Constraints peculiar to forest canopy habitats helped shape epiphyte natural history, but the selective forces are not always apparent. Some characters and lineages appear to have been affected more than others; for instance, iteroparity is nearly routine in the group, whereas breeding mechanisms and modes of pollen and seed dispersal are much more diverse. Multiple paths to similar ends complicate the search for a common theme. Enough data are available, however, to offer tentative judgments on several aspects of the plant life cycle that permit success in tree crowns. This chapter deals with such aspects and the factors responsible for their existence.
Pollination: identity of vectors
If breeding systems differentiate terrestrial from canopy-based pte-ridophytes, the fact remains unreported. Comparisons of angiosperms are easier, and pollination has been studied in numerous epiphytic flowering plants, especially neotropical Orchidaceae. Pollinators of these taxa tend to be more species-constant and specialized than those serving nearby terrestrials, although sharing of pollinators is sometimes possible. Avians are especially important in northern South America, where nectar-feeding birds and the flora they serve reach unparalleled diversity. Large, heavily ornithophi-lous families include Bromeliaceae, Ericaceae, Gesneriacese, and Lorantha-ceae; birds frequently pollinate Cactaceae, Marcgraviaceae (Norantea), and Rubiaceae (Ravnia, Manettia) as well. The relationship between epiphytism and avian pollination is particularly apparent in families with diverse floral syndromes. Ornithophily in Bignoniaceae is rare except in two epiphytic genera: Gibsoniothamnus is entirely pollinated by birds, Schlegelia partly so (Gentry and Dodson 1987b). Chiropterophily is well developed in some Central American vrieseas (Bromeliaceae) and Marcgravia but is definitely a minor relationship overall. Even less common is the pollination by small rodents of two epiphytic Costa Rican Blakea species (Lumer and Schoer 1986).
Insect visitors are extremely diverse. Pleurothallidinae, a neotropical assemblage of about 3,800 mostly epiphytic orchids, along with largely paleotropical Bulbophyllum and many small-flowered relatives, are generally pollinated by flies (e.g., Bradesia, Drosophila, tachnids). Moths service certain Cactaceae and orchid taxa (e.g., Epidendrum, Polyradicion)-, scarab beetles pollinate Costa Rican aroids (Bawa et al. 1985). Inflorescenses of Philodendron bipinnatifidum heat up and emit fruitlike scents sought out by Dynastinae beetles (Gottsberger 1986). The best known pollination liaison is that between male euglossine bees and the Cymbidieae and Maxillarieae - more about this later. Families with inconspicuous reproductive organs are Araliaceae, Moraceae, Piperaceae, Myrsinaceae, and Urticaceae (the last two families are only marginally epiphytic); those of Melastomataceae are similarly drab except for Blakea and Topobea which produce large bright flowers (Gentry and Dodson 1987b). Anemophily is rare in canopy-based flora, if it exists at all: The best candidates are some Peperomia, the two epiphytic grasses, and a couple of mistletoes.
Access to particular pollination syndromes has varied among the epiphytes owing to diverse origins and disparate habitats. Little can be said about phylogenetic constraints, but plant size, population structure, height in the canopy, type of forest, local climate, and the co-occurrence of taxa with the same floral syndromes have clearly influenced reproductive patterns. In humid forest at La Selva, Costa Rica, Bawa et al. (1985) discovered that epiphytic and nonepiphytic vegetation alike shared vectors that were largely restricted to the same canopy level. Surveys of 143 trees were made in order to identify where major groups of pollinators were most active. Precise data were collected for 58 woody species, and floral characteristics were used to infer primary visitors for the rest. In the subcanopy, where epiphytes were most numerous, pollinator diversity was richest. Whereas 44.2% of the canopy overhead was attended by medium-size to large bees (e.g., Euglossa, Centris), only 17.6% of the subcanopy was so visited; the rest was serviced mostly by hummingbirds and such insects as small bees, beetles, butterflies, and sphinx moths.
Production of a few showy flowers presenting high caloric, unusually fragrant, or otherwise readily detectable and specialized rewards is common in epiphytes and probably mandated by small plant size and diffuse dispersion (Ackerman 1986). If these plants were utilizing generalized vectors or were less attractive to pollinators, competition with more floriferous trees and lianas could reduce fecundity. In effect, the modest resources available to scattered epiphytes for sexual reproduction have to be invested not only in seeds of appropriate size and mobility, but also in faithful, if not exclusive, pollinators. Suitable animal vectors are wide-ranging trapliners and those foragers susceptible to floral deception. Most canopy-dwelling Bromeli-aceae, Cactaceae, Ericaceae, Gesneriaceae, Melastomataceae, Rubiaceae, and many orchids are visited by such animals. Common qualities of these epiphytes are long reproductive periods effected by individual plants that produce each day for weeks or months a few conspicuous, high-reward flowers protected from generalized foragers. Three deception systems also foster reliable visitation; Orchidaceae illustrate all three: Large canopy-based Epi-dendrum is predominantly a food-deception group, and some Araceae and Dracula mimic the substrates of fungus gnats. Several slipper orchids offer nesting cues. Mate mimicry, although best known in pseudocopulatory terrestrials, also occurs in epiphytic Trichoceros and Oncidium henekenii.
An understanding of breeding systems and their advantages to particular taxa is generally hindered by lack of data on pertinent habitat characteristics and the difficulty of measuring meaningful genetic structure within populations (Loveless and Hamrick 1984). Except for a few cases like that offered below to illustrate the consequence of habitual outcrossing versus autogamy in two Tillandsia species, no correlations among epiphytes have been demonstrated between breeding system and habitat type or various life history characteristics. Moreover, even if a single mating pattern were to serve all canopy-adapted flora, genetic structure would vary if for no other reason than the influence of seed dispersal. Not only do seeds move independently of pollen, but embryos carry two rather than one allele for each gene.
Orchidaceae demonstrate how disparate even those breeding mechanisms maintained by related epiphytes can be. Most family members are self-compatible (Dressier 1981), but quite a few regularly outcross, sometimes with floral contrivances featuring almost legendary complexity and specificity. Best known of the allogamous forms in tropical forests are the cymbidioids (e.g., Catasetum, Stanhopea, Coryanthes) whose flowers attract pollen vectors via complex combinations of volatiles. Males of a single local euglos-sine bee population often visit a single plant species - occasionally even just one among two or more intraspecific chemotypes - to forage not for food but for the fragrance that reputedly aids subsequent mate procurement and/ or reproduction (Williams and Whitten 1983). Such foragers are attracted to no other orchid although members of other plant families are used as nutrient sources. The euglossines that service the cymbidioids are also nota-
ble for long lives and ability to forage over great distances - additional qualities that have promoted orchid and bee speciation and maintained plant hyperdispersion (Benzing in press). Horticulturists have long exploited orchids for their extraordinarily weak reproductive barriers that reflect histories of ethological isolation. Most insect pollinators of Orchidaceae are, however, weaker fliers than the euglossines, and quite a few may not be particularly faithful. Publicity accorded the most spectacular coevolved orchids has obscured the fact that the vast majority engage in more mundane sexual liaisons, and some do not attract pollinators at all.
Autogamy obliged by cleistogamy has been reported in a number of Orchidaceae (Dressier 1981). Occasional flowers may be chasmogamous. Regularly self-pollinating populations tend to be abundant and widely substratum-compatible, often to the point of earning a reputation for weediness (e.g., Caularthron bilamellatum and Spathoglottis plicata). Both allogamy and autogamy foster profligacy in some other epiphytic taxa. As a group, Tillandsia tends to be allogamous, yet some of the most widespread species are regular, if not obligate, selfers. Ball moss (Tillandsia recurvata) is a selfer that ranges across most of tropical and subtropical North and South America; in heavily infested tree crowns, its biomass may exceed that of host foliage (Fig. 7.3). By way of contrast, the most broadly distributed of all bro-meliads is Spanish moss (T. usneoides) whose flowers fail to yield capsules in at least certain southeastern United States populations unless cross-pollinated. The only published study of breeding system influence on isozyme pattern revealed much homozygosity within, but great variation among, local populations of T. recurvata. Tillandsia ionantha, a less widely distributed outcrossing Mesoamerican taxon, possessed more alleles per locus but little variation in allele frequency among populations (Soltis, Gilmartin, Rieseberg, and Gardner 1987). Cleistogamous T. capillaris must rank among the most inbred of all members of this predominantly epiphytic genus (Gilmartin and Brown 1985). Related taxa probably exhibit less consistent genetic structures. Wide-ranging species such as T. paucifolia feature self-compatible populations in some regions (e.g., South Florida) but will not self-fertilize elsewhere (e.g., northern South America). Catopsis nutans is dioecious in Central America; South Florida populations produce perfect flowers only.
Marcgraviaceae appear to be mostly autogamous (Gentry and Dodson 1987b). Another more diverse group of purportedly self-fertilizing epiphytes is that associated with ants. Numerous ant nest-garden species and certain ant-fed ant-house epiphytes regularly produce fruits without visitation by pollinators (e.g., Hydnophytum spp., Anthurium gracile, Epiphyllum phyl-
lanthus, Aechmea mertensii), a practice possibly fostered by their aggressive zoobionts (Madison 1979a). Any resulting homozygosity may pose fewer problems for obligate ant nest-garden species than for plants that must accommodate to more diverse rooting media. Additional factors apparently override myrmecophily in dictating mating capacity in some situations. A Costa Rican Tillandsia caput-medusae proved to be self-incompatible in a pollinator-free greenhouse, whereas a less colorful, equally ant-prone collection from southern Mexico fruited heavily in the same setting. Gentry and Dodson (1987b) believe that self-compatibility and autogamy are much more common in epiphytes than in moist, tropical-lowland terrestrials. Ack-erman (1986) agrees and considers these features to be complementary, if not obligatory, for epiphytes with hyperdispersed populations or specialized pollinators such as trapliners.
Was this article helpful?