Introduction

Pinus is the most diverse genus of conifers with at least 111 extant species which are almost entirely confined to the Northern Hemisphere (Price et al. 1998; Richardson and Rundel 1998; Gernandt et al. 2005). Its natural distribution extends from the arctic and subarctic regions of North America and Eurasia to the subtropical and tropical regions of Central America and Asia; P. merkusii is the only species whose natural range extends south of the equator (Critchfield and Little 1966; Mirov 1967; Price et al. 1998; Richardson and Rundel 1998). Pinus evolved at least during the Early Cretaceous, about 130 million years ago, based on the oldest confirmed fossil record which is a permineralized seed cone of P. belgica (Alvin 1960; Millar 1998; Richardson and Rundel 1998). This genus likely evolved in the middle latitudes with the northeastern United States and Western Europe as candidates for its center of origin (Millar 1998). By the end of the Cretaceous, Pinus had diversified into two subgenera–Strobus (soft pines or haploxylon, characterized by one fibrovascular bundle in the needle) and Pinus (hard pines or diploxylon, with two fibrovascular bundles in the needles). Since then they have radiated across the northern continents and as the climate cooled down, most species have become well adapted to the temperate climates (Francini Corti 1958; Millar 1998; Richardson and Rundel 1998). Several reports suggest that a second center of Pinus diversity is the sub-tropical and tropical regions, specifically Mexico, which currently has at least 47 species (about 35 of which are endemic) and thus represents the largest record in terms of the number of species from any region of comparable size (Mirov 1967; Farjon and Styles 1997; Perry et al. 1998; Vargas-Mendoza et al. 2011). The life history of pines is considered to also have diverged early into two evolutionary strategies, probably in response to competition with the emerging angiosperms during the Cretaceous (Keeley and Zedler 1998; Keeley 2012). One is an adaptive shift toward abiotically stressful environments such as low nutrient soils and extreme cold or heat. The other is toward fire-prone landscapes with diverse fire regimes. These two adaptive modes generally follow the split between soft and hard pines (Keeley 2012).

Starting from seed, the time required for pines to reach sexual maturity (i.e., the stage of first regular formation of pollen and seed cones) varies considerably among species. This ranges from those with 10 years or less as in P. attenuata, P. clausa, P. contorta, P. greggii, P. pungens and P. rigida, to those with 30 years or more as in P. flexilis, P. edulis, P. resinosa and P. lambertiana but with the majority falling within these two extremes (Krugman and Jenkinson 1974; Strauss and Ledig 1985; Lanner 1998; Dvorak et al. 2000a b, c, d, e, f, g, h, i, j, k). Species in high fire frequency habitats usually begin their reproduction earlier than others (Keeley and Zedler 1998; Keeley 2012). Pines are monoecious with pollen (male) and seed (female) cones borne separately on the same tree. Pollen cones are usually located on older branches (tertiary or higher) in the lower part of the crown as lateral buds. Seed cones are usually located at the distal end of the more vigorous main branches in the upper part of the crown as subterminal lateral buds (Fletcher 1992; Owens 2006). These two types of cone are usually borne on separate shoots, although both types of cones occur on the same shoot in some species (Owens 2006). This spatial separation of reproductive structures is one of the mechanisms in pines that reduce self-fertilization by increasing the chance for seed cones to receive pollen from other trees.

The pine reproductive cycle involves many developmental stages including initiation and differentiation of pollen and seed cones, meiosis, gametogenesis, pollination, pollen germination, fertilization, embryogenesis, seed and cone maturation, cone drying and opening, and seed dispersal. What makes it even more interesting is that most of these stages occur for several weeks or months and together, these extend the entire reproductive process to 3 years in most pines. However, the process occurs in 4 years in a few species (Singh 1978). Therefore, on a tree or even on a branch, it is possible to observe buds or seed cones representing the first, second and/or third years of growth (Fig. 1). As a result, pine reproductive cycle is perhaps the most complicated of the land plants. Therefore, considering its evolutionary, ecological and economic values, it is necessary to have a thorough and broader understanding of the pine reproductive process and the variations that may occur among species and under different environments. Essential in studies of pine reproductive biology is to determine the length of the reproductive cycle of a species. Cone initiation is commonly used as the beginning of the pine reproductive cycle, while the end is usually at seed maturation and dispersal (Bramlett 1977; Singh 1978; Gifford and Foster 1989). This may involve the study of vegetative and reproductive bud initiation and development and the identification of various types of buds (e.g. long shoot buds, short shoot buds, pollen cone or seed cone buds) (Bramlett 1977). Understanding of the development of each bud type from initiation to mature structures is necessary so that previous years’ seed cones are not mistaken as part of the current year’s development (Figs. 1, 2) (Owens 2006). There may be subtle differences between immature and mature seed cones that may occur on the same branch that were initiated in different years. This is particularly important when trying to determine the factors affecting seed cone and seed production for a particular species. Therefore, knowledge of the developmental stages associated with these buds and cones has several practical applications including allowing one to induce or enhance cone production and confidently recognize the optimum times of pollen shedding and collection, seed-cone receptivity, pollination, and stages of ovule, seed, and seed cone development and maturity (Owens 2006). This information in turn enables better design of experiments and techniques to facilitate breeding and genetic improvement of pines.

Fig. 1
figure 1

A polycyclic long-shoot at the end of the growing season showing a mature third-year cone, seed cone and a lateral branch at the base. Above these are two polycyclic shoots (1 and 2 indicating that in each of 2 years the long-shoot terminal bud formed two complete cycles rather than one (mono cyclic). The lowermost bears two sets of mature seed cones, one set terminating each of the two cycles for that year (1 and 2) and the uppermost bearing two sets of 1-year-old seed cones, one set terminating each of the two cycles for that year (1 and 2). In one year each sequence is usually shorter than the ones formed before and each sequence is usually terminated a lateral seed cone. At the tip is a long-shoot terminal bud as shown in Fig. 2. It is difficult to trace back the phenology of such polycyclic shoots) (Source: Owens 2006. The reproductive biology of lodgepole pine. Forest Genetics Council of British Columbia, Extension Note 07, p9)

Full size image
Fig. 2
figure 2

Diagram of a dormant (monocyclic) long-shoot terminal bud from a sexually mature lodgepole pine (Pinus contorta var. latifolia) showing the time at which the axillary buds were initiated in the axils of some cataphylls during long-shoot bud development. Months on the left indicate the time periods when the type of axillary bud was initiated and structures on the right are the times when different types of axillary buds differentiated and developed before winter dormancy began. The column in the center of the long-shoot bud indicates the category of bud types usually found along the length of a long-shoot bud in a reproductively mature lodgepole pine. In juvenile trees the distal axillary buds would be long shoots rather than seed cones and in the lower portion all axillary buds would form dwarf shoots rather than pollen cones. In polycyclic shoots this entire sequence or part of it would be repeated in one growing season (Source: Owens 2006. The reproductive biology of lodgepole pine. Forest Genetics Council of British Columbia, Extension Note 07, p13)

Full size image

The reproductive process in pines is best understood when the developmental stages are correlated with season and climate, commonly called phenology. It is generally known that the timing of occurrence of the various reproductive stages in a species is influenced by several factors including climate (temperature, rainfall, day length, etc.), geography (latitude, elevation, aspect, etc.), and genetics (provenances, genotypes, etc.). Therefore, in any report for a species, there could be days or weeks variation in one or more of the developmental stages. It becomes even more confusing with the information from individuals of a species that are grown outside their natural distribution range and so this review will be limited to examples growing within their native range. Pines in north temperate regions are generally subjected to a distinct and relatively long dormant period followed by a relatively short growing season. Growth following the dormancy period generally results in synchronized male and female development. This is particularly important during pollination where dehiscence of the majority of the pollen cones occurs and forms a thick “cloud” that results in sufficient pollen deposition in receptive seed cones, and thus increases seed production. Pollen availability and seed-cone receptivity in north temperate pines usually occur within a relatively short period, which is 1–2 weeks on a tree and a month or so in a stand or seed orchard (Owens 2006). In contrast, most pines in the tropical region are subjected to longer growing season but shorter or indistinct dormant period. This is generally due to a warmer climate, which is characterized by a narrow range of temperature and day length fluctuations and so seasonality is mainly determined by the duration and severity of the dry season. As a result, pollination is asynchronous, i.e., it occurs earlier in the season in some cones or trees to about a month later in others (in the same species) and so the duration of pollination generally lasts for months. This likely produces sparse amount of pollen per day and so reproductive success in terms of pollination rate or seed set is likely low. According to Critchfield (1966), pollen shedding is under less rigid environmental control in tropical pines than those in the northern regions. This corroborates with the reports on the continuous and long duration of pollination in tropical pines like P. merkusii (Pousujja et al. 1986) and P. caribaea (Dvorak et al. 2000a), and occurrence of multiple flushes of cone production in P. tecunumanii (Dvorak et al. 2000k).

The reproductive cycle of Pinus contorta var. latifolia (lodgepole pine) is commonly used as a representative of temperate hard pine species because detailed information about its developmental processes is available (O’Reilly and Owens 1987, 1989; Owens 2006). The trees used in the study were located at a moderate elevation in the middle of the species range and thus a good representative for the group. The reproductive phenology of lodgepole pine varies with geographical distribution (O’Reilly and Owens 1987), but the developmental sequence and associated details are generally similar and so a single general description of the reproductive cycle is acceptable for other north temperate hard pines, and thus is used in this review. For temperate soft pines, P. monticola (western white pine) is used here as a representative also because detailed information on its developmental stages is available (Owens 2004; Owens and Fernando 2007). Detailed information on the reproductive biology of the temperate pines from the southern United States is also available such as P. taeda (loblolly pine), which is the most ecologically and economically important tree in the southern U.S. (and perhaps, the entire country). It is a good representative for the group and various aspects of its reproductive biology are also known (Bramlett 1977; Greenwood 1980, 1986; Skinner 1992). Also, its reproductive cycle is very similar to lodgepole pine but the dormant periods are shorter (Skinner 1992). For pines growing in the tropics, generally only information on the timing of some broad processes like pollination and cone maturation and collection, and seed production are available. Therefore, there is limited information on the developmental stages involved in the reproductive process of pines from the tropics. Examples of information that are lacking include timing of pollen and seed cone initiation and early development, stages of the male and female gametophyte before, during and after pollination, duration of seed-cone receptivity, and stages and duration related to dormancy at first and second years of growth. There is also not one species that stands out from the literature as a representative, particularly one in which the process has been examined in detail. Pinus merkusii is the only pine species native to the southern hemisphere, but information on its reproductive biology also only pertains to broad processes, and the same is true for the Mexican and Central American pines (see Dvorak et al. 2000a,b, c, d, e, f, g, h, i, j, k). Therefore, with only limited data available for pines from the tropics, information from several species has to be included. In this review, the term “tropical pines” is taken to include all pines as they occur naturally in the tropical region, including those at high elevational belts where sub-tropical (but not temperate) climatic conditions prevail. This review aims to highlight the importance of developmental analyses in our understanding of the pine reproductive process, particularly on how it occurs in species from regions with contrasting environments. However, since detailed developmental information is only available from a few species, a comprehensive analysis of this subject is currently not possible.

Pollen and seed cone initiation on year one

The overwhelming majority of pines undergo a reproductive cycle of about 27 months that takes three calendar years to complete. A reproductive cycle that is based on lodgepole pine is presented in Fig. 3. Typically, pollen and seed cone initiation occurs in the late summer of year one, pollination occurs in spring of year two, and gametophyte development resumes followed by fertilization, embryo and seed development which all occur in year three. The reproductive process starts with cone-bud initiation and ends with seed maturity and seed dispersal. In lodgepole pine, initiation of the reproductive structures occurs in the long shoot buds (LSB), which are formed in spring and contain a mix of vegetative and reproductive apices including the soon to develop pollen and seed cone buds (Fig. 2). Pollen cone buds are initiated in July or August (summer) and appear, based on longitudinal sections of LSBs, as mounds of meristematic cells that later differentiate by producing microsporophyll primordia along their flanks (Figs. 2, 3a). The seed cones are initiated about a month after pollen cone-bud initiation and differentiate also after about 1 month, usually in September (fall), by producing bract-scale primordia (Figs. 2, 3a). The cone scales arise in the axils of the bracts but remain fused to the bract, while the ovule primordia are borne on the adaxial surface of the scale, two ovules per scale. Pollen and seed cone buds of lodgepole pine over-winter at the microsporocyte and megasporocyte stages, respectively. These cells are diploid (also called microspore and megaspore mother cells, respectively) and represent the stage immediately before meiosis (Owens 2006). In temperate hard pines in the southern U.S. such as P. taeda, P. elliottii, P. glabra and P. palustris, pollen and seed cone primordia are also initiated in the summer (Greenwood 1980; Skinner 1992). Pollen and seed cones emerge from buds in late winter as a result of short winter dormancy period or warmer winter conditions (Dorman and Barber 1956). Since megasporocytes were observed by February 23 (Skinner 1992), which is just after winter dormancy, it is likely that these have been initiated before dormancy and thus, it is the stage where the reproductive buds over-wintered, as in lodgepole pine. In temperate soft pines like western white pine, the timing of pollen and seed cone initiation is generally similar to lodgepole and loblolly pines except that seed cone bud differentiation occurs in the 2 year of growth just before pollination (Owens 2004). One possible advantage of this is during seed cone induction experiments, which can be done in the same year as pollination. In temperate hard pines, experimental seed cone induction must be done in the summer of the year before pollination (Owens and Blake 1985; Owens 2006).

Fig. 3
figure 3

a–j. Diagram of the reproductive cycle in Pinus contorta var. latifolia (lodgepole pine) with emphasis on the two dormant stages which occur between years one and two, and years two and three. The first dormant period (left) could be extended in northern or high elevation zones and shortened or omitted in tropical regions altering their lengths. The length of the second dormant period (right) appears to be relatively constant in pines from contrasting environments (Figure modified from: Owens 2006. The reproductive biology of lodgepole pine. Forest Genetics Council of British Columbia, Extension Note 07, p 5)

Full size image

Lodgepole and loblolly pines generally have similar phenologies in their first year of growth. However, one major difference is that the winter dormant period is shorter in loblolly pine, beginning generally in late fall and ending January or February of the following year. If this shorter dormancy period is influenced by the relatively warmer climate, then it must be even shorter in tropical pines. Reproductive bud dormancy is also often correlated with the accumulation of food reserves and cone crop potential, which is well documented in some north temperate pine species (Owens 1995). The timing of pollen and seed cone initiation in Mexican and Central American pines is not known, as well as their correlation with lateral shoot elongation, stages of vegetative bud development and accumulation of food storage reserves in reproductive structures. The onset and duration of dormancy in year one and the associated developmental stages need to be established. Information on these aspects will be useful to better understand cone and seed crop production in sub-tropical and tropical pines.

Pollen and seed cone development on year two

Pollen and seed cones of lodgepole pine become morphologically distinct and emerge from the LSB in early spring and late spring to early summer of the second year of growth, respectively (Figs. 2, 3b). The microsporocytes in pollen cone buds undergo meiosis followed by mitosis to form five-cell mature pollen grains over a period of only a few weeks. As pollen matures in the spring, pollen cones dry, microsporangia split open, and pollen is shed almost completely within a day or two (Owens 2006). Seed-cone receptivity in pines varies but it generally ranges from 2–10 days (Singh 1978; Owens et al. 2005). The major factor that regulates pollen development and maturation is temperature (i.e., higher temperature hastens both developmental processes), whereas the major factors that affect pollen shedding are temperature and humidity. In contrast, the primary factor that regulates seed cone development and receptivity is temperature, whereas humidity has negligible effect. For optimal pollination, peak pollen release should coincide with peak seed-cone receptivity. However, a wet spring may not affect the rate of seed cone development but it can delay pollen release by a week or two. This condition often leads to dichogamy, which is the temporal separation of the sexes in terms of pollen release and seed-cone receptivity (Fig. 4). Therefore, dichogamy occurs when pollen release and receptivity do not coincide. If peak pollen release precedes peak seed-cone receptivity, it is called protandry. Protogyny occurs when peak seed-cone receptivity precedes peak pollen release, which may due to environmental effects, as described above. Also, pines in coastal California are generally protogynous (Critchfield 1980), perhaps, due to the high amount of moisture in the air that delays pollen release. Dichogamy has only been reported in some pine species but in most of these, they are protandrous (Ledig 1998). Dichogamy is typically a genetically controlled reproductive strategy in angiosperms that facilitates cross-pollination, which is likely also the case in pines. However, complete dichogamy or no overlap in pollen availability and ovule receptivity which occurs in some angiosperms has not been reported in pines.

Fig. 4
figure 4

Representation of the variation in temporal separation (Dichogamy) of pollen shedding and seed-cone receptivity that may occur in pines (Figure Courtesy of Dr. John N. Owens)

Full size image

Many of the developmental details described above are not available for pines from the tropics. However, if the assumption is that they are similar with north temperate pines, then the hot and humid climate in the tropics will likely create a less ideal condition for optimum pollination. Warm temperatures in the tropics will hasten both pollen and seed cone development but the high humidity may delay or prevent the occurrence of peak pollen release. This will likely allow pollen and seed cone initiation to occur most of the year but just a few each day and with each cone having its own “developmental” cycle, shedding of pollen and seed-cone receptivity will likely not coincide. As a result, asynchronous pollen release and seed-cone receptivity will produce sparse amounts of pollen per day and very low seed set. Based on this, pollination success in tropical pines may be lower than north temperate pines where thick pollen clouds are common and often visible over forests or seed orchards. Pinus kesiya growing in its natural habitat in sub-tropical regions of Thailand consistently produced a few (<10) seeds per cone and this has been attributed to low pollination success (JN Owens, personal communication). The low pollination success was due to individual seed cones being receptive for the usual 1 week but the soggy pollen cones were either not shedding pollen or shedding pollen in a very small amount over several weeks. There was no synchronization of pollen cone dehiscence and seed-cone receptivity (JN Owens, personal communication). Based on the generally longer duration of pollination in Mexican and Central American pines, which ranges from 1–4 months (Table 1), it indicates that pollen cone dehiscence is asynchronous. However, it remains to be seen how much of this is due to timing of cone initiation and development, humidity, rain or other climatic factors. Whereas seed cone or ovule receptivity occurs only for a few days on a tree in north temperate pines (Owens et al. 2005), it will also be interesting to determine its duration in sub-tropical and tropical pines. This information will help better define the optimum timing of artificial or supplementary pollinations, as well as its correlation with pollen shedding and thus the strength of dichogamy.

Table 1 Timing of pollination and cone maturation in representative Mexican and Central American pines
Full size table

Pollen and ovule development in relation to pollination (year two)

At pollination, the developing ovules in lodgepole pine contain megasporocytes (one megasporocyte per ovule) which undergo meiosis and it is about this time that they are receptive to pollination (Fig. 3b). It will be interesting to know if tropical pines are also pollinated at this developmental stage or if variation exists. Pollination also coincides with the appearance of pollination drop, which is a liquid secretion that emanates from the ovule. Pines have saccate pollen grains that are captured by the pollination drops at the micropylar end of the inverted ovule; the sacci of the pollen allows it to float up the pollination drop and together with the recession of the drop, it gets inside of the ovule (Singh 1978; Gifford and Foster 1989). The megasporocytes undergo meiosis, which results in the formation of four haploid megaspores per megasporocyte, but only one survives and it is called the functional megaspore (Fig. 3c, d). This becomes the female gametophyte that starts out by undergoing five free nuclear divisions producing 16 nuclei (Singh 1978). This is the stage where the ovule remains for the rest of the year (Fig. 3e). Therefore, the ovules (and the seed cone) are not fully developed at pollination but the structures necessary to capture and house the pollen grains are in place (Owens 2004, 2006). Generally, three to five pollen grains could fit in the pollen chamber in an ovule and this number is considered optimum for seed cone retention (Owens et al. 2005; Owens and Fernando 2007). It has been considered that at least 20 % of the fertile ovules need to be pollinated to prevent cone drop (Sarvas 1962; Sweet 1973; Owens et al. 2005; Owens and Fernando 2007). It has also been suggested that pollen grains or developing pollen tubes have a substance that when recognized by the ovules, triggers the development of the functional megaspores into female gametophytes by undergoing several rounds of free nuclear divisions (McWilliam 1959; Sarvas 1962; Sweet 1973). It will be interesting to identify the substance or molecule from the pollen or pollen tube that is responsible for stimulating ovule development, particularly since it is considered to be unlikely any of the classical plant hormones (Sweet 1973). Nevertheless, as shown in lodgepole and loblolly pines, unpollinated ovules do not continue development and if their number is greater than 80 %, the entire seed cone will abort and eventually falls off. This appears to be a mechanism that ensures food resources are not wasted in keeping a seed cone that is not sufficiently pollinated and thus, has lower potential in contributing progenies for the next generation.

Whereas pollination in north temperate pines occurs 1 to 2 weeks around June, temperate pines from the southern U.S. shed their pollen earlier. In fact, one of the earliest of the southern U.S. pines is P. clausa, which sheds its pollen in late December (Dorman and Barber 1956). In P. radiata, one of the earliest of the western temperate pines, pollen shedding starts in late January (Duffield 1953). It is also interesting to note that the timing of pollination in the southern U.S. pine area extending from Texas to Virginia is strongly correlated with their latitudinal distribution (Zobel and Goddard 1954; Dorman and Barber 1956; Krugman and Jenkinson 1974). Generally, pollination in P. elliottii is January to February, P. palustris and P. taeda are February to April, and P. echinata is March to April (Dorman and Barber 1956; Baker and Langdon 1990; Skinner 1992). In comparison with north temperate pines, the duration of pollination in these species is longer which indicates asynchrony in pollen shedding and that the dormancy period is shorter.

For many pines in the tropical region, peak pollination occurs earlier in the season than those in the southern and western U.S. (Critchfield 1966). Pollination coincides with the dry season in most regions of Mexico and Central America (Farjon and Styles 1997; Dvorak et al. 2000a, b, c, d, e, f, g, h, i, j, k). In Pinus caribaea, which represents one of the earliest of the Central American pines to shed pollen, pollination is late October to early November in eastern Nicaragua (Critchfield 1966) and appears to extend until April (Dvorak et al. 2000a). Peak pollination occurs in October for P. ayacahuite (Little 1962), November to December for P. pringlei (Shaw 1909), January in P. kesiya (Fernando and Owens, personal observations), and March for P. durangensis (Martinez 1948). Report that pollen shed in some tropical pines occurs almost throughout or most part of the year is not uncommon, as in the case for P. merkusii (Pousujja et al. 1986) and P. caribaea (Dvorak et al. 2000a). Multiple flushes of cone production have also been reported for Mexican and Central American pines including P. tecunumanii (Dvorak et al. 2000k). A summary of the timing of pollination and cone maturation in several Mexican and Central American pines are presented in Table 1. Pollen shed is not well synchronized in the pines from these areas based on the long durations of pollination and as a consequence, small amounts of pollen may be shed at any one time (since only a small proportion of pollen cones shed their pollen). The thick pollen cloud that is produced by north temperate pines which is necessary for good seed set is likely absent in tropical pines. In P. caribaea, seed production is generally low and this has been considered as a consequence of poor synchronization between pollen release and seed-cone receptivity (Dvorak et al. 2000a). Based on Table 1, pollination in Mexican and Central American pines that undergo the typical reproductive cycle occurs much earlier compared to the north temperate pines. Interestingly, the development of their seed cones and ovules appears to still proceed slowly over the entire year two and extends through year three since pollination to cone maturation is mostly 22–29 months. This is generally similar to those in lodgepole pine except for the two extremes, i.e., P. maximinoi and P. greggii which have 12 and 33 months, respectively (Dvorak et al. 2000c, d, e, f, g). This suggests that seed cone and ovule development is generally not significantly facilitated by the warmer conditions in the tropics.

After pollination in north temperate pines, the seed cone scales close, the entire seed cone bends down, and the cones undergo slow development. During their first year winter dormancy, the ovules are only in the form of small primordia with no internal differentiation whereas in year two they have become well differentiated ovules, except that their female gametophytes are not yet fully formed. In the ovules, pollen grains germinate within 1 to 2 weeks after pollination and usually several pollen tubes grow to about one-third or half of the length of the nucellus, and then they undergo dormancy (Fig. 3e). The female gametophytes are still at free nuclear stage at this point (32 nuclei) and this is where the ovules undergo winter dormancy for the second time (Gifford and Foster 1989). Dormancy at this stage may be necessary for the accumulation of food reserves which are crucial for the formation and maturation of the characteristically large female gametophytes in spring. Where temperatures are warmer as in the southern U.S., seed cone development may continue, although very slowly, through the winter as in P. elliottii in Florida, and P. attenuata and P. ponderosa at low elevations in the Sierra Nevada (Dorman and Barber 1956), and P. taeda in North Carolina (Skinner 1992). There is no report on specific seed cone or ovule stages after pollination and before seed maturation in Mexican and Central American pines. Although it is logical to assume that the warmer conditions where the pines are growing will allow development to continue, it seems that whatever process that occurs during this time has no major effect on the overall duration since the period from pollination to cone maturation still generally happens within the same time frame as the north temperate pines. This indicates that freezing temperature is not the primary reason for the characteristically slow female gametophyte development in north temperate pines. Slow female gametophyte development occurs in pines from both temperate and tropical regions.

As indicated above for north temperate pines, the pollen that has already germinated in the ovule also undergoes dormancy. This dormancy starts about a month after pollination in year two and ends just before fertilization in year three. However, the factors that regulate this dormancy or pause in pollen tube growth have not been broadly considered prior to this review. These factors are low or freezing temperature, drought or water stress, and lack of signal from the female gametophyte. Dormancy may also allow the pollen tube to accumulate the needed food reserves to help sustain their growth and formation of sperm at springtime. It is interesting to note that although the pause in pollen tube growth coincides with the winter season in temperate region, it also occurs in tropical pines (Robbins 1983). This is also supported by the facts that the duration of the reproductive process in temperate and tropical pines is generally the same. This suggests that the low or freezing temperature associated with the north temperate climate is not likely the main reason for the delay. Limited water availability during summer or winter in temperate species and dry season in tropical pines are also unlikely to be the main factor since the onset of spring or wet season does not immediately cause resumption of pollen tube growth; wet conditions also occur between summer and winter but this does not induce pollen tube growth either. Germinated pollen in the nucellus of the ovule is known to contain abundant starch grains (Singh 1978; Fernando et al. 2005) which indicates availability of food reserves. Therefore, it is more likely that since the female gametophyte has not formed the eggs at pollination and throughout the rest of year two, that there is no point for the pollen tube to continue its growth and so sperm formation and delivery have to be delayed until the egg is formed, which occurs in year three. On the other hand, as already indicated, the presence of germinated pollen in the nucellus is necessary for seed cone development since female gametophytes in ovules with no germinating pollen do not continue development (McWilliams 1959; Sarvas 1962; Sweet 1973; Owens et al. 2005; Fernando et al. 2005). This indicates that the pollen tube provides a signal for the female gametophyte to continue development. Once fully formed, the female gametophyte (complete with archegonia with egg cells) may provide the signal to the germinated pollen to resume its development. Therefore, there appears to be some intricate signaling mechanisms that coordinate male and female development and in the process, it also acts as a resource conservation strategy. It appears that the slow reproductive process in pines is not directly constrained by temperature and perhaps, water availability, but rather, it seems to be a trait that is rooted deeply in their ontogeny.

Ovule and seed development on year three

Reproductive development in the following spring (1 year after pollination) is relatively rapid. The female gametophyte composed of 32 free nuclei undergoes several mitotic divisions to form 1000–2000 free nuclei (Singh 1978; Gifford and Foster 1989). The female gametophyte then undergoes cellularization (Fig. 3f) starting with the nuclei along the periphery of the enlarged functional megaspore cell. During later stage of cellularization, the archegonia are initiated from female gametophyte cells at the micropylar end of the ovule (Fig. 3g). Archegonia are each initially composed of neck cells and a large central cell, but the latter divides to produce the ventral canal and egg cells. Pine ovules generally produce two to seven archegonia (Lill 1976; Singh 1978; Gifford and Foster). The presence of multiple archegonia per ovule is an important feature of the mating system in pines and may explain how species of the genus could achieve high levels of genetic variation (Ledig 1998). This will be elaborated further below in relation to embryo formation.

Fertilization and polyembryony (year three)

After over-wintering in north temperate pines, ovule development resumes from free nuclear stage to formation of the archegonia and eventually, the egg. Pollen tube growth also resumes and it elongates toward the mature archegonium and it is during this stage that its generative nucleus divides to form two sperm cells. The pollen tube tip enters the archegonium via the neck cells, discharges its contents into the egg cytoplasm, and fertilization occurs (Fig. 3g). However, only one of the two sperms from a pollen tube participates in fertilization. Fertilization takes place in the spring or early summer about 12 months after pollination (in lodgepole pine) which is followed immediately by embryo formation and accompanied by rapid growth of the seed cone (Owens 2006; Owens and Fernando 2007). Although it has not been directly established, fertilization and the rest of the process described above for temperate pines proceed similarly in tropical pines in the third year of development. However, their phenology and rate of development likely varies.

The egg in each of the several archegonia per ovule in pines may be fertilized and each may develop into a proembryo (Singh 1978; Gifford and Foster 1989). The number of proembryos formed may be equal to or less than the number of archegonia in each ovule depending on how many eggs become fertilized. This process and stage of embryo formation is called simple polyembryony where each proembryo is sired by a pollen from the same or a different tree. Thus, the several archegonia in an ovule may be fertilized by the same and/or different male parents, resulting in proembryos that may be genetically different from each other. In pines and some other conifers, they undergo another stage of embryo formation, which is called cleavage polyembryony where each of the proembryos produced through simple polyembryony splits into four genetically identical early embryos (Fig. 3h). The selective advantage of cleavage polyembryony is not known. Nevertheless, this is followed by a process of intense competition and selection that results in all but one early embryo surviving and then finally, usually only one mature embryo per ovule (Fig. 3i). The embryo possessing the greatest growth activity because of its genotype gains an advantage due to its foremost position with respect to its competitors, and the rest are eliminated since practically always only one embryo remains in a mature seed (Buchholz 1926) (Fig. 3i). This developmental selection usually takes place in earlier stages of development so that by the time the largest embryo has begun to form its cotyledons, only small traces of the aborted embryos remain. This post-fertilization selection mechanism is considered to be primarily responsible in the maintenance of genetic diversity in pines, particularly since they appear to lack pre-pollination selection mechanism like those in angiosperms. Self-pollination leading to self-fertilization occurs in pines but it generally leads to reduced seed set as a result of selected abortion of selfed embryos during development—this mechanism is referred to as inbreeding depression. Inbreeding depression is well documented in various species of pine (see Remington et al. 2000; Williams 2007; Chancerel et al. 2013).

Seed development, maturity and dispersal (year three)

Loblolly pines growing in seed orchards in South Carolina had an average seed efficiency of about 50 % (Skinner 1992). Seed efficiency is the percentage of filled seed per cone divided by the seed potential or the number of fertile ovules per cone. Seed efficiency in lodgepole pine is similar to that of loblolly pine, whereas about 80 % for western white pines growing in seed orchards in their natural habitats in British Columbia, Canada (Owens 2004; Owens et al. 2005; Owens and Fernando 2007). This appears to be about as high as seed efficiencies can be expected to be in these pines indicating that the most important factor in reducing filled seed production is pollination success, i.e., low pollination success results in low seed set. Other factors also reduce seed production but most can be traced back to the degree of pollination. For example, in lodgepole pine, if too few ovules per cone are pollinated the seed cone aborts soon after pollination which may significantly reduce total seed cone and seed production (Owens et al. 2005). This is true for western white pine and loblolly pine, and likely the same for other species including the tropical pines. Whereas pollen shedding in southern U.S. pines follow a strong latitudinal pattern (Texas to Virginia), no strong relationship with latitude was found in term of their seed maturation, which generally occurs around the first 3 weeks of October (Dorman and Barber 1956). It appears that early pollination does not necessarily result in early seed maturation.

Seed cones and seeds of most pines mature relatively rapidly during the third year of growth. This occurs from late summer and fall in most species while it is during late winter in others. Seed cones turn brown as they mature and cone scales usually open to release the fully developed seeds (Fig. 3j). In some pines (e.g., P. muricata and P. radiata), the seeds mature during the fall, but the cones are not fully developed until late winter and thus cause a delay in seed dispersal (Burns and Honkala 1990; Young and Young 1992). In P. coulteri, P. sabiniana and P. torreyana, seeds are gradually dispersed from slowly-opening cones (Lanner 1998). In some species there is even a longer delay in seed dispersal, which could be several years, particularly in the case of strongly serotinous species like P. attenuata which could remain closed for 15–20 years (Vogl 1973). In general, serotiny is the retention of seeds in closed-cones for one or more years, but could open rapidly when exposed to high temperatures (Critchfield 1957). In species where seed retention occurs less than a year (e.g., P. oocarpa [in its southern geographic range], P. tecunumanii), they are called partially serotinous (Farjon and Styles 1997; Lanner 1998). Serotiny also varies in terms of its degree, i.e., in a few species (e.g., P. contorta var. latifolia, P. rigida), some populations are serotinous, whereas other populations are non-serotinous (Keeley and Zedler 1998c). This and other mechanisms to delay seed dispersal have a selective basis since they may reduce predation by limiting the time that seeds are on the ground prior to predictable rainfall. This is a trait that has been considered to have evolved in the hard pines (or subgenus Pinus) as an adaptation to fire (Keeley and Zedler 1998; Keeley 2012). Pines bearing serotinous cones will release their seeds when a facilitating fire occurs, which could be 5–10 years, but could be as long as 35–40 years, depending on the species and conditions (Mason 1932). The heat from fire or xeric sites is required to soften the resin that glued the cone scales together (Krugman and Jenkinson 1974; Keeley and Zedler 1998). Serotiny is also partly due to the formation of abundant woody tissues at the base of cone scales (Owens 2006). The California closed-cone pines (e.g., P. coulteri, P. muricata, and P. radiata) illustrate the different complexities of seed release. Examples of other closed-cone pines from Mexico and Central America include P. greggii, P. jaliscana, P. patula, and P. pringlei (Barnes and Styles 1983; Keeley and Zedler 1998). Most of the soft pines (subgenus Strobus) do not exhibit serotiny (Keeley 2012).

Three-year reproductive cycle

Whereas the majority of pines undergo a reproductive cycle that spans three calendar years, a few species undergo even longer reproductive cycle. The major difference is the time lapse between pollination and fertilization, which is about 2 years in the species that exhibit a reproductive cycle that spans four calendar years; as described in the previous section, this is about 1 year in most pines. Examples are sub-tropical and tropical pines such as P. leiophylla, P. maximartinezii, and P. pinea (Francini Corti 1958, 1962, 1969; Dalimore and Jackson 1966; Mirov 1967; Singh 1978; Donahue and Mar-Lopez 1995; Dvorak et al. 2000f; Mutke et al. 2005). Generally, the timing of the reproductive stages is as follows: Year One–pollen and seed cone buds are initiated toward the end of the calendar year; Year Two–pollen cones mature during the early part of the year, whereas the ovules in seed cones develop slower and so are at megasporocyte at pollination. Female meiosis (megasporogenesis) occurs not long after pollination and remains at functional megaspore stage for the rest of the year; Year Three–the megaspore develops slowly as it undergoes free nuclear divisions for most part of the year, while the pollen tube remains dormant in the nucellus throughout this time; and Year Four–the female gametophyte composed of free nuclei develops quickly into a multicellular structure in the early part of the year, form mature archegonia (around June), pollen tube growth resumes, and fertilization takes place. Within a couple of months, the embryos have formed their cotyledons and seeds have matured. The seeds are either shed right away or towards the latter part of the year. The species involved tend to grow at high elevations, northern sites or very dry sites and so the additional dormant period may be induced by drought or water stress. Although this lengthy reproductive process correlates with bigger female gametophytes (commonly known as the pine nuts) and very large seeds, it also occurs in some pines that exhibit the typical length of reproductive cycle (Singh 1978). Interestingly, judging on the stage of seed cone enlargement and external resin exudations, fertilization in P. maximartinezii may occur in year two (Donahue and Mar-Lopez 1995). This indicates that embryo development in this species may be the stage that gets extended for 1 year. This latter type of longer reproductive process has also been reported in non-Pinus genera such as Juniperus (Ottley 1909; Kotter 1931), Calitris (Baird 1953) and Chamaecyparis (Owens and Molder 1975). Considering that no detailed developmental analyses has been done recently on the three pine species, it will be helpful to confirm the exact stages that are involved and the variation that may exists so as to have a better understanding of this unusually extended form of pine reproductive cycle.

Summary

In general, pine reproductive cycle spans three calendar years. It has three growth periods separated by two dormant periods. The growth periods are: (a) pollen and seed cone bud initiation and early development that occur in year one; (b) pollination, pollen germination and pollen tube growth into the nucellus of the ovule, and early female gametophyte development that occur in year two; and (c) resumption of gametophyte development, fertilization, embryo and seed development, and seed maturation and dispersal that occur in year three. The two dormant periods occur, in broader terms, between initiation and pollination, and pollination and fertilization. The length of the first dormant period appears to vary with climatic conditions, whereas the second is relatively unaffected by climate. From cone initiation to seed maturity, it takes about 27 months to complete. Pollination to cone maturation could be about 16 months, whereas pollination to fertilization is about 13 months. These durations could be interpreted as different if one is not familiar with the entire reproductive process. These types of reports also create a problem when comparing species or provenances if one is not careful in distinguishing the reproductive stages. It is likely difficult for everyone to adhere to a common system since research aims vary and so it is suggested that reports should clearly indicate the stages being considered with an acknowledgment of the stages that have not been considered, which in this case is often the initiation stages in year one. In a few pines, the reproductive process spans four calendar years. The major difference is the extended delay between pollination and fertilization, which is 2 years, instead of one (or about 13 months) as in most pines. However, this may vary between species and thus, a re-examination is necessary.

Detailed developmental analysis of the reproductive process in tropical pines is needed. However, considering the diversity of tropical pines, it might be necessary to designate one species as a representative and focus the research effort on it. Although this might not be a popular option because it involves fundamental research and its significance may not be immediately appreciated, its long term contribution will be invaluable. On the other hand, choosing a model species is a complicated task in itself. In this review, some of the developmental stages and timing of their occurrences in tropical pines were deduced from generally limited information and in comparison with what is known from other species. Therefore, it is imperative that these have to be validated. In spite of this, it is clear that tropical pines exhibit longer pollination period due to asynchrony. However, although asynchrony of pollen release might not result in the huge quantity of pollen at one particular time like those that creates thick pollen clouds in north temperate species, such staggered pollen production over an extended period of time may increase the chances of cross-pollination. This is particularly important if there is also asynchrony in seed-cone receptivity. It will be interesting to determine the correlation of these to the genetic diversity of seeds produced through natural pollination. Therefore, it is hypothesized that a possible adaptive value of the asynchrony in tropical pines is in terms of capturing genetic diversity.

It has been suggested by Keeley and Zedler (1998) that the lengthy reproductive cycle of pines limits the evolution of their life-history, which is apparently a constraint to having an annual growth habit. They also pointed out that it is not known whether this long reproductive cycle is an irreversible trait that is rooted deeply in their ontogeny or if it is a trait maintained by current selection. Based on the information presented in this review, it appears that in species with the typical three-year reproductive process, the length is generally the same in pines from temperate and tropical regions. This indicates that the constraint is not directly imposed by their growing conditions, at least not directly due to temperature. Tropical pines may skip the first long dormant period associated with north temperate species because of the generally warmer conditions, but the reproductive process is still about as long. The dormant period between cone initiation and pollination can be facilitated in tropical pines, but the length of dormancy between pollination to fertilization appears to be the same. The adaptive significance of the long pause between pollination and fertilization is not known, but the pine perennial habit appears to be intertwined with at least this dormancy stage. Nevertheless, the characteristically slow pollen tube growth and female gametophyte development in pines appears to be more likely a feature that is rooted deeply in their ontogeny rather than due to any current selection pressure that they are experiencing. The slow pace of pollen tube growth in pines coupled with slow female gametophyte development is amazing in comparison to the seemingly instantaneous pace of the development of the corresponding structures in angiosperms (Williams 2012). The study of Pinus sexual reproduction (conifers or gymnosperms in general) offers the opportunity to probe deeper into the resilient reproductive strategy that characterizes the most successful non-angiosperm genus of seed plants. The long generation time of pines coupled with their lengthy reproductive cycles severely imposes limitations in breeding, genetic improvement, and functional genomics. However, a better understanding of its reproductive cycle, particularly the details of the developmental processes involved, may allow the development of research or management strategies to circumvent some of these limitations. It is hoped that this review will encourage more research in this area of plant biology.