The evolutionary ecology of vegetative dormancy in mature herbaceous perennial plants

Authors


*Correspondence author. E-mail: dormancy@gmail.com

Summary

1. I present an evolutionary ecology interpretation of vegetative dormancy in mature herbaceous perennials. This kind of vegetative dormancy has been noted for at least 40 years, but has only recently become a topic of study.

2. Vegetative dormancy may be considered in a life-history context. Both vegetative dormancy and mortality typically decrease with increasing size. Vegetative dormancy’s relationship to reproduction is more complex, because some species increase flowering and fruiting after dormancy while others do the opposite.

3. If vegetative dormancy is adaptive, then it is most likely a bet-hedging trait. Dormancy-prone plants are often long-lived, and in such organisms, bet-hedging traits should counter the effects of environmental stochasticity on adult survival. This adaptive context may vary by life span, because in shorter-lived plants, fitness is most sensitive to changes in reproduction rather than survival.

4. Vegetative dormancy could evolve if the costs of sprouting ever outweigh the benefits. The benefits of sprouting include: (i) photosynthesis and (ii) the opportunity to flower and reproduce. The costs include: (i) greater chance of herbivory, (ii) greater need for limiting nutrients, and (iii) greater maintenance costs. The many losses of photosynthesis among plants suggest that these benefits may not always outweigh the costs.

5. Vegetative dormancy may be an evolutionary step towards the loss of photosynthesis. Many non-photosynthetic plants acquire carbon from their mycorrhizal fungi. Many autotrophic, dormancy-prone plants also acquire some carbon from their mycorrhizal fungi. Further, non-photosynthetic plants often become dormant to an even greater extent than autotrophic, dormancy-prone plants.

6.Synthesis Vegetative dormancy often occurs in clades with non-photosynthetic, myco-heterotrophic plants, with implications for the evolution of traits involved in carbon nutrition. The links between vegetative dormancy, other life-history traits, mycorrhizas and the loss of photosynthesis should provide exciting directions for further research in plant evolutionary ecology. Particularly needed is an assessment of the physiology of vegetative dormancy, including whether the mycorrhiza is a carbon source in all dormancy-prone plant species. Equally important is a better understanding of the genetic relationships among photosynthesis, myco-heterotrophy and dormancy.

Introduction

The phenomenon hereafter referred to as ‘vegetative dormancy’ is a condition in which a herbaceous perennial plant does not sprout above ground for one or more growing seasons, living instead only underground (Lesica & Steele 1994). Vegetative dormancy is common and has been noted in at least 10 families and 52 species across the plant kingdom, in ferns, orchids, legumes, sedges and buttercups, among others (Table 1). Within populations, vegetative dormancy may be quite common with over half of some populations dormant in any given year (Table 1).

Table 1.   A list of noted dormancy-prone plant species, whether they are mixotrophic/myco-heterotrophic, whether they include or are closely related to non-photosynthetic plants, and the trophic status of their mycorrhizal fungal partners. Numbers under ‘Dormancy level’ are either noted lengths of dormancy in years, or percentage of the population dormant in any given year, whichever was presented in the reference. Blanks indicate that the level of dormancy was not measured, although the phenomenon was noted. Labels under the columns ‘Mycotrophy’ and ‘Loss of photosynthesis’ indicate whether myco-heterotrophy or mixotrophy have been observed and whether non-photosynthetic individuals have been observed within the species (S), the genus (G), or not at all (–). Labels under ‘Trophic status of mycorrhizal fungi’ indicate associations with fungi that are typically arbuscular mycorrhizal (AM), ectomycorrhizal (EM), saprotrophic (S), of unknown trophic status although their identity is known (U), or unknown (?)
Family and genusSpeciesDormancy levelDormancy referenceMycotrophyLoss of photosynthesisTrophic status of mycorrhizal fungi
  1. Labels under ‘Mycotrophy’ column derived from Abadie et al. (2006), Bidartondo et al. (2004) and Gebauer & Meyer (2003). Labels under ‘Trophic status of mycorrhizal fungi’ derived from Bidartondo et al. (2004), McCormick et al. (2004, in press), Rasmussen (1995), Selosse et al. (2004) and Shefferson et al. (2007).

Asclepiadaceae
Asclepiasmeadii Alexander et al. (1997)?AM
Asteraceae
Solidagomissouriensis1–10 YearsMorrow & Olfelt (2003)?AM
Caryophyllaceae
Silenespaldingii10%
1–6 Years
Lesica & Steele (1994)
Lesica & Crone (2007)
?AM
Cyperaceae
Bolboschoenuslaticarpus Hroudováet al. (2007)?AM
maritimus Hroudová & Zákravský (1995)?AM
planiculmis Hroudováet al. (2007)?AM
Scirpuslacustris Squires & van der Valk (1992)?AM
maritimus20–25 YearsSquires & van der Valk (1992)?AM
Fabaceae
Astragalusscaphoides10%Lesica & Steele (1994)?AM
Liliaceae
Calochortuslyallii18%Miller et al. (2004)?AM
macrocarpus26%Miller et al. (2004)? AM
Trilliumgrandiflorum Hanzawa & Kalisz (1993) AM
Lythraceae
Lythrumsalicaria Gilbert & Lee (1980)?AM
Ophioglossaceae
Botrychiumdissectum21%Montgomery (1990)?GAM
hesperium6%
4 Years
Lesica & Steele (1994)
Lesica & Ahlenslager (1996)
?GAM
paradoxum3 YearsLesica & Ahlenslager (1996)?GAM
watertonense1 YearLesica & Ahlenslager (1996)?GAM
Orchidaceae
Cephalantheradamasonium Abadie et al. (2006)SGSGEM, U
longifolia20.6%
3 Years
Shefferson et al. (2005a)
Kull (2002)
SSGEM, U
rubra2 YearsKull & Tuulik (1994)GSGEM, S, U
Cleistesbifaria30%Kéry et al. (2005)??
divaricata3 YearsGregg (1991)??
Coeloglossumviride1 YearWillems & Melser (1998)??
Corallorhizaodontorhiza85%McCormick et al. (in press)SGSGEM
Cypripediumacaule5 YearsGill (1989)?EM, U
andrewsii40.5%Shefferson (2006)??
calceolus4.8%Kull (1995)
Shefferson et al. (2005a)
?U
candidum33.8%Shefferson (2006)?U
macranthon var. rebunense T. Kawahara (pers. comm.)?U
parviflorum32.0%Shefferson et al. (2001)?EM, U
reginae12.3%Kéry & Gregg (2004)?EM, U
Dactylorhizasambucina1%Tamm (1972)GU
Epipactisalbensis11 YearsRydlo (1995)GG?
atrorubens3 YearsKull & Tuulik (1994)SSGEM, U
helleborine3 YearsLight & MacConnaill (1991)SSGEM, U
microphylla Abadie et al. (2006)GSGEM
Herminiummonorchis2 YearsWells et al. (1998)??
Isotriamedeoloides2 YearsMehrhoff (1989)??
Liparisliliifolia2 YearsWhigham & O’Neill (1991)?S
Listeraovata3%, 2 yearsTamm (1972)SGEM, U
Microstylismonophyllos Elveland (1993)??
Neotineaustulata54%Shefferson & Tali (2007)?U
Ophrysapifera2 YearsWells & Cox (1991)?U
sphegodes46%Hutchings (1987a)?U
Orchismascula17%
1 Year
Tamm (1972), Inghe & Tamm (1988)U
militaris8 YearsHutchings et al. (1998)?U
morio1 YearWells et al. (1998)?U
simia2 YearsWillems (1982)?U
Platantherabifolia3 YearsVakhrameeva & Denisova (1988)SEM, U
chlorantha3 YearsVakhrameeva & Denisova (1988)GS, U
Spiranthesspiralis1 YearWells (1967)?U
Ranunculaceae
Callianthemummiyabeanum2 YearsNishikawa et al. (2005)?AM

Vegetative dormancy in the sense presented in this study is a kind of dormancy that occurs in geophytes and has been referred to as ‘dormancy’, ‘prolonged dormancy’ and ‘adult whole-plant dormancy’ (Gill 1989; Lesica & Steele 1994; Shefferson et al. 2005a). The term ‘dormancy’ often refers to conditions in which metabolism slows or shuts down for some period as a response to cyclic or stochastic stress (Danks 1987; Dahms 1995). No such decline in metabolism has ever been recorded in vegetatively dormant plants, and there is doubt that it should exist (Lesica & Steele 1994). Instead, here, vegetative dormancy denotes a time during which the plant does not sprout. Consequently, vegetative dormancy can be seen as an extension of over-winter dormancy, in which buds form on the plant but fail to sprout (Kull 1995). Unlike such seasonal dormancy, vegetative dormancy is not cyclical and implies neither the cessation of meristem activity nor a predictable return to sprouting (Rohde & Bhalerao 2007).

Vegetative dormancy has been discussed in the literature for at least 40 years (Wells 1967; Rabotnov 1969). Much of this literature has focused on orchids, although it is unclear whether this is due to vegetative dormancy being unusually common in the Orchidaceae, or to an orchid bias among plant demographers. Earlier plant demographers have often mistaken vegetative dormancy for observer error (Gilbert & Lee 1980). Tamm (1948, 1972, 1991) noted gaps in sighting in his classic studies of population dynamics of several species of European meadow orchids, all of which can become vegetatively dormant, but attributed those gaps to observer error. Some authors have suggested that lack of sprouting is at least sometimes due to the actions of herbivores eating the buds prior to sprouting and/or observation (Tamm 1972). Others have postulated that vegetative dormancy may depend in some way on mycorrhizal interactions (Wells 1967; Gill 1989), although the exact nature of this relationship has rarely been defined. Even in recent studies of dormancy-prone plants (i.e. plant species that undergo vegetative dormancy), finding plants long thought to be dead has surprised plant population ecologists (Morrow & Olfelt 2003). Often, vegetative dormancy is merely noted in studies of other aspects of plant population biology and so detailed discussion on it is not added (e.g. Hanzawa & Kalisz 1993). However, vegetative dormancy may also be found more often in orchids because of a potentially real tendency for it to occur in that group, just as non-photosynthetic species and deceit-pollinated species are over-represented in this group (Leake 1994; Jersákováet al. 2006; Otero & Flanagan 2006).

Why should an herbaceous perennial plant become vegetatively dormant? In this review, I offer both micro- and macroevolutionary perspectives on vegetative dormancy for further research. I suggest and provide evidence for the following hypotheses: (i) vegetative dormancy is a life-history trait with both ecological and evolutionary relationships to fitness; (ii) vegetative dormancy is likely to vary with both age and expected life span; (iii) vegetative dormancy is probably a bet-hedging trait in response to stochastic environmental variation; and (iv) vegetative dormancy may be an evolutionary step towards the loss of photosynthesis and the development of a parasitic, myco-heterotrophic lifestyle.

Vegetative dormancy as life-history adaptation

Relationships with survival and fitness

Studies of the evolutionary dimensions of vegetative dormancy often focus on its relationships to survival. In long-lived organisms, fitness is most sensitive to changes in adult survival, making survival a proxy for fitness (Sæther & Bakke 2000), and dormancy-prone species are often long-lived. Perenniality and iteroparity are associated with decreased variance in reproductive success because long life spans are associated with greater overlap among generations (Goodman 1984). Further, high, stable survival rates provide organisms with a means of buffering against environmental stochasticity (Morris et al. 2008). Thus, whether vegetative dormancy is adaptive or maladaptive should depend on the nature of its relationship to survival.

Empirical evidence suggests that, accounting for size, vegetative dormancy may function adaptively by buffering plant life histories from stochasticity. In a classic field study of a population of the relatively short-lived orchid Ophrys sphegodes, vegetative dormancy was associated with higher mortality, suggesting a fitness cost to vegetative dormancy via survival (Hutchings 1987a,b). Likewise, in Cypripedium parviflorum, vegetative dormancy was associated with lower survival (Shefferson et al. 2003). However, increased size results in the availability of a greater resource base and the spreading of risk across a wider area (Mágori et al. 2003). Thus larger plants should exhibit higher, more stable survival (Bierzychudek 1982; Callaghan et al. 1992; Alpert & Stuefer 1997; Hutchings 1999) and should exhibit a lower tendency towards vegetative dormancy (Shefferson 2006). In the relatively long-lived Cypripedium system, both vegetative dormancy and survival vary by size, the former negatively and the latter positively, leading to a negative correlation where there is no real trade-off (Shefferson 2006). In one of the few experimental studies of vegetative dormancy conducted, Shefferson et al. (2005a) showed that vegetative dormancy appears to buffer survival from stress in the short term in two long-lived perennial orchids, Cypripedium calceolus and Cephalanthera longifolia. In that study, plants in two populations of the former species and three of the latter responded to defoliation with increased vegetative dormancy in the next year but no change in survival relative to controls. Further, in the dormancy-prone Trillium grandiflorum (Liliaceae), defoliation resulted in an increased probability of transitioning from flowering in the year of defoliation to vegetative dormancy in the next (Knight 2003). In both this and the preceding instances, I assume that defoliation has not extended to buds developing into future years’ sprouts, as might happen in some instances of herbivory (Tamm 1972). Thus, vegetative dormancy is an age-related response to stressful environmental conditions.

To be adaptive, vegetative dormancy must both result in greater overall fitness and be a trait that can be selected for. A study of the goldenrod Solidago missouriensis (Asteraceae) suggests that the variance in the tendency to become vegetatively dormant may be partially genetic in dormancy-prone taxa (Morrow & Olfelt 2003). With genetic variation present, greater overall fitness should be observed either as increased mean fitness or decreased variance in fitness in dormancy-prone individuals, and this variation should be transferred to the next generation. Since current evidence suggests that decreased variance in fitness is probably more likely than increased arithmetic mean fitness, vegetative dormancy should function as a bet-hedging trait and should maximize geometric mean fitness (Gillespie 1974, 1977; Slatkin 1974). This situation is possible if genotypes resulting in increased levels of vegetative dormancy are associated with milder declines in fitness during stressful periods than in less dormancy-prone genotypes (Seger & Brockman 1987; Philippi & Seger 1989). In this case, vegetative dormancy should evolve similarly to seed dormancy, dispersal and other adaptations to unpredictable environmental variation, all of which are linked by a common and arguably interchangeable body of theory (Venable & Brown 1988; Philippi & Seger 1989).

In nature, vegetative dormancy often varies with climatic and microsite factors. Common patterns include correlations between demographic measures of vegetative dormancy on the one hand, such as percentage of the population dormant or transition probability from flowering in one year to dormant in the next, and environmental variables on the other, such as seasonal or annual precipitation and temperature, and soil depth and moisture content, although the signs of these correlations vary among studies (Zákravský & Hroudová 1994; Shefferson et al. 2001; Kéry & Gregg 2004; Shefferson 2006; Hroudováet al. 2007). Parallel patterns in vegetative dormancy among nearby populations of both orchids and mariposa lilies reinforce the notion that climatic factors may drive dormancy (Miller et al. 2004; Shefferson et al. 2006; but see Shefferson & Tali 2007). However, tests of climatic correlations in such studies have often been simple, and the highly correlated nature of climatic variables has prevented the isolation of the specific factors driving dormancy in these systems. More complex modelling of relationships between complex climatic variables, vegetative dormancy and survival are necessary to flesh this out, as is in situ experimentation. A PCA-style analysis of a large suite of environmental variables across many dormancy-prone taxa may offer the most comprehensive answer to the question of what triggers this phenomenon.

Relationships with life span

Vegetative dormancy’s relationships to survival and other life-history traits should differ between long-lived and short-lived perennials. In short-lived plants that undergo vegetative dormancy, the relatively greater importance of reproduction than of adult survival to fitness should result in greater impacts on reproduction than expected in the case of long-lived plants. In the relatively short-lived Silene spaldingii (Caryophyllaceae), vegetative dormancy in one year resulted in a higher probability of flowering in the next year than in plants that had sprouted but not flowered in the first year, suggesting that vegetative dormancy may result in greater benefits to reproduction (Lesica & Crone 2007). Further, in an observational assessment of vegetative dormancy and survival trends among multiple populations of the short-lived orchid Neotinea ustulata, survival was most clearly affected by climatic conditions while vegetative dormancy appeared to fluctuate independently according to microsite conditions (Shefferson & Tali 2007). These trends contrast with those of long-lived, dormancy-prone plants, such as Cypripedium species, where dormancy varies in parallel with climatic factors, while survival varies independently among populations (Shefferson 2006). Although further work with other dormancy-prone species is required for generalization, these studies nonetheless suggest that life span may indeed be key to understanding basic life-history relationships between vegetative dormancy and other traits.

Vegetative dormancy’s relationship to life span suggests that it may have a relationship to whole-organism senescence (i.e. ageing). At the whole-organism level, senescence refers to physiological deterioration after reproductive age has been reached, and is typically observed as increased age-specific mortality beyond a given age (Finch 1990). Although some argue that true senescence is caused by internal factors alone (Roach 1993), the actual trigger for senescence may be at least partially external in organisms in which size, rather than age, provides population structure. Long-lived plants generally exhibit life histories suggesting an ‘escape’ from senescence (Silvertown et al. 2001), and sometimes even negative senescence (Vaupel et al. 2004), in which age-specific mortality continually declines with age. In such plants, this may result from indeterminate growth and clonality (Vaupel et al. 2004; Rose et al. 2007), although it may be impossible to distinguish these patterns from the mortality plateaus predicted by theoretical models of senescence (Hamilton 1966; Mueller & Rose 1996). If senescence occurs in these plants, then vegetative dormancy may increase with age during senescence because of deteriorating physiological quality decreasing the ability of the organism to handle environmental stress. This may be particularly true in Ophrys sphegodes (Orchidaceae), in which dormancy and mortality are positively correlated (Hutchings 1987b). Thus, we may find that vegetative dormancy is highly likely early in life, becomes less likely as age and size increase into and through maturity, and finally increases again as the plant becomes senescent.

Relationships with reproduction

Both positive and negative relationships have been noted between vegetative dormancy and reproduction. These relationships also appear contingent on size, primarily in relation to whether the plant attains critical threshold sizes for reproduction (Lacey 1986). In the wild, dormancy-prone plants may reproduce to a lower extent than they are capable of, because such plants are often long-lived and delay reproductive events (Bierzychudek 1982). If a plant is forced to reproduce beyond its normal level, as would happen in an experimental pollination study, reproductive trade-offs may be more easily observed because of the increased demand for resources (Bell & Kofopanou 1986). In an observational assessment of life-history strategies in the orchid Cypripedium parviflorum, flowering appears relatively cheap, with flowering plants more likely to continue flowering in the next year than vegetatively sprouting or dormant individuals (Shefferson et al. 2003). However, the evidence also suggests that flowering is conditioned on size such that as size increases, so does flowering (Shefferson 2006). Likewise, fruiting also appears to bear little cost in this system, but nonetheless increases with size (Shefferson & Simms 2007). Fruiting costs to sprouting were observed in an experimental study of Cypripedium acaule (Orchidaceae), where a pollination treatment led to increased vegetative dormancy in the next year relative to controls (Primack & Stacy 1998), reinforcing the benefits of the experimental approach over the observational census. Such reproductive costs are likely to be both age- and size-dependent, with small, young plants most likely to exhibit them while older, larger plants are able to reproduce more and so are also more likely to mask reproductive costs via the greater variance in flowering and/or fruiting.

Given the inconsistent relationship between vegetative dormancy and reproduction across studies, one may question whether a direct relationship even exists. Life-history traits that trade-off against one another may nonetheless exhibit positive phenotypic correlations due to both variability in resource acquisition and complex allocation pathways among life-history traits (van Noordwijk & de Jong 1986; de Jong & van Noordwijk 1992). Observational studies are particularly prone to such problems, but even experimental studies are not immune because of the difficulty in controlling all possible indirect interactions among life-history traits (Reznick 1985; Bell & Kofopanou 1986). Thus, the most common method used to infer life-history relationships in vegetative dormancy, namely the assessment of correlations among demographic rates in wild populations, is not enough to infer these relationships. A more experimental approach, involving manipulation of life-history traits via pollination, defoliation or other means, as well as breeding designs that can parse out the intrinsic, genetic from the extrinsic, environmental influences on vegetative dormancy rates may result in stronger inferences. Model or weedy plants known to undergo vegetative dormancy such as Silene spaldingii (Caryophyllaceae) (Lesica & Crone 2007) and Lythrum salicaria (Lythraceae) (Gilbert & Lee 1980) may offer the best systems for such efforts.

The costs and benefits of sprouting: a macroevolutionary perspective

The question of why a plant should evolve vegetative dormancy may be considered via an assessment of the costs and benefits of sprouting. All else being equal, the main benefits of sprouting are twofold: (i) the production of energy via photosynthesis and (ii) sexual reproduction via flowering. A plant cannot do either in a growing season during which it is vegetatively dormant. However, there are costs to sprouting: (i) a greater chance of herbivory because of the greater visibility of a sprouting plant, (ii) increased need for limiting nutrients to grow above-ground tissue, and (iii) greater maintenance costs to support above-ground tissue in addition to below-ground tissue. For vegetative dormancy to be adaptive, some of the time the costs of sprouting must overcome the benefits.

Non-photosynthetic plants provide an interesting perspective on vegetative dormancy in a cost–benefit framework. Non-photosynthetic plants are plants in which photosynthesis has been evolutionarily lost (Leake 1994; Wolfe & de Pamphilis 1998), although non-functional photosystems still often exist (Archibald & Keeling 2002; Barbrook et al. 2006). Approximately, 400 species of non-photosynthetic plants are myco-heterotrophic, meaning that the primary source of carbon nutrition in these plants is a mycorrhizal fungus rather than photosynthesis (Leake 1994), while the remainder are dicots that directly parasitize plants (de Pamphilis et al. 1997; Bungard 2004). The breakdown of the typically mutualistic nature of the mycorrhiza in myco-heterotrophic plants is common and has evolved independently numerous times across the plant kingdom (Leake 1994; Taylor & Bruns 1997; Bidartondo 2005). Because non-photosynthetic plants acquire their energy via mycorrhizal interactions, one of the main benefits of sprouting no longer applies – the production of energy via photosynthesis. Such plants often only sprout to flower and reproduce, and annual sprouting rates can be as low as 15% (McCormick et al. in press). Further, competition for limiting nutrients may be muted by mycorrhizal interactions (Chiariello et al. 1982; Fitter et al. 1998; Brooker et al. 2008).

I propose that vegetative dormancy may be an intermediate phase along the evolutionary path to the loss of photosynthesis and the evolution of a myco-heterotrophic lifestyle. Dormancy-prone plants have several important similarities to non-photosynthetic, myco-heterotrophic plants. First, they appear to be mycorrhizal specialists in that they are often both obligately mycorrhizal and exhibit high specificity towards their fungal hosts (Shefferson et al. 2005b, 2007). Non-photosynthetic plants are generally extreme specialists on their fungal hosts (Taylor & Bruns 1997; Bidartondo 2005; Ogura-Tsujita et al. 2009), and this specialization contrasts greatly with the generalism often noted in the breadth of mycorrhizal fungi associating with most photosynthetic plants (Molina et al. 1992; Taylor et al. 2002). Secondly, many potentially dormancy-prone plant species show isotopic evidence of acquiring carbon and nitrogen from ectomycorrhizal fungi, a phenomenon known as ‘mixotrophy’ (Bidartondo et al. 2004; Julou et al. 2005). Most plants are mycorrhizal, with orchids thought to be obligately so in the wild, and fungal carbon is an important and often dynamic factor in such symbioses (Dearnaley 2007; van der Heijden et al. 2008; Hynson et al. 2009). Thirdly, because vegetatively dormant plants do not photosynthesize, they may already be facultatively myco-heterotrophic. This may be particularly true of dormancy-prone orchids, which begin life as myco-heterotrophic protocorms and seedlings and so have purely myco-heterotrophic periods in ontogeny (Rasmussen 1995).

If vegetative dormancy is an evolutionary step towards the loss of photosynthesis and the transition to myco-heterotrophy, then non-photosynthetic, myco-heterotrophic species may be closely related to photosynthetic, dormancy-prone species more often than expected by chance. This pattern may occur if the common ancestor of both such species was photosynthetic and dormancy-prone, but occasionally occurred in a non-photosynthetic, myco-heterotrophic form (e.g. via mutations at major loci coding for photosynthesis and mycorrhizal symbiosis). If some of these non-photosynthetic individuals became reproductively isolated from the rest of the population, either through dispersal away from the population or via the establishment of new flowering times based on altered nutritional dynamics, then a new, non-photosynthetic species may arise. This should be particularly evident in young clades, although in older clades such phylogenetic patterns may be obscured by other evolutionary events.

Several lines of evidence support this hypothesis. First, although data are limited, non-photosynthetic, myco-heterotrophic plants also may not sprout every growing season, and so can be said to undergo vegetative dormancy as well (McCormick et al. in press). Secondly, many genera of plants that include non-photosynthetic species also include autotrophic, dormancy-prone species. Examples include the fern genus Botrychium, which includes the non-photosynthetic Botrychium lunaria (Schmid & Oberwinkler 1994) as well as several dormancy-prone species. Thirdly, some clades with dormancy-prone plants include not only non-photosynthetic, myco-heterotrophic species, but also predominantly myco-heterotrophic species that have lost most but not all of their photosynthetic ability, as in the orchid genus Limodorum (Girlanda et al. 2006; Fig. 1). Fourthly, some dormancy-prone species occasionally occur as ‘albino mutants’ in which photosynthesis no longer occurs, including Cephalanthera longifolia (Abadie et al. 2006). Fifthly, some of the most dormancy-prone photosynthetic plant species associate with potentially ectomycorrhizal fungi, which would make excellent carbon providers for a myco-heterotrophic lifestyle (Bruns et al. 2002). An example is Cypripedium acaule, the pink lady’s slipper orchid, which may become vegetatively dormant for 20 consecutive years or more (Gill 1996), and associates at least some of the time with ectomycorrhizal members of the basidiomycete genus Russula (Shefferson et al. 2005b). Lastly, photosynthetic plants belonging to clades in which some species are non-photosynthetic and myco-heterotrophic also show evidence of mixotrophy, in which at least some carbon is derived from mycorrhizal sources (Abadie et al. 2006; Tedersoo et al. 2007) (Table 1, Fig. 1).

Figure 1.

 Occurrence of (a) vegetative dormancy, (b) myco-heterotrophy and mixotrophy, (c) the loss of photosynthesis and (d) trophic status of mycorrhizal fungi in the orchid tribe Neottieae sensu stricto (Orchidaceae) with Palmorchis trilobulata as an outgroup. The phylogeny was derived from Bateman et al. (2005) and Abadie et al. (2006). In (a), taxa known to undergo dormancy are marked with asterisks (*), taxa likely to be dormancy-prone although demographic data are lacking are marked with crosses (†), and those for which dormancy is unlikely although demographic data are lacking are not marked with any signs. Labels derived from Abadie et al. (2006), Girlanda et al. (2006) and Table 1.

A further expectation of the hypothesis that vegetative dormancy is a step in the evolutionary path to the loss of photosynthesis is that vegetative dormancy should be ancestral to the transition to a purely non-photosynthetic lifestyle in clades that include both photosynthetic, dormancy-prone and non-photosynthetic, myco-heterotrophic plants. The orchid family provides some evidence to support this prediction. Vegetative dormancy is found throughout the family and probably evolved early in its evolutionary history. In contrast, the loss of photosynthesis is thought to have occurred multiple times in recent evolutionary history, leading to most non-photosynthetic orchid taxa being most closely related to photosynthetic taxa (Bidartondo 2005). An example is the monophyletic tribe Neottieae (Fig. 1), which includes non-photosynthetic, myco-heterotrophic species such as Cephalanthera austinae (Taylor & Bruns 1997) and Neottia nidus-avis (McKendrick et al. 2002), as well as many photosynthetic species in which occasional non-photosynthetic ‘mutants’ arise (Abadie et al. 2006). Mixotrophy is likely to be ancestral to photoautotrophy in Neottieae (Abadie et al. 2006; Fig. 1), and the commonness of vegetative dormancy in the tribe suggests that it may have evolved at approximately the same time (Fig. 1).

Further directions

A relationship between vegetative dormancy and mycorrhizas has long been hypothesized (Wells 1967), though parasitism of the mycorrhizal fungus by the plant is a recent suggestion (Gill 1989; Montgomery 1990). The evidence suggests that an ‘inverse’ parasitism, in which a macroscopic organism parasitizes a microbe, may be at the heart of vegetative dormancy. However, a number of questions still need to be addressed to resolve this most basic of questions.

The most pressing need is undoubtedly that for further research on the nutritional and physiological mechanisms governing vegetative dormancy. For example, the form of carbon transferred between dormancy-prone plant and mycorrhizal fungus should be identified, as should the effect of mycorrhizal networks on carbon nutrition. If the mycorrhizal fungus is indeed a source of carbon for the dormant plant, then a dormant plant may act as an epi-parasite on other plants in the local community should the fungus also be ectomycorrhizal or arbuscular mycorrhizal (Gebauer & Meyer 2003). However, while dormant plants might obtain carbon from their mycorrhizal fungal partners, stored carbon in plant tissues may also be remobilized. Insufficient stored reserves and insufficient mycorrhizal carbon acquisition during dormancy may contribute to the heightened mortality of dormant Orchis ustulata (Hutchings 1987b), while in the sedge genus Bolboschoenus dormancy appears to include little or no mobilization of stored resources until sprouting occurs (Hroudová & Zákravský 1995; Hroudováet al. 2007). Such differences among taxa suggest the potential for the parallel evolution of dormancy in taxa with differing life-history strategies and growth forms, and suggest that dormancy may exhibit multiple physiologies among different taxa.

Related to plant nutrition and physiology, a pressing need exists for research on the extrinsic and intrinsic triggers of vegetative dormancy. If vegetative dormancy is simply bud dormancy that occurs across the entire plant, then we might expect both the environmental triggers and cues and the intrinsic hormonal regulatory mechanisms behind vegetative dormancy to be equivalent to those governing bud dormancy. Since vegetative dormancy occurs across the entire plant, it seems likely that emergent properties at the whole-organism level may alter these relationships. An experimental approach with different environmental treatments applied to a dormancy-prone plant would offer the best hope of isolating the mechanisms regulating dormancy.

The genetic and co-evolutionary dynamics of the mycorrhizal symbiosis in both non-photosynthetic and photosynthetic dormancy-prone plants needs to become the subject of greater investigation. In particular, do dormancy-prone plants provoke a defensive response from their mycorrhizal fungi? Do such plants select for fungi resistant to parasitism? If the ecological link between vegetative dormancy and mixotrophy is particularly tight, then what sorts of phylogenetic relationships among vegetative dormancy, microspermy, mixotrophy, myco-heterotrophy and the loss of photosynthesis should occur? More mechanistically, are the losses of photosynthesis and the acquisition of a purely myco-heterotrophic lifestyle governed by mutations in one pleiotropic gene or does a more complex relationship exist?

Finally, more demographic work needs to be conducted across the plant kingdom to identify how commonly vegetative dormancy occurs. In particular, focus on the Ericaceae would help to answer many evolutionary questions, because the Ericaceae include a high number of well-studied non-photosynthetic, myco-heterotrophic species within the monophyletic subfamily Monotropoideae (Bidartondo 2005), as well as non-photosynthetic, holoparasitic species such as Mitrastema yamamotoi (Barkman et al. 2004). A focus on the Adder’s tongue family, Ophioglossaceae, which includes both dormancy-prone taxa and non-photosynthetic, myco-heterotrophic taxa, may also help to clarify the subject.

Acknowledgements

I would like to thank Y. Abe, T. Hattori, T. Koizumi, J. Nagata, D. Roach, the Wildlife Ecology and Microbial Ecology laboratories at the Forestry and Forest Products Research Institute in Tsukuba, Japan, and the faculty and staff of the Odum School of Ecology, University of Georgia, USA, for logistical support during the writing of this manuscript. Special thanks to P. Alpert, D. Gill, T. Kartzinel, and two anonymous referees, who provided critical reviews of earlier drafts of this manuscript.

Ancillary