During his voyage on the Beagle, Charles Darwin industriously collected natural history specimens and sent them home from various ports along the way (Briggs, 2009). On his return to Great Britain, he immediately began supervising work on the scientific results of the expedition, employing the help of specialists to record his fossils, fishes, mammals, birds and plants. Among the 209 plants that Darwin collected on the Galapagos Islands and sent to J. D. Hooker was a new species of Cucurbitaceae, Sicyos villosus J. D. Hooker (Fig. 1). Unlike his bird collections, Darwin labelled his plants by island, which is why we know that this cucurbit came from Charles Island, now Floreana. On the herbarium label, the species is described as ‘In great beds injurious to vegetation.’ Since then, this species has never been found again, despite intense search efforts of several botanists. Based on one fruit from a bag attached to Darwin’s 175-year-old specimen in the Cambridge herbarium (Fig. 1, insert), one of us (H.S.) generated nuclear and chloroplast sequences to find out the affinities of this vanished mystery plant. We also tested its relationship to another endemic Galapagos cucurbit, the Santa Cruz gourd, Sicyocaulis pentagonus Wiggins, which is known from five collections on Santa Cruz and Isabela, but has not been collected in the past 35 years.
Based on molecular data (Kocyan et al., 2007; Schaefer et al., 2008, 2009), the New World Sicyoeae comprise a clade of about 150 species in 20 genera (Apatzingania, Brandegea, Cyclanthera, Echinocystis, Echinopepon, Elateriopsis, Frantzia, Hanburia, Linnaeosicyos, Marah, Microsechium, Parasicyos, Pseudocyclanthera, Rytidostylis, Sechiopsis, Sechium, Sicyocaulis, Sicyos, Sicyosperma, Vaseyanthus). Herbarium specimens of 76 species representing all these genera were sequenced for the chloroplast regions rpl20–rps12 and trnL/trnL–F, and the entire nuclear ribosomal RNA intergenic spacer region (ITS1–5.8S–ITS2), following the methods described in Kocyan et al. (2007) and Schaefer et al. (2008). The genus Sicyos was represented with 41 of its c. 60 species. Trees were rooted on Nothoalsomitra (Schaefer et al., 2009). Herbarium vouchers and GenBank accession numbers are given in the cited papers and in a phylogenetic study of Sicyos (P. Sebastian, H. Schaefer, and S. Renner, in preparation). Sequences were edited with Sequencher (v.4.9; Gene Codes, Ann Arbor, MI, USA) and aligned by eye, using MacClade v.4.06 (Maddison & Maddison, 2000). The aligned plastid matrix comprised 1818 nucleotides, the aligned ITS matrix comprised 644 nucleotides. Analyses of the separate plastid and nuclear data partitions produced congruent phylogenetic estimates, and the data were therefore concatenated. The complete alignment is available from P.S. Maximum likelihood (ML) analyses as well as ML bootstrapping relied on RAxML v.7.0.4 (Stamatakis et al., 2008; available at: http://phylobench.vital-it.ch/raxml-bb/) and used the GTR + G model.
To obtain absolute ages for the divergences between the Galapagos Islands species and their mainland relatives, we used Bayesian time estimation with an uncorrelated-rates model as implemented in beast v.1.4.8 (Drummond & Rambaut, 2007). The alignment used for dating excluded all gapped positions and comprised 2113 nucleotides and 79 species. We again used the GTR + G model with four rate categories. Fossil calibration came from the pollen Hexacolpites echinatus from the Oligocene of Cameroon (Salard-Cheboldaeff, 1978; Muller, 1985), which is the oldest known hexacolpate Sicyoeae-type pollen. The most conservative assignment of this pollen is to the split between Linnaeosicyos versus the remaining New World Sicyoeae (Schaefer et al., 2009). The Oligocene epoch ranges from 33.9 to 23 Ma, and the stratum containing Hexacolpites has not been precisely dated; we used an age of 28.5 ± 6 Ma because an analysis of Cucurbitaceae divergence times using first the upper then the lower boundary of the Oligocene found no significant difference (Schaefer et al., 2009). We also constrained the root of the tree to 37 ± 3 Ma based on Schaefer et al. (2009). Markov chain Monte Carlo runs extended for five million generations, sampling every 1000th generation. Of the 5001 posterior trees, we excluded the first 1000 as burn-in. Convergence was checked using Tracer v.1.4.1 (Rambaut & Drummond, 2007). The estimated covariance parameter was 0.68, and the 95% highest posterior density (HPD) did not enclose 0, justifying the hypothesis of non-autocorrelated rate variation. Results under a strict clock model were similar to those obtained with the relaxed clock model.
A chronogram for 79 species of Sicyoeae–Cucurbitaceae, including 68% of the c. 60 species currently assigned to Sicyos, shows that both Galapagos cucurbits belong to this widespread genus, which has also diversified in the Hawaiian archipelago and in Australia and New Zealand (Fig. 2). Darwin’s Sicyos villosus is closest to species from North America and Mexico, while the Santa Cruz gourd, Sicyocaulis pentagonus, is closest to species from Peru and Ecuador. The molecular clock suggests that the divergence from their respective closest relatives occurred 4 (± 2) Ma in Sicyos villosus, and 1.4 (± 1.2) Ma in Sicyocaulis pentagonus. Thus, the two species arrived on the Galapagos archipelago through non-anthropogenic long-distance dispersal from different continental source populations and at times that match the geological age of the islands: the Galapagos Islands are the product of hotspot activity 930 km west of the Ecuadorian coast and are at least 3–4 Myr old (Hickman & Lipps, 1985). Most plant species (up to 60%) appear to have arrived via birds, and the closest floristic ties are with Ecuador and Peru, followed by Central America and Mexico (Porter, 1976). It is also known that storm petrels migrate between Peru and the Galapagos (Tomkins, 1982) and that these and other seabirds nest in habitats where Sicyos occurs (Marks, 1992), suggesting that the spiny fruits (Fig. 1) may have been carried by birds.
Floreana, where Darwin collected Sicyos villosus, was settled in 1807 and continuously inhabited with only short breaks of a few years’ abandonment during the 19th century. The settlers brought livestock, and by the time Darwin visited (in 1835) they owned approximately 2000 head of cattle (Steadman, 1986). Markham, who visited Floreana in 1880, found it ‘in undisturbed possession of the so-called wild cattle … donkeys, dogs, pigs, and other animals that had been left to run wild on the abandonment of the island by the former inhabitants’ (Steadman, 1986, p. 62). A plausible explanation for the disappearance of Darwin’s gourd could be that it was grazed to extinction. And yet many Galapagos plants may be adapted to grazing: prior to the arrival of feral animals some 2–3 million giant tortoises (Geochelone nigra) lived on the archipelago and fed on its vegetation (Fowler de Neira, 1985; Coblentz & Baber, 1987). Tortoises, however, are inefficient grazers compared with goats and cattle, and the reach of the animals is different. There are many examples of Galapagos plants being apparently or obviously grazed to extinction or extreme rarity (A. Tye, personal communication, 18 January 2010). However, the speed of the disappearance of a plant described as abundant in 1835 is surprising.
Another plausible cause for the decline of Sicyos villosus (and possibly also Sicyocaulis pentagonus) could be Cucurbitaceae-specific viruses (e.g. cucumber mosaic virus, water melon mosaic virus, zucchini yellow mosaic virus) introduced with cucurbit crops cultivated by the settlers. The dramatic decline of Sicyos australis in New Zealand, a species now almost entirely restricted to small offshore islands where no cucurbit crops are grown, has been attributed at least partly to these viruses (Delmiglio & Pearson, 2006). Similar inadvertent introduction of cucumber viruses could have occurred on the Galapagos Islands. Finally, it is possible that the extinction of Darwin’s Galapagos gourd is only apparent and due to infrequent and uneven botanical collecting. In the course of this project, we contacted four botanists who have resided on the Galapagos Islands and made systematic collections there (Henning Adsersen, Ole Hamann, Henk van der Werff and Alan Tye). All of them stressed that there are still some areas that have never been subject to a full botanical survey.
As illustrated by the sole specimen in existence of Sicyos villosus, Darwin’s unique natural history collections continue to shed light on the origin and destruction of the Galapagos biota even 175 years after his visit. And although Darwin himself did not comment on the likely mode of transport of Sicyos seeds or fruits, he was acutely aware of the role of birds in plant dispersal (Briggs, 2009) and may well have collected the spiny seeds of Sicyos villosus (Fig. 1, inset) with this hypothesis in mind.