Author for correspondence: Malcolm C. Press Tel: +44 (0)114 222 4111 Fax: +44 (0)114 222 0002 Email: firstname.lastname@example.org
I. Introduction 1
II. Parasitism: direct consequences 2
III. Dynamics of parasite–host interactions: host range, preference and selection 2
IV. Impacts of parasitic plants on the plant community 3
V. Impacts of the plant community on parasite populations 5
VI. Impacts of the parasite on other trophic levels 6
VII. Impacts of the parasite on the abiotic environment 11
VIII. Concluding remarks 12
Parasitic plants have profound effects on the ecosystems in which they occur. They are represented by some 4000 species and can be found in most major biomes. They acquire some or all of their water, carbon and nutrients via the vascular tissue of the host's roots or shoots. Parasitism has major impacts on host growth, allometry and reproduction, which lead to changes in competitive balances between host and nonhost species and therefore affect community structure, vegetation zonation and population dynamics. Impacts on hosts may further affect herbivores, pollinators and seed vectors, and the behaviour and diversity of these is often closely linked to the presence and abundance of parasitic plants. Parasitic plants can therefore be considered as keystone species. Community impacts are mediated by the host range of the parasite (the diversity of species that can potentially act as hosts) and by their preference and selection of particular host species. Parasitic plants can also alter the physical environment around them – including soil water and nutrients, atmospheric CO2 and temperature – and so may also be considered as ecosystem engineers. Such impacts can have further consequences in altering the resource supply to and behaviour of other organisms within parasitic plant communities.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Parasitic plants are a taxonomically diverse group of angiosperms that rely partially or completely on host plants for carbon, nutrients and water, which they acquire by attaching to host roots or shoots using specialist structures known as haustoria and by penetrating host xylem and/or forming close connections with phloem. The site of attachment to the host classifies the parasite as either a root or shoot parasite, whereas the presence or absence of functional chloroplasts defines the parasite further as being either hemiparasitic or holoparasitic, respectively (Musselman & Press, 1995).
Parasitic plants are common in many natural and seminatural ecosystems from tropical rain forests to the high Arctic (Press, 1998), accounting for 1% of angiosperm species (∼3–4000) within c. 270 genera and more than 20 families (Nickrent et al., 1998; Press et al., 1999). They occur in many life forms, including annual and perennial herbs (e.g. Rhinanthus spp. and Bartsia spp.), vines (e.g. Cuscuta spp. and Cassytha spp.), shrubs (e.g. Olax spp. and mistletoes) and trees (sandlewoods, e.g. Okoubaka aubrevillei, which grows up to 40 m tall; Veenendaal et al., 1996).
Parasitism often severely reduces host performance, which leads to changes in competitive interactions between host and nonhost plants and a cascade of effects on community structure, diversity, vegetation cycling and zonation (Pennings & Callaway, 2002). Impacts on the plant community are enhanced further because parasitic plants simultaneously parasitise and compete with co-occurring plants; their own productivity and populations are therefore dependent on both the ‘quality’ of the hosts that they parasitise and the strength of competition from neighbouring plants. Additionally, the uptake of host solutes can have consequences for organisms of other trophic levels (such as herbivores and pollinators), and co-occurring organisms may also be affected by the impacts of parasitic plants on the abiotic environment, including impacts on nutrient cycling, soil water relations, local temperature and atmospheric CO2 concentrations. Importantly, such major impacts can occur even when parasitic plants are minor components of the ecosystem.
Despite the profound effects that parasitic plants have on the communities in which they occur, they are still often ignored in community theory (highlighted by Pennings & Callaway, 2002). With this in mind, this review examines the numerous interactions that parasitic plants have with host and nonhost plant communities, with other organisms (including herbivores, pollinators, mycorrhizal fungi and other parasites) and discusses parasitic plant impacts on the abiotic environment to highlight the far-reaching consequences of these interactions for community structure and function. Because this review is primarily concerned with community level interactions, we only briefly review the direct impacts of parasitism on individual host plants; more detail of these direct impacts and their physiological basis can be found in, for example, Stewart & Press (1990), Press & Graves (1995), Watling & Press (2001) and Phoenix & Press (2005).
III. Dynamics of parasite–host interactions: host range, preference and selection
Community-level impacts of parasitic plants depend greatly on which species are parasitised. Ultimately, this is dependent on parasite host-range (diversity of hosts rather than geographical range), preference for particular host species, and the degree to which preferred species can be selected or ‘foraged’ for.
1. Host range: parasitic plants are usually generalists
Examples of wide host range are documented for both root and shoot parasites. In the former case, Castilleja spp., for instance, are known to parasitise more than 100 different hosts from a variety of families (Press, 1998), whereas Rhinanthus minor has approximately 50 different host species from 18 families within European grasslands, and in a dune system study, single R. minor plants have been found to parasitise up to seven different host species simultaneously (Gibson & Watkinson, 1989). Although shoot parasites tend to have a smaller host range than do root parasites (Norton & Carpenter, 1998), broad host ranges are still apparent, such as with Cuscuta spp. (dodders) with hosts that number in the hundreds (Kelly et al., 1988; Musselman & Press, 1995), whereas the tropical rain forest mistletoe Dendrophthoe falcata has approaching 400 known host species (Narasimha & Rabindranath, 1964; Narayanasamy & Sampathkumar, 1981; Joshi & Kothyari, 1985).
Parasitic plants that can only utilise one or few host species are the exception rather than the rule, and perhaps the most notable among the root parasites is Epifagus virginiana (Orobanchaceae) which only parasitises Fagus grandifolia (Musselman & Press, 1995). Among shoot parasites, mistletoes provide some examples of narrow host range, including the dwarf mistletoe Arceuthobium minutissimum (Viscaceae), which only parasitises the pine species Pinus griffithii (syn. wallichiana) (Kuijt, 1969), and epiparasitic mistletoes (e.g. Phoradendron scabberimum), which only grow on other mistletoes (Musselman & Press, 1995).
2. Host preference: when generalists are specialists
Intriguingly, despite the large host range of the majority of parasitic plants, many also show high levels of host preference, such that while many different plant species within a community can act as hosts, the majority of hosts are taken from just a subset of those available (e.g. Orobanchaceae: Werth & Riopel, 1979; Gibson & Watkinson, 1989; Santalaceae: Joel et al., 1991; Krameriaceae: Musselman & Dickison, 1975; Olacaceae: Musselman & Mann, 1978). In this way, parasitic plants are not true generalists and can behave more as specialists.
Why parasites may have a different host preference in different locations is not known, although for mistletoes, changes in host susceptibility to infection between different regions has been suggested as one mechanism (Snyder et al., 1996). Host preference may also depend on the diversity of potential hosts available; mistletoes of the Loranthaceae show a low host preference in heterogeneous tropical rain forests and high host preference in less diverse temperate forests. This may occur because preference for a particular (and perhaps better) host is more possible in a less diverse system where the preferred host is therefore a larger component of the community (Norton & Carpenter, 1998).
3. Host selection
Selection of or foraging for preferred hosts can operate in a number of ways, both spatially and temporally. A particular host species may appear to be ‘preferred’ simply as an artefact of its abundance, i.e. an abundant host species is used more because it is more likely to be encountered by the parasite. Even so, true host preference – when a host is used disproportionately to its abundance – appears to be a common occurrence among both root and shoot parasitic plants. For such preference to operate, the parasite may need chemical cues from suitable hosts to trigger germination (e.g. Bouwmeester et al., 2003) and/or haustorial development (Matvienko et al., 2001; Tomilov et al., 2004). Because rapid attachment following germination is critical for many parasites, many have adapted to follow such chemical cues. Host preference therefore results because the parasite is less likely to germinate and/or produce haustoria away from the triggering host species (Musselman & Press, 1995). Chemical cues may also play a role in the active foraging seen in the stem parasites Cuscuta subinclusa and C. europea. These parasites display nastic movements that allow them to forage for hosts, rejecting (growing away from) or accepting (coiling around) the stem of hosts following contact, but before any penetration of the host shoot is made (Kelly, 1990, 1992). The mechanisms underpinning these responses, however, remain elusive.
IV. Impacts of parasitic plants on the plant community
Impacts on the community are often considerable and occur because: (i) impacts on the host are great; (ii) major impacts occur even where the parasite is a minor component of the ecosystem; and (iii) a single parasite may impact on a large area of the ecosystem. Over one season, for instance, a single Cuscuta plant may form thousands of connections with many host species and may cover an area greater than 100 m2 (Kelly, 1990), resulting in considerable impacts on the plant community despite its being perhaps less than 5% of vegetation biomass (Pennings & Callaway, 1996).
1. Plant community biomass
Because reductions in host growth are often greater than the increases in parasite growth, reductions in plant community productivity are often observed. Rhinanthus species, for instance, have been shown to reduce total productivity in European grasslands by between 8 and 73% (Davies et al., 1997), whereas dwarf mistletoes (one of the most destructive pathogens of commercially viable trees) can reduce volume growth of Douglas fir, for instance, by up to 65% (Mathiasen et al., 1990).
Interestingly, Joshi et al. (2000) have shown that community biomass reductions by Rhinanthus are smaller in grasslands that have greater functional diversity. They proposed that higher plant diversity could buffer the effects of overexploitation of individual host species such that less sensitive species will compensate for loss of biomass of more sensitive species. Further, Matthies and Egli (1999) have shown that host biomass is reduced the most under low nutrient conditions, suggesting that community-level impacts may also be greatest where resources acquired by the parasite (such as nutrients) are limiting.
2. Plant community diversity
Impacts on community structure can also be great. Primarily, impacts on host performance shift the competitive balances from host species toward nonhost species and ultimately result in community change. Very often, the most heavily parasitised species are competitive dominants, in which case parasitism facilitates the maintenance of competitively subordinate species (Press, 1998). The preference (by choice or chance) of Rhinanthus spp. for grass hosts, for instance, is well known to reduce grass biomass and facilitate an increase in forb abundance (Davies et al., 1997). Introduction of Rhinanthus is therefore used as an effective management tool to restore high-fertility/low-diversity pastures to high-diversity meadows (Westbury & Dunnett, 2000). In the case of shoot parasites, the salt marsh studies of Pennings & Callaway (Pennings & Callaway, 1996, Callaway & Pennings, 1998) have shown that Cuscuta salina has preference for the host Salicornia virginica. Because this host is the community dominant, its suppression in areas where Cuscuta is abundant facilitates the expansion of the competitively subordinate species Limonium californicum and Frankenia salina. This in turn increases community diversity.
Conversely, where preferred hosts are competitively subordinate, parasitism can reduce abundance of subordinate species, allow greater dominance of the most abundant species and hence reduce community diversity. Such a case was observed in sand dune systems, where Gibson & Watkinson (1989) showed that Rhinanthus minor– known usually to increase diversity – tended to reduce diversity by preferentially parasitising subordinate species. Further, supposed preferred host species may not necessarily decline in abundance where the abundance of other potential hosts is great enough to ‘hide’ the preferred host from the parasite. For instance, N-rich legume species are well known to be good (preferred) hosts for Rhinanthus spp., but in the study of Davies et al. (1997), Rhinanthus actually increased, rather than reduced, legume abundance in European grasslands. Davies et al. (1997) proposed that the high density of grasses overrides host preference, so that grasses are parasitised more and suppressed more because their roots are far more likely to be encountered than roots of preferred legumes.
By facilitating coexistence and diversity through limitation of competitive dominants, parasitic plants can be considered as ‘keystone species’ (Paine, 1969; Pennings & Callaway, 1996; Smith, 2000). Certainly, parasitic plants fit the keystone species definition of exerting a major influence on community assemblages out of proportion to their own abundance or biomass. Paine (1969) coined the ‘keystone species’ term from observations of the predatory starfish, Pisaster, which increases the diversity of mussel bed communities by consuming dominant species of mussel and hence facilitates the coexistence of subordinate mussel species. The keystone species term has since been used to describe the central role played by a variety of species within communities, from sea otters and fish (Estes & Palmisano, 1974; Power, 1995) to succulent trees and Sphagnum mosses (Midgley et al., 1997; Mitchell et al., 2002). In many cases, parallels with the action of parasitic plants are clear.
In addition to the effects of parasitism, annual parasites may further increase diversity through facilitation of invasion; for example, an increase in bare ground following die-back of Rhinanthus alectorolophus at the end of the season was seen to facilitate weed invasion and led to increased community diversity (Joshi et al., 2000). Interestingly, facilitation of invasion was less in more diverse communities, indicating that a negative feedback mechanism may operate: once the community reaches a certain level of diversity, invasion may no longer be facilitated; should community diversity decline again, invasion will once again increase.
3. Vegetation cycling and zonation
The effects of parasitic plants on community structure are often dynamic and will change depending on environmental conditions or the performance of the parasite itself. Parasitic plants can therefore impact and regulate both vegetation cycling and zonation. At the simplest level, an aggressive parasite can drive a preferred host locally extinct; this may, in turn, result in the parasite also becoming locally extinct. The originally suppressed preferred host is then able to return, and following this, the parasite can then re-establish on the new host plants. Such population cycling has similarities with some predator–prey cycles (see e.g. Krebs et al., 1995). Perhaps the best example of such cycling in parasitic plants is provided by Cuscuta salina described previously (Pennings & Callaway, 1996). Not only does Cuscuta facilitate invasion of subordinate species, but this process also initiates vegetation cycling because Cuscuta populations decline following Salicornia suppression, which in turn reduces the facilitation of Limonium and Frankenia invasion and allows Salicornia to return.
Such cycling interactions may also explain why some parasites, such as Rhinanthus minor, appear to ‘move through’ vegetation. Patches heavily infested with Rhinanthus will quickly decline in grass (preferred host) abundance, leaving neighbouring uninfected patches with higher grass abundance more suitable for establishment of the next generation of Rhinanthus seedlings, and the Rhinanthus patch will appear to ‘move’ over time. The vegetation left behind will recover rapidly (Gibson & Watkinson, 1992) and will once again become suitable for Rhinanthus.
Such interactions between hosts and parasites are often constrained by environmental factors that can influence the virulence of the parasite and the competitiveness of hosts and nonhosts. Through this, parasitic plants can regulate the zonation of vegetation. Again, Cuscuta salina provides and excellent example of this. Whereas Salicornia dominates in the lower part of the salt marsh, Arthrocnemum subterminale dominates at higher elevations (Callaway & Pennings, 1998) and the two compete strongly at their abrupt ecotone (Pennings & Callaway, 1992). Cuscuta preferentially parasitises Salicornia, conferring a competitive advantage to Arthrocnemum, and effectively stops Salicornia from invading into the Arthrocnemum zone (Fig. 1). Because Cuscuta patches are dynamic, this probably makes the Salicornia–Arthrocnemum ecotone less abrupt. Further, the competitive advantage provided to Arthrocnemum by Cuscuta is seen to be much greater at lower elevations within the marsh. Here, Arthrocnemum is much less competitive in these more saline areas, so the benefit of being released from Salicornia competition by Cuscuta is much greater. This suggests that the advantage of parasitism to a subordinate species should be greatest where it is most at a competitive disadvantage (i.e. where competition is most asymmetrical) (Callaway & Pennings, 1998).
V. Impacts of the plant community on parasite populations
At a simple level, greater abundance and/or performance of preferred or good hosts will enhance the performance (growth and reproduction) of the parasite. Root hemiparasites, for instance, are particularly common in grassland systems because grasses are often preferred hosts, having abundant root systems (i.e. easy to locate) and fine roots that are easy to penetrate. Similarly, Cuscuta shows greater biomass and reproduction within patches of preferred/good hosts (Kelly et al., 1988; Kelly, 1990).
The age of the hosts selected by the parasite may also impact on its own population dynamics. Seel and Press (1996), for instance, observed that Rhinanthus minor produced significantly less biomass when parasitising 6-month-old Poa alpina hosts than when parasitising mature plants. Further, Rhinanthus attached to Poa that had been previously parasitised grew better than Rhinanthus attached to Poa not previously parasitised. It was suggested that the parasite benefited from previous parasitism of the host because this reduced host flowering that would otherwise represent a loss of resources. Impacts of hosts on parasite communities clearly not only depend on what is parasitised but also when parasitism occurs.
Because parasites compete with hosts for resources, competition from the host can also affect parasite populations. With some hemiparasites, competition with hosts for light is believed to restrict the parasites to low-productivity environments (Matthies, 1995). In high-productivity environments, increased shading of these partially autotrophic plants may reduce their competitiveness. This theory is supported by the work of Matthies (1995), who showed that shading from host plants reduced biomass of the root hemiparasites Rhinanthus seratonis and Odonites rubra by 30%, and also by the work of Joshi et al. (2000), in which survival of Rhinanthus alectorolophus in grassland communities was inversely correlated with community leaf-area index. It has been suggested therefore that parasite-induced reductions in host and community biomass represent an advantage to some hemiparasites because this also reduces competition for light.
In extreme cases, where resource limitation results in parasite death, host access to limiting resources can almost completely control parasite distribution. Summer drought in southwest Australian heaths restricts Olax phyllanthi to patches of deep-rooted hosts that have access to the water table (Pate et al., 1990a). Further, where environmental conditions vary within a community, the parasite may prefer hosts in areas where the hosts experience less environmental stress. For instance, Miller et al. (2003) suggest that in a semiarid flood plain in southern Australia, eucalyptus (E. largiflorens) are poor hosts for the mistletoe Amyema miquelii in areas of greater water and/or salinity stress.
Beyond this, parasite performance may depend on the diversity of its host community. Joshi et al. (2000) found that both growth and reproductive effort of Rhinanthus alectoroluphus was greatest when growing in plant communities of high functional diversity (Fig. 2). This may occur because: (i) high functional diversity facilitates a mixed diet believed to be beneficial to some parasitic plants (Marvier, 1998a); (ii) high diversity enhances the chance of the parasite finding a good host; and/or (iii) the parasite benefits form greater host biomass (resource size) where higher diversity leads to greater community productivity (Joshi et al., 2000).
Host preference can result in aggregation of the parasites around preferred hosts; such aggregation can occur at the level of the host, patch or community. The majority of mistletoes within a population, for example, may be found on just a few host individuals – a conseqeuence of the mistletoe seed dispersal mechanisms (see Section VI.3) – with most other hosts of the same species harbouring no or few parasites (Aukema, 2003). Mistletoes also aggregate at the community and landscape scale. Isolated trees are unlikely to become infected, and migration of the parasite through the landscape may be slow until a new site eventually reaches some threshold of mistletoe density and then readily attracts its avian seed dispersers (Aukema, 2003).
Finally, parasites can directly impact on their own populations through parasitism of members of their own population (self-parasitism). This is seen in the case of Olax phyllanthi, where physiologically superior individuals acquire resources from inferior Olax plants and which may therefore explain the rapid self-thinning which takes place in Olax populations during early postfire succession (Pate et al., 1990b).
VI. Impacts of the parasite on other trophic levels
It is not only plants within communities that can be heavily affected by parasitic plants. Many other organisms, including birds and insect herbivores, other parasites and mycorrhizal fungi can be affected, either directly or indirectly. This wide range of impacts occurs because many parasitic plants can have both top-down effects (e.g. as a natural enemy of the host) and bottom-up effects (e.g. as a keystone resource). For instance, herbivorous insects and mammals consume parasitic plant foliage; frugivorous birds consume mistletoe berries; fungi and insects can take advantage of host plants weakened by parasitic plants (Parker & Riches, 1993; Aukema, 2003); and parasitic plants can compete with other consumers where the host is a shared (and potentially limiting) resource (Pennings & Callaway, 2002).
1. Interactions with herbivores
In the case of indirect effects, hosts weakened by a parasitic plant may be more susceptible to insect attack. In the case of dwarf mistletoes (Arceuthobium spp.), the increased susceptibility of host trees may result from their increased water stress because the parasite transpires readily – despite its relatively small surface area – even under water-limited conditions (Fisher, 1983). A resulting reduction in host resin exudation may be one mechanism for increased host susceptibility (Parker & Riches, 1993; Aukema, 2003). Conversely, herbivores may feed less on parasitised hosts, possibly because of competition between herbivore and parasite for the host resource. Puustinen & Mutikainen (2001), for example, observed that parasitism by Rhinanthus serotinus reduced feeding of the snail Arianta arbustorum on Trifolium repens hosts (this being an indicator of the competition for resources between snail and parasitic plant). However, when feeding on cyanogenic and acyanogenic Trifolium hosts was compared, the saving in leaf area consumed of cyanogenic Trifolium over acyanogenic individuals was lower in parasitised hosts, i.e. parasitism appeared to reduce the benefits of cyanogenisis in alleviating herbivory. In natural ecosystems, this could prove particularly costly for the acyanogenic plants because cyanogenesis is energetically expensive.
Further, where parasites and herbivores compete for the same host resource, the performance of the parasite may be reduced where hosts experience heavy herbivory. Salonen and Puustinen (1996) observed that partial defoliation of the host Agrostis capillaris could reduce flowering of the parasite Rhinanthus serotinus.
Parasitic plants themselves can be attractive food sources for herbivores. In the case of mistletoes, their fruit is often available year round, their flowers provide abundant nectar and their foliage is often rich in nutrients (Watson, 2001). Indeed, fruit, flowers and foliage of mistletoes are known to be food for some 66 families of birds, 30 families of mammals and even one fish species (Watson, 2001), in addition to an unknown diversity of insect herbivores. The quality of the parasite as a resource to these herbivores can be greatly affected by host nutrient status. Although nutrient-rich hosts benefit both holo- and hemiparasites, enhanced nutrition of the parasite can also cause it to be more attractive to insect herbivores and to increase the population growth of those herbivores further. Survival of the aphid, Nearctaphis kachena, for instance, when feeding on the root-hemiparasite Castilleja wightii, was positively correlated with the N concentration of the host plant (Marvier, 1996). Castilleja performance was therefore poorer on N-rich hosts because of resource competition with the larger aphid population. In this case, hosts with high N content were poorer hosts for Castilleja in the presence of herbivores. Further, N-rich hosts were therefore better as indirect host for aphids than they were as direct host for Castilleja because N-rich hosts increase Castilleja aphid populations at the expense of Castilleja performance.
The host plant may not only affect parasite–herbivore interactions through uptake of nutrients, but also through uptake of host secondary metabolites that may have antiherbivory properties. The uptake of host alkaloids by root-hemiparasitic Orobanchaceae has been well documented (e.g. Stermitz et al., 1989; Schneider & Stermitz, 1990; Mead & Stermitz, 1993; Marko & Stermitz, 1997) and reductions in herbivory or herbivore performance when feeding on the alkaloid-acquiring parasites have been observed (Marko & Stermitz, 1997; Mead et al., 1992). Loveys et al. (2001) observed that fruit of the root hemiparasite Santalum acuminatum (the quandong) contained a natural insecticidal compound acquired from neighbouring Melia azadarach hosts. The uptake of such compounds from the host was proposed to be beneficial because a bioassay using the apple moth (Epiphyas postvittana) showed that its larvae suffered higher mortality when feeding on fruit of Santalum growing near Melia hosts. This may also explain the observation of commercial growers that Santalum growing near Melia have fruit that suffer less insect attack.
In addition to such direct benefits, the parasite may also gain indirect benefits from uptake of secondary metabolites. In the case of Castilleja indivisa, Adler (2000) observed that this root hemiparasite not only gained from reduced herbivory by insect larvae when acquiring alkaloids from ‘bitter’ lupine hosts (compared with alkaloid-free ‘sweet’ lupine hosts), but the reduced herbivory of floral parts increased the visitation by hummingbird pollinators (Fig. 3). In turn, the Castilleja parasite showed greater seed production and hence gained increased fitness from both reduced herbivory and increased pollination.
The proposed benefits of the ‘mixed diet’ mentioned previously (see Section IV) may also extend to host mediated impacts on herbivory. Marvier (1998a) observed that a mixed diet of legume and nonlegume hosts not only enhanced the growth of Castilleja wightii (compared with double legume or double nonlegume hosts) but also resulted in slower growth of aphid colonies feeding on the Castilleja. Such benefits may provide one reason for the maintenance of a broad host range by many parasitic plants growing in natural communities (i.e. in the presence of herbivores) despite certain hosts appearing to be far more beneficial to parasite performance in pot studies (i.e. in the absence of herbivores).
Parasitic plants may also mimic or use their host's foliage to avoid herbivory. Protective cryptic mimicry in mistletoes (Barlow & Wiens, 1977) – where the mistletoe foliage appears similar to its host's – may protect the mistletoe against vertebrate herbivores that use visual cues for feeding selection. Such mimicry may be important (and indeed appears more frequently) in mistletoes that are nutritionally better food sources than their hosts, i.e. those with higher nitrogen and protein concentrations than their hosts (Ehleringer et al., 1986). Conversely, mistletoes with lower tissue nutritional quality than their hosts may benefit from advertising this fact by not mimicking host foliage (Ehleringer et al., 1986). Certainly, mistletoes with lower tissue quality than their hosts are more likely to lack mimicry (Ehleringer et al., 1986; Bannister, 1989). However, the relationship between mistletoe foliar quality and mimicry for herbivore defence is open to debate because the relationship between foliar N and mimicry is apparent in populations of mistletoes (in New Zealand) which probably evolved in the absence of – and therefore without selection pressure from – herbivorous mammals (Bannister, 1989). Further debate arises because levels of herbivory may not necessarily be lower in mimic compared with nonmimic mistletoe species (Canyon & Hill, 1997), and because mistletoe leaf N may be directly related to host N (Glatzel, 1987; Canyon & Hill, 1997). A simple direct relationship between mistletoe nutritional quality and mimicry cannot be assumed.
2. Interactions with pollinators
Two groups of parasitic plants show particularly close interactions with pollinators and seed dispersers: the mistletoes and the Rafflesiaceae. The latter, particularly species of the genera Rafflesia and Rhizanthes, consist almost entirely of endothermic flowers and lack stems and leaves. High respiration rates and endothermy combine to create flowers that are up to 9 K warmer than surrounding ambient air and have considerably elevated local CO2 concentrations. These factors, in combination with the release of volatiles, which give the flowers the odour of faeces or carrion, result in the attraction of blowflies that pollinate these parasitic flowers (Patiño et al., 2000, 2002). The endothermy and high respiration rates are metabolically costly, but these costs are ultimately passed to the hosts that provide substrates for respiration (Patiño et al., 2002). In the case of Rhizanthes, the effective mimicry of faeces and carrion further enhances pollination by the blowflies because oviposition can be stimulated, which increases the time the blowflies spend inside the flower while searching for somewhere to lay.
Mistletoes show close interactions with pollinators and their seed vectors, associations that can be considered truly mutualistic. Mistletoes therefore may act simultaneously as parasites and mutualists in natural communities. Many mistletoes in the Loranthaceae are dependent on birds to open flower buds and act as pollinators, and therefore often have large, odourless flowers of bright colour to attract these pollinators (Watson, 2001). The explosive action on flower opening insures transfer of pollen to the bird and in return allows the bird access to a previously untapped nectar supply which is often available in large quantities and particularly rich in sugars (Stiles & Freeman, 1993; Baker et al., 1998). Such luxurious provision of carbohydrate-rich nectar may be made possible because the substrates are provided by two sources: the host and the partially autotrophic mistletoe. It appears that for both mistletoes and the Rafflesiaceae, the parasitic habit allows energetically expensive mechanisms for the attraction of pollinators.
3. Interactions with seed dispersers
For most mistletoes, birds act as seed dispersers, and in some instances the same species may act as both pollinator and seed disperser (Kuijt, 1969; Robertson et al., 1999). Indeed, many of the bird species are highly specialised to consume mistletoe berries (Restrepo et al., 2002), and even in those mistletoes where initial seed dispersal is by hydrostatic explosion, birds can play a subsequent role in transporting seeds further (Watson, 2001). Fruits are often adapted for bird dispersal: they are usually large; conspicuously coloured; and are often high in soluble carbohydrate, minerals, lipids and fats, and can have an abundance of amino acids (Chiarlo & Cajelli, 1965; Godschalk, 1983; Lamont, 1983). As with nectar rewards, the provision of such energetically expensive fruit may be facilitated by the parasitic habit, which will confer much of the production cost to the host. The close association of avian frugivores with mistletoes may be further enhanced because these fruit dispersers may find the fruit reward available all year round. This is achieved through discontinuous ripening of a single mistletoe (prolonging duration of fruit provision) and asynchrony in peak fruiting time between individual mistletoes of the same population or between mistletoes of separate population (Watson, 2001).
A key feature of the frugivore–mistletoe association is that the behaviour of the seed disperser is modified by the mistletoe to enhance successful seed dispersal (cf. blowfly oviposition stimulation to enhance pollination of Rhizanthes). To aid seed dispersal, a sticky viscin coats the mistletoe seed, allowing it to adhere to host branches following defecation or regurgitation by the avian vector (Reid et al., 1995; Aukema, 2003). Indeed, this effect is the origin of the name ‘mistletoe’, which approximates to ‘dung-stick’, a name based on the early observations that mistletoes appear where bird droppings are deposited on trees. Further, because defecated or regurgitated seed may stick to the bird's bill or abdomen, the birds will engage in bill or abdomen wiping to dislodge the seed, and because such behaviour often takes place on suitable host branches, the seed is effectively stuck to the host by the bird. The chance of seed being deposited on suitable hosts is enhanced further because a suitable host already parasitised by mistletoes will carry the mistletoes fruit reward and attract further avian mistletoe dispersers.
Mistletoes therefore are among the few examples of plants with directed dispersal, ensuring that seed is often moved to suitable hosts (Aukema, 2003). Perhaps this explains why mistletoe seeds germinate readily in most situations without the need for a specific chemical germination cue (Norton & Carpenter, 1998).
Beyond this, where the same frugivore species disperse seeds of both mistletoe and host, novel tripartite mutualistic associations can develop. Such a case occurs with Townsend's solitaires (Myadestes townsendi) that forage for seed of the mistletoe Phoradendron juniperinum and its juniper host, Juniperus monosperma (van Ommeren & Whitham, 2002). The mistletoe provides a stable and prolonged resource of fruit, whereas the juniper fruit supply is much more variable: mistletoe berry production therefore most strongly regulates the abundance of the avian frugivores, and far more of these birds are attracted to juniper stands infected with mistletoe than to uninfected juniper stands. The junipers ultimately benefit, because the mistletoes attracts greater populations of the juniper/mistletoe shared seed dispersal agent (the Townsend's solitaires), the end result being that mistletoe-infected juniper stands have higher juniper seedling densities (van Ommeren & Whitham, 2002). However, there are trade-offs because at very high mistletoe densities, the negative physiological impacts of the mistletoe on juniper hosts will outweigh any positive effect of attraction the Townsend's solitaires and, further, at such high mistletoe densities, the attraction of Townsend's solitaires may be detrimental to the juniper because this will serve to enhance the mistletoe population further (Fig. 4).
The close host–parasite–vector association means that parasite seed vectors may have patchy distributions determined by the parasite and its host. The abundance of Chilean mockingbirds (Mimus thenca), for instance, is strongly associated with the prevalence of Tristerix aphyllus, a mistletoe parasitic on columnar cacti (Martínez del Rio et al., 1996), which in turn is restricted to north-facing slopes populated by its cacti hosts. This association also provides an example of where the behaviour of seed dispersal agents can be used by hosts to reduce the chance of mistletoe infection (rather than being used by the parasite to increase infection). In this case, the Tristerix mistletoe reduces the reproductive effort in the cacti hosts, but the resulting selection pressure appears to select for cacti with longer spines that deter the mistletoes’ avian seed vectors (Martínez del Rio et al., 1995; Medel, 2000; Medel et al., 2004). As highlighted by Medel et al. (2004), the aggregation of parasites within a community is therefore not only dependent upon the attraction of seed vectors to already infected hosts, but also on host resistance traits.
4. Interactions with other (nonplant) parasites of the host
Where two parasites share the same host, either one parasite may facilitate the establishment of the other (e.g. by weakening the host), or there may be direct competition between the two parasites for host resources (Petney & Andrews, 1998). Interactions between parasitic plants and other parasitic organisms have been little studied, but it can be predicted from theory that attack by multiple parasites will be more detrimental to the host and that the most aggressive of the two parasites will benefit to the detriment of the other. These predictions are supported by Puustinen et al. (2001), who studied dual parasitism of Trifolium pratense by the root hemiparasite Rhinanthus serotinus and the cyst nematode Heterodera trifolii. Simultaneous parasitism by both parasites reduced Trifolium biomass more than parasitism by either parasite alone. Further, Heterodera appeared to be a more aggressive parasite than Rhinanthus because the reduction in Trifolium biomass was greater under parasitism by the cyst nematode than under parasitism by the root hemiparasite. The competitive advantage of Heterodera over Rhinanthus was confirmed under dual parasitism conditions where attachment to the Trifolium host did not enhance Rhinanthus growth if the host was also parasitised by Heterodera, while conversely, parasitism by Rhinanthus did not reduce the number or size of cysts produced by Heterodera.
5. Interactions with soil microbes
Parasitic plants can have considerable impacts on soil organisms, even though their direct contact with the soil system through roots may be minimal or nonexistent. Both root and shoot parasites, for instance, can reduce the mycorrhizal associations of host plants. Gehring and Whitham (1992) found that colonisation of arbuscular mycorrhizal (AM) fungi on Juniper monosperma tree roots was negatively associated with mistletoe (Phoradendron jumiperum) density. This may be driven by two mechanims: either AM fungi increase resistance of the juniper host to mistletoe infection and/or (because mistletoes and mycorrhizas compete for host photosynthate) mycorrhizal infection rates will be lower where mistletoes are stronger competitors for plant carbon. The latter seems more likely because it was also observed that reductions in mycorrhizal associations caused by mistletoe infection were greater in female trees than in males, presumable because female trees invest more photosynthate in reproductive structures, therefore increasing the competition for photosynthate between mistletoes and AM fungi (Gehring & Whitham, 1992).
Similar parasite–mycorrhiza interactions have been observed for root hemiparasitic plants. Rhinanthus minor has been shown to reduce AM colonisation of Lolium perenne by about 30% (Davies & Graves, 1998). Again, this reduction in colonisation may be explained if the AM fungi are weaker competitors than the hemiparasite for host carbon. This was further supported by the observation that Rhinanthus appeared to benefit from AM colonisation of the host, showing greater growth and reproductive output on AM colonised Lolium. Clearly, the AM fungi did not significantly reduce the acquisition of host carbon by Rhinanthus while Rhinanthus may have benefited from enhance nutrition of AM colonised hosts. Indeed, because mycorrhizal stimulation of plant productivity was much greater for Rhinanthus than for Lolium, the indirect benefits of AM fungi to Rhinanthus were greater than their direct benefits to Lolium. Similarly, Salonen et al. (2000) observed that the hemiparasite Melampyrum had greater growth and produced more flowers when parasitising Pinus sylvestris colonised by ecto-mycorrhizal (EM) fungi than when parasitising nonmycorrhizal Pinus. Because EM symbiosis increased the growth of Pinus, it was proposed that greater photosynthate could be made available to the Melampyrum (despite competition with EM fungi) because of the greater photosynthetic leaf area of the larger mycorrhizal host (Salonen et al., 2000).
A further way in which parasitic plants may impact on soil microbes is through inputs of their particularly nutrient-rich litters. Although such impacts are yet to be directly measured, parasite litter is known to impact on nutrient cycling (discussed in Section VII) (Quested et al., 2002, 2003a,b) so we should also expect considerable impacts on soil organisms. Because nutrient-rich litter can support greater, more active microbial populations (Beare et al., 1990), and because litter quality is known to affect fungal community composition and the balance between bacterial and fungal components of the soil system (Wardle, 2002), nutrient-rich parasite litter may have similarly large effects on the soil biota.
6. Effects on the diversity of other (nonplant) organisms
Given the role of parasitic plants as a keystone resource within communities and their considerable impact on the diversity of co-occurring plants, it is perhaps unsurprising that they also have profound effects on the diversity of other organisms. In addition to their importance as a food resource for birds and invertebrates, mistletoes can alter the structure of the habitat for many organisms that live on or within the host. ‘Witches brooms’ and mistletoe clumps are used extensively as nesting or roosting sites for birds, either as structural support for nests or to aid in concealment and may provide hibernation sites or shelter in hot weather for mammal species such as pine martens, porcupines and squirrels (reviewed by Watson, 2001). The silviculture practise of sanitising forest stands by removing infected trees (and therefore also removing witches brooms) can therefore be to the detriment of wildlife using these structures (Bull et al., 2004). Also, mistletoe foliage may be used in nest lining, perhaps intriguingly because the foliage of some species may have antibacterial properties and may stimulate the immune function of fledglings (Watson, 2001). Further, through increasing the chance of host mortality, mistletoes can create a more heterogeneous mosaic of habitat structure (Bennetts et al., 1996). Given these numerous roles, it is perhaps unsurprising that mistletoes have been shown to increase the diversity of forest insects and birds, the former also potentially increasing the abundance and diversity of the later. Such mechanisms are apparent in, for instance, Colorado Ponderosa pine forests, where bird number and diversity are positively correlated with the level of dwarf mistletoe (Arceuthobium vaginatum) infestation despite this species of mistletoe rarely being used as a food source by birds (Bennetts et al., 1996). In this case, the number of cavity nesting birds is also greater in heavily mistletoe-infested sites, probably because more tree snags are available as a result of greater tree mortality in heavily infested stands.
Similar impacts for root parasites have yet to be reported. However, because root parasites often alter the diversity of co-occurring plants and can increase habitat heterogeneity through their own patchy distribution, impacts on the diversity of other organisms such as invertebrate herbivores seem likely.
VII. Impacts of the parasite on the abiotic environment
In addition to being considered as keystone species, parasitic plants can also be seen as ecosystem engineers (organisms that modulate the availability of resources by causing physical state changes in biotic and abiotic materials) (Jones et al., 1994). Within this definition, their role as autogenic engineers (which change the environment through their own physical structure) has been discussed above, for instance, where mistletoes are used as nesting sites for birds, or where the die-back of Rhinanthus opens gaps in grassland communities, thus facilitating the invasion of weeds (Joshi et al., 2000). However, parasitic plants can also play a major role as allogenic engineers, which change the environment by transforming materials from one physical state to another. This role is perhaps best exemplified by their impacts on nutrient cycling, particularly by root hemiparasites. These plants often occur in nutrient-poor communities, and it is becoming increasingly apparent that their effects on nutrient cycling within these systems can be considerable. The transformation of materials that occurs in this allogenic engineering process is the unlocking of nutrients from more recalcitrant or less available forms into more labile, available forms.
Parasitic plants typically have much higher concentrations of foliar nutrients than their hosts (reviewed by Lamont, 1983; Pate, 1995), typically being two- to fourfold greater for N and P in root hemiparasite foliage (Quested et al., 2002, 2003a,b) and up to 20-fold greater for K in mistletoes (Lamont, 1983). Further, because nutrient resorption efficiency is low, litter may have similar concentrations of nutrient as living foliage (Quested, 2002, 2003a,b). In a study of seven annual and perennial species of sub-Arctic root hemiparasitic Orobanchaceae, of 64 co-occurring species, only plants with an alternative N source (N-fixers and carnivorous plants) had equivalently high concentrations of N in litter (Quested et al., 2003b). It was apparent therefore that these hemiparasites could represent a considerable point source of nutrients. This was confirmed with litter-fall studies using Bartsia alpina, which was seen to increase annual litter-N input to soil by 42% within a 5-cm radius of its stem. These litter inputs are considered to be of heightened importance because they decompose faster and release nutrients more rapidly than litter of co-occurring species, and, further, may stimulate the decomposition of more recalcitrant litters of co-occurring species when mixed (Quested et al., 2002; Quested et al., 2005). In the nutrient-limited environments where such parasites often occur, the potential for impacts on the nutrition of co-occurring plants is clear. Indeed, a bioassay study showed that, compared with litter inputs of other co-occurring species, Bartsia alpina litter could considerably increase foliar N concentration (twofold increase) and growth (∼50% increase) of two commonly co-occurring species, Betula nana and Poa alpina (Quested et al., 2003a). Clearly, such impacts on growth are likely to affect the competitive balance between species, with those plants most able to access parasite litter nutrients benefiting the most.
The importance of this release of nutrients may be further heightened because host species are often slow-growing, long-lived and are often evergreen with nutrient-poor, slowly decomposing litter. As such, the acquisition of nutrients from such hosts and its release in more labile form as parasite litter represents the unlocking of tightly and long-held nutrients (Press, 1998). Also, this process may act to concentrate nutrients in the vicinity of the parasite, but because many hemiparasites are clonal (such as Bartsia alpina, which can form rhizomes > 50 cm in length; Nilsson & Svensson, 1997), there may also be a significant redistribution of nutrients unlocked from host species, possibly making them more available to more host and nonhost plants. Such redistribution will be further enhanced where reciprocal parasitism occurs (parasites attached to each other). In such cases, resources acquired by one parasite can become shared between parasites (Prati et al., 1997) that will then become redistributed through the senescent leaves of both individuals.
In the case of shoot parasites such as mistletoes, there will be little redistribution of nutrients spatially because parasite litter will fall below the host tree; however, this parasitism will still unlock host tree nutrients and redistribute it to understorey plants. As yet, however, we know of no study that has determined whether understorey vegetation below mistletoe-infected trees differs from the understorey vegetation of uninfected trees. Finally, given that some parasitic plants may be long-lived (e.g. for more than 100 yr; Molau, 1990), the impacts of this continuous enhancement of nutrient inputs may build up to have considerable impacts on local biogeochemical cycling.
In addition to their impacts on biogeochemical cycling and nutrient availability, parasitic plants may also impact on water availability as result of their very high rates of transpiration (Ehleringer & Marshall, 1995) and further impact on host water relations. By increasing the whole-tree water use, for instance, mistletoes may reduce soil water potentials, and so reduce the availability of this resource to host and nonhost species alike (Sala et al., 2001). Further, Marvier (1998b) suggested a similar mechanisms to explain why the root hemiparasite Triphysaria pussilus does not release subordinate dicots from competitive exclusion when vigorously parasitising dominant prairie grasses. In this case, it was proposed that where Triphysaria grows particularly well on its preferred grass hosts, its high transpiration rates result in reduced soil water potentials, and hence effectively outcompetes dicots for this limited resource. Marvier (1998b) highlighted that in such circumstances, the successful exploitation of preferred dominant host species is counterintuitively not good for subordinate nonhost species.
Impacts on nutrients and soil water may help to maintain a heterogenous ‘patchy’ distribution of these key resources. This, in turn, may enhance biodiversity of co-occurring species at the ecosystem scale because the point-to-point differences in resource supply will allow coexistence of different plant species, each most suited to (or a superior competitor within) each patch with a particular composition of resources (Tilman, 1997).
Other abiotic factors which can be influenced by parasites include atmospheric CO2 concentrations and floral temperatures, which, in the case of the Rafflesiaceae, may aid in attraction of pollinating insects (see description of Rafflesia and Rhizanthes in Section VI.2). However, whereas the endothermic flowers of the Rafflesiaceae can be considerably warmer than ambient air temperatures, the high transpiration rates of some other parasitic plants may make parasite foliage considerably cooler. This has yet to be studied in parasites of natural ecosystems, but certainly the considerable transpiration of the crop parasite Striga hermonthica can cool its canopy temperature by 7°C below that of its host (sorghum) and ambient air (Press et al., 1989). Although 7°C represents a considerable cooling, to date, the significance of this for other organisms, such as insect herbivores, is unknown.
VIII. Concluding remarks
Parasitic plants are a diverse group of organisms with regard to their taxonomy, morphology and biogeography. In this review, we have demonstrated that they can play key roles in determining community structure and function and should be considered as both keystone species and allogenic and autogenic ecosystem engineers. The combination of both top-down and bottom-up effects means that they can have considerable impact on multiple trophic levels within communities, affecting population dynamics, diversity and distributions of co-occurring host and nonhost plants, invertebrates, birds and mammals. Further, despite their minimal contact with the soil system, they may also impact greatly on the soil biota and soil resources: this can have further consequences for co-occurring organisms. Parasitic plants are clearly major and key components of many ecosystems, given the considerable extent of their impacts (even when minor components of ecosystems), the diversity of ecosystems in which they occur, and the diversity of organisms with which these parasites interact. Parasitic plants should not be ignored in community study or theory.