Flower form and pollinator diversity in the Middle Eocene.
Flower form and pollinator diversity in the Middle Eocene of British Columbia and Washington. M. Sc.Thesis, Department of Botany, University of Alberta, 1991. DOI
Summary
Study area.
The floral and insect material analyzed in this study are from sediments deposited during the Middle Eocene in a series of lakes located in central British Columbia and northern Washington. The specimens are compression-impression fossils in fine grained, laminated shales from eleven localities. Potassium-argon dating and biostratigraphic correlation dates the sites at 48 – 50 m. y.
A diverse flora has been described from this area, with the most comprehensive studies done on the compression flora at Republic and the permineralized material from Princeton. Insects and fish have also been described. Paleoecological studies have analyzed micro-habitat and seasonal distribution and several conclusions on paleoclimate and paleogeography have been achieved.
Staminate inflorescences
Diversity was high among forty specimens of staminate inflorescences (“catkins”) from British Columbia (Princeton and McAbee) and Republic, Washington. Four multicharacter similarity groups were generated for seventeen of the specimens using cluster analysis; the remaining specimens (“outliers”) were highly diverse and did not consistently associate with any of the four groups. Groups I and II consisted of complex racemes with large triangular primary bracts, the two groups differing in width and density of the catkin, angle of the primary bracts and pollen sac size. Group III specimens were narrow spikes with multiple oval bracts and exserted stamens and Group IV specimens had the smallest and least dense bracts, the most exserted stamens and largest pollen sacs of all other groups. Groups I and II had characteristics similar to the Betulaceae; three catkins associated with leaves of the fossil Betulaleopoldae are in these groups. Group III showed affinities with the Juglandaceae. There was a striking locality factor in the grouping. Groups I and II contain only specimens from Princeton; Republic is represented in Groups III and IV and in the majority of outliers. The higher diversity at Republic cannot be accounted for by differences in age or in depositional habitat, and may be the result of microhabitat differences.
Flower form
Fifty specimens of flowers and inflorescences from British Columbia (Horsefly, One Mile Creek, Falkland, McAbee) and Republic were classified into multicharacter similarity groups using cluster analysis and discriminant function analysis. The characters with the lowest variability were the number of tepals (mode of 5), the degree of connation of the perianth (about 60%) and the length/width ratio of both perianth and flower (about 1). Cluster analysis generated 6 similarity groups among twenty-six of the specimens; using discriminant function analysis, the characters that were most important in differentiating between the groups proved to be perianth width, perianth connation and shape of the tepals. An additional five floral form categories were added: a group identified by cluster analysis based on androecial characters and 4 outlying specimens. Groups 1 and 3 consisted of large, robust perianths of Florissantiaphysalis and Pistillipollianthuswilsonii. Groups 4, 5 and 6 were small, bowl-shaped, partially connate perianths that varied independently in the three predictor variables; Group 7 differed from these three groups primarily in perianth size. The remaining flower form categories include a small stereomorphic, highly connate form (Group 2), a large zygomorphic flower (Group 8), a number of bowl-shaped perianths with numerous, exserted stamens (Group 9), a staminate raceme (Group 10) and a connate perianth with a sub-apical constriction (Group 11).
The sample of flowers from British Columbia-Republic shares many floral characters and many floral form categories with other samples of Paleocene-Eocene flowers, notably those from the Claiborne formation of Tennessee; however, each area has distinct forms as well, indicating that there may be differences in the diversity of pollinating agents. DOI
Insect diversity
Insect specimens in the four anthophilous orders (Coleoptera, Hymenoptera, Diptera, Lepidoptera) were examined from nine localities (Driftwood Creek, Quilchena, Horsefly, One Mile Creek, Tulameen Road, Princeton Firehall, McAbee, China Creek and Republic) in British Columbia and Washington. Thirty taxa (at the family or superfamily level) were represented in 129 specimens, with the most common families being Bibionidae and Ichneumonidae. Almost half of the coleopteran specimens were isolated elytra and identification was not possible. Coleoptera was the most diverse of the four orders, with specimens identified in 10 families and 2 superfamilies. Fifteen of the families identified are new records for the Middle Eocene of British Columbia-Republic, bringing the diversity of insects in these four orders in the area to 34 families. Nine families (Lucanidae, Cantharidae, Cleridae, Argidae-Diprionidae, Scoliidae, Pompilidae, Rhagionidae, Geometridae, Noctuidae) represent the earliest occurrences of the taxa in North America and five (Cantharidae, Cleridae, Argidae-Diprionidae, Geometridae, Noctuidae) are the earliest records of the respective famililies in the stratigraphic record. Significant for the study of paleoanthecology are the new records for lepidopterans. The insect diversity of British Columbia-Republic (34 families in the four orders) is lower than at the more southerly, contemporaneous Green River (59 families); only 20 families are present in both areas.
Pollinator diversity
Based on a literature review of the feeding habits of insects, at the family level, in the four anthophilous orders (Coleoptera, Hymenoptera, Diptera and Lepidoptera) and of their stratigraphic occurrence, a “short list” of potential pollinators of Middle Eocene flowers was compiled. At least 28 families of pollinating insects were available to flowers analyzed in this study; the most important are Mordellida, Syrphide, Bibionidae, Scoliidae, Vespidae, Geometridae and Noctuidae. In addition, Acroceridae, Papilionidae and Lycaenidae are present at Green River and Anthophoridae is found in Middle Eocene Germany. Fossil Nemestrinidae and Bombyliidae, present in Mesozoic strata in Europe, may have been important pollinators in North America as well. There is circumstantial evidence that several apoid hymenopterans (Melittidae, Fidellidae, Megachilidae and Meliponinae) had evolved by the epoch. Bird and bat pollinators are absent from sediments in or previous to the Middle Eocene, but there is a possibility that small marsupials and mammals were pollinators.
Pollination syndromes and pollinating agents
For each of the flower form categories, the most probable pollinating agents were selected from the range of anthophilous families present at British Columbia-Republic in the Middle Eocene. The larger pollinator fauna available in North America and world-wide at the time is also considered. The pollinators appropriate to categories 1 and 3 are large Scarabaeidae, Cantheridae, Cerambycidae and Chrysomelidae, short-haustellate Syrphidae, Empididae and Bibionidae and non-specialized Scoliidae and Vespidae, as well as large generalist hymenopterans. The optimal pollinators for category 2 flowers are long-haustellate, hovering Syrphidae, as well as alighting Geometridae and Noctuidae. The predicted optimal pollinators for categories 4, 5, 6 and 7 are Mordellid beetles and small generalist coleopterans, such as Elateridae, Tenebrionidae and Curculionidae, small hymenopterans with no modifications for feeding on flowers (Cynipidae, Ichneumonidae, Braconidae and Formicidae) and small dipterans (Bibionidae, Syrphidae, Empididae, Mycetophilidae, Sciaridae). When the Green River insect fauna is considered as well, a wide range of small coleopteran, hymenopteran and dipteran families are added to the pollinator spectrum. Among Category 9 flowers, the most probably pollinators are small coleopterans (Mordellidae, Elateridae, Tenebrionidae, Curculionidae); exsertedness of the stamens may indicate wind-dispersal of pollen. Wind as a pollinating agent is the most probably scenario for the non-amentiferous category 10 inflorescence. The optimal pollinators available to the highly connate corolla of category 11 are Syrphidae, Vespidae, Scoliidae, Noctuidae and Geometridae, and, present at Green River, Sphecidae, Acroceridae and four lepidopteran families; other Bombylioidea and Anthophoridae are present by the Middle Eocene in Europe.
In the large leguminous flower of category 8, a proboscis of 15 – 20 mm is necessary to reach the base of the flower. Apoid hymenopterans, which pollinate extant flowers of similar form, are not present as fossils by the Middle Eocene in North America. I propose that an alternate group of long-tongued pollinators, already present in the fossil record, pollinated flowers with this syndrome and was later competitively replaced by evolving apoid bees. Appropriate pollinators include Syrphidae, found at British Columbia-Republic, Acroceridae, present at the coeval Green River, and Nemestrinidae and Bombyliidae present in Mesozoic strata in Europe; an anthophorid bee, the earliest record of apoid hymenopterans, is present in the Middle Eocene of Europe.
The fossil record of pollinators is sufficient to account for flower form seen in the specimens from British Columbia and Washington analyzed in this study. It is not necessary to infer, from phylogenetic or biogeographical evidence, additional pollinators not found as fossils. In the absence of a diverse bee fauna, syrphid flies, bibionids, bombylioids and non-apoid hymenopterans may be significant pollinating agents for complex flowers.