Summary Sexually deceptive orchids are predicted to represent a special case of plant speciation where strong reproductive isolation may be achieved by differences in floral scent. In this study of Australian sexually deceptive Chiloglottis orchids, we performed choice experiments to test for wasp pollinator specificity in the field; identified the compounds involved in pollinator attraction by gas chromatography with electroantennographic detection (GC‐EAD), gas chromatography with mass selective detection (GC‐MS), chemical synthesis and behavioural bioassays; and mapped our chemical findings on to a phylogeny of the orchids. Field experiments confirmed pollination is a highly specific interaction, but also revealed a pool of nonpollinating ‘minor responder’ wasps. Six novel compounds, all 2,5‐dialkylcyclohexan‐1,3‐diones, called ‘chiloglottones’, were discovered to be involved in pollinator attraction. Bioassays confirmed that pollinator specificity has a strong chemical basis, with specificity among sympatric orchids maintained by either different single compounds or a variation in a blend of two compounds. The phylogenetic overlay confirmed that speciation is always associated with pollinator switching and usually underpinned by chemical change. If the chemical differences that control reproductive isolation in Chiloglottis have a strong genetic basis, and given the confirmed pool of potential pollinators, we conclude that pollinator‐driven speciation appears highly plausible in this system.

Caladenia is very unusual in that it contains species that attract pollinators by two different strategies, food and sexual deception. Among the sexually deceptive species, baiting for pollinators has shown that within populations orchid species are typically pollinated by a single species of thynnine wasp. However, some wasp species can be pollinators of more than one species of orchid usually when their ranges do not overlap. There is a trend for closely related orchids to exploit wasps from the same genus, with different lineages of orchids often pollinated by different genera. Very little is known about pollination of food-deceptive Caladenia species, although it is evident they attract a suite of generalist food-seeking insects. Food-deceptive species have a higher pollination rate than do sexually deceptive species. Studies of population genetics and pollen movements are few, although they suggest a pattern of fine-scale genetic structuring within populations, owing to predominantly restricted seed dispersal and low genetic differentiation among populations as a consequence of rare long-distance seed-dispersal events. Both evolutionary and ecological research of Caladenia will greatly benefit from a better understanding of the insect species involved in pollination, their ecological requirements and the ecological and genetic consequences of food and sexual deception.

Pollination by sexual deception in orchids is characterised by a high degree of pollinator specificity, which may account for the rarity of natural hybrids within the group. Only one such hybrid has been formally recognised in Australia, Chiloglottis × pescottiana R.S.Rogers, which has intermediate floral morphology between its putative parents, C. valida D.L.Jones and C. trapeziformis Fitzg. In this paper, genetic and morphometric analyses confirm the hybrid origin of this taxon. Allozyme analysis of C. trapeziformis and C. valida revealed fixed allelic differences at four ‘diagnostic’ loci and significant frequency differences at three other loci. In all cases, C. × pescottiana exhibited fixed heterozygosity at the diagnostic loci. Multidimensional scaling of both the genetic data and seven morphometric traits revealed distinct clusters of C. trapeziformis and C. valida while C. × pescottiana formed an intermediate cluster between the two parents. To test for genetic compatibility between C. trapeziformis, C. valida and C. × pescottiana, a series of reciprocal artificial crosses were performed. In all cases, the percentage of capsules formed was at least as great for between-species crosses as for within-species selfs and crosses (range 75–100%). No significant differences in the percentage of seed with normal embryos was detected between self- and cross-pollinations within C. trapeziformis (range 77–81%), C. valida (range 59–74%) and C. × pescottiana (range 30–51%), but the percentage of normal embryos was notably lower in C. × pescottiana. The cross C. trapeziformis female by C. valida male produced significantly more normal embryos (90%) compared with the reverse cross (46%). Artificial backcrosses of C. × pescottiana to C. valida and C. trapeziformis had lower percentages of normal embryos when C. × pescottiana was the pollen donor (39–43%) rather than recipient (62–68%), suggesting reduced pollen viability in the latter taxon. The size of F2 embryos in C. × pescottiana seed capsules was smaller than the embryos of both C. valida and C. trapeziformis. Despite confirmation of hybridisation, little evidence for backcrossing was found. Thus, while the specific pollinator relationships may occasionally break down in these sexually deceptive orchids, reduced viability of hybrid pollen and F2 seed, and inefficient pollination of the hybrid, may minimise introgression. It is concluded from the available evidence that hybridisation has not been a major evolutionary factor in the diversification of sexual deception worldwide.

Australian sexually deceptive orchids are typically highly pollinator specific, each species having a single unique hymenopteran pollinator species. Pollinator specificity in six of the nine described species in the Chiloglottis gunnii Lindl. complex was investigated by using field pollinator-choice tests, with Chiloglottis taxa translocated within and among biogeographical regions. Specific pollinators revealed the existence of five undescribed cryptic taxa in the C. gunnii complex, three within C. pluricallata D.L.Jones and two within C. valida D.L.Jones, in addition to the six described species. Of the 11 Chiloglottis taxa, 10 had a single thynnine-wasp pollinator throughout their sometimes large distributions, whereas one, C. valida, had a second pollinator in parts of its distribution. Eleven pollinators belonged to the genus Neozeleboria and one to Eirone. Pollinator-choice testing showed that cross-attraction of pollinators occurs between three geographically isolated Chiloglottis taxa on the New South Wales (NSW) New England Tableland and taxa in the South Eastern Highlands of NSW and Victoria. The data suggested there is sharing of chemical attractants and supported the recognition of at least five odour types within Chiloglottis, each encompassing one to three orchid taxa and their pollinators. The following two broad generalisations are made: (1) there is no cross-attraction of pollinators among sympatric Chiloglottis species, i.e. sympatric orchid taxa do not share attractant odours; and (2) all Chiloglottis species have different specific pollinators, although they may share attractant odours allopatrically. Some 28 thynnine-wasp species were attracted as minor non-pollinating responders to Chiloglottis taxa; five of these were pollinators of other Chiloglottis species. These wasps were much more taxonomically diverse than the pollinators, belonging to six genera, and suggest that some orchid-odour components are widely shared within the sex pheromones of the Thynninae.

Orchid species belonging to the sexual-deception pollination syndrome exhibit highly specific, usually one-to-one, relationships with their pollinators. This specificity is mediated by the orchid’s mimicry of the sex-attractant pheromones emitted by females of the pollinator species. Chiloglottis valida D.Jones sensu lato is a widespread, sexually deceptive, terrestrial orchid found in south-eastern New South Wales, and eastern and southern Victoria from sea level to at least 1600 m in the Australian Alps. Flowers from 38 C. valida s.l. populations from throughout this area were compared in field choice experiments for the specificity of attracted pollinator species. Four potential pollinator wasps in the thynnine genus Neozeleboria Rohwer were attracted. The data demonstrate the existence of two attractant odour types among C. valida s.l. and its pollinators, and support the recognition of two partially sympatric cryptic species in the orchid, each with two potential pollinators. The copheromone pollinator pairs replace each other on the altitudinal gradient, albeit with some overlap. In alpine areas the pollinators of the two cryptic orchid species are themselves sibling species within Neozeleboria monticola Turner s.l. The results indicate that C. aff. valida, the sister species of C. valida s.s., has two geographically replacing pollinators.

No abstract

No abstract

Divergence in sexually deceptive orchids is thought to occur through shifts in the attraction of specific pollinators, a process that is mediated by changes in the floral odours that lure sexually excited male insects. We investigated the origin of reproductive isolation in a sexually deceptive species complex of Chiloglottis R.Br. (Orchidaceae: Diurideae). Two geographically separated montane regions in eastern Australia were sampled, each containing sympatric pairs of orchid taxa presently found under the name, Chiloglottis pluricallata. Behavioural tests confirmed at least three distinct orchid taxa that specifically attract different pollinators. An artificial crossing experiment among two taxa from one region demonstrated their interfertility, and confirmed isolation to be a function of pollinator attraction. A phylogeographic analysis using amplified fragment length polymorphisms (AFLPs) indicated that samples from each geographical region are most closely related, a pattern consistent with in situ or sympatric divergence. However, an extensive population genetic study on two taxa from one region failed to entirely reject the possibility of intertaxon gene flow. Although clear genetic differentiation of the taxa is evident in two out of three sites where both grow in direct sympatry, overall, the two taxa are not strongly distinguished by AFLP markers. The reconstruction of a simple bifurcating pattern of divergence may be confounded by a combination of contemporary population-level processes operating within each taxon, the retention of ancestral polymorphism or intertaxon gene flow.