Pollen : ovule ratio and its relationship with other reproductive traits in some Passiflora species ( Passifloraceae )

Amela García, M.T., Miguez, M.B. & Gottsberger, G. 2014. Pollen: ovule ratio and its relationship with other reproductive traits in some Passiflora species (Passifloraceae). Anales Jard. Bot. Madrid 71(2): e009. The pollen/ovule (P/O) ratio has been used as predictor of the reproductive system of angiosperms (lowest P/O values correspond to obligate autogamous species while the highest correlate with obligate xenogamous species) but it does not keep for all the taxa. The relation of P/O index with pollen and stigma size, flower diameter, pollen vitality, stigmatic area/pollen-bearing area of the pollinators ratio (SA/PBA), taxonomy and breeding system was analyzed in four Passiflora species with different degree of compatibility, pollination system and taxonomic placement. P/O of the self-compatible species (with shorter anthesis, scarce pollinator visits or minor longevity) was lower than P/O of the selfincompatible ones. The P/O values of the outcrossers could be related with the highly efficient pollination performed by the frequently visiting Xylocopa species; these bees transport Passiflora pollen in a usually monospecific mass on the thorax, resembling grouped pollen dispersal. All the taxa exhibited pollen with unimodal size and high vitality, except for P. misera, in which pollen size was bimodal, with similar quantities of grains in both classes, the large being more vital than the small ones. Correlations of P/O were positive with flower size, negative with stigma area/pollen-bearing area of the pollinator, null with stigma area and not significant with pollen size. The P/O ratio did not reflect taxonomic affinities; this index and the breeding system were more related with pollinator type, anthesis, flower size and SA/PBA than with pollen grain or stigma size.

In most cases, P/O ratio falls within the ranges calculated by Cruden (1977) or keeping the tendency (P/O of autogamous species lower than P/O of xenogamous species) (Schoen, 1977;Spira, 1980;Wyatt, 1984;Preston, 1986;Graumann and Gottsberger, 1988;Ortega Olivencia and Devesa Alcarez, 1993;Weller & al., 1998;Christ & al., 2001;Lehnebach and Riveros, 2003;Medán, 2003), but some exhibit higher (Choteau & al., 2006;Devoto and Medán, 2008;Hokche and Ramírez, 2008) or lower (Wild & al., 2003;Galloni & al., 2007) values than expected.Rather than breeding system, in several taxa P/O ratio reflects better pollination mechanism (Galloni & al., 2007) or pollination efficiency (Choteau & al., 2006).P/O ratio was significantly associated with pollination mechanism in species with unknown breeding system (López & al., 1999) and in 21 xenogamous legumes (Etcheverry & al., 2012).Lower P/O ratios usually coincide with precise or highly specialized pollen transfer mechanisms (Small, 1988;Jürgens & al., 2002;Choteau & al., 2006).This has been exemplified in the Mimosoideae (Cruden, 1977).The extreme is represented in Asclepiadoidea (Cruden, 1977;Wyatt & al., 2000).Lack of adjustment to expected values of P/O were also discussed in relation to pollination ecology (Affre and Thompson, 1998).Cruden (2000) admitted that this index might be confusing, especially for xenogamous species with low P/O ratio, many of which have large pollen grains and/or large stigmas, based on the relation between P/O ratio, pollen size and the ratio of stigma area to the pollen-bearing area of the pollinators (SA/PBA) (Cruden and Miller-Ward, 1981).The stigma area is expected to be large enough to receive sufficient pollen grains to fertilize a large number of ovules (Cruden and Miller-Ward, 1981); the SA/PBA was used as a measure of the likelihood of a pollen grain reaching a stigma, and it was hypothesized that P/O would be inversely related to SA/PBA (Cruden and Miller-Ward, 1981).
P/O ratio has not been examined in Passiflora L. yet.Preliminary records in species with known breeding system (Amela García, 1999) suggest that this index is lower than expected by the predictions of Cruden (1977).The aim of this work was to explore the relationship between P/O ratio, pollen and stigma size, flower size, pollen vitality, pollen-bearing area and pollen transfer efficiency in Passiflora species with different degree of self-compatibility and from different infrageneric categories, to know how these floral traits interrelate and corroborate if the P/O ratio predicts the breeding system or reflects taxonomic affinities in Passiflora.
Buds ready to open were collected and fixed in ethanol 70° for pollen and ovule counts, pollen and stigma size measurements, whereas other buds were dried between paper sheets for pollen vitality tests.

Pollen and ovule counts
For pollen grain number estimation, anthers were opened under a dissecting microscope and all the pollen was transferred to plastic tubes containing 5 ml of 0.9% NaCl solution.Pollen grains were counted with a cell counter and analyzer system CASY (Schärfe System, Reutlingen, Germany).The number obtained was multiplied by the number of anthers of the bud.The ovules were counted dissecting the ovary under a stereoscopic microscope.The P/O index (ratio between pollen grains and ovules of a bud) defined by Cruden (1977) was calculated, the values obtained were compared with the ranges given by that author and with the breeding system, which had been determined with controlled pollinations (Amela García, 1999).

Pollen size and vitality
The pollen grain size was measured with the automatic cell counter and analyzer system mentioned above.This device gives the diameter of the figure according to the position it is, so it is somewhat like an average between the equatorial and polar axis in these subespheroidal grains (Amela García & al., 2002).Pollen vitality was estimated by optical contrastfluorescence microscopy (Greissl, 1989) and by staining the main pollen constituents, lipids in all of them (Amela García, 1999), with Sudan IV (Kerns and Inouye, 1993).

Stigma area and pollen-bearing area of the pollinators
Passiflora species have capitate stigmas, whose most resembling geometric form is an ellipse, so the longer and the shorter axis were measured with an ocular micrometer of a dissecting microscope, converted to mm and then the ellipse area was calculated.The deposition site of the pollen of Passiflora species on their pollinators is the thorax dorsum, which can be resembled to a rectangle, so the length and the width were measured and then the rectangle area was calculated.Dimensions were taken at naked eye with a ruler made of graph paper, on specimens deposited at the Museo Argentino de Ciencias Naturales "B.Rivadavia".Pollinators were chosen based on previous works (Amela García, 1999;Amela García and Hoc, 1997, 1998a, 2001).Specimens of the pollinators of P. suberosa could not be measured as they were not available for us.Afterwards, the stigma area/pollen bearing area was calculated for each species.

Flower size
The major diameter (distance between the apexes of opposite sepals) was measured on fresh material with graph paper at naked eye in these actinomorphic flowers.

Statistical analysis
The data were subjected to ANOVA and correlation analysis using the statistical program InfoStat 9.0 (2009).

P/O ratio and breeding system
P/O ratios differed significantly between the selfincompatible and the self-compatible species: F=50.66, p<0.0001, n=89 (Table 1); P/O of self-incompatible species were higher (3 folds) than those of the self-compatible ones.

Pollen grain size and vitality
The pollen grain size was rather constant within each species, considering the little standard deviation (Table 2) and differed significantly between some species (F=308.11,p<0.0001, n=94) (Table 2); P. misera exhibited greater variation, in this species a tendency to a bimodal distribution was detected (Fig. 1).Pollen vitality was high in all species with both tested staining methods (Table 3); in P. misera, vitality was notoriously higher in the bigger grains (Table 3).

P/O, pollen grain size, flower size, stigma area and SA/PBA
The correlation between P/O and pollen grain size was not significant (Fig. 2A).

P/O, breeding system, pollinators, anthesis and plant longevity
The lower P/O ratios of P. foetida and P. suberosa as compared to the ones of P. caerulea and P. misera are in accordance with the self-compatibility of the former and the self-incompatibility of the latter (Amela García, 1999), and the tendency coincides with the one discovered by Cruden (1977).Nevertheless, P/O values were lower than expected by the predictions of Cruden (1977) for all the species, irrespective of their breeding system; however, P/O of selfcompatible species resulted more closely to Cruden's values, but the ones of xenogamous species were substantially lower than those proposed by him.Based on the ranges given for xenogamous taxa, Cruden and Miller-Ward (1981) argued that species with very efficient pollen transfer mechanisms had lower P/O ratios than expected as compared to species with equivalent breeding systems but less efficient pollen flow.The specialization between bees and flowers results in a more efficient pollination with less pollen loss (Pijl, 1954).Probably the low P/O ratio in P. caerulea and P. misera is related with the efficient pollen transfer performed by Xylocopa species (Amela García and Hoc, 1997 and Amela García unpublished.data, respectively), becoming unnecessary a major pollen production.Their reproductive system could be assimilated to the one of "sweepstakes" (Cruden, 1977), as, although pollen is not clustered by abundant pollenkit or viscin threads nor grouped in pollinia, the transport in mass on the pollinators thorax, comprising the majority or even the whole pollen load in this body part, would be comparable with those agglutinations.Every time a pollinator visits a flower it takes a considerable amount of grains Values with different letters differ significantly at p<0.01.Numbers within each cell represent the average±standard deviation and n=number of anthers or flowers (number of plants).
from the anthers (usually getting most of the notothorax covered with pollen), and leaves a similar quantity of pollen through its wide contact with the stigmas when they are deflected.Cruden and Miller-Ward (1981) found that P/O was negatively correlated with the pollen-bearing area of the pollinators and with pollen grain size.An additional coincidence with these authors' observations is that most of the xenogamous species with the lower P/O ratio have larger pollen grains: P. caerulea (this work) and P. mooreana (Amela García and Hoc, 1998b).
Still, P/O ratios of self-compatible species were a bit lower than expectations of Cruden (1977).An inverse relation between P/O ratio and pollinator reliability was anticipated by Cruden and Miller-Ward (1981) but low P/O ratio of a self-compatible orchid also coincided with low pollinator visitation rate (Lehnebach and Riveros, 2003).Pollinator constancy of P. foetida is high (Amela García and Hoc, 1998a).In contrast, scarce frequency of pollinators to P. suberosa might be compensated by its longer anthesis and its higher degree of self-compatibility (Amela García, 2008) as compared to P. foetida (Amela García and Hoc, 1997).In all the Passiflora species studied so far, stigma receptivity lasts the whole anthesis (Amela García, 1999), so anthesis   duration is a direct indicator of receptivity in them.A positive relation between P/O ratio, degree of outcrossing and flower life span occurred between different populations analyzed by Wyatt (1984) and species analyzed by Choteau & al (2006).A tendency in these traits may be seen in the Passiflora species analyzed: anthesis duration of P. foetida<P.suberosa similar to P. misera<P.caerulea.Among the vines studied, there is also a trend between degree of self-compatibility, P/O and longevity.The selfcompatible species, with lower P/O ratio, are annuals (the whole plant or the aerial stems) or biannuals (except from P. suberosa, that lives longer) whilst the self-incompatible ones, with higher P/O ratio, render successfully more years.This tendency, discovered by Cruden (1977), was further confirmed in various taxa (Small, 1988;Plitman and Levin, 1990;Galloni & al., 2000;Jürgens & al., 2002;Choteau & al., 2006) with isolated exceptions (López & al., 1999).

P/O, pollen grain size, pollen vitality and breeding system
Although the pollen grains of the examined Passiflora species were large (P.foetida and P. caerulea) or mediumsized (P.misera and P. suberosa), considering the categories established by Erdtman (1952), no significant correlation was found between P/O and pollen grain size, in contrast to predictions by Cruden & Miller-Ward (1981).This relation varies among taxa: positive but weak for Asclepiadaceae (Wyatt, 1984), null for Genisteae (López & al., 1999) and negative for other taxa (Cruden and Miller-Ward, 1981;Wyatt, 1984;Cruden, 2000).Götzenberger & al. (2007)   concluded from 311 angiosperms that P/O depends on pollen grain number but not size.In Passiflora, pollen grain size is related with taxonomic position (Amela García & al., 2002) rather than with breeding system.As in P. misera, more than one pollen grain size classes were encountered in the same anther of several Caryophylloideae but, unlike Passiflora, number of grains in each size class was dissimilar (Jürgens & al., 2012).A bimodal pollen size distribution with the larger grains viable and the smaller ones aborted was also detected in a conifer (Aizen and Rovere, 1995).In the species with two pollen size classes, in which the smaller ones are not vital, the useful P/O ratio would be minor than the estimated one, nearly the half in the examined Passiflora species.So pollen vitality, i.e., proportion of pollen grains capable of germination, should be taken in account in pollen or P/O counts.
All the Passiflora species studied exhibited high pollen vitality, regardless of their breeding system.Relatively high degrees of pollen vitality have been detected for selfincompatible species (Wild & al., 2003;Hoc and Amela García, 1998) as well as self-compatible ones (Hoc and Amela García, 1999;Lehnebach and Riveros, 2003).Low levels of pollen vitality occur in species with different degrees of inbreeding and similar pollen longevities were found in two out of four Cyclamen species with the most opposite levels of inbreeding (Affre and Thompson, 1998).So pollen vitality seems to have no relation with breeding system.
In any case, the high pollen vitality suggests that the majority of the grains would be able to fertilize the conspecific ovules, and, as the transfer is directional (by means of constant visitors), a great part of the grains would reach their destiny.Besides, in P. caerulea, vitality is not affected by rain (Amela García and Hoc, 1997), as in well protected anthers (Hoc and Amela García, 1999) but not in very exposed ones (Hoc & al., 1995).In addition, the high vitality benefits the pollinators that feed their larvae with pollen, most Passiflora spp.pollinators carried Passiflora pollen in their hind legs (Amela García, 1999), being profitable the major part of the pollen load.In the species with large vital grains and small non vital ones, there would be a tendency to diminish the energetic cost of the pollen loss that the hymenopterans use for nesting, offering a certain non vital quantity; but that kind of grains also can reach the stigmas, as they are in the same anthers as the big ones, so this strategy would be inefficient.

P/O, flower size, stigma area and SA/PBA
The positive correlation between flower size with P/O was also found in 14 Vicia spp.(Ortega-Olivencia & al., 1997) but not in 21 papilionoid legumes (Etcheverry & al., 2012).P/O ratio was not correlated with stigma area in the Passiflora species analysed, perhaps because stigma area was not so different between them.On the contrary, P/O ratio was inversely related to the stigmatic area in two species of Stylidaceae and eight species from three genera of Proteaceae (Cruden, 1997).In the last ones, taxa with higher P/O produced larger number of smaller pollen grains and had smaller stigmas (Cruden, 1997).The co-occurrence of large stigmas with many ovules was corroborated in several species (Cruden and Miller-Ward, 1981).
In spite of the different degree of autogamy of the studied species and the minor than expected P/O ratio in the self-incompatible and with more efficient pollination ones, P/O index was inversely and strongly related with the SA/PBA, as in the 19 xenogamous angiosperms studied by Cruden and Miller-Ward (1981).It is interesting to note that the pollinators that do not always pollinate, due to their behavior on the flowers, i.e., X. artifex in P. caerulea and Pseudaugochloropsis graminea in P. foetida (Amela García andHoc, 1997 and1998a, respectively), have smaller pollen-bearing areas and, in consequence, a major SA/PBA resulted, i.e., a lower probability of pollination.

P/O and sexual system
Hermaphroditic species exhibit lower P/O ratios and larger pollen grains than sexually dimorphic congeners (Weller & al., 1998); lower P/O ratios have been computed for more than a hundred hermaphroditic taxa as compared to those with other sexual systems (see Cruden, 2000, Fig. 4).Having this in mind, the hermaphroditic flowers of Passiflora would be expected to have low P/O ratios; however, other factor needs to be considered: a certain proportion of the flowers of at least 13 Passiflora spp.do not bend the styles during anthesis, thus the stigmas do not get in contact with the pollen (either of the same flower or carried by a pollinator), turning the sexual system functionally andromonoecious (Amela García and Hoc, 2011 and cites therein).Considering this sexual system, the P/O ratio would be greater, as pollen grain number varies as a function of the percentage of male flowers (Cruden, 2000), either they are morphologically or functionally staminate.P/O from andromonoecious species are greater than from homoecious ones (Cruden, 2000).

P/O and taxonomy
The P/O index was lower than expected in P. caerulea and P. misera whereas it approached the expected values for P. foetida and P. suberosa, so the P/O index did not reflect taxonomic affinities in the species analysed.Wyatt & al (2000) concluded that pollen grain and ovule number did not reflect taxonomy but ecogeographical pressures in 38 species from 4 tribes of Asclepiadoidea.Lack of consistency between phylogenetic relationships and P/O was also detected by Mione and Anderson (1979) cited in Cruden (2000) and Affre and Thompson (1998), who found P/O more related with ecology.Instead, a tendency to xenogamy or autogamy within various taxa at different levels (genus or tribe) of Polemoniaceae (in which P/O ratio is a good predictor of the breeding system) was perceived by Plitmann and Levin (1990).P/O variations were associated to genus but not to tribe in 21 papilionoid Legumes (Etcheverry & al., 2012).Jürgens & al (2002) suggested that taxonomy is indirectly related with P/O as a consequence of the combination of features typical of each taxonomic group.P/O, like other traits related to pollination, are open to selection; as random evolution also takes part in phylogeny, P/O could not necessarily reflect taxonomic affinities, as suggested by Affre and Thompson (1998) and Galloni & al. (2007).

Fig. 1 .
Fig. 1.Frequency distribution of pollen grains in pollen size classes of four Passiflora species.

Table 1 .
Breeding system (determined by experimental pollinations), pollen and ovule number, P/O ratio and estimated breeding system by P/O ratio of four Passiflora species

Table 2 .
Pollen grain size and flower size of four Passiflora species

Table 3 .
Pollen vitality of four Passiflora species with two different techniques: fluorescence and main constituents n=number of pollen grains tested; l=large grains; s=small grains.

Table 4 .
Stigma area of four Passiflora species, thorax area of their pollinators and stigma area/pollen-bearing area (SA/PBA)