Flotation preferentially selects saccate pollen during conifer pollination


Author for correspondence:
Andrew B. Leslie
Tel: +1 7737951170
Email: aleslie@uchicago.edu


  • Among many species of living conifers the presence of pollen with air bladders (saccate pollen) is strongly associated with downward-facing ovules and the production of pollination drops. This combination of features enables saccate pollen grains captured in the pollination drop to float upwards into the ovule. Despite the importance of this mechanism in understanding reproduction in living conifers and in extinct seed plants with similar morphologies, experiments designed to test its effectiveness have yielded equivocal results.
  • In vitro and in vivo pollination experiments using saccate and nonsaccate pollen were performed using modeled ovules and two Pinus species during their natural pollination period.
  • Buoyant saccate pollen readily floated through aqueous droplets, separating these grains from nonbuoyant pollen and spores. Ovules that received saccate pollen, nonsaccate pollen or a mixture of both all showed larger amounts and higher proportions of saccate pollen inside ovules after drop secretion.
  • These results demonstrate that flotation is an effective mechanism of pollen capture and transport in gymnosperms, and suggest that the prevalence of saccate grains and downward-facing ovules in the evolutionary history of seed plants is a result of the widespread use of this mechanism.


In contrast with angiosperm pollination, in which pollen germinates on the exposed stigmatic surface of the carpel and grows through tissue in order to fertilize ovules, gymnosperm pollen must usually reach the inside of the ovules in order to germinate (Owens et al., 1998; Chandler & Owens, 2004– for exceptions, see Colangeli & Owens, 1989; Owens et al., 1995). Therefore, gymnosperm pollination requires a physical mechanism that can transport pollen grains that have first landed on the outside of the ovules. Sacci, or inflated air bladders, on many gymnosperm pollen grains may play an essential role in one such mechanism. Although sacci have often been interpreted as aids for aerial pollen dispersal, because they lower grain density and settling velocity (mentioned in Wodehouse, 1935; Proctor et al., 1996; Schwendemann et al., 2007), they may instead function primarily as floats that aid in the movement of pollen into the ovule during pollen capture (Doyle & O’Leary, 1935; Doyle, 1945; Tomlinson et al., 1991; Tomlinson, 1994). Living gymnosperms that produce saccate pollen (c. 375 species in the conifer families Pinaceae and Podocarpaceae) share a correlated suite of ovular characteristics, including ovules that open downwards with respect to gravity at the time of pollination and ovules that produce aqueous secretions or utilize rainwater as ‘pollination drops’– a transient drop of liquid on the tip of the gymnosperm ovule (Fig. 1a–d; Doyle & O’Leary, 1935; Doyle, 1945; Tomlinson et al., 1991; Tomlinson, 1994; Takaso & Owens, 1995; Owens et al., 1998; Tomlinson & Takaso, 2002). Saccate pollen grains (which are buoyant in water) that encounter the pollination drop will then float upwards through the drop and into the downward-oriented ovule (Fig. 1c,d; Runions & Owens, 1996; Salter et al., 2002), a process that is believed to concentrate buoyant pollen and exclude nonbuoyant pollen and spores (Tomlinson, 1994). That pollen sacci are involved primarily in this aqueous pollen capture, rather than aerial pollen dispersal, is suggested by the absence of sacci in pteridophytes with wind-dispersed spores and wind-pollinated angiosperms, and the secondary loss of sacci (or functional sacci) in various derived species of Pinaceae and Podocarpaceae in which ovules do not face downwards with respect to gravity (Tomlinson, 1994; Runions et al., 1999; Leslie, 2008).

Figure 1.

Pinus nigra ovulate structures. (a) Pinus nigra cone at pollination consisting of numerous, helically arranged ovule scales, each proximally bearing two downward-facing ovules. Bar, 1.0 mm. (b) Detail of Pinus nigra cone at pollination with red- (Tsuga canadensis) and blue- (Pinus nigra) stained pollen adhering to downward-facing ovules. Arrow indicates the micropylar arms of one ovule covered in pollen. Bar, 1.0 mm. (c) Pinus nigra ovule before application of an artificial pollination drop. Bracket identifies the ovule and the arrow indicates naturally deposited saccate pollen grains adhering to one of the two micropylar arms of the ovule. (d) Pinus nigra ovule after application of an artificial pollination drop. Arrow indicates saccate pollen that has floated to the apex of the drop, which, in this case, appears to be in the center of the drop because the cone is held horizontally under a microscope.

Although laboratory experiments have shown that flotation readily moves saccate pollen through water drops (Tomlinson, 1994; Runions & Owens, 1996; Salter et al., 2002) and can separate saccate from nonsaccate pollen in isolated ovules (Runions & Owens, 1996), the effectiveness of this pollination mechanism in actual pollination has been ambiguous. Some studies have concluded that pollen flotation is very effective at concentrating saccate pollen (Runions & Owens, 1996), whereas others have suggested that it is effective only in combination with rainwater washing through cones (Greenwood, 1986), or even that it is not necessary for efficient pollination (McWilliam, 1958; Lill & Sweet, 1977). A significant and potentially confounding issue that may explain some of these differences is the relatively rapid withdrawal of the pollination drop back into the ovule after it encounters pollen (Doyle & O’Leary, 1935; Owens et al., 1998; Mugnaini et al., 2007). Although the mechanism by which withdrawal occurs is uncertain (Mugnaini et al., 2007), a retreating pollination drop could potentially carry nonbuoyant pollen with it and negate any separation that initially occurred (Greenwood, 1986). Resolving this issue and identifying the exact role of flotation in pollination are not only important in conifer reproductive biology, but are also crucial to our understanding of seed plant reproductive evolution in a much broader sense. Saccate pollen is a conspicuous element of the fossil pollen record from as early as the late Carboniferous (Millay & Taylor, 1974), and many of the diverse groups of seed plants that have produced it have also had ovule morphology consistent with pollen flotation (Leslie, 2008; Hernandez-Castillo et al., 2009), suggesting that this mechanism may have a deep history in gymnosperm reproductive biology. This study presents combined observations from in vitro laboratory experiments and controlled in vivo pollination studies from two species of Pinus (P. nigra and P. mugo) to determine how effectively a flotation-based pollination mechanism transports saccate pollen inside ovules and discriminates between saccate and nonsaccate pollen, as well as how pollination drop retraction affects these processes.

Materials and Methods


Saccate pollen was collected from fresh pollen cones of Pinus nigra Arnold (average grain diameter of 57 μm not including sacci) and Pinus mugo Turra (49 μm) growing at the University of Chicago, IL, USA. One-half of each sample was stained with a 1% solution of toluidine blue, washed with water several times to remove excess stain, and allowed to dry. The grains were stained so that they could be seen easily and counted accurately, especially when inside ovules. Nonsaccate pollen was collected from fresh pollen cones of Tsuga canadensis Carrière (average grain diameter, 72 μm) growing at the Morton Arboretum in Lisle, IL, USA. Pollen was stained with a 1% solution of safranin O following the same procedure as for saccate Pinus pollen. Naturally occurring nonsaccate pollen was used in this study instead of saccate pollen rendered nonbuoyant by chemical treatment (as in some previous studies; see Lill & Sweet, 1977; Greenwood, 1986), because preliminary experiments have shown that these treatments produce pollen that initially floats before sinking. Smaller nonsaccate spores of Lycopodium clavatum from the Sigma-Aldrich company (average spore diameter, 42 μm) were also stained with safranin O following the same procedure as in the pollen grains.

In vitro experimental pollination

Saccate Pinus pollen, nonsaccate T. canadensis pollen or a mixture of these were packed into a Corning 10 μl glass capillary tube (interior diameter, 0.3 mm) that was bent to simulate an inverted Pinus ovule (Fig. 2a). A syringe connected to the back of the tube was used to push a column of water through it, simulating the secretion of a pollination drop by the ovule (Fig. 2b). The water column could then be gradually pulled back through the column, simulating the effects of pollination drop withdrawal. A black and white CCD video camera (Sony model XC-37 1/2′′ monochrome CCD camera; Sony Electronics Inc., San Jose, California, USA) mounted on a Leica MZ6 stereomicroscope (Leica Camera Inc., Allendale, New Jersey, USA) was used to record pollen movement through the water column.

Figure 2.

 Pollen separation in artificial ovules. (a) Glass capillary tube containing a mixture of unstained nonsaccate Tsuga canadensis pollen (yellow) and stained saccate Pinus nigra pollen (blue) before the injection of a water column. (b) Glass capillary tube after the injection of a water column. Blue saccate grains have floated to the top. Bar, 1.0 mm.

In vivo experimental pollination

As seed cones emerged from buds (15–19 May 2008 and 2009), groups of P. nigra and P. mugo cones were assigned to separate batches that each received a different type of pollen treatment. The batches varied in size, but were large enough (typically 15–30 cones) to provide sufficient pollinated ovules for further study. To test the behavior of saccate pollen alone, one batch received only naturally deposited wind-borne saccate pollen; nonsaccate pollen was not artificially applied. To test the behavior of nonsaccate pollen and spores, batches of cones received either stained or unstained nonsaccate T. canadensis pollen or stained nonsaccate L. clavatum spores. These grains were artificially applied using a glass pipette with a rubber bulb to dust small amounts of pollen or spores onto ovules. These cones were then covered with Nitex® bolting cloth (pore size, 35 μm; Wildlife Supply Company; Yulee, Florida, USA) to prevent wind-borne saccate pollen from entering the cone during pollination.

To test pollen discrimination by flotation, stained nonsaccate T. canadensis pollen was applied to a batch of cones in both Pinus species which then remained uncovered so that ovules would also receive naturally deposited conspecific wind-borne saccate pollen. Ovules therefore received a variable mixture of nonsaccate and conspecific saccate pollen. To test whether pollen staining affected the results, an additional batch of cones from both species received a mixture of unstained nonsaccate T. canadensis pollen and stained saccate Pinus pollen. These cones were then covered with Nitex® bolting cloth to prevent unstained wind-borne saccate pollen from entering the cones.

Data collection

Several (typically five) cones per batch were collected from each species on alternating days during pollination (19–29 May 2008 and 2009), and 10–25 ovules per cone were examined and dissected using a Leica MZ6 stereomicroscope. For each ovule, pollen grains adhering to the micropylar arms and lodged in the micropyle (Fig. 1b,c) were tallied and recorded as the number of pollen grains occurring on the outside of the ovule. These grains were then removed and the integument was carefully dissected using a thin needle to expose the nucellus (tissue directly covering the megagametophyte) and the inner surface of the integument that is directly adjacent to the nucellus. Pollen grains adhering to, or pressed against, either the nucellus or the inner integument were tallied and recorded as the number of pollen grains occurring inside the ovule. Ovules that received no pollen grains were not included in this study.


In vitro pollination

Both saccate Pinus pollen and nonsaccate T. canadensis pollen were wetted and any saccate pollen grains, whether stained or unstained, rapidly floated upwards through the water column. By contrast, nonsaccate pollen grains remained at the downward-facing opening of the capillary tube (Fig. 2b, and Supporting Information Video S1). When the water column was retracted up the tube, it often carried a small number of nonsaccate grains with it. However, most grains adhered to the sides of the capillary tube and were not drawn up by the water column. This simulated drop retraction occurred over a period of c. 1 min, which is faster than natural rates of drop withdrawal (Doyle & O’Leary, 1935; Mugnaini et al., 2007).

In vivo pollination in control ovules

In P. nigra and P. mugo ovules that received only unstained, wind-borne saccate pollen, grains accumulated on the micropylar arms for up to 10 d after bud break. Ovules received a highly variable amount of pollen, ranging from 1 to 27 grains (see Table S1 for sample statistics), but almost all pollen grains were found on the outside of ovules collected before May 26 (Fig. 3a,b). On average, only 6–14% of the total pollen grains each of these P. nigra or P. mugo ovules received were found inside. However, those cones sampled 10 or more days after bud break (designated here as ‘late-sampled cones’) showed a marked increase in both the proportion and absolute number of pollen grains inside ovules (Fig. 3a,b– 27 May, 29 May). On average, between 44% and 73% of the total pollen grains received by each ovule were found inside; this corresponds to three or more pollen grains, on average, inside these ovules, as compared with less than one, on average, inside ovules sampled earlier. In late-sampled cones, there was also a strong positive correlation (Spearman’s ρ = 0.76, < 0.01; see Fig. S1a) between the total number of pollen grains per ovule (pollen outside plus pollen inside) and the number of pollen grains inside the ovule (Fig. 4a).

Figure 3.

 Mean number of naturally deposited saccate pollen grains inside (closed columns) and outside (open columns) of ovules that received pollen during pollination days in Pinus nigra (a) and Pinus mugo (b). Pollen grains were recorded as outside if they were found adhering to the outer surface of the integument or lodged in the micropyle, and inside if they were found on the nucellus or the inner integument tissues directly adjacent to the nucellus. The number of ovules sampled per day ranged from 21 to 67; 95% confidence intervals around the mean of each column were calculated on the basis of 10 000 resampled means. (See Supporting Information Table S1 for standard deviations around sample means.)

Figure 4.

 Number of pollen grains inside late-sampled ovules compared with the total number of pollen grains outside and inside these ovules. (a) Pooled data for all Pinus ovules (P. nigra and P. mugo) receiving naturally deposited unstained conspecific saccate pollen. (b) Pooled data for all Pinus ovules receiving only unstained nonsaccate Tsuga canadensis pollen. (c) Pinus mugo ovules receiving only stained Lycopodium clavatum spores. The line in each graph represents the limiting case where all recorded pollen grains occurred inside the ovule. Large circles, > 5 ovules; medium circles, 2–5 ovules; small circles, 1 ovule.

By contrast, late-sampled P. nigra and P. mugo ovules that received only nonsaccate T. canadensis pollen had few grains inside them – between 0.63 and 1.4 pollen grains on average. Although staining had a statistically significant impact on the amount of nonsaccate pollen inside these ovules, in both cases only a small number of grains entered ovules (Table S2). These ovules showed a weak, but significant, correlation (Spearman’s ρ = 0.28, < 0.01; Fig. S1b) between the total number of pollen grains per ovule and the number of grains inside the ovule (Fig. 4b). Late-sampled ovules that received stained L. clavatum spores were only recovered from several P. mugo cones, but showed a pattern similar to T. canadensis pollen even when present at much higher numbers per ovule (Fig. 4c).

In vivo pollination in test ovules

Pinus nigra and P. mugo ovules that received both nonsaccate T. canadensis and saccate Pinus pollen had more saccate pollen grains than nonsaccate pollen grains inside late-sampled ovules (Fig. 5a,b). Saccate pollen was also found inside ovules in much higher proportions (an average of 37–48% of the total amount of saccate pollen per ovule was found inside) than was nonsaccate pollen (12–15% of the total amount). Similar results were seen regardless of the staining procedure used for each type of pollen (Fig. 5a,b). Similar results were also seen in cones that were left uncovered (Fig. 5a) and cones that were bagged to exclude wind-borne pollen (Fig. 5b). Bagged P. mugo ovules that received stained saccate P. nigra pollen (Fig. 5b) contained only slightly less saccate pollen inside, on average, than those receiving natural, conspecific P. mugo pollen (2.7 grains per ovule and 3.5 grains per ovule, respectively).

Figure 5.

 Mean numbers of pollen grains inside (right) and outside (left) late-sampled ovules receiving a mixture of nonsaccate Tsuga canadensis and saccate Pinus pollen. (a) Data from P. nigra (= 172) and P. mugo (= 92) ovules receiving unstained saccate pollen and stained nonsaccate pollen. (b) Data from P. nigra (= 86) and P. mugo (= 50) ovules receiving stained saccate pollen and unstained nonsaccate pollen; 95% confidence intervals around the mean of each column were calculated on the basis of 10 000 resampled means. (See Supporting Information Table S1 for standard deviations around sample means.)


Flotation preferentially transports saccate pollen relative to nonsaccate pollen into downward-oriented ovules, as in vitro experiments have vividly demonstrated here (Fig. 2b and Video S1) and elsewhere (Runions & Owens, 1996; Salter et al., 2002). This process occurs entirely within water drops, rather than on their surfaces, as suggested by some previous studies (Tomlinson, 1994). Selective pollen flotation also occurs in vivo: consistently more saccate pollen than nonsaccate pollen is found inside Pinus ovules – regardless of the staining or pollination procedure (Figs 4a,b, 5a,b). Despite this, nonsaccate grains inside ovules are not rare [41% of ovules that received both saccate and nonsaccate pollen had some (usually one) nonsaccate grains inside them], which suggests that retreating pollination drops often carry a small amount of nonbuoyant pollen with them. Flotation therefore does not fully exclude nonbuoyant grains, but does discriminate against them, because a much smaller proportion of these grains are transported inside the ovules relative to saccate pollen (Fig. 4b). As many angiosperm pollen grains are susceptible to damage by water (Mao & Huang, 2009), the secretion of a pollination drop may further discriminate against contaminant angiosperm pollen, although this effect would also be present in gymnosperms that produce pollination drops but do not use flotation. The dominant role of flotation in saccate pollen transport is further illustrated by the fact that ovules were unable to discriminate between different species of saccate pollen, as P. nigra saccate grains were found inside P. mugo ovules in large numbers (Fig. 5b). Taken together, the results presented here suggest that the effectiveness of flotation as a pollination mechanism is a result of its ability to efficiently transport and preferentially concentrate saccate pollen inside ovules, rather than its ability to actively exclude all contaminant pollen or spores. Although some studies have concluded that rainwater washing through cones is primarily responsible for this effect (Greenwood, 1986), precipitation that occurred during the collecting period of this study did not alter significantly the distribution of pollen grains on ovules. This suggests that the secretion of a pollination drop alone is sufficient for the operation of this mechanism.

An efficient flotation mechanism enables many species to use morphological structures that catch wind-borne pollen before drop secretion, thereby lengthening the duration of pollen capture and increasing both the total amount of pollen received and the probability of long-distance pollen sampling. In P. nigra and P. mugo, for example, a newly secreted pollination drop will encounter and potentially transport up to 10 d worth of pollen accumulation (in some cases up to 30 grains) adhering to the micropylar arms. These micropylar arms play a similar role in Picea pollination (Runions & Owens, 1996), and their presence highlights how flotation can enable these taxa to ‘scavenge’ (Tomlinson et al., 1991) pollen from around the ovules. Similar elongated bifid integuments are even present in several fossil taxa in conjunction with saccate pollen – including early conifers (Kerp et al., 1990) and the unrelated Triassic corystosperm Umkomasia (Thomas, 1933; Klavins et al., 2002).

As a flotation-based mechanism could theoretically be utilized by any gymnosperm with buoyant pollen and the correct ovule orientation, pollen flotation might be expected to have a significant history in seed plant reproduction. Indeed, saccate pollen was produced by almost one-half of all the higher level seed plant groups whose pollen is known (based on orders listed in Taylor et al., 2009 and treating angiosperms as one group of seed plants), suggesting that this mechanism has had a widespread role in seed plant reproduction through time. Although ovule orientation in fossil plants is difficult to interpret, even the earliest Carboniferous and Permian conifers with saccate pollen also possess ovules that are inverted, or reflexed, towards the cone axis, as in living Pinaceae and Podocarpaceae that utilize flotation (Mapes, 1987; Kerp et al., 1990; Hernandez-Castillo et al., 2009). Multiple nonconifer gymnosperm groups (including Carboniferous callistophytes, Permian glossopterids, Triassic corystosperms and the Jurassic plant Caytonia) produced saccate pollen (Harris, 1964; Millay & Eggert, 1970; Yao et al., 1995; Taylor et al., 2009) and had reflexed or pendant ovules that probably faced downwards (Harris, 1964; Rothwell, 1981; Klavins et al., 2002), although the orientation of glossopterid ovules is controversial (Taylor & Taylor, 1992; Doyle, 2006). The reproductive biology of these groups is especially important in understanding seed plant evolution, as many of them (glossopterids, corystosperms and Caytonia) have been implicated in angiosperm origins on the basis of their reproductive morphology (Frohlich, 2003; Doyle, 2006; Frohlich & Chase, 2007; Doyle, 2008). The reproductive structures of Caytonia, in particular (which consist of ovules surrounded by a reflexed enclosing structure, or ‘cupule’, with an opening at the base), have been linked with the anatropous ovules of angiosperms in recent theories of angiosperm evolution (Frohlich & Chase, 2007) and phylogenetic analyses incorporating fossil data (Hilton & Bateman, 2006; Doyle, 2006, 2008). The overall reproductive morphology of Caytonia, coupled with the results of this study, suggests that its reflexed ovular orientation is most probably related to the use of a flotation-based pollination mechanism. Although this does not provide evidence for or against a close relationship with angiosperms, acceptance of these phylogenetic hypotheses should be recognized to imply that the anatropous orientation of angiosperm ovules records the ancestral use of a distinctly gymnospermous flotation-based pollination mechanism.


I would like to thank Kunso Kim and the Morton Arboretum for allowing me to collect Tsuga pollen. Kevin Boyce, Peter Crane, Michael LaBarbera, Fred Ruddat and David Jablonski contributed helpful discussions of ideas and comments on the manuscript. I would also like to thank Mark Geary II for his patience during pollination days. James Doyle, John Runions and an anonymous reviewer provided helpful comments and suggestions.