<<

Novelties of the flowering tube underlie diversification of a key life history stage

Joseph H. Williams*

Department of Ecology and Evolution, University of Tennessee, Knoxville, TN 37996

Edited by Peter R. Crane, University of Chicago, Chicago, IL, and approved June 2, 2008 (received for review January 3, 2008) The origin and rapid diversification of flowering has puzzled angiosperm lineages. Thus, I performed hand and evolutionary biologists, dating back to Charles Darwin. Since that timed collections on representatives of three such lineages in the time a number of key life history and morphological traits have field [Amborella trichopoda, polysepala, and Aus- been proposed as developmental correlates of the extraordinary trobaileya scandens; see supporting information (SI) Text, Meth- diversity and ecological success of angiosperms. Here, I identify ods for Studies]. several innovations that were fundamental to the evolutionary Each of these has an extremely short fertilization lability of angiosperm reproduction, and hence to their diversifi- interval—pollen germinates in Ͻ2 h, a pollen tube grows to an cation. In gymnosperms pollen reception must be near the egg in Ϸ18 h, and to an egg in 24 h (Table 1). The window for largely because sperm swim or are transported by pollen tubes that fertilization must be short because the egg cell is already present grow at very slow rates (< Ϸ20 ␮m/h). In contrast, pollen tube at the time of pollination (Table 1) and this is also the case for growth rates of taxa in ancient angiosperm lineages (Amborella, species within a much larger group of early-divergent lineages Nuphar, and ) range from Ϸ80 to 600 ␮m/h. Com- (Table S2 and references in ref. 11). Early-divergent angio- parative analyses point to accelerated pollen tube growth rate as sperms have far shorter fertilization intervals than any gymno- a critical innovation that preceded the origin of the true closed sperm (Fig. 1) (12) except for Gnetales: intervals are 6–8 days carpel, long styles, multiseeded ovaries, and, in monocots and in Gnetum (13) and 10–36 h in Ephedra (14, 15). , much faster pollen tube growth rates. Ancient angio- I used the data from Fig. 1 and Table 1 to map fertilization sperm pollen tubes all have callosic walls and callose plugs (in timing onto five major hypotheses of plant relationships contrast, no gymnosperms have these features). The early associ- (differing mainly in the placement of Gnetales) (16). I coded ation of the callose-walled growth pattern with accelerated pollen fertilization intervals as either ‘‘short’’ (i.e., Amborella/Ephedra tube growth rate underlies a striking repeated pattern of faster interval of Յ2 days), ‘‘moderate’’ (/Gnetum interval of and longer-distance pollen tube growth often within solid path- 2–7 days), or ‘‘long’’ (i.e., typical gymnosperm interval of Ͼ7 ways in phylogenetically derived angiosperms. Pollen tube inno- days). The common ancestor of extant seed plants was fixed as vations are a key component of the spectacular diversification of long (14). carpel (flower and ) form and reproductive cycles in flowering All five analyses infer that the common ancestor of extant plants. angiosperms (angiosperm CA) had a short fertilization interval, as summarized in Fig. 2 (see also SI Text, Methods for Phyloge- heterochrony ͉ key innovation ͉ modularity ͉ origin of angiosperms ͉ netic Reconstruction of Fertilization Timing, and Fig. S1). If evo-devo Gnetales is sister to pines, conifers, all other gymnosperms, or all other seed plants, two separate evolutionary transitions from lowering plants, or angiosperms, are thought to have origi- long to short fertilization intervals are inferred: one in the Fnated in an environment where rapid reproduction was common ancestor of Ephedra (or Gnetales) and one in the advantageous (1). Virtually all of their most defining features, angiosperm CA. However, if Gnetales is sister to angiosperms a including the , closed carpel, highly reduced male and single origin of the short fertilization interval in their common female gametophytes, double fertilization, sexually formed ancestor is supported. polyploid , and an exceptionally short pollination-to- fertilization interval (progamic phase), are thought to have Developmental Origins of Rapid Fertilization Syndromes in evolved under selection for a faster reproductive cycle (1–6). To Seed Plants understand what changes may have contributed to speeding the In seed plants, pollination (pollen reception by ovule or carpel) progamic phase, I undertook a series of comparative analyses of and fertilization (sperm fusion with egg cell) are key landmarks

the interacting ontogenies that determine fertilization timing in for the development of the male gametophyte within sporophytic EVOLUTION seed plants. tissues. The plesiomorphic condition is that pollination occurs A survey of Ͼ900 studies, covering 130 seed plant families and near the beginning of megagametogenesis (14) and fertilization 717 taxa, indicates that the time interval between pollination and occurs after a long period of development of a large female fertilization (hereafter, ‘‘fertilization interval’’) ranges from 10 h gametophyte (13–15). grade angiosperms retained the to Ͼ12 months in gymnosperms and from 15 min to Ͼ12 months plesiomorphic seed plant feature of pollination near the onset of in angiosperms (Fig. 1). Because angiosperms generally have megagametogenesis (refs. 4, 17, and Table 1). Because all much shorter fertilization intervals, it has long been supposed angiosperms have exceptionally small female gametophytes (17), that shortening of the fertilization process accompanied the origin of angiosperms (1–5). But two alternative hypotheses have also been proposed: the short fertilization interval may have Author contributions: J.H.W. designed research, performed research, analyzed data, and evolved earlier in history, before the origin of angiosperms (6), wrote the paper. or later, in one or more derived angiosperm lineages (7). Our The author declares no conflict of interest. understanding of early angiosperm relationships, based on re- This article is a PNAS Direct Submission. cent molecular phylogenetic analyses (8–10), provides the op- *E-mail: [email protected]. portunity to distinguish among these alternatives. This article contains supporting information online at www.pnas.org/cgi/content/full/ Rates, timing, and duration of the fertilization process have 0800036105/DCSupplemental. not yet been characterized in newly defined ‘‘basal grade’’ © 2008 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800036105 PNAS ͉ August 12, 2008 ͉ vol. 105 ͉ no. 32 ͉ 11259–11263 Downloaded by guest on September 26, 2021 0.4 a ia Angiosperms e n a e Gymnosperms h s p ra a m y 0.3 y /B e a a il * a e ll /N a a rs s rs r m a e r b b i e d o e d u e r a m o m n h a g if e t c o h o tr u e t c k n h e a b p b s i m o 0.2 y in o p n in m u a u c ri ll C G C E G P A N C A Illi T A

0.1

Proportion of taxa Proportion of 0 7d 5d 5h 8h 6m 2m 1m 72h 14d 36h 21d 48h 24h 12h <2h 12m >12m Time between pollination and fertilization Origin of extant angiosperms Fig. 1. Variation in fertilization timing among seed plants. Data represent the earliest time reported from pollination until a pollen tube enters the Fig. 2. Origins of rapid fertilization syndromes among major seed plant micropyle of the ovule. The proportion of angiosperms in each time category lineages. Fertilization intervals are mapped onto the ‘‘Gnepine’’ phylogenetic is represented in color (801 reports) and that of gymnosperms in black (109 hypothesis (16) under the assumption that the seed plant ancestor had a long studies) (hence, the bars in the graph sum to two, not one). Dark blue, light fertilization interval. The polarity of terminal taxa was determined from a blue, and magenta represent monocots, eudicots, and all other angiosperms, much larger analysis by using data from Fig. 1 (SI Text, Methods for Phyloge- respectively. h, hours; d, days; m, months. See SI Text, Fertilization Intervals of netic Reconstruction of Fertilization Timing). Fertilization intervals are short Seed Plants and Table S1. (black), moderate (dark gray), long (light gray), or equivocal (hashed line). *, excluding Pinaceae.

fertilization in the angiosperm CA is inferred to have become shifted earlier in time as a result of an abbreviation of female the pollen tube pathway and reducing pollen germination time, gametophyte ontogeny (1–4). In contrast, Ephedra retains the whereas slow pollen tube growth rate is conserved. Basal grade angiosperms all have faster pollen tube growth presumed plesiomorphic state of late fertilization timing (after Ϸ ␮ development of a large female gametophyte), but is specialized in rates than any gymnosperm, ranging from 80 to 600 m/h (Table 1), and comparable rates were calculated from other that pollination occurs long after onset of megagametogenesis (15). studies (Table S2). Thus, the angiosperm CA is inferred to have Thus, quite different heterochronic processes underlie changes in achieved its short fertilization interval by accelerating pollen relative onset and offset timing of the progamic phase: a short tube growth. Importantly, all early-divergent angiosperms have fertilization interval arose in Ephedra primarily by delay of polli- pollen tubes with callose (1,3-␤-glucan) secondary walls. Callose nation, and in the angiosperm CA, largely by earlier fertilization. plugs of the pollen tube are also present in Amborella, Nuphar, In virtually all gymnosperms, pollen hydration and germina- and Austrobaileya (Fig. 3). Rapid pollen tube growth and the tion takes two or more days and the active period of pollen tube callose wall and plugs are clearly conspicuous synapomorphies of growth (excluding dormant periods) takes five days or more (13, angiosperms, yet these traits have so far received little attention. 14, 18). Maximum in vivo growth rates of gymnosperm pollen In summary, the angiosperm CA and Ephedra both have a short tube tips calculated from the literature (Table S2) are Ϸ1 ␮m/h ␮ ␮ fertilization interval characterized by rapid pollen germination and in Zamia (Cycadales), 2 m/h in Ginkgo (Ginkgoales), 5 m/h in sperm transport to the egg via narrow pollen tubes (albeit through ␮ Gnetum (Gnetales), and 6 m/h in Agathis (Coniferales). These different tissues). However, shortening the fertilization interval in Ͻ ␮ rates are similar to in vitro growth rates that range from 1 m/h Ephedra involved streamlining of a basically gymnospermous set of ␮ in Pinus (19) to 10–20 m/h in various other gymnosperms (20, progamic phase traits, whereas in the inferred angiosperm CA 21). Thus, the plesiomorphic long gymnosperm fertilization many novelties of the pollen tube and carpel acted to speed the interval is characterized by slow pollen germination and slow tip fertilization process. Although morphological links between Gn- growth of pollen tubes. etales and angiosperms continue to be found in the fossil record In contrast, Ephedra has rapid pollen germination (1–2 h) (22) (23), at least some of their similarities appear to have different and its pollen tubes reach the archegonium in as little as 10–16 developmental bases (15, 24, 25). h (14). But Ephedra pollen tube growth rates are similar to other gymnosperms. I calculated in vivo rates of 14 ␮m/h in E. distachya Relationship Between Pollen Tube Growth and and 6–19 ␮m/h in E. trifurca (Table S2); in vitro rates of 4 ␮m/h Carpel Evolution are reported for other Ephedra species (22). Ephedra pollen Pollen tube growth rates are mainly determined by how rapidly germinates directly on the surface of the female gametophyte tube walls can be constructed and by the nature of their and a narrow pollen tube must grow only a short distance interaction with maternal tissues. Gymnosperm pollen tube tips (60–230 ␮m) to the egg within a pathway entirely composed of contain cellulose, esterified pectins, and callose (19). Their archegonial neck cells of the female gametophyte (14, 15). Thus, lateral tube walls are thick and primary, comprising mostly the short fertilization interval in Ephedra results from shortening esterified pectins and cellulose. Tip callose is transient so lateral

Table 1. Developmental characteristics of the progamic phase of basal grade angiosperms Minimum pt pathway Pt growth rate, Taxon Egg cell present* Fertilization interval, h† length Ϯ SD, mm‡ ␮m/h§

Amborella trichopoda Yes 13.7, 24 1.32 Ϯ 0.14, n ϭ 16 80 Ϯ 24 (dir) 96 (mp) Nuphar polysepala Yes 8, 13 4.71 Ϯ 1.33, n ϭ 20 589 (mp) Austrobaileya scandens Yes 18, 24 4.88 Ϯ 0.73, n ϭ 16 271 (mp)

*Egg cell present or not at time of pollination. †Time interval between pollination and first penetration of ovule, or first evidence of sperm in female gametophyte, respectively. ‡Minimum distance from stigmatic surface to egg along pollen tube (pt) pathway. §Calculated directly from in vivo pollen tube lengths (dir) (see SI Text, Methods for Pollination Studies), or indirectly by dividing minimum pathway length by shortest fertilization interval (mp). The latter is a useful metric for comparative studies. h, hours; n, number of maternal plants.

11260 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800036105 Williams Downloaded by guest on September 26, 2021 Pollen tubes of basal grade angiosperms appear similar to those of most model angiosperms: all have prominent callosic secondary walls and callose plugs (Fig. 3). Pollen tubes of Amborella usually form only a single plug midway to the ovule (Fig. 3A). In Nuphar (Fig. 3B) and other (refer- ences in Table S2), callose plugs are formed frequently. In Austrobaileya (Fig. 3C) plugs are found only early in develop- ment or sometimes not at all. Interestingly, Nymphaealean species have much faster pollen tube growth rates than other early-divergent angiosperms (Table 1 and Table S2). Studies of derived angiosperms often find a positive relationship between pollen tube growth rates and frequency of callose plug formation (36–38). Callose walls are associated with rapid growth patterns, but they also have other properties. Callose is relatively imperme- able (27, 28) thus cross-talk with maternal tissues is confined to the very tip of the growing tube. Furthermore, callose plugs seal off the pollen tube so the sperm-containing apical portion advances independently of the pollen grain and old tube (39). These features may reduce the risk of damage and allow pollen tubes to grow longer distances (39). Gymnosperm pollen tubes typically do not grow long distances because they maintain a continuous cytoplasm from pollen grain to pollen tube tip throughout development and advance slowly through solid nucellar tissue by causing cell death and absorption of intervening cells (18, 19). Pollen tubes of conifers must digest a 0.1- to 3-mm-thick nucellar covering of the egg (14, 40). Longer pollen tubes up to Ϸ9-mm-long are found in several phyloge- netically derived conifers (e.g., Agathis, Tsuga) (18, 40). In these taxa, pollen germinates on an integument, , or scale, and pollen tubes grow slowly to the egg for upward of six weeks. A broad survey of early-divergent angiosperms indicates they generally also have relatively short pollen tube pathways (Fig. 4, Table S3), ranging from Ͻ0.5 mm in Trithuria to Ϸ15 mm long in Hydnora. By using an averaging algorithm that accounts for phylogeny (41), the mean (Ϯ SD) ancestral minimum pollen tube pathway length of angiosperms (shortest distance to first ovule) is reconstructed as 1.58 Ϯ 0.49 mm (see SI Text, Methods for Fig. 3. Pollen tube growth patterns in basal grade angiosperms. (A) Ambo- Pollen Tube Pathway Length Analyses). In a separate analysis rella trichopoda pollen tubes grow between stigmatic hairs (sh), but only one ancestral ovule number is reconstructed as one per carpel (see or a few enter a narrow stylar canal (sc), grow to the ovarian cavity (oc) and Table S3), similar to previous studies (11). Thus, the ancestral around the downward facing single ovule (o). (B) Nuphar polysepala pollen angiosperm pollen tube pathway was short because stigmas were tubes grow intercellularly below the stigmatic surface (data not shown) then near the sole ovule. This conclusion is also supported by studies enter a wide stylar canal that opens into a multiovulate ovarian cavity (cw, carpel wall). (C) Pollen tubes of Austrobaileya scandens grow into a massive showing that the gynoecia of fossil from Early Creta- stigmatic secretion, then enter individual carpels through a cleft (cl) between ceous deposits are very small and without obvious styles (11, 42). the two stigmatic lobes (sl; only one shown) and follow the open stylar canal The plesiomorphic short pollen tube pathway of angiosperms to the ovarian cavity (composite image). (D–I) Callose plugs from Amborella (D was formed by an open (ascidiate) carpel that was sealed by and E), Nuphar (F and G), and Austrobaileya (H and I). All stained with aniline secretions (11). Pollen tubes of Amborella, Nuphar, and Aus- blue (A, double-stained with DAPI). (Scale bars: A, 100 ␮m; B, C, 500 ␮m; D–I, trobaileya all grow within relatively open, secretion-filled spaces 10 ␮m.) (Fig. 3) before penetrating the solid layer of nucellar cells that is only 1–3 cells thick above the egg cell (17). Thus, the origin of EVOLUTION the angiosperm pollen tube pathway involved transfer of the walls lack callose and plugs do not form (19). In contrast, pollen reception apparatus from ovule to carpel (), de- angiosperm pollen tube tips are made up almost entirely of velopment of a secretory pathway connecting stigma to ovule, esterified pectins (26–28). Just behind the tip, lateral tube walls and reduction of the plesiomorphic thick nucellar apex to an become reinforced by deesterification of tip pectins (26, 28) and extremely thin layer. secretion of a callose secondary wall (29, 30). This unique pollen The data above suggest that early angiosperm pollen tubes tube wall structure is thought to allow faster growth rates than acquired novel callosic secondary walls and a slightly faster possible in other tip-growing cells because (i) the pectic tip is growth rate within a short pathway composed mainly of secre- more plastic and rapidly extensible, (ii) the mature tube cell wall tions. Despite having the structural features that could facilitate has greater resistance to tension stress because of deesterifica- much faster and longer distance pollen tube growth, the fastest tion of pectins and secretion of callose, and (iii) callose plugs help pollen tube growth rates among early-divergent angiosperms are maintain positive turgor pressure in the growing tip (27, 31). Ͻ3% of maximum sustained rates known in derived lineages There is evidence that building an amorphous callose-based wall (Table 1 and Table S2) and their longest pollen tube pathways is faster and more energy efficient than biosynthesis of a are Ͻ3% of the known maximum (Fig. 4 and Table S2). cellulose microfibril-based wall (29, 32). Furthermore, silencing Interestingly, slow male gametophyte development in these taxa of genes involved in either pectin modification or callose syn- is generally mirrored by slow metabolic rates of their associated thesis reduces pollen competitive ability (33–35). sporophytes (43). Thus it seems likely that in early angiosperm

Williams PNAS ͉ August 12, 2008 ͉ vol. 105 ͉ no. 32 ͉ 11261 Downloaded by guest on September 26, 2021 Amborella trichopoda AM Hydatella filamentosa1 1000 Brasenia schreberi Cabomba aquatica Maize 1 NY Cabomba australis Cabomba caroliniana Cereus Nuphar polysepala Barclaya longifolia Ondinea purpurea Nymphaea odorata Nymphaea10 tetragona Victoria cruziana Austrobaileya scandens Maize 5 Trimenia moorei Trimenia neocaledonica AU Trimenia papuana Illicium anisatum 100 Illicium floridanum Illicium henryi Illicium micrantha Kadsura20 japonica Kadsura longipedunculata Schisandra chinensis Schisandra sphenanthera Ceratophyllum demersum Acorus calamus Acorus gramineus Tofieldia calcyculata Petunia MO Gymnostachys anceps Cichorium Orontium aquaticum Pothos30 longipes Agathis Alisma lanceolatum 10 Hydrocleys nymphoides Butomus umbellatus Hydrocharis morsusranae Scheuchzeria palustris Nuphar Austrobaileya ulvaceus Triglochin maritima Groenlandia densa Potamogeton alpinus Tacca40 chantrieri Geranium Red Oaks Tamus communis Sciaphila albescens Hedyosmum brasiliense Hedyosmum mexicanum Amborella Hedyosmum nutans Ascarina lucida CH Ascarina rubricaulis Chloranthus erectus 1 Chloranthus japonicus Triticum typical Chloranthus50 serratus Chloranthus spicatus Nymphaea Sarcandra chloranthoides Sarcandra glabra Gnetum Sarcandra irvingbaileyi gymnosperm Myristica subalulata Liriodendron tulipifera Michelia fuscata vitiensis range baccata Eupomatia60 laurina Annona muricata Artabotrys hexapetalus Ephedra Asimina triloba 0.1 Monodora myristica Pollen tube pathwayPollen length (mm) tube Calycanthus floridus 100 1000 10000 0.1 1 10 Idiospermum australiense Gomortega keule Gyrocarpus americanus Hernandia cordigera Sparattanthelium70 speciosum Cassytha filiformis Persea americana Cinnamomum camphora Laurus nobilis Litsea neesiana Lindera benzoin Doryphora sassafras Hortonia angustifolia Hedycarya dorstenioides Kibaropsis80 caledonica Peumus boldus EM Steganthera ilicifolia Tambourissa peltata Tambourissa purpurea Fertilization interval (h) Tambourissa sieberi Wilkiea huegeliana Wilkiea macrophylla Siparuna andina Lactoris90 fernandeziana Hydnora johannis Aristolochia sempervirens Fig. 5. The relationship between the fertilization interval and pollen tube Asarum europaeum Saruma henryi Houttuynia cordata Saururus cernuus pathway length in seed plants (both axes log scale). Typical gymnosperms are Saururus loureiri Manekia naranjoana Zippelia begoniaefolia Macropiper100 excelsum restricted to the darkest shaded box (12, 14), except for extremes represented augustum Piper brevipedunculatum Piper peltata caperata by green diamonds. Basal grade angiosperms (purple diamonds) fall above the Peperomia clusaiefolia Canella alba Drimys winteri Tasmannia membranea hand-fitted gymnosperm constraint line, but are conservative relative to the Zygogynum baillonii Zygogynum120 pancheri Zygogynum pomiferum Zygogynum stipitatum majority of phylogenetically derived angiosperms (only extremes are shown, Dicranostigma franchetianum Hypecoum procumbens Euptelea polyandra Circaeaster agrestis blue diamonds). For Agathis, Gnetum, and red oaks, a light shaded diamond Akebia quinata Sinofranchetia chinensis Menispermum canadense Nandina domestica includes only the total period of active pollen tube growth, whereas dark Berberis120 cretica Berberis vulgaris foetida ED Thalictrum dipterocarpum diamonds represent the total fertilization interval. Data and references are in Xanthorhiza simplicissima Nelumbo nucifera Platanus hybrida Opisthiolepis heterophylla Table S2. Tetracentron sinense Trochodendron130 aralioides Sabia japonica Pachysandra terminalis Sarcococca ruscifolia Gunnera tinctoria Myrothamnus moschata Liquidambar orientalis Corylopsis willmottiae Cercidiphyllum japonicum Daphniphyllum139 macropodum reducing the potential for diversification of pollen reception 0 51015 Minimum stigma-egg distance (mm) structures. To lengthen pollen tube pathways, developmental time must be increased (Agathis), at a cost of longer exposure of Fig. 4. Pollen tube pathway lengths of early-divergent angiosperms. Black pollen tubes to the environment. Gymnosperm progamic phase bars, minimum distance from stigma to nearest egg (in apical ovule) along evolution is thus severely constrained by the tight linkage of pollen tube pathway; gray bars, additional minimum distance to furthest egg developmental time and pollen tube pathway length (Fig. 5). (in basal-most ovule). Lack of a gray bar indicates either one ovule/carpel or all A critical step in early angiosperm prehistory was the origin of at same level. Dashed lines represent maximum pollen tube pathway lengths of gymnosperms, in which pollen typically germinates on the nucellus their unique pollen tube wall growth pattern. The callose wall (3 mm) or rarely on other parts of ovule or cone (9 mm). Taxa and traits are and plug preceded the origin of novelties such as true closed listed in Table S3 in same order. Major clades: AM, Amborella; NY, Nympha- carpels, solid styles, and deep multi-ovulate ovaries, as well as the eales plus Hydatellaceae; AU, ; MO, Monocots; CH, Chloran- evolution of extreme traits such as fertilization intervals as short thaceae; EM, Eumagnoliids; and ED, Eudicots. as 15 min, pollen tube pathway lengths as long as 500 mm, and pollen tube growth rates Ͼ1,000-fold faster than those of gymnosperms (Fig. 5). Angiosperm pollen tube wall innovations history both pollen tube metabolic rates and maternal provi- gave pollen tubes the capacity for rapid and long-distance sioning of pollen tubes evolved slowly. growth, increasing the evolutionary potential of both pollen tubes and the tissues they interact with. Progamic Phase and Angiosperm Diversity Because angiosperms lack the gymnosperm developmental During the Early Cretaceous angiosperms and ephedroids (Gn- constraint of slow and static pollen tube growth rates, develop- etales) were both diversifying in similar habitats at similar mental time and pollen tube pathway length became evolution- latitudes, but today the 65 species of Ephedra are confined to arily dissociated (55) in derived groups. Wheat and red oaks have semiarid habitats, whereas angiosperms have radiated into, and similarly short pollen tube pathways but fertilization intervals of speciated within, virtually every environment on earth (44, 45). 25 min and 12 months, respectively, whereas Amborella and some A host of key innovations have been proposed as correlates of the varieties of maize share 24-h fertilization intervals even though tremendous diversity and ecological success of angiosperms, and maize pollen tubes travel well over 200 times further (Fig. 5). virtually all involve structural or life history traits of the sporo- Ultimately, rapid and long-distance pollen tube growth enabled phyte (1, 5, 11, 46–53). However, based on the data presented the decoupling of the pollen reception apparatus from the here, male gametophyte (pollen tube) innovations that evolved egg-bearing apparatus. Released from this functional constraint, early in angiosperm history were ‘‘developmental enablers’’ (54) carpels were able to diversify in size, facilitating the coevolution for some of the most important reproductive traits of flowering of flower and pollinator form and probably also the evolution of plant sporophytes. fruit form and function. Thus, angiosperm fertilization biology Gymnosperm reproductive evolution is constrained because is distinguished not only by many novelties and extreme traits, their slow and relatively invariant pollen tube growth rates but also by much greater independence (modularity) of their impose a trade-off between pollen tube developmental time and developmental processes. pollen tube pathway length (Fig. 5). To shorten developmental Elevated reproductive trait diversity, and perhaps increased time, pollen must be received nearer the egg (Ephedra), thereby modularity as well (56), are strongly linked to the elevated

11262 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0800036105 Williams Downloaded by guest on September 26, 2021 taxonomic diversity of flowering plants. Rapid reproduction, due heart of the tremendous ‘‘reproductive flexibility and opportun- in large part to accelerated pollen tube growth rates, is a ism’’ that Stebbins (57) and others have described as the critical fundamental life history strategy shared by many of the most factor in angiosperm success. diverse herbaceous clades, such as grasses and asters (Fig. 5). Moreover, the great developmental flexibility of angiosperm pollen tubes expanded the potential for pollen competition and ACKNOWLEDGMENTS. I thank Tanguy Jaffre´(Noumea, ) and Andrew Ford (Atherton, ) for logistical support, T. Arias, A. Becker, maternal responses to its effects (48), in turn, speeding the A. Herget for laboratory assistance, P. Crane, A. Doust, T. S. Feild, W. E. evolution of prezygotic forms of mating systems and reproduc- Friedman, and P. Stevens for comments, and P. Diggle, S. McCormick, S. Russell, tive barriers. Gymnosperms generally lack strong prezygotic and A. Staehelin for discussions. This work was supported by University of barriers (12). Pollen tube growth rate innovations truly lie at the Tennessee and National Science Foundation Grant DEB 0640792.

1. Stebbins GL (1974) Flowering Plants: Evolution above the Species Level (Harvard Univ 30. Li HJ, Bacic A, Read SM (1999) Role of a callose synthase zymogen in regulating wall Press, Cambridge, MA). deposition in pollen tubes of Nicotiana alata Link et Otto. Planta 208:528–538. 2. Takhtajan A (1976) Origin and Early Evolution of Angiosperms, ed Beck CB (Columbia 31. Li Y-Q, Moscatellis A, Cai G, Cresti M (1997) Functional interactions among cytoskele- Univ Press, New York), pp 207–219. ton, membranes, and cell wall in the pollen tube of flowering plants. Int Rev Cytol 3. Doyle JA (1978) Origin of angiosperms. Ann Rev Ecol Syst 9:365–392. 176:133–199. 4. Favre-Ducharte M (1979) Time relations and sexual reproduction in Cichorium and 32. Kudlicka K, Brown RM (1997) Cellulose and callose biosynthesis in higher plants. 1. other angiosperms as compared with Archegoniates. Phytomorphology 29:166–178. Solubilization and separation of (133)- and (134)-beta-glucan synthase activities 5. Haig D, Westoby M (1989) Selective forces in the emergence of the seed habit. Biol J from mung bean. Plant Physiol 115:643–656. Linn Soc 38:215–238. 33. Herrmann MM, Pinto S, Kluth J, Wienand U, Lorbiecke R (2006) The PTII-like kinase 6. Donoghue MJ, Scheiner SM (1992) Ecology and Evolution of Plant Reproduction,ed ZmPtiIa from maize (Zea mays L.) co-localizes with callose at the plasma membrane of Wyatt R (Chapman and Hall, New York), pp 356–389. pollen and facilitates a competitive advantage to the male gametophyte. BMC Plant 7. Maheshwari P (1950) An Introduction to the Embryology of Angiosperms (McGraw– Biol 6:22. Hill, New York). 34. Nishikawa SI, Zinkl GM, Swanson RJ, Maruyama D, Preuss D (2005) Callose (beta-1,3 8. Jansen RK, et al. (2007) Analysis of 81 genes from 64 plastid genomes resolves glucan) is essential for Arabidopsis pollen wall patterning, but not tube growth. BMC relationships in angiosperms and identifies genome-scale evolutionary patterns. Proc Plant Biol 5:22. Natl Acad Sci USA 104:19369–19374. 35. Bosch M, Hepler PK (2006) Silencing of the tobacco pollen pectin methylesterase 9. Moore MJ, Bell CD, Soltis PS, Soltis DE (2007) Using plastid genome-scale data to resolve NtPPME1 results in retarded in vivo pollen tube growth. Planta 223:736–745. enigmatic relationships among basal angiosperms. Proc Natl Acad Sci USA 104:19363– 36. Dumas C, Knox RB (1983) Callose and determination of pistil viability and incompati- 19368. bility. Theor Appl Gen 67:1–10. 10. Saarela JM, et al. (2007) Hydatellaceae identified as a new branch near the base of the 37. Mulcahy GB Mulcahy DL (1983) Pollen: Biology and Implications for Plant Breeding, eds angiosperm phylogenetic . Nature 446:312–315. Mulcahy DL, Ottaviano E (Elsevier, Amsterdam), pp 29–33. 11. Endress PK (2001) The flowers in extant basal angiosperms and inferences on ancestral 38. Snow AA, Spira TP (1991) Pollen vigour and the potential for sexual selection in plants. flowers. Int J Plant Sci 162:1111–1140. Nature 352:796–797. 12. Willson MF, Burley N (1983) Mate Choice in Plants: Tactics, Mechanisms, and Conse- 39. Lord E (2000) Adhesion and cell movement during pollination: Cherchez la femme. quences (Princeton Univ Press, Princeton, NJ). Trends Plants Sci 5:368–373. 13. Friedman WE, Carmichael JS (1996) Double fertilization in Gnetales: Implications for 40. Owens JN, Catalano GL, Morris SJ, Aitkenchristie J (1995) The reproductive biology of understanding reproductive diversification among seed plants. Int J Plant Sci 157:S77– Kauri (Agathis australis). 2. Male gametes, fertilization, and cytoplasmic inheritance. S94. Int J Plant Sci 156:404–416. 14. Singh H (1978) Embryology of Gymnosperms (Borntraeger, Berlin). 41. Webb CO, Ackerly DD, Kembel SW (2007) Phylocom: Software for the Analysis of 15. Friedman WE (1990) Sexual reproduction in Ephedra nevadensis (Ephedraceae): Fur- Community Phylogenetic Structure and Trait Evolution, Version 3.41. Available at ther evidence of double fertilization in a non-flowering seed plant. Am J Bot 77:1582– http://www.phylodiversity.net/phylocom/. Accessed September 26, 2007. 1598. 42. Friis EM, Pedersen KR, Crane PR (2000) Reproductive structure and organization of 16. Burleigh JG, Mathews S (2007) Assessing systematic error in the inference of seed plant basal angiosperms from the early Cretaceous (Barremian or Aptian) of western Por- phylogeny. Int J Plant Sci 168:125–135. tugal. Int J Plant Sci 161:S169–S182. 17. Williams JH, Friedman WE (2004) The four-celled female gametophyte of Illicium 43. Feild TS, Arens NC, Doyle JA, Dawson TE, Donoghue MJ (2004) Dark and disturbed: A (; Austrobaileyales): Implications for understanding the origin and early new image of early angiosperm ecology. Paleobiology 30:82–107. evolution of monocots, eudicots, and eumagnoliids. Am J Bot 91:332–351. 44. Crane PR, Lidgard S (1989) Angiosperm diversification and paleolatitudinal gradients 18. Fernando DD, Lazzaro MD, Owens JN (2005) Growth and development of conifer pollen tubes. Sex Plant Reprod 18:149–162. in Cretaceous floristic diversity. Science 246:675–678. 19. Derksen J, Li YQ, Knuiman B, Geurts H (1999) The wall of Pinus sylvestris L. pollen tubes. 45. Rydin C, Pedersen KR, Friis EM (2004) On the evolutionary history of Ephedra: Creta- Protoplasma 208:26–36. ceous fossils and extant molecules. Proc Natl Acad Sci USA 101:16571–16576. 20. Hoekstra FA (1983) Pollen: Biology and Implications for Plant Breeding, eds Mulcahy 46. Ehrlich PR, Raven PH (1964) Butterflies and plants—A study in coevolution. Evolution DL, Ottaviano E (Elsevier Science, Amsterdam), pp 35–41. (Lawrence, Kans) 18:586–608. 21. Yatomi R, Nakamura S, Nakamura N (2002) Immunochemical and cytochemical detec- 47. Raven PH (1977) Suggestion concerning Cretaceous rise to dominance of angiosperms. tion of wall components of germinated pollen of gymnosperms. Grana 41:21–28. Evolution (Lawrence, Kans) 31:451–452. 22. El-Ghazaly G, Rowley J, Hesse M (1998) Polarity, aperture condition and germination 48. Mulcahy DL (1979) The rise of angiosperms: A genecological factor. Science 206:20–23. in pollen grains of Ephedra (Gnetales). Plant Syst Evol 213:217–231. 49. Bond WJ (1989) The tortoise and the hare—Ecology of angiosperm dominance and 23. Friis EM, et al. (2007) Phase-contrast X-ray microtomography links Cretaceous gymnosperm persistence. Biol J Linn Soc 36:227–249.

with Gnetales and Bennettitales. Nature 450:549–553. 50. Dilcher D (2000) Toward a new synthesis: Major evolutionary trends in the angiosperm EVOLUTION 24. Carlquist S (1996) Wood, bark, and stem anatomy of Gnetales: A summary. Int J Plant fossil record. Proc Natl Acad Sci USA 97:7030–7036. Sci 157:S58–S76. 51. Gorelick R (2001) Did insect pollination cause increased seed plant diversity? Biol J Linn 25. Rudall PJ, Bateman RM (2007) Developmental bases for key innovations in the seed- Soc 74:407–427. plant microgametophyte. Trends Plants Sci 12:317–326. 52. Verdu´M (2002) Age at maturity and diversification in woody angiosperms. Evolution 26. Li YQ, Chen F, Linskens HF, Cresti M (1994) Distribution of unesterified and esterified (Lawrence, Kans) 56:1352–1361. pectins in cell walls of pollen tubes of flowering plants. Sexual Plant Reprod 7:145–152. 53. Moles AT, et al. (2005) A brief history of seed size. Science 307:576–580. 27. Parre E, Geitmann A (2005) More than a leak sealant. The mechanical properties of 54. Donoghue MJ (2005) Key innovations, convergence, and success: Macroevolutionary callose in pollen tubes. Plant Physiol 137:274–286. lessons from plant phylogeny. Paleobiology 31:77–93. 28. Geitmann A, Steer M (2006) The architecture and properties of the pollen tube cell 55. Raff RA (1996) The Shape of Life (Univ of Chicago Press, Chicago). wall. Plant Cell Monogr 3:177–200. 56. Yang AS (2001) Modularity, evolvability, and adaptive radiations: A comparison of the 29. Schlu¨pmann H, Bacic A, Read SM (1994) Uridine-diphosphate glucose metabolism and hemi- and holometabolous insects. Evol Dev 3:59–72. callose synthesis in cultured pollen tubes of Nicotiana alata Link et Otto. Plant Physiol 57. Stebbins GL (1981) Why are there so many species of flowering plants? Bioscience 105:659–670. 31:573–576.

Williams PNAS ͉ August 12, 2008 ͉ vol. 105 ͉ no. 32 ͉ 11263 Downloaded by guest on September 26, 2021