TREE vol. 7, no. 8, August 1992 185-190 12 Slack, N.G. (1992) Biol. Consetv. 59, 233-241 13 Siiderstriim, L., HallingbBck, T., Gustafsson, L., Cronberg, N. and Hedenls, L. (1992) Biol. Conserv. 59, 265-270 14 Wyatt, R. (1992) Biol. Conserv. 59, 99-l 07 15 Longton, R.E. (1992) Biol. Conserv. 59, in Plants: Regulation and Function (van 89-98 Groenendael, J. and de Kroon, H., eds), 16 During, H.J. and Van Tooren, B.F. pp. 153-176, SPB Academic Publishing (1987) Trends Ecol. Evol. 2, 89-93 20 Shaw, J. (1990) Evolution 44, 17 Miles, C.J. and Longton, R.E. (1990) 441-446 Bat. J. Linn. Sot. 104, 149-173 21 Hedderson, T.A. (1992) Biol. Conserv. 18 Bremer, P. and Ott, E.C.J. (1991) 59, 113-120 Lindbergia 16, 3-18 22 Hodgetts, N.G. (1992) Biol. Conserv. 19 During, H.J. (1990) in Clonal Growth 59,259-264 Orchid Pollination Biology Orchids display many unsurpassed floral specializations, as both rewarders and frauds in their interaction with animal pollinators. Accumulating evidence indi- cates that their floral evolution is driven bg pollinator traits and that expenditure for maximized sexual reproduction is parcelled out over their lifetimes in strategies for coping with pollinator and resource limi- tations. Recent advances in orchid polli- nation 6ioIogy center mainly on floral evolutionary processes, pseudocopulation and other deceptive pollination systems, and flower and fruit production in relation to costs of sexual reproduction. One of Darwin��� s passions was orchids. With their ��� endless vari- ation��� in floral contrivances for sexual reproduction, these plants provided him with a wide range of products of natural selection��� . The emergence of orchid pollination biology is thus intimately linked with the emergence of evolutionary biology. Subsequent research has continued to reveal far-reaching adaptations in orchid-pollinator interactions while shedding light on crucial issues for our understanding of plant evolution in general. Orchidaceae is the most species- rich plant family, containing per- haps 25000 species. Typically, its members are animal-pollinated tropical herbs that flourish in moist forests a smaller proportion is restricted to temperate regions2. The family is characterized by a fusion of the stamens and styles into a bisexually functional organ, the column, which maximizes pollen export and import via some specific body-part of the pollinator. The tip of the column presents pollen grains in coherent masses, the pollinia, which usually attach as one entire Anders Nilsson is at the Dept of Systematic Botany, PO 80x 541, S-751 2 I Uppsala, Sweden. 0 1992, Elsewer Science Publishers Ltd (UK) ��� package��� via column-derived viscid matter onto the departing flower visitor. The pollinia are dispensed intact or as subunits onto the stigma of a single or a few flowers. The stigma captures the pollen-masses from the animal upon contact, an ��� all-or-nothing��� event that can lead to the production of over two million tiny, dust-like wind- dispersed seeds that are thus often fathered by a single donor plant. Most orchids provide a pollinator reward (e.g. nectar) like other angio- sperms, but 8000-10 000 species act by deceit, i.e. are frauds that provide no reward3. In both groups, evolution has frequently led to elaborate extremes in floral traits. Nectar is assumed to be the original and principal floral reward in or- chids, and its position varies from completely superficial to hidden at the base of long corolla tubes (spurs). Though some orchids are self pollinating (autogamous) and a few are pollinated by birds, most are pollinated by insects. Recent work supports the idea that natural selec- tion is the principal process behind orchid floral evolution and that the orchid-pollinator relationship is best viewed as a very asymmetric one, with most of the evolution occurring on the plant side. Asymmetric plant-pollinator relationships Orchids need pollinators, but are pollinators dependent on orchids? In general, the more specialized or inaccessible the reward, the narrower the cadre of pollinators. The extremes in nectary length are found among the hawkmoth- pollinated star orchids of Madagas- car, some species having spurs 30-40 cm long. About 90 species in five genera have spurs exceeding L. Anders Nilsson 8 cm in length. For a guild of such orchids the sole evolutionary agent is the generalist forager hawkmoth Panogena lingens4. Angraecum arachnites is associated only with a long-tongued morph within P. lingens5. Darwin��� hypothesized an evolutionary race between the spur length in Angraecum sesquipedale and the tongue length in its hawkmoth pollinator: only those individual orchids that compel individual hawkmoths to push against the column, due to their spurs being longer than the hawkmoths��� tongue, experience successful pollen export and import thus, any evolution towards longer-tongued hawkmoths would drive selection for longer orchid spurs (Fig. 1). Manipulations of spur length on moth-pollinated Platan- thera orchids corroborate Darwin��� s hypothesis for flower depth evol- ution: spur shortening reduces plant fitness, and the greatest plant fitnesses are achieved when spurs are longer than the moths��� tongues6f7. Any evolution in tongue length is driven by competition between individual moths for ac- cess to nectar in various plants this also drives any orchid-moth interactional evolution. Pollination ecotypes adapted to the traits of local pollinator(s) are produced by natural selectiot?, and column structures are subject to strong selection in response to the limited distribution of scaleless surfaces for pollinia attachment on the lepi- dopterans9. In contrast, there is no evidence that these pollinators are dependent specifically on the orchids for nectar. Pollinating bees are evidently not dependent on orchids. Neotropical fragrance-offering orchids depend 255

. TREE vol. 7, no. 8, August 1992 POlllnator tongue Floral tube length Fig. I. The hypothesis for the evolution of deep flowers in orchids. Long-tongued pollinators are not compelled to contact the sexual organs of short-tubed flowers, which therefore receive fewer pollinations than long tubed flowers. The latter thus pollinate freely with each other via the hawkmoth pollinators. Any evolution towards longer-tongued hawkmoths would drive selec- tion for longer orchid spurs. Reproduced horn Ref. 7, with permission. on fragrance-collecting male euglos- sine bees for sexual reproduction, but the bees are not dying, migrating, or deferring fragrance- seeking activity when the or- chids are not in bloom���. ��� Since the orchid-euglossine relationship can be highly specific, mutational change in the production of key attractive compounds (fragrance casting) provides a possibility for sympatric speciation in the orchids in which unique chemotypes attract novel beesl��� J2. Male territorial display activity is stimulated by access to fragrance13, but usually only small fractions of euglos- sine populations engage in orchid pollination��� 4. Similarly, oil-collect- ing mellittid bees exploit oil- offering Disperis orchids much less than their principal oil host plants15. In fragrance-offering di- oecious Catasetum, where forcible pollinia loading on bees at the male flowers triggers aversion reactions in bees, which subsequently prefer the different display of the female pollinia-receiving flowers, the male and female floral displays have become very unlike���. ��� Deceptive orchids are apparently nonmutualists, although hypotheti- cal benefits to pollinators are not lacking17. However, the fraud orchids are wholly dependent on their pollinators since without them their sexual reproduction is usually nil. Fraud orchids therefore have been driven into far-reaching specific floral specializations by pollinator traits for instance, chemically at- tracted male lepidopterans pol- linate nonrewarding Epicfendrum panicuIaatum in search of pyr- rolizidine alkaloids that are used as precursors in pheromones for courtship and defence18. The pollinator-driven speciation in fraud orchids is illustrated by Ophrys. The male aculeate hymenopteran pollinators have in- nate species-specific preferences for chemical attractants Ophrys taxa serviced by different male aculeates produce different floral fragrance bouquets19. Speciation in Ophlys is always a consequence of change in pollinator even intra- specific taxa are pollinated by dif- ferent families of bees*O. Fragrance casting and subsequent selective fine-tuning of floral traits may ex- plain why phylogenetically disparate hymenopterans pollinate closely related orchid taxa. Clearly, floral evolution in orchids - rewarders as well as frauds - is governed by pollinators��� traits in very asymmetric relationships. Two such pollination systems that have received par- ticular attention recently are food deception and pseudocopulation. Food frauds Many fraud orchids are food- deceptive. They detain insects with a variety of dummy signals and nectarless or pollenless structures. Food frauds either mimic certain co-blooming rewarding flowers or function without models. These two strategies result in different evol- utionary outcomes. In mimic-model systems, fitness in the mimic orchid is sensitive to variation in relative frequency, density and display of models. In Orchis caspia and Cephalan- thera rubra, fruit set is higher in the presence of their food-flower models21*22. However, it is not clear if this is due to mimic-model interaction or simply results from the pollinator abundance near food plants. Typically, mimic-model sys- tems engage solitary bees which have a strong innate preference for foraging on only one or a few plants such bees thus have the host plant fidelity required to lead orchids into successful food-flower deception. Associative learning by pollinators of a model controls the selection on the mimic, whose variation in floral traits is thus constrained by the model. Sometimes several models are mimicked by one fraud. Thelymitra and Diuris species mimic guilds of spring-flowering pollen rewarders and legumes, respectively23,24. Food-fraud orchids without a model, however, often bloom greg- ariously and exhibit display poly- morphisms. Generalist fpolylectic) foragers such as bumblebees act as pollinators and have constrained flowering to periods when emerg- ent, naive bees can be deceived25. Instinctive ��� reaction images��� are crude but associative learning is excellent in polylectic insects, causing selection for both flam- boyant floral displays that initially elicit instinctive reactions and exten- sive intraspecific floral variation that offsets subsequent associative learning and thus increases the number of visits before abandon- ment26,27. More detailed work is needed on variation in orchid repro- ductive success in both nonmodel and mimic-model deception. Pseudocopulation The phenomenon of pseudo- copulatory pollination is exclusive to orchids. It involves male in- sects - mostly aculeate hymenop- terans. Male aculeates usually emerge before their females, a trait generally believed to maximize mating success. Blooming in pseudo- copulatory orchids occurs when male competition is at a peak, that is when there is a relative deficit of virgin females28. The main question concerns the nature of the exact floral attractant mechanism that has the power to excite the males sexually. The flower-lip display of these orchids is dark brown, brick or maroon and often has a distinct pilosity and central speculum, resembling the body and crossed wings of the female insect. Early studies in Ophrys indicated that while visual and tactile cues are important, floral fragrances play a key role in male attraction29. Male aculeates are attracted instinctively to their females��� pheromones, and floral fragrances in Ophrys contain compounds identical to those in the pheromonal secretions of the re- 256