1
Preprint similar to pp. 1196-1203 in Evolution 67 (2013)
Brief Communication
Floral Paedomorphy Leads to Secondary Specialization in Pollination of Madagascar Dalechampia (Euphorbiaceae)
W. Scott Armbruster 1, 2, 3,7, Joongku Lee 4,6, Mary E. Edwards5, and Bruce G. Baldwin4
1 School of Biological Sciences, University of Portsmouth, Portsmouth PO1 2DY, UK
2 Institute of Arctic Biology, University of Alaska, Fairbanks, AK 99775-7000, USA
3 Department of Biology, NTNU, NO-7491 Trondheim, Norway
4 Department of Integrative Biology and Jepson Herbarium, University of California, Berkeley, CA 94720-2465, USA
5 Geography and Environment, University of Southampton, Southampton SO17 1BJ, UK
6 Present address: Korea Research Institute of Bioscience and Biotechnology, 125 Gwahak-ro, Yuseong-gu, Daejeon 305-806, South Korea
7 To whom correspondence should be addressed:
[email protected]
The traditional evolutionary interpretation of Von Baer's "laws" of embryology is that retention of early developmental forms into adulthood (paedomorphosis) leads to the evolution of simpler or more generalized morphology and ecology. Here we show that paedomorphosis can also be involved in an increase in ecological specialization, in this case of plant-pollinator relationships. A paedomorphic transition from generalized pollination (by several functional types of pollinators) to specialized pollination (by one or a few species in one functional type) occurred in a clade of endemic Madagascar vines (Dalechampia spp., Euphorbiaceae). This evolutionary transition involved staminate flowers that fail to develop "normally," instead holding mature pollen inside virtually unopened, bud-like flowers. This paedomorphic morphology restricts reward access to "buzz-pollinating bees, specifically Xylocopa sp. (carpenter bees), which can remove pollen by sonication. This is one of very few reports of paedomorphic specialization, and, as far as we are aware, the first documented case of a rapid reversal to specialized pollination in a lineage of plants that had previously switched from specialized to generalized pollination in conjunction with dispersing to a new region.
Keywords: buzz-pollination, flower development, heterochrony, neoteny
Von Baer's "laws" of embryology (von Baer 1828) posit that generalized forms precede specialized forms in development. This simple assertion contributed importantly to the thinking of Darwin, Haeckel, de Beer, and other early evolutionary biologists (de Beer 1930; Gould 1977). One extension of these "laws" is that evolutionary changes leading to retention of early developmental forms into adulthood (paedomorphosis) are associated with the evolution of generalization or simplification of form and function (de Beer 1930; Hardy 1954; Gould 1977). This view has been surprisingly resilient, and it is still generally accepted. For example, Gould (1977) endorsed the classical view of the evolutionary significance of paedomorphy in stating that it provides: 1) an "escape from specialization in the replacement of inflexible adult structures by generalized juvenile forms" (p. 285), leading to greater evolutionary potential (see also Hardy 1954), and 2) a mechanism for "large and rapid evolutionary shifts not requiring extensive genetic reorganization" (p. 285; see also Takhtajan 1969). Although biologists no longer interpret von Baer's "law" as inviolate, it remains a generalization that paedomorphosis is an important basis for rapid evolutionary change towards simpler or more generalized ecology and morphology (Lahaye et al. 2005; Joss 2006; Box and Glover 2010). Except for habitat specialization via paedomorphy in amphibians (see Wake 1991), there are very few examples of paedomorphosis leading to specialization. We report one such case here, where paedomorphy is involved in the evolution of specialization in plant-pollinator relationships.
Heterochronic disturbances, including paedomorphosis and peramorphosis (accelerated/extended development prior to sexual maturity), are possible promoters of major evolutionary transitions (von Baer 1828; Gould 1977), such as new ecological relationships with mutualists. For example, a number of switches between types of pollinators and between cross-pollination and self-pollination in several plant lineages appear to have been facilitated by heterochronic changes in flower development (Box and Glover 2010). Such transitions remain poorly understood, despite their importance in the origins and maintenance of biodiversity (Thompson 1996). Nearly all studies of paedomorphic transitions between pollination systems have described shifts from cross-pollination to self-pollination (Lord 1982; Minter and Lord 1983; Hill at al. 1992; Gallardo et al. 1993; Stewart and Canne-Hilliker 1998; Ehlers and Pedersen 2000; Porras and Munoz 2000; Sherry and Lord 2000; Georgiady and Lord 2002; Box and Glover 2010; Li and Johnston 2010) or shifts from more specialized to more generalized cross-pollination (e.g., in Stylidiaceae (Laurant et al. 1998) and orchids (Box et al. 2008)). We have found a surprising exception to this dominant trend: paedomorphy appears to be involved in a transition from generalized to specialized pollination.
We report here the discovery, in a clade of Madagascar euphorb vines, of secondarily specialized adaptations for pollination by bees that collect pollen by sonication (buzz pollination). We hypothesize that the transition occurred by paedomorphosis, and test this hypothesis by comparing late floral development in related "normal" and buzz-pollinated species.
Materials and Methods
Study System
Dalechampia L. (Euphorbiaceae) occurs throughout the lowland neotropics and paleotropics west of Wallace's Line and comprises ca. 130 species of twining vines and monopodial shrubs. One major biogeographic event in this group's evolutionary history was the relatively recent colonization of Madagascar from Africa (Armbruster and Baldwin 1998). Most species in the neotropics and all species in Africa and Asia are pollinated by female megachilid (Megachilidae) or apid (Apidae) bees that collect floral resin for use in nest construction (Armbruster and Steiner 1992; Armbruster 1993). The only published pollination observations from Madagascar report absence of any reward beside pollen and occurrence of generalized pollination by pollen-feeding beetles (Scarabidae: Cetoniini; Cerambycidae), flies, and pollen-collecting bees (Armbruster and Baldwin 1998; Armbruster et al. 1993). This represents a switch from specialized to generalized pollination upon dispersal to an area lacking an abundance of the "normal" pollinators (large resin-collecting bees; Armbruster and Baldwin 1998).
Methods
Field observations of floral morphology, flower development, and pollinator behavior were made in Madagascar in December 1990--January 1991 (Kirindy Forest, Menabe, Toliara; Ambila Lemaitso, Toamasina; Ampijoroa, Mahajanga; Mahajanga, Mahajanga), January--February 1996 (Ramena, Antsiranana; Ankarana, Antsiranana; Ampijoroa, Mahajanga; Mahajunga, Mahajanga), and November 2006 (Kirindy Forest, Menabe, Toliara). Additional observations consistent with our own were provided by JM Olesen, MO Randriambahiniarime, and VA Ojo based on fieldwork at Kirindy Forest, Menabe, November 2004. Observations on behavior of male and female Xylocopa sp. were made on D. bernieri Baill. near Ramena, Antsiranana, in January 1996. Plant voucher specimens were collected and deposited with ALA, MO, and TAN. Pollinating bees were identified to genus using keys in Michener (2007). Additional observations of floral morphology and floral development that bear on this study were conducted in Tanzania (Armbruster and Mziray 1987; Armbruster et al. 1993), South Africa (Armbruster and Steiner 1992), and Gabon (Armbruster et al. 2005), as well as South America (see review in Armbruster 1993).
Thin-layer sectioning was performed at the School of Biological Sciences, University of Portsmouth, on paraffin-embedded flowers collected from greenhouse material of D. bernieri and D. parvifolia Lam. grown in nearly identical conditions near one another on the same bench in the Institute of Arctic Biology Greenhouse, University of Alaska, Fairbanks. These taxa were chosen because they 1) exhibit the contrasting floral morphologies of interest, 2) were available as live flowering material in cultivation, and 3) are reasonably close phylogenetically (see below). Thin-sections were examined under a compound microscope fitted with a digital camera. Measurements of floral organ and cell dimensions were made using these photomicrographs and a standard image analysis system. Raw data are deposited on Dryad (www.datadryad.org).
Molecular trees based on concatenated chloroplast DNA (3' trnK intron) and 18S-26S nuclear ribosomal external and internal transcribed spacer (ETS and ITS) sequences were generated from single individuals of 81 taxa (88 populations) of Dalechampia and two closely related outgroup species in Plukenetia L. and Tragia L. using Bayesian inference (sequence and tree data deposition: GenBank [http://www.ncbi.nlm.nih.gov], accession numbers GQ463738-GQ463956, voucher data included (further details at http://www.pnas.org/content/106/43/18085/suppl/DCSupplemental); TreeBASE [http://www.treebase.org], submission number 13531). Taxon sampling represented all major subgeneric groups and critical diversity in pollinator reward and blossom morphology within Dalechampia (Armbruster 1993; Armbruster and Baldwin 1998). DNA extractions, PCR, and sequencing followed standard procedures (Doyle and Doyle 1987; Steele and Vilgalys 1994; Baldwin and Markos 1998; Baldwin and Wessa 2000; Armbruster et al. 2009). Sequences were aligned in SeaView 4.0 (Galtier et al. 1996) using MUSCLE (Edgar 2004) and adjusted manually using Simmons's similarity criterion (Simmons 2004). Four independent Bayesian Markov chain Monte Carlo (MCMC) analyses were conducted using MrBayes 3.1 (Ronquist and Huelsenbeck 2003), each using three "cold" and one "heated" simultaneous chains. Evolutionary models selected for phylogenetic analysis based on the Akaike information criterion in MrModeltest (Nylander 2004) were GTR+G for the 3' trnK intron, GTR+I+ for the ITS region, and HKY+ for the ETS. Each MCMC analysis was terminated at 10 106 generations, when the average standard deviation of split frequencies (SDSF) was
