference Galloway2001a,b) and Etterson and Galloway (Reference Etterson and Galloway2002), the (prezygotic) light environment of the pollen parent had significant effects on seed germination in the winter annual/strict biennial Campanula americana. However, the expression of paternal effects on germination percentage and rate, and also on seed mass, depended on the maternal light (high, medium, low) environment. In animals, environmentally (level of competition) induced (adaptive) paternal effects have been unequivocally demonstrated in the broadcast-spawning marine invertebrate the solitary ascidian Styela plicata (Crean et al., Reference Crean, Dwyer and Marshall2013).
Androdioecy refers to the coexistence of males (female-sterile, thus produce only pollen) and hermaphrodites (produce both pollen and ovules/seeds) in a breeding population (Pannell, Reference Pannell2000). We have found only three cases in which germination percentages of male-sired seeds (MS) were compared with those of (outcross) hermaphrodite-sired seeds (HS). In Datisca glomerata MS = HS (Riesberg et al., Reference Riesberg, Philbrick, Pack, Hanson and Fritsch1993), Fraxinus ornus MS = HS (Verdu et al., Reference Verdu, Montilla and Pannell2004) and Laguncularia racemosa Ms > HS (Landry and Rathcke, Reference Landry and Rathcke2007). In the trioecious (population consisting of males, females and hermaphrodites) cactus Pachycereus pringlei, germination percentage of seeds of females pollinated by males was higher than that of seeds in the other pollination treatments, i.e. seeds of females × males germinated to higher percentages, but not significantly (Sosa and Fleming, Reference Sosa and Fleming1999).
To summarize, in 37 of the 59 studies (62.7%) discussed above on the influence of the paternal parent on seed germination, we conclude that the father had a positive influence on germination, whereas in the other 22 studies it did not (Table 1, see footnote to table).
Positive (+) or no positive (0) influence of the paternal parent on seed germination [percentage and/or rate (speed)] of F1 offspring (seeds) in 59 studies on 45 species of angiosperms [3 monocots (M) and 42 eudicots] in 38 genera and 27 families
*Several of the studies (e.g. Schmitt and Antonovics, Reference Schmitt and Antonovics1986; Bertin et al., Reference Bertin, Barnes and Guttmann1989; Helernum and Schaal, Reference Helenurm and Schaal1996) had more than one outcome for a taxon, i.e. positive influence versus no positive influence of paternal parent on seed germination. In those cases, even if only one of the outcomes was positive we show a positive influence for the paternal parent.
Influence of paternal parent on seed size and development
As with seed dormancy/germination, most studies on the parental effects on seed size have shown that the paternal influence on seed development and seed mass is small or non-existent, and when there is a paternal effect on seed size it is usually considerably smaller than that of the maternal parent (e.g. Antonovics and Schmitt, Reference Antonovics and Schmitt1986; Marshall and Ellstrand, Reference Marshall and Ellstrand1986; Mazer et al., Reference Mazer, Snow and Stanton1986; Mazer, Reference Mazer1987b; Marshall, Reference Marshall1988, Reference Marshall1991; Nakamura and Stanton, Reference Nakamura and Stanton1989; Pittman and Levin, Reference Pittman and Levin1989; Fenster, Reference Fenster1991; Richardson and Stephenson, Reference Richardson and Stephenson1991; Lyons, Reference Lyons1996; Lacey et al., Reference Lacey, Smith and Case1997; Shaw and Byers, Reference Shaw, Byers, Mousseau and Fox1998; de Jong and Scott, Reference de Jong and Scott2007; Holland et al., Reference Holland, Chamberlain, Waguespack and Kinyo2009; Diggle et al., Reference Diggle, Abrahamson, Baker, Barnes, Koontz, Lay, Medeiros, Murgel, Shaner, Simpson, Wu and Marshall2010; de Jong et al., Reference de Jong, Hermans and van der Veen-van Wijk2011; Li et al., Reference Li, Peng, Shi, Wang, Liu and Wang2015; Pélabon et al., Reference Pélabon, Hennet, Bolstad, Albertsen, Opedal, Ekrem and Armbruster2016). However, although maternal genotype explained 29.3% of the variation in seed size in reciprocal crosses among four accessions of Arabidopsis thaliana, the paternal genotype explained ‘… a substantial proportion of the variation (10.4%)’ (House et al., Reference House, Roth, Hunt and Kover2010). A recent study by Pires et al. (Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016) showed that paternal effects on the control of seed development in Arabidopsis thaliana exists but are buffered by the maternal genome, i.e. genomic imprinting, an epigenetic mechanism for the parent-of-origin-specific [female or male (monoallelic)] expression of alleles that, in plants, occurs mostly in the endosperm of the developing seed (Vinkenoog et al., Reference Vinkenoog, Bushell, Spielman, Adams, Dickinson and Scott2003; Köhler et al., Reference Köhler, Wolff and Spillane2012; Gehring, Reference Gehring2013; Pires, Reference Pires2014; Rodrigues and Zilberman, Reference Rodrigues and Zilberman2015; Pires et al., Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016).
More specifically, MEDEA (MEA) is an imprinted maternally expressed gene essential for normal seed development. Thus, seeds of Arabidopsis thaliana that maternally inherit a loss-of-function mea allele exhibit excessive cell proliferation and abort when the pollen parent is accession Ler (Grossniklaus et al., Reference Grossniklaus, Vielle-Calzada, Hoeppner and Gagliano1998; Pires et al., Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016). Pires et al. (Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016) have shown that when mea ovules are pollinated by some other accessions of A. thaliana the paternal effect is released and the seeds develop normally, i.e. mea seed abortion can be paternally suppressed. They suggest that MEA acts as a maternal buffer against the paternal genome on seed development and conclude that this provides support of the (intragenomic) parental conflict theory in angiosperms whose seeds are sired by multiple donors.
According to the parental conflict theory (Haig and Westoby, Reference Haig and Westoby1989, Reference Haig and Westoby1991; Moore and Haig, Reference Moore and Haig1991) [the predominant theory of several (theories) on genomic imprinting, for example see Rodrigues and Zilberman, Reference Rodrigues and Zilberman2015], the maternal parent has equal interest in all of her sib and half-sib progeny, and thus it would be to her best interest to devote equal resources to all of them (maternal family). On the other hand, for plants that receive pollen from more than one father, it would be in the best interest of each of the multiple paternal parents, who do not father all sibs and half-sibs of the maternal family, to get preferential treatment (i.e. relatively more resources) for the progeny he sires, i.e. to have more than an equal share of the resources devoted to his offspring. Thus, there is conflict between maternal and paternal genes within offspring over how the resources from maternal tissue via the endosperm are devoted to the developing embryos. The results of a study on seeds of Arabidopsis thaliana by Scott et al. (Reference Scott, Speilman, Bailey and Dickinson1998) provide support for the parental conflict theory. Thus, seeds with a double dose of paternal genes [2x × 4x (1m : 2p endosperm) produce large endosperms and embryos, while maternal plants with a double dose of genes [4x × 2x (2m : 1p endosperm)] produce small endosperms and embryos. According to Pires et al. (Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016), MEA activity buffers the paternal effects from maximizing growth ‘… suggesting that they were likely shaped by parental conflict’. In sum, then, the study by Pires et al. (Reference Pires, Bemer, Müller, Baroux, Spillane and Grossniklaus2016) shows that paternal effects on seed development do exist but are buffered by the maternal genome.
General conclusion
Although undoubtedly the mother has much more influence than the father on dormancy and germination (and development and size-mass) of offspring seeds produced, the pollen parent sometimes can have an effect on variation in these life history traits. However, in general, the influence of the father on these stages of the plant life cycle is minimal compared with that of the mother.
Author ORCIDs
Carol C. Baskin 0000-0001-7680-154X
Appendix. Relative performance
A meaningful way to compare the effect of one-pollen donor versus multiple-pollen donors on seed germination is to use a measure of relative performance (RP):
$${\rm RP} = \lpar {W_{\rm m}- W_{\rm s}} \rpar /W_{{\rm max}}\lsqb {W_{\rm m},W_{\rm s}} \rsqb ,$$where W m is mean germination percentage of seeds (progeny) resulting from multiple-donor pollinations, and W s is the mean germination percentage of seeds resulting from a single-donor pollination.
Using this equation, values will be equidistant from 0, i.e. they are bound between –1.0 and + 1.0, when W s > W m (to –1) and W s ≤ W m (to + 1). Thus, a positive value indicates that seeds from multiple-donor pollinations germinated better than those from a single-donor pollination, and the closer the value is to 1.0 the greater the difference in performance. A negative value indicates that seeds from a single-donor pollination germinated better than those from multiple-donor pollinations, and the more negative the value the greater the single-donor benefit.
Relevant to the study by Sork and Schemske (Reference Sork and Schemske1992) discussed in the present paper, the above equation can be used to calculate inbreeding depression of seed germination by substituting W o (germination of outcrossed seeds) for W m and letting W s be germination of selfed seeds instead of germination of a single-donor pollination.
