Pollination mutualisms in Caryophyllaceae
In a recent article inaugurating the new evolution section of New Phytologist, Verne Grant (2004) suggests that innovative shifts in our understanding of speciation and other processes often arise from the merging and reintegration of ideas already present within existing yet isolated paradigms. The origin and persistence of plant–pollinator mutualisms from antagonistic interactions seems fertile for such ‘hybridization and introgression’, and for concerted, international collaboration in mounting the empirical evidence needed to test existing theory rigorously in multiple species and populations. We need a transformation analogous to the ‘renaissance’ in speciation theory spawned by integrating molecular, genecological and phylogenetic approaches (Rieseberg, 2004; Rausher, 2005). In the case of pollination mutualisms, however, such a synthesis will also require the combined efforts of scientists who understand the natural history and evolution of the interacting partners from both botanical and entomological perspectives. In this feature issue of New Phytologist, six research papers, and a review and metasurvey of pollination and seed parasitism, highlight new advances in our understanding of plant–pollinator relationships in the Dianthus and Silene lineages of the Caryophyllaceae, including antagonistic as well as mutualistic elements.
‘Pollination mutualisms and antagonistic interactions are dynamic, influenced by the community and historical contexts in which they arise. Thus, linking theoretical and empirical approaches will be critical to extending our understanding of the evolution and persistence of nursery pollination mutualisms.
Pollinator diversity in the carnation family (Caryophyllaceae)
The Caryophyllaceae family is exceptionally diverse; the two largest genera, Silene (sensu lato) and Dianthus, comprise nearly half of the species (approx. 1000) and occur on five continents (Oxelman et al., 2001; Judd et al., 2002). Varied pollination syndromes (sensu Fenster et al., 2004) characterize caryophyllaceous species, as is evident from the metasurvey data of over 50 scientists and 30 species analyzed in the paper by Kephart et al. (pp. 667–680) in this feature issue. One theme that has arisen consistently in the literature is the relative importance of nocturnal vs diurnal pollinators and of lepidopterans vs hymenopterans. Although much important work still remains, in 55.5% of the species surveyed it was lepidopterans that were the major pollinators. Nocturnal moths emerged as the most important of lepidopteran pollinators in both native habitats (Kephart et al.), and in European Silene latifolia ssp. alba (S. alba) naturalized in North America (Young, 2002). Yet frequent visitation need not imply effective pollination and gene dispersal, and two papers in this issue help address this inadequacy. In S. alba, Barthelmess et al. (pp. 689–698) use genetic markers to show that nocturnal moths contributed most to interpopulation gene flow regardless of distance. In contrast, by measuring per visit pollen deposition on stigmas as well as visitation frequency, Bloch et al. (pp. 699–706) found two satyrid butterflies to be the most important pollinators in a guild of only five lepidopteran species that visit Dianthus carthusianorum. Surprisingly, though nectar-gathering butterflies are prevalent in many survey populations (e.g. S. acaulis, S. dioica, S. regia), clear specialization for diurnal butterfly pollination occurs only in Dianthus (Jennersten, 1988; Bloch et al.)
Although bees are major pollinators of only 29.6% of survey taxa, bumblebees (i.e. Bombus) also visit moth-pollinated species, and are the most common diurnal pollinators (Table 1, Kephart et al.). Bombus spp. are diverse in morphology (e.g. proboscis length), however, leading to pollination behaviors on Silene and Viscaria ranging from effective pollination to nectar robbing (Jennersten, 1988; Jürgens et al., 1996). The relative roles of bees and butterflies in populations of diurnally pollinated species also vary spatially and temporally, and merit more intensive study in multiple populations. For example, butterflies were once considered key pollinators of the red, tubular corollas of Viscaria vulgaris until later studies revealed the greater importance of Bombus (Jennersten, 1988; Jennersten & Nilsson, 1993). Butterflies, flies and bees also visit predominantly hummingbird-pollinated S. regia and S. virginica (Menges, 1995; Fenster & Dudash, 2001).
The importance of flies as pollinators remains ambiguous. Hoverflies (Syrphidae) and bee flies (Bombylidae) are common visitors to many Caryophyllaceae (Table 1, Kephart et al.), but only S. stockenii and S. integripetala appear exclusively or largely visited by bombylids (Talavera et al., 1996, Oxelman, unpublished). Syrphids and other dipterans are often viewed as pollen-collecting generalists and poor pollinators, but their effectiveness may be underestimated, and is still little understood (Larson et al., 2001).
Nursery pollination and seed predation by nocturnal moths
Three papers in this issue, including a comprehensive review, focus on ‘nursery’ pollination by nocturnal moths whose larvae grow protected within developing fruits of caryophyllaceous hosts. Previously, Dufay and Anstett (2003) reviewed 13 nursery pollination systems (e.g. Lithophragma–Greya, Lophocereus–Upiga, Trollius–Chiastocheta) and a few others have since been described, for example an obligate mutualism between the tropical tree Glochidion and Epicephala moths (Kawakita et al., 2004). A major goal of this feature issue was to explore the conflicting antagonistic vs mutualistic interactions of Silene and allied genera with Hadena and Perizoma moths. Prior study had revealed high variability in the outcomes of the reciprocal exploitation. For example, Perizoma affinitata exhibits the host specificity characteristic of nursery mutualisms, transfers pollen of S. dioica effectively while leaving many undamaged seeds in infested fruits, and has morphological and behavioral traits that might facilitate active pollination (Westerbergh, 2004). In contrast, the often antagonistic Hadena–Silene interactions appear largely unspecialized and lack the regulation of damage (e.g. fruit abortion) found in figs and yuccas (Pettersson, 1991; Dufay & Anstett, 2003). This wide variability in host specificity and the potential outcome of host–predator interactions in the Caryophyllaceae make it a particularly worthy group for studying the evolution of nursery mutualisms (Dufay & Anstett, 2003; Westerbergh, 2004). Thus, the review by Kephart et al. in this feature focuses on the extent of specialization between pollinating moth seed predators and their caryophyllaceous hosts, the evidence that moths act as selective agents on floral traits, and the characteristics of moths and plants that might facilitate or regulate the interaction.
The most widely studied relationship of nocturnal moths to caryophyllaceous plants is that of Hadena bicruris. Brantjes (1976) observed and manipulated their ‘hyperparasitic’ behavior on Silene latifolia nearly 30 years ago. In this issue, Dotterl et al. (pp. 707–718) extend this research, using gas chromatography, electrophysiology and wind tunnel bioassays, to identify the scent compounds that most strongly elicit a response from naïve adults of H. bicruris. Lilac aldehydes, the most abundant compounds emitted from Silene flowers, attracted 90% of moths tested, implying a potential behavioral adaptation to this host plant. Moreover, in populations of S. latifolia that include the anther smut, Microbotryum violaceum, Biere & Honders (pp. 719–727) demonstrate that oviposition by female H. bicruris is significantly greater for uninfected flowers. Thus, the moths may increase their chances of rearing larvae in fruits with adequate seed resources by selecting against diseased flowers. However, predation is also highest on healthy flowers, which increases the fitness cost for both moth and plant, reducing opportunities for mutualistic interactions to evolve.
Specialization, floral traits and the evolution of mutualism
Both asymmetry and variation in the specificity of interactions occur in some of the most widely known pollination mutualisms: fig–fig wasps and orchid–euglossine bees (Bronstein et al., 2004). The extraordinary reciprocal specificity of plant associations with globeflower flies, fig wasps and yucca moths illustrates the potential for monophagous seed predators to evolve close 1 : 1 nursery pollination mutualisms; it also contrasts sharply with eusocial insects (e.g. Bombus), which are sometimes copollinators but whose foraging periods generally exceed the flowering seasons of most plant species (Thompson & Pellmyr, 1992; Dufay & Anstett, 2003). Analysis of Caryophyllaceae–moth data from the metasurvey revealed little evidence of exclusive 1 : 1 mutualistic relationships: although most Hadena and Silene species exhibit specificity for only one or two partner species, some plants host up to nine species of Hadena. However, Hadena seed predators are major pollinators of some species of Dianthus and Silene (Kephart et al.). Close specialization is possible for nocturnal moth–Caryophyllaceae interactions, given the high diversity not only of Silene–Dianthus but of Hadena, whose species exceed 145 in nearctic and palearctic regions, feeding almost exclusively on Caryophyllaceae hosts (Troubridge & Crabo, 2002). Detailed field studies and phylogenetic analyses of both plant and moth partners are needed to understand their interrelationships, the relative costs and benefits of their interaction, and the ecological contexts that constrain or allow for the evolution of these nursery pollination mutualisms.
Several articles in this feature issue provide direct and indirect evidence for selection on floral traits mediated by diurnal or nocturnal pollinators. Statistical analysis of metasurvey data revealed significant associations between nocturnal pollination and several traits that may represent evolutionary responses to the behavior of night-active moths (i.e. white flower color, scent, evening dehiscence of anthers; Kephart et al.). With respect to floral scent emission, presently over 30 species of Dianthus, Saponaria and Silene are now characterized, revealing differences in the chemical signatures of nocturnal and diurnally pollinated taxa (Jürgens et al., 2002, 2003; Jürgens, 2004). Of nocturnally pollinated Silene, at least two additional taxa are now known to emit lilac aldehydes similar to those that attract and elicit dose-dependent antennal responses in Hadena bicruris (Dötterl et al.; Dötterl et al., unpublished).
To ascertain how floral traits affect plant fitness in diurnally visited populations of Silene dioica in northern Sweden, Giles et al. (pp. 729–739) conducted a phenotypic selection experiment using a multitrophic system. Bombus queens are the main pollinators of S. dioica on a series of islands where prevalence of the fungal pathogen Microbotryum violaceum increases with island age. The authors found strong, and potentially rapid, directional selection imposed by pollinator preferences for floral traits that were predicted to affect visibility to pollinators and pollen/spore capturing ability. In a rather elegant experiment, Giles et al. excavated healthy and infected plants from a highly diseased natural population, manipulating them prior to measurement to produce symptom-free female flowers. Their study supports the prediction that floral phenotypes that attract pollinators also reduce plant fitness when the visitors transmit spores instead of pollen, and they neatly differentiate between the effects of stigma exsertion and petal size on pollinator preference. Interestingly, the direction of selection shifts among spatially and temporally heterogeneous populations depending on whether pollinators are carrying predominantly pollen (e.g. young populations) or spores (aging populations with well-established fungal infestations). In Silene acaulis, in which Microbotryum infection is also influenced by bumblebee pollinators, Marr (pp. 741–751) found equivalent rates of infection for both females and hermaphrodites, supporting an earlier finding that spore deposition by pollinators was similar between sexes. The high variability observed in outcrossing rates (0.5–1.0) over short spatial scales may reflect shifts in pollinator activity related to flowering time in this alpine species. Resource heterogeneity also influenced seed production in S. acaulis, but Marr found no correlation between outcrossing rates and disease frequency, implying that pollinators are not altering their behavior in the presence of the disease.
Conclusions and approaches for understanding and conserving mutualisms
Pollination mutualisms and antagonistic interactions are dynamic, influenced by the community and historical contexts in which they arise. Thus, linking theoretical and empirical approaches will be critical to extending our understanding of the evolution and persistence of nursery pollination mutualisms. To discern effectively the direction of selection for reciprocal interactions, we must gather comparable data sets that sample and manipulate key variables in multiple populations, species and years, over broad geographic regions. First, to determine losses in fitness as a result of the presence of seed predators, we need more accurate estimates of the number of seeds remaining in capsules and consumed by moth larvae. Currently, our knowledge of damage by moth pollinator-seed predators during primary and secondary attacks is based primarily on estimates of fruit capsule destruction (e.g. Wolfe, 2002), and only rarely is the role of parasitoids considered (Elzinga et al., 2005). Moreover, even for classic nursery mutualisms (e.g. figs, yuccas), records of seed losses are often not clearly attributed to pollinating vs nonpollinating seed predators (Thompson & Pellmyr, 1992). Second, even in cases where researchers have identified the most common visitors of caryophyllaceous species (e.g. Jürgens et al., 1996, Kephart et al.), we typically lack species-specific data on their relative effectiveness in pollen transfer and on the foraging behaviors that influence their contributions to seed production or to pollen-mediated gene flow (Barthelmess et al.). Analyzing multiple components of pollinator importance is also essential (e.g. Bloch et al.), yet inconsistent definitions of pollinator effectiveness and efficiency remain problematic, making accurate assessments and comparisons difficult in many plant taxa (Inouye et al., 1994; Gross, 2005). Third, although researchers have identified some of the important ecological contexts for selection and trait evolution (e.g. Microbotryum infection; Biere & Honders; Giles et al.), we need comprehensive data on population isolation, pollen limitation and pollinator scarcity to discern the evolutionary consequences of these factors relative to nursery pollination (Holland & Fleming, 2002; Ashman et al., 2004; Elzinga et al., 2005; Kephart et al.). Even in taxa with moderately broad geographic distributions (e.g. S. dioica), the major pollinators are variable in their presence and relative importance in natural populations (Carlsson-Granér et al., 1998; Westerbergh, 2004), influencing the potential for emerging mutualisms to persist over time.
This feature highlights some of the most recent findings of leading researchers studying pollination, trait evolution and seed predation in the Caryophyllaceae. However, these papers also illustrate the limitations of our current knowledge and suggest new directions for collaborative efforts. Elucidating the origin and persistence of pollination mutualisms will require merging not only different approaches (e.g. genetic, ecological), but also empirical data sets gathered systematically with respect to specific, a priori measurement outcomes for species in multitrophic systems whose interacting populations are spatially and temporally variable. The ecological contexts of pollination will likely condition the sign and the trajectory of these interactions; hence it will be important to document the role of abiotic factors and Allee effects as well as the presence of fungal and parasitoid associates, frequent copollinators, and ‘cheaters’, which threaten mutualisms by accruing benefits without incurring the essential costs. Recent modeling efforts are also advancing our understanding of the role of competitive asymmetry in stabilizing and conserving mutualisms (Bronstein et al., 2004). Thus, an important goal of greater interdisciplinary collaboration might be to facilitate the design of innovative approaches for conserving species interactions amidst habitat fragmentation and population isolation.