GUEST EDITORIAL Life, death and fossilization on Gran Canaria – implications for Macaronesian biogeography and molecular dating Cajsa Lisa Anderson1*, Alan Channing2 and Alba B. Zamuner3 The increasing efforts to integrate methods and results in historical biogeography with molecular phylogenetic dating are extremely important, because ignoring temporal information will obscure biogeographical patterns and give erroneous results (Donoghue & Moore, 2003; Ree & Sanmartı́n, 2009). Today’s dating methods yield sufficiently realistic divergence times between organism lineages to be incorporated into biogeographical studies. These new methods (reviewed by e.g. Anderson, 2007) are not dependent on a strict molecular clock and they allow for calibration with multiple age constraints from the fossil record or geological evidence. Unless combined with temporal information, biogeograph- ical inference is highly sensitive to the problems of pseudo- congruence (i.e. two groups showing similar patterns but with a different temporal origin are unlikely to have been affected by the same biogeographical events) and pseudo-incongruence (i.e. distributional noise obscuring a common pattern) (see e.g. Vanderpoorten et al., 2007). Divergence times can be used to discriminate between alternative biogeographical scenarios – such as in the old debate between dispersal and vicariance – by comparing the divergence times of the disjunct taxa with the timing of the geographical barrier. Through an integration of methods and results in dating and biogeography we can gain information on how speciation and biodiversity change over time owing, for example, to climatological and other geological events, (mass-)extinctions and new dispersal routes. During the last few years, new analytical methods have been developed (e.g. Nylander et al., 1Real Jardı́n Botanico, CSIC, Plaza de Murillo 2, 28014 Madrid, Spain, 2School of Earth and Ocean Sciences, Cardiff University, Park Place, CF10 3YE, Cardiff, UK, 3Departamento de Paleobotánica, Facultad de Ciencias Naturales y Museo, UNLP, 1900 La Plata, Argentina *Correspondence: Cajsa Lisa Anderson, Real Jardı́n Botanico, CSIC, Plaza de Murillo 2, 28014 Madrid, Spain. E-mail: cajsa.lisa.anderson@gmail.com ABSTRACT The Canaries have recently served as a test-bed island system for evaluating newly developed parametric biogeographical methods that can incorporate information from molecular phylogenetic dating and ages of geological events. To use such information successfully, knowledge of geological history and the fossil record is essential. Studies presenting phylogenetic datings of plant groups on oceanic islands often through necessity, but perhaps inappropriately, use the geological age of the oldest island in an archipelago as a maximum-age constraint for earliest possible introductions. Recently published papers suggest that there is little chance of informative fossil floras being found on volcanic islands, and that nothing could survive violent periods of volcanic activity. One such example is the Roque Nublo period in Gran Canaria, which is assumed to have caused the extinction of the flora of the island (c. 5.3–3.7 Ma). However, recent investiga- tions of Gran Canaria have identified numerous volcanic and sedimentological settings where plant remains are common. We argue, based on evidence from the Miocene–Pliocene rock and fossil records, that complete sterilization of the island is implausible. Moreover, based on fossil evidence, we conclude that the typical ecosystems of the Canary Islands, such as the laurisilva, the Pinus forest and the thermophilous scrubland, were already present on Gran Canaria during the Miocene–Pliocene. The fossil record we present provides new information, which may be used as age constraints in phylogenetic datings, in addition to or instead of the less reliable ages of island emergences or catastrophic events. We also suggest island environments that are likely to yield further fossil localities. Finally, we briefly review further examples of fossil floras of Macaronesia. Keywords Canary Islands, fossil, Gran Canaria, laurisilva, Macaronesia, Miocene, molec- ular dating, Pliocene, Roque Nublo, Tetraclinis. Journal of Biogeography (J. Biogeogr.) (2009) 36, 2189–2201 ª 2009 Blackwell Publishing Ltd www.blackwellpublishing.com/jbi 2189 doi:10.1111/j.1365-2699.2009.02222.x 2008; Ree & Smith, 2008; Sanmartı́n et al., 2008), which are exciting and promising for historical biogeography research. The model-based methods, importantly, as opposed to previ- ous parsimony methods, provide the possibility of incorpo- rating external data, such as palaeogeographical and palaeontological information (e.g. distance between areas or the fossil occurrence of a lineage in an area at a certain time). The new methods provide more possibilities – but they also introduce a larger demand for accurate additional data. It is well known that the most important factor in molecular dating, regardless of the method used, is the quantity and quality of age constraints (e.g. Bremer et al., 2004; Britton, 2005). It is a sensible assumption that biogeographical inferences are just as dependent on geological reconstructions and age constraints for obtaining reasonably reliable results. Ignoring fossil data hence leads to less well-constrained analyses and therefore to less reliable results. Unfortunately, we feel that at present the geological processes (and their time-scale) that can lead to, for example, vicariance events are often being misjudged or over-simplified. For example, the actual split of continents is a long process involving several events, and the subsequent drifting will lead to vicariance on a continuous time-scale, varying for different organism groups. Furthermore, fossil records that can constrain and/or corroborate hypotheses are frequently overlooked. We believe that the great rigour and effort put into creating biological datasets needs to be extended to the use of the geological data that biogeographical inferences are dependent on. Volcanic oceanic island chains, such as the Hawaiian and Macaronesian islands, are very commonly used as examples when exploring new biogeographical methods and models (e.g. Ree & Smith, 2008; Sanmartı́n et al., 2008; Whittaker et al., 2008). There are several good reasons for using Macaronesia as a test-bed system. There is a high degree of endemism on the islands, both within Macaronesia and between the islands, and the lineages can frequently be traced back to mainland Africa or the Mediterranean region. Furthermore, the litho- and chronostratigraphy of volcanic islands are often relatively well known, even though their history is quite complex (see e.g. Carracedo, 1999; Whittaker et al., 2008, and references there- in). In the absence of a fossil record, molecular dating of phylogenies relies on other, less reliable, temporal information. Some authors apply a strict molecular clock on ‘average rates’ from their own data, and others use ages or ‘average rates’ from other studies of organism groups (e.g. Kim et al., 1996; Emerson, 2003; Caujapé-Castells, 2004; Oberprieler, 2005; Dı́az-Pérez et al., 2008). Some studies avoid absolute dating, or discuss relative dating results in the light of geological and climatological data (e.g. Navascués et al., 2006). When geological information is used, it is based either on the date of the emergence of an island (e.g. Böhle et al., 1996; Percy et al., 2004; Kim et al., 2008) or on a catastrophic event as a maximum age for a clade or the whole flora of an island (Emerson, 2003). Maximum ages are always a problem in dating, as they in effect put a hard bound on a crown group’s age (most dating methods do not allow for overriding constraints, that is, finding a solution outside the boundaries of the age con- straints). As opposed to minimum ages they are not open to the possibility of a ‘ghost range’ (the time interval between the first fossil occurrence and the actual divergence) of as yet undiscovered fossils or a mistakenly calculated island age. When applied in biogeography, putting a maximum constraint on the crown-group node implicitly means there was one single colonization event, and diversification started immedi- ately upon arrival. From these considerations it follows that an erroneous maximum age could cause underestimates of all divergence times in the phylogeny. In the case of the Canary Islands, where numerous (now submerged) seamounts allow for the existence of earlier islands, (e.g. Geldmacher et al., 2001; Ancochea & Huertas, 2003), and when using the aerial age of Fuerteventura, the oldest island of today (21 Myr old), we are possibly severely underestimating the age of dispersal events of plant lineages from the mainland (e.g. the five plant groups dated by Kim et al., 2008). In this paper we explore two assumptions about oceanic volcanic islands in general, and about Gran Canaria in particular, of direct relevance to the biogeographical inter- pretation of island floras. First, we address the assumption that catastrophic volcanic events can potentially sterilize whole islands. We take as an example the development of the Roque Nublo stratocone volcano, which is suspected to have sterilized Gran Canaria c. 5.3–3.5 Ma. Second, we consider whether there is little or no chance of discovering informative (plant) fossils on volcanic islands (e.g. Whittaker et al., 2008; Rodrı́guez-Sánchez et al., 2009). In this paper we present direct evidence from the fossil record of Gran Canaria and other Macaronesian islands that disproves the second assumption and, in combination with an evaluation of the strata of the Roque Nublo group, sheds doubt on the first. This will provide valuable data for future research, both for empirical studies and for refining island biogeography models and assessing their validity. ROQUE NUBLO AND THE STERIL IZATION HYPOTHESIS Early geology and windows of opportunity The geology of Gran Canaria has been intensively studied for more than two centuries (Carracedo et al., 2002). This has resulted in detailed accounts of the island’s stratigraphy, a densely sampled isotope geochronology (e.g. Pérez-Torrado et al., 1995, 1997; van den Bogaard & Schmincke, 1998; Guillou et al., 2004) and the publication of detailed geological maps (e.g. Mapa Geologica de España, Instituto Geológico y Minero de España). The early subaerial volcanism of Gran Canaria, following emergence at c. 14.5 Ma, commenced with a short (c. 0.5 Myr) basaltic shield-building phase (e.g. Schneider et al., 2004) (Fig. 1a). Towards the end of this period the main shield volcano collapsed, forming the Caldera de Tejeda (e.g. van den C. L. Anderson et al. 2190 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd (a) Pleistocene–Holocene sediments Landslides Cone sheet Edge of Tejeda caldera Recent volcanism Post Roque Nublo Roque Nublo Las Palmas and Arguineguín Formations erosional phase Intracaldera Formation Fataga Group unclear, probably reduced extrusive volcanism and locally erosional phase Mogán Group Shield basalts 0 5 10 14 El Tablero Formation Ayacata Formation Tenteniguada Formation Tirajana Formation epiclastics pyroclastics Riscos de Chapín Formation Rincón de Tejeda Formation 3.5 4 4.5 5.5 5 3 Montaña Horno Formation Miocene, Fataga Group, fossil locality c. 13 Ma Las Palmas detritic Formation, fossil locality c. 4.5 - 4 Ma Fossil localities associated with Roque Nublo volcanism 1 1 2 2 (b) (c) 0 10 20km N 3 3-7 4 5 6 7 Figure 1 (a) Simplified map of the geology of Gran Canaria, showing the distribution of pre-Roque Nublo (Miocene), Roque Nublo (Pliocene) and post-Roque Nublo (Pliocene–Holocene) strata on Gran Canaria. The map is based on Carracedo et al. (2002); colours are mainly adopted from Mapa Geológico de España (Instituto Geológico y Minero de España) and (chrono)stratigraphy from van den Bogaard & Schmincke (1998). (b) Distribution of the main formations of the Roque Nublo Cycle. The map and stratigraphy are based on Pérez- Torrado et al. (1995); colours are mainly adopted from Mapa Geológico de España (Instituto Geológico y Minero de España), with chronostratigraphic synthesis from Pérez-Torrado et al. (1995), Guillou (2003) and Guillou et al. (2004). For a description of the various phases, see text. (c) The main areas with fossil localities described in this paper. 1, Barranco de Mogán–Azulejos; 2, Las Cuevas del Guincho; 3, Barranco de Tirajana; 4, Berrazales–El Hornillo; 5, Soria; 6, Embalse de Cueva de las Niñas, Pajonales; 7, Casa Forestal de Pajonales. Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2191 ª 2009 Blackwell Publishing Ltd Bogaard & Schmincke, 1998). Post-caldera volcanism was characterized by intense and violent eruptions, resulting in large volumes of ignimbrites and silicic lavas. Two main magmatic phases, the Mogán Group (c. 14–13.3 Ma) and the Fataga Group (c. 12.4–9.9 Ma), were separated by the com- positionally transitional Montaña Horno Formation (van den Bogaard & Schmincke, 1998). Hiatuses of c. 30–140 kyr between ignimbrite eruptions of the Mogán Group (evidenced by isotopic dating, erosional unconformities and soil forma- tion; van den Bogaard & Schmincke, 1998) provided windows of opportunity for plant introductions from mainland Africa and the Mediterranean as well as from the older Canarian islands (Carine, 2005). These earliest volcanic phases were followed by a long period of erosion (c. 3 Myr), with only minor eruptions on the northern slopes of the island. The erosion caused a radial pattern of deep barrancos, which controlled the distribution of the products of later volcanic activity (e.g Pérez-Torrado et al., 1995). Large volumes of sediments formed by erosion were deposited predominantly in the south (Arguineguı́n Forma- tion) and in the north-east (Las Palmas Detritic Formation) as well as off the coast (Schneider et al., 1998). The long period of quiescence was followed by three further phases of volcanism: Roque Nublo, post-Roque Nublo and recent volcanism (Fig. 1a,b). Extinction? The hypothesis that Roque Nublo volcanism was a catastrophe that killed all life on Gran Canaria probably stems from the work of Hausen in 1962 (Rothe, 2008), and has since been used as the earliest possible age of (re-)introduction and evolution of the laurisilva and pine-forest communities, for example by Marrero & Francisco-Ortega (2001), Emerson (2003) and Whittaker et al. (2008). If this assumption is wrong (and we argue that it is) and is used as an age constraint for integrated phylogenetic dating and biogeography, the reconstructions will be erroneous. Indeed, the Roque Nublo volcanism was violent and protracted (lasting c. 2 Myr) and produced a volume of eruptive products estimated at c. 100–200 km3 at an average eruption rate of 0.1 km3 kyr)1. However, as we demonstrate, the volcano did not erupt continuously throughout this interval. Distributions of the various formations of the Roque Nublo cycle provide evidence that at any one time during the evolution of the volcanic sequence, large areas of the island were relatively unaffected by volcanism. We envisage that early Roque Nublo volcanism was more likely to have been responsible for habitat fragmentation than for complete extinction. Below we present a synthesis of the evolution of the Roque Nublo cycle and how the various phases affected the flora of the island. The current distributions of the formations of the Roque Nublo sequence are illustrated in Fig. 1b (see also figure 2 in Pérez-Torrado et al., 1995). The geographical extent and hypothetical geomorphology of the Roque Nublo strato- cone prior to its gravitational collapse are provided in figure 4 of Pérez-Torrado et al. (1995). The Roque Nublo event(s) Broadly speaking, from c. 5.3 to 3.7 Ma volcanism was restricted to the development of scattered cinder cones (El Tablero Formation) and effusive (rather than explosive) basaltic lava flows (Riscos de Chapı́n Formation). Subse- quently, from c. 3.7 to 3.1 Ma, more explosive volcanic activity (Tirajana Formation) built an asymmetrical stratovolcano cone with shallow northern and steep southern slopes, which at its maximum development was c. 2.5–3 km high (Pérez- Torrado et al., 1995). The volcanic cone eventually became unstable, and in the period c. 3.5–3.1 Ma experienced large gravitational flank collapses that formed the Ayacata Forma- tion (Funck & Schmincke, 1998; Guillou, 2003; Guillou et al., 2004). The latest phase of activity, which was mainly intrusive, developed within the amphitheatre-shaped collapse scars between 3.11 and 2.87 Ma (Guillou et al., 2004). The volcanic stratigraphy of the sequence is thus characterized by basaltic lava flows in its basal part, crudely bedded tuffaceous phonolitic rocks and breccia sheets interbedded with lava flows in its lower middle part, thick massive breccia sheets in its upper middle part, and a few small lava flows in its upper part (van den Bogaard & Schmincke, 1998). Initial cinder-cone eruptions and basaltic lavas were of limited lateral extent, whereas later more voluminous basaltic lava flows were channelled down deep palaeo-barrancos (palaeo-valleys) (draining the north-east and eastern sectors of the island), which had formed in the volcanic hiatus prior to Roque Nublo volcanism (Pérez-Torrado et al., 1997). Detailed observation of the evolution of the Roque Nublo sequence provides clear evidence for long periods of quiescence. The evolution of the chemical composition of Roque Nublo cycle volcanism, from basaltic to phonolitic, indicates magmatic differentiation of subvolcanic magma chambers. This process requires considerable periods of volcanic quiescence (e.g. Frisch & Schmincke, 1969). Isotopic dating of the volcanic sequence, although not complete at a flow-by-flow level, indicates that hiatuses in volcanic activity in any one location were of the order of tens of thousands of years (e.g. Pérez- Torrado et al., 1995; van den Bogaard & Schmincke, 1998; Guillou et al., 2004). Both geographical localisation of the deposition of volcanic products and hiatuses would appear to allow ecosystem recovery. Only two phases of the Roque Nublo sequence, the emplacement of the Tirajana Formation ignimb- rites and the massive Ayacata Formation flank collapses, appear to be of such magnitude that they might plausibly have led to an island-wide extinction event. As detailed below, we consider that even these dramatic events would fail to cause mass extinctions. Potentially catastrophic eruptions The Roque Nublo ignimbrites (Tirajana Formation) originated from hydrovolcanic eruptions as rising magma came into contact with groundwater at high levels in the crust (Pérez- Torrado et al., 1997). Eruptions originated from multiple C. L. Anderson et al. 2192 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd successive vents with estimated diameters of c. 300 m. Succes- sive eruptions broke through closed vents created by magma solidification in the vent conduits of preceding eruptions (Pérez-Torrado et al., 1997). The relatively wide vents together with the incorporation of rock components from the old vent conduits meant that the volcano could not sustain vertically buoyant eruption columns, and instead eruptions were char- acterized by tephra fountains (Pérez-Torrado et al., 1997). Ignimbrite flows originating from these vents contained a lot of water and were relatively cool (c. 300 �C in proximal areas, but dropping below 100 �C distally). During initial eruptions, ignimbrites were confined to palaeo-barrancos (Pérez-Torrado et al., 1995). Between valley systems, the ridges and plateaus were relatively unaffected by volcanic deposition, as during the preceding basaltic activity. Where barrancos broadened close to the palaeo-coast, ignimbrite flow deposits thinned to c. 2 m, and the separation of water and rock components produced lahar-like flows (Pérez-Torrado et al., 1997). Unconformities and conglomerates interbedded within the ignimbrites in these areas provide evidence of quiescent periods between eruptive events (Pérez-Torrado et al., 1997). As volcanism continued, barrancos were eventually filled and overtopped. Volcanic material then formed broad aprons that extended from the Roque Nublo crater area chiefly to the northern half of the island (Funck & Schmincke, 1998). The location and asym- metry of the cone formed during this period (Pérez-Torrado et al., 1995) again allowed large sectors of the island to remain as viable habitat. Potentially catastrophic collapses The Roque Nublo stratovolcano was subject to at least three collapse events (Ayacata Formation), which are separated chronologically and stratigraphically by periods of scarp erosion and volcano regrowth (Carracedo & Day, 2002). These collapses were of only moderate size (estimated volumes of the order of 20–70 km3) and dominantly affected the south and west of the island (Carracedo & Day, 2002). The largest collapse (affecting the south flank at c. 3.5 Ma; Funck & Schmincke, 1998) locally filled barrancos with up to 500 m of avalanche debris, as well as depositing thinner debris layers on intervening plateau surfaces. However, mountain crest/ridge environments remained debris-free (Mehl & Schmincke, 1999). Toreva blocks (kilometre-scale mega-blocks that retain their internal stratigraphy) were rafted laterally (1–2 km) during the collapse event but underwent only minor rotation (e.g. Belousov et al., 2001). The geometry of sector-collapses allows large geographic areas of the island to remain unscathed following each event, and stratigraphical intervals between collapses provide time for vegetation recovery. Survival In summary, Roque Nublo volcanism was protracted and in part violent. However, the volcano did not continuously erupt (either violently or effusively) during this interval (e.g. Pérez- Torrado et al., 1997), and no single pyroclastic eruption is implicated as devastating the entire island. We see ample evidence both in the volcanic rock record and in the geometry of debris avalanche deposits for the survival of very broad and diverse habitats and ecosystems throughout the Roque Nublo cycle. Although the flora of the island did face many challenges, leading to random survival and mosaic habitats, a complete extinction of all species on the island seems impossible. To date, geologists and biologists are unsure of the extent to which the island’s biota was affected. It would be incorrect to suggest that all papers referring to the Roque Nublo eruptions as killing Gran Canaria’s vegetation assume complete extinc- tion. Navascués et al. (2006), in addressing the extinction of pine forests, suggested the possibility of a few ridge-top refugia from which recolonization could occur. Araña & Carracedo (1980) took the view that all the vegetation of the island went extinct, although there was a possibility that some coastal species managed to survive. Marrero & Francisco-Ortega (2001) estimated an approximate extinction rate of 50% if volcanism comparable to Roque Nublo were to occur today. FOSSIL FLORAS OF GRAN CANARIA The published fossil record of Gran Canaria, to date, comprises floras that, although pointing to diverse vegetation on the island throughout the Neogene, are confined to only rudi- mentary plant descriptions (Schmincke, 1967, 1968; Garcı́a- Talavera et al., 1995; Pérez-Torrado et al., 1995; Schneider et al., 2004). In order to address this we have reinvestigated previously published localities, and explored additional potential palaeobotanical targets, for example unconformities with evidence for subaerial weathering, soil/palaeosoil forma- tion and epiclastic (re-worked volcanics) and clastic sediments (fluvial, alluvial and lacustrine). This has yielded c. 30 plant fossil horizons of Miocene–Pliocene age; that is, fossils from both before and during the Roque Nublo cycle. The localities are spread around the island, and represent several different vegetation types. Below we summarize the main findings and localities in a broadly chronological order. Miocene thermophilous scrubland Garcı́a-Talavera et al. (1995) briefly described a Miocene flora from the Barranco de Mogán, south-western Gran Canaria. The flora pre-dates Roque Nublo, and occurs in the Montaña Horno Formation, which is bracketed between the Mogán and Fataga Groups and dates to c. 13.3–13.0 Ma. The fossil twigs, leaves and fruits were not described in detail, but some tendencies were distinguished. The four leaf morphotypes described were all small and suggested to have been leathery, belonging to smaller shrubs. This was indicative of palaeo-vegetation adapted to semi-arid conditions and characteristic of the present scrub on drier slopes and barrancos of the Canary Islands. The authors listed extant taxa of these environments, for example Bystropogon, Chamaecytisus, Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2193 ª 2009 Blackwell Publishing Ltd Echium, Cistus, Carlina, Maytenus and Kleinia, but did not make direct morphological/anatomical comparisons that would indicate possible taxonomic affinities of the fossil material. We investigated further exposures within the same strati- graphic interval (but located further to the WNW, locality 1 in Fig. 1c). Here, ash and tuff fall-deposits subjected to later epithermal alteration contain at least 10 leaf morphotypes preserved as moulds and by carbonate permineralization (see Figs S1 & S2 in Supporting Information). Some of these appear to correspond morphologically to the four illustrated by Garcı́a-Talavera et al. (1995). Based on leaf morphology, the overall assemblage appears to be dominated by leaves of the large woody eudicots charac- teristic of the present-day thermophilous scrub vegetation. Venation and cuticular patterns are most often preserved, but cell structures are less common. Based on leaf morphology and initial microscopic examinations of venation and epidermal characters we note morphological similarities between the fossil leaves and some of the genera mentioned by Garcı́a- Talavera et al. (1995), and possibly also Euphorbia and Crassulaceae. However, detailed anatomical investigations of material from both fossil localities are required for correct phylogenetic placement. Laurisilva vegetation transported by lahar-like flows and deposited at Las Cuevas del Guincho The Las Cuevas del Guincho coastal section (locality 2 in Fig. 1c) exposes the Middle Member of the Las Palmas Detritic Formation (Schneider et al., 2004) formed between c. 4.5 and 4 Ma (Pérez-Torrado et al., 2002). This dominantly sedimen- tary unit was formed by the accumulation of volcanic and debris-flow materials around the north-east margin of the Roque Nublo stratovolcano. The flora here includes disartic- ulated and fragmentary leaves, and twigs and fruits that have been transported by lahar-like flows. The leaves are preserved as moulds and compressions and by permineralization, often with detailed cuticular patterns and cell structures (see Fig. S3). As yet we have no secure taxonomic descriptions for our material, but at least 10 distinct leaf morphotypes of eudicots and Lauraceae can be distinguished. Several different monocot leaves are also present, and at least one fern. Coarse conglomeratic horizons in the area contain abundant evidence of entrained vegetation, including tree trunk moulds and permineralized wood fragments. The Pliocene thermophilous scrub of Barranco de Tirajana Deposits relating to the earliest explosive phase of the Roque Nublo stratovolcano preserve perhaps the most dramatic evidence of volcanism impacting upon vegetation. Pérez- Torrado et al. (1995) reported a fossiliferous pyroclastic agglomerate that lies directly above a lava flow dated at c. 3.9 Ma. It locally marks the base of the Tirajana Formation and can be traced for 7 km along the Tirajana valley (south- east of the central stratovolcano) (locality 3 in Fig. 1c). The agglomerate has abundant vegetation impressions at various places along its base. Reinvestigations of this section have revealed the presence of a diverse flora comprising abundant fragmentary leaves and trunk and branch moulds, less common permineralized branch/trunk fragments, plus less frequent articulated eudicot foliar branches and palm fronds. Based on morphology, and initial optical- and scanning- electron-microscope investigations of anatomical features, the articulated specimens appear to represent Euphorbia (Fig. 2a,b) and Arecaceae (Phoenix) (Fig. 2c,d). External moulds of woody stems and branches from the locality are often collapsed; however, three-dimensional examples with faithful replication of stem features such as ribs, leaf/branch scars and protuberances are not uncommon. At least four stem morphotypes with morphological features comparable to the large woody perennial shrubs that are typical of Macaronesian thermophilous scrubland are present (see Fig. S4). Tetraclinis forests and laurisilva of Berrazales–El Hornillo, Pajonales and Soria The volcanologist Hans Ulrich Schmincke was the first to discover and describe a Miocene–Pliocene flora on Gran Canaria (Schmincke, 1967, 1968). He reported several localities with well-preserved twigs and leaves, and even 10-m-long tree trunks that were associated mainly with the Roque Nublo breccias (Tirajana/Ayacata Formations). Two specific localities were mentioned. At Berrazales, small dolomitized stems were found, often with root or shoot scars and cellular preservation. At Pajonales he described well-preserved leaves of laurel-like plants and impression fossils reminiscent of palm leaves, and stems from a ‘bamboo-like’ plant. In the Berrazales area we investigated extensive fossiliferous deposits occurring in the valley to the north and east of El Hornillo (locality 4 in Fig. 1c). Here, a sedimentary sequence 60–100 m thick comprises coarse boulder conglomerates above a landscape unconformity cut into early Roque Nublo basaltic lavas and felsic to rhyolitic ashes, ignimbrites and lavas of the Miocene Fataga Group. The sedimentary sequence is mapped as the epiclastic member of the Tirajana Formation and is thus younger than c. 3.9 Ma (Pérez-Torrado et al., 1995). Towards the top it is dominated by fluvial channel conglomerates, gravels, sands and silts. The fluvial sequence is capped by rubble-based basaltic lavas related to the post- Roque Nublo rift volcanism. The latter commenced c. 3.5 Ma, but in the Berrazales area appears to have been active in the period c. 2.7–1.7 Ma (Guillou et al., 2004). Typically, plants occur as casts/moulds of in situ tree stumps with associated prostrate logs within coarse breccias. Sediments within moulds that fill the cavity created by the decay of trunks/branches often contain taxonomically identifiable permineralized wood and bark fragments. In situ and in-growth-position stumps are particularly common directly above the unconformity between early Roque Nublo basaltic lavas within the El Hornillo valley. C. L. Anderson et al. 2194 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd At this and many other localities, carbonate permineralized root systems are observed to encrust the weathered top and vertical fissure/joint surfaces of the basaltic lavas below the unconformity. Transported trunks (Fig. 3a), branches (Fig. 3b), twigs, leaves (Fig. 4a–f), fruits/capsules (of Lauraceae and/or eudicots) (Fig. S5), and monocot stems and leaves are common within the finer-grained fluvial sediments above the basal conglomerates. Cellular preservation of wood, bark and leaf tissues by carbonate permineralization is common throughout the El Hornillo sequence (Figs 3 & 4). Within silt lenses, leaves are preserved as external moulds (Fig. 4a), sometimes lined with clays, with partially permineralized veins. Laterally extensive horizons containing iron-stained tubes with dominantly vertical orientations mark probable root horizons that occur at various levels throughout the section. The leaf assemblage, often with well-preserved morphology, leaf venation and cuticular characters, appears to be dominated by members of the broad-leaved sclerophyllous genera of today’s Macaronesian laurisilva. Preliminary interpretations, yet to be confirmed, include genera of Lauraceae, and the eudicot genera Arbutus (Ericaceae) (Fig. 4c,e), Ilex (Aquifoli- aceae) (Fig. 4b,d) and Hedera (Araliaceae). Articulated per- mineralized fern fronds, possibly of the genus Asplenium, are less common elements of the flora (Fig. 3g,h). Wood fragments (Fig. 3e,f) include the gymnosperm Tetraclinis (Cupressaceae). This conifer was an element of (a) (b) (c) (d) Figure 2 Fossils from Barranco de Tirajana (locality 3, Fig. 1c). (a) Articulated leaves of Euphorbia preserved as moulds with partial carbonate permineralization of tissues. The leaf arrangement in rosette-like groups is reminiscent of several Euphorbia species that have rosettes at the tip of branches and inhabit the thermophilous scrubland of today. (b) Detail of leaf with prominent mid-vein and partially permineralized epidermis. (c) Field photograph of fossil rachis of a palmate palm frond. The rachis is triangular in cross-section, with near- parallel pairs of leaflet/pinnae attachment point scars. These are orientated almost perpendicularly to the rachis long axis. The parallel configuration of scars indicates the former presence of tightly longitudinally folded pinnae. A dried rachis of the frond of Phoenix canariensis acts as a scale. (d) Block with rachis mould on front face and strap-like pinnae with parallel venation on top surface. Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2195 ª 2009 Blackwell Publishing Ltd (a) (e) (f) (g) (h) (b) (c) (d) Figure 3 (a–f) Typical field and microscopic features of Pliocene wood of the Roque Nublo Cycle, Gran Canaria. (a) Coarse epiclastic sediments with large, transported, branching trunk preserved as external mould. El Hornillo (locality 4, Fig. 1c) (road section). (b) Partially permineralized branch fragment. Typical preservation includes permineralized wood plus bark. El Hornillo (Berrazales area). (c,d) Char- coalified Pinus wood. Forestal de Pajonales area (locality 7, Fig. 1c). (c) Longitudinal fracture-section illustrating tracheids with bi-, but dominantly uni-seriate bordered pits. (d) Transverse section showing transition from early to late wood and resin duct. (e,f) Radial sections of carbonate permineralized conifer wood cf. Tetraclinis. El Hornillo (road section). (e) Tracheids with files of uni-seriate bordered pits and rays (two cells high). (f) Ray (eight cells high) with cupressoid crossfield pitting. (g,h) Fragmentary pinnate fern, possibly of the genus Asplenium. El Hornillo (road section). (g) The pinnae alternate along the rachis. Pinnae are unstalked, longer than their width, slightly asymmetrical and ‘eared’ towards the base. Margins are smooth. Veins are mostly simple, sometimes forked. (h) Scanning electron microscope detail of pinnule surface, showing the main venation that bifurcates down to the ear and along the rest of the pinnule. C. L. Anderson et al. 2196 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd the widespread European laurisilva during the Palaeogene, but became almost extinct during the Neogene climate change (e.g. Kvaček et al., 2000; Kvaček, 2007, and references therein). The presence of Pliocene Tetraclinis fossils therefore suggests a more humid climate on Gran Canaria at this time. Today relict populations occur in Malta, south-east Spain and north- west Africa. This is the first evidence that this genus had a distribution that included Macaronesia. It also appears to be the first direct palaeobotanical evidence of extinction on Gran Canaria. (a) (b) (c) (d) (e) (f) Figure 4 Typical preservation features of lauraceous and eudicot leaves from the El Hornillo–Berrazales area (locality 4 in Fig. 1). (a) Leaf specimen (cf. Ocotea) with preserved mid rib, petiole, parts of smooth margin and net venation. (c,e) Carbonate permineralized eudicot leaf (cf. Arbutus) with axial and abaxial surfaces exposed. (c) Abaxial surface (left side of image) with epidermal cells and stomata between polygonal vein network. Adaxial surface (right side of image) lacks stomata. (e) Detail of abaxial epidermis, polygonal epidermal cells and numerous stomata. (b,d) Carbonate permineralized leaf (cf. Ilex) with preservation of cuticular and epidermal features. (d) Epidermis with stomata and lobed epidermal cells. (f) Transverse section through leaf illustrating anatomical preservation. Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2197 ª 2009 Blackwell Publishing Ltd Tetraclinis wood and evidence of laurisilva genera have also been observed at further localities to the south within deposits mapped as Roque Nublo collapse deposits (Ayacata Forma- tion), for example around Soria (locality 5 in Fig. 1c) and Embalse de Cueva de las Niñas in the Pajonales area (locality 6 in Fig. 1c). The pine forest of Pajonales The Pajonales area has extensive exposures of Roque Nublo volcanic breccias (Tirajana Formation) and landslide deposits (Ayacata Formation) occurring within palaeo-barrancos cut into Middle Miocene ashes, tuffs and ignimbrites. Small outcrops of fluvial conglomerates and cross-bedded sands containing pods of silts are preserved within shallow (c. 20-m- deep) valleys between the two volcanic units. In one fossil locality in Forestal de Pajonales (locality 7 in Fig. 1), at about 1400-m altitude, charcoalified Pinus wood (Fig. 3c,d) occurs within Roque Nublo volcaniclastics. We speculate that this locality belongs to another ecosystem, resembling the pine-dominated high-altitude forests of today. The Pinus wood has abundant epithelial cells surrounding resin ducts, which may suggest affinity with the only indig- enous species of the genus present in the western islands of Macaronesia, the endemic Pinus canariensis. Other outcrops in Forestal de Pajonales have yielded, as yet unidentified, charcoalified plant fragments, probable monocot stems, partially dolomite permineralized wood and twigs plus root horizons. FURTHER MACARONESIAN FOSSIL FLORAS AND POTENTIALLY FOSSILIFEROUS STRATA It is fair to say that fossil floras of the Macaronesian oceanic volcanic islands are so far virtually unexplored. However, the absence of a fossil record is far from reality, and plant remains have been reported from several Macaronesian islands beside Gran Canaria. We briefly review records from other Canary Islands, Madeira and Azores in Appendix S1. What we have learned from our initial explorations of Gran Canaria is that the volcanism of the Macaronesian islands has a much greater potential for fossilization than was previously thought. Good potential targets include islands that had protracted subaerial volcanic development phases punctuated by erosional episodes. We believe that many fossil localities are waiting to be discovered. IMPLICATIONS It could be argued that the fossils we have recorded are those of plants killed in events leading to the ‘sterilization’ of Gran Canaria (Emerson, 2003). However, we have observed evidence of plant growth in the form of in situ stumps of trees and root horizons at many localities. These, and the presence of numerous stacked plant fossil horizons containing trans- ported but taxonomically similar plants, to us instead repre- sent the persistence of vegetation through the period. Equally it could be argued that we are recording successive waves of recolonization from other islands, as would be anticipated if sterilization had occurred (Carine, 2005). However, we see little evidence of floral turnover, because in localities with stacked fossil horizons contained within sediments derived from the same source region, the same vegetation returns following each destructive event. Palaeobotanical exploration of Gran Canaria is in its rather belated infancy (given that the first report of fossils is from the late 1960s). Already an emerging picture provides evidence that three ecosystems typical of the Canary Islands now were also present during the Mio-Pliocene. Laurisilva and Pinus- dominated forest ecosystems were present on Gran Canaria during major late Miocene/early Pliocene volcanic events, supporting the concept of these elements of the island’s flora as Miocene relicts (Vargas, 2007). At lower elevations, sclero- phyllous scrub vegetation was most probably present during the Miocene. As of today, however, these observations are insufficient to pinpoint the clearly intricate patterns of vegetation evolution and historical biogeography that biolo- gists are trying to elucidate (Rodrı́guez-Sánchez et al., 2009). Further research (particularly relating to plant taxonomy and taphonomy) is clearly needed to address these and the countless other questions that the presence of a fossil flora is bound to raise. If our tentative interpretations of plant genera gain support from further investigations, this could push back the age of the introduction of several plant lineages in Gran Canaria, when compared to molecular clock studies using island ages (e.g. Aeonium, Kim et al., 2008; Echium, Mansion et al., 2009) or the Roque Nublo extinction hypothesis as age constraints. Future studies of pre-existing collections and new discoveries should provide important data for inferences of phylogeogra- phy and biogeography. One major problem when reconstructing Canary Island biogeography is the large numbers of probable extinctions on Lanzarote and Fuerteventura. These islands have undergone extensive erosion, and have therefore lost many potential habitats. The endemic pine forests and laurisilva might have been there when the islands still had high mountains, which are necessary for the moist north-east winds to drop their rain. Some of the biogeographical patterns we recognize through molecular phylogenies are most probably obscured by these extinctions. Several phylogenies suggest that numerous plant groups have spread from the mainland to Tenerife or Gran Canaria and then westwards (e.g. Adenocarpus, Percy & Cronk, 2002; Bystropogon, Trusty et al., 2005; Descurania, Goodson et al., 2006; Cistus, Guzmán & Vargas, 2009), whereas in reality they might have spread from the mainland to the eastern islands much longer ago. It is notable that the plant groups exhibiting this pattern often have a preference for higher- altitude habitats. The highly eroded Selvagens archipelago is c. 30 Myr old, and the Dacia seamount, which is regarded as a former island, is in the age range c. 9–47 Ma. Numerous other seamounts allow for the possibility that several other Maca- C. L. Anderson et al. 2198 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd ronesian islands with overlapping ages have existed (e.g. Geldmacher et al., 2001, 2005). One implication of this is that the laurisilva could actually be a ‘true ancient’ flora, contem- porary with the large European distribution of this ecosystem before the Miocene–Pliocene climate changes, rather than a more recent element (see references in e.g. Vargas, 2007; and Rodrı́guez-Sánchez et al., 2009). When using the aerial age of Fuerteventura, the oldest island of today (21 Myr old), we are possibly severely underestimating the dispersal of Macarone- sian plant lineages from the mainland. We hope that this report will generate further palaeonto- logical and palynological interest in Macaronesia. So often within our separate disciplines we hold key data/knowledge for others, but frequently we do not manage to integrate them. With the new methods in historical biogeography we not only can, but should begin to, integrate the fossil and rock records with phylogenetics, molecular dating and biogeography. ACKNOWLEDGEMENTS We thank H.-U. Schmincke, I. Sanmartı́n Bastida, J. Fuertes Aguilar, R.J. Whittaker, F. Ronquist, L. Sánchez Pinto, Z. Kvaček, A. Hemsley, D. Edwards, B. Oxelman and M. Thulin for help, encouragement and comments. The manuscript benefited substantially by comments from two anonymous referees. The research was covered by grants from Sederholms Resestipendium, Uppsala University, and the P. E. Lindahls Foundation Natural Sciences, Royal Swedish Academy of Sciences, to C.L.A. REFERENCES Ancochea, E. & Huertas, M.J. (2003) Age and composition of the Amanay Seamount, Canary Islands. Marine Geophysical Researches, 24, 161–169. Anderson, C.L. (2007) Dating divergence times in phylogenies. Comprehensive Summaries of Uppsala University Disserta- tions, Acta Universitatis Upsaliensis, Uppsala. Araña, V. & Carracedo, J.C. (1980) Canarian Volcanoes, Vol. 3. Gran Canaria. Rueda, Madrid. Belousov, A., Belousova, M. & Schmincke, H.-U. (2001) Structure and emplacement mechanisms of the Roque Nublo debris avalanche, Gran Canaria (Canary Islands). XXVI EGS General Assembly, Nice, France. Geophysical Research Abstracts, v.3. van den Bogaard, P. & Schmincke, H.-U. (1998) Chronostra- tigraphy of Gran Canaria. Proceedings of the Ocean Drilling Program, Scientific Results, 157, 127–140. Böhle, U.R., Hilger, H.H. & Martin, W.F. (1996) Island colonization and evolution of the insular woody habit in Echium L. (Boraginaceae). Proceedings of the National Academy of Sciences USA, 93, 11740–11745. Bremer, K., Friis, E.M. & Bremer, B. (2004) Molecular phy- logenetic dating of asterid flowering plants shows early Cretaceous diversification. Systematic Biology, 53, 496–505. Britton, T. (2005) Estimating divergence times in phylogenetic trees without a molecular clock. Systematic Biology, 54, 500– 507. Carine, M.A. (2005) Spatio-temporal relationships of the Macaronesian endemic flora: a relictual series or window of opportunity? Taxon, 54, 895–903. Carracedo, J.C. (1999) Growth, structure, instability and col- lapse of Canarian volcanoes and comparisons with Hawaiian volcanoes. Journal of Volcanology and Geothermal Research, 94, 1–19. Carracedo, J.C. & Day, S. (2002) Canary Islands: classic geology in Europe, Vol. 4. Terra Publishing, Harpenden, UK. Carracedo, J.C., Pérez Torrado, F.J., Ancochea, E., Meco, J., Hernán, F., Cubas, C.R., Casillas, R., Rodrı́guez Badiola, E. & Ahijado, E. (2002) Cenozoic volcanism: II. The Canary Islands. The geology of Spain (ed. by W. Gibbons and T. Moreno), pp. 438–472. The Geological Society, London. Caujapé-Castells, J. (2004) Boomerangs of biodiversity?: the interchange of biodiversity between mainland North Africa and the Canary Islands as inferred from cpDNA RFLPs in genus Androcymbium. Botanica Macaronésica, 25, 53–69. Dı́az-Pérez, A., Sequeira, M., Santos-Guerra, A. & Catalán, P. (2008) Multiple colonizations, in situ speciation, and volcanism-associated stepping-stone dispersals shaped the phylogeography of the Macaronesian red fescues (Festuca L., Gramineae). Systematic Biology, 57, 732–749. Donoghue, M.J. & Moore, B.R. (2003) Toward an integrative historical biogeography. Journal of Integrative and Compar- ative Biology, 43, 261–270. Emerson, B.C. (2003) Genes, geology and biodiversity: faunal and floral diversity on the island of Gran Canaria. Animal Biodiversity and Conservation, 26, 9–20. Frisch, T. & Schmincke, H.-U. (1969) Petrology of clinopy- roxene-amphibole inclusions from the Roque Nublo volca- nics, Gran Canaria, Canary Islands (petrology of the Roque Nublo Volcanics I). Bulletin of Volcanology, 33, 1073–1088. Funck, T. & Schmincke, H.-U. (1998) Growth and destruction of Gran Canaria deduced from seismic reflection and bathymetric data. Journal of Geophysical Research, 103, 393– 415. Garcı́a-Talavera, F., Sánchez-Pinto, L. & Socorro, S. (1995) Vegetales fósiles en el complejo traquı́tico-sienı́tico de Gran Canaria. Revista de la Academia Canaria de Ciencias, VII, 77–91. Geldmacher, J., Hoernle, K., van den Bogaard, P., Zankl, G. & Garbe-Schonberg, D. (2001) Earlier history of the ‡70-Ma- old Canary hotspot based on the temporal and geochemical evolution of the Selvagen Archipelago and neighbouring seamounts in the easttern North Atlantic. Journal of Volca- nology and Geothermal Research, 111, 55–87. Geldmacher, J., Hoernle, K., van den Bogaard, P., Duggen, S. & Werner, R. (2005) New 40Ar/ 39Ar age and geochemical data from seamounts in the Canary and Madeira volcanic provinces: support for the mantle plume hypothesis. Earth and Planetary Science Letters, 237, 85–101. Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2199 ª 2009 Blackwell Publishing Ltd Goodson, B.E., Santos-Guerra, A. & Jansen, R.K. (2006) Molecular systematics of Descurainia (Brassicaceae) in the Canary Islands: biogeographic and taxonomic implications. Taxon, 55, 671–682. Guillou, H. (2003) The Plio-Quaternary volcanic evolution of Gran Canaria based on new unspiked K–Ar ages and magnetostratigraphy. Eos, Transactions, American Geophys- ical Union, 84. Fall Meeting Supplement, Abstract V11I-06. Guillou, H., Pérez Torrado, F.J., Hansen Machin, A.R., Car- racedo, J.C. & Gimeno, D. (2004) The Plio-Quaternary volcanic evolution of Gran Canaria based on new K–Ar ages and magnetostratigraphy. Journal of Volcanology and Geo- thermal Research, 135, 221–246. Guzmán, B. & Vargas, P. (2009) Long-distance colonization of the Western Mediterranean by Cistus ladanifer (Cistaceae) despite the absence of special dispersal mechanisms. Journal of Biogeography, 36, 954–968. Kim, S.-C., Crawford, D.J., Francisco-Ortega, J. & Santos- Guerra, A. (1996) A common origin for woody Sonchus and five related genera in the Macaronesian islands: molecular evidence for extensive radiation. Proceedings of the National Academy of Sciences USA, 93, 7743–7748. Kim, S.-C., McGowen, M.R., Lubinsky, P., Barber, J.C., Mort, M.E. & Santos-Guerra, A. (2008) Timing and tempo of early and successive adaptive radiations in Macaronesia. PLoS ONE, 3, e2139. Kvaček, Z. (2007) Do extant nearest relatives of thermophile European Cenozoic plant elements reliably reflect climatic signal? Palaeogeography, Palaeoclimatology, Palaeoecology, 253, 32–40. Kvaček, Z., Manchester, S.R. & Schorn, H.E. (2000) Cones, seeds, and foliage of Tetraclinis salicornioides (Cupressaceae) from the Oligocene and Miocene of western North America: a geographic extension of the European Tertiary species. International Journal of Plant Sciences, 161, 331–344. Mansion, G., Selvi, F., Guggisberg, A. & Conti, E. (2009) Origin of Mediterranean insular endemics in the Boragi- nales: integrative evidence from molecular dating and ancestral area reconstruction. Journal of Biogeography, 36, 1282–1296. Marrero, A. & Francisco-Ortega, J. (2001) Evolución en islas. La forma en el tiempo. Naturaleza de las Islas Canarias. Ecologı́a y conservación (ed. by J.M. Fernández-Palacios and J.L. Martı́n Esquivel), pp. 141–150. Turquesa Ediciones, Santa Cruz de Tenerife. Mehl, K.W. & Schmincke, H.-U. (1999) Structure and emplacement of the Pliocene Roque Nublo debris avalanche deposits, Gran Canaria, Spain. Journal of Volcanology and Geothermal Research, 94, 105–134. Navascués, M., Vaxevanidou, Z., González-Martı́nez, S.C., Climent, J., Gil, L. & Emerson, B.C. (2006) Chloroplast microsatellites reveal colonization and metapopulation dynamics in the Canary Island pine. Molecular Ecology, 15, 2691–2698. Nylander, J.A.A., Olsson, U., Alström, P. & Sanmartı́n, I. (2008) Accounting for phylogenetic uncertainty in bio- geography: a Bayesian approach to dispersal–vicariance analysis of the thrushes (Aves: Turdus). Systematic Biology, 57, 257–268. Oberprieler, C. (2005) Temporal and spatial diversification of Circum-Mediterranean Compositae-Anthemideae. Taxon, 54, 951–966. Percy, D.M. & Cronk, Q.C.B. (2002) Different fates of island brooms: contrasting evolution in Adenocarpus, Genista and Teline (Genisteae, Leguminosae) in the Canary Islands and Madeira. American Journal of Botany, 89, 854–864. Percy, D.M., Page, R.D.M. & Cronk, Q.C.B. (2004) Plant– insect interactions: double-dating associated insect and plant lineages reveals asynchronous radiations. Systematic Biology, 53, 120–127. Pérez-Torrado, F.J., Carracedo, J.C. & Mangas, J. (1995) Geochronology and stratigraphy of the Roque Nublo cycle, Gran Canaria, Canary Islands. Journal of the Geological Society (London), 152, 807–818. Pérez-Torrado, F.J., Martı́, J., Mangas, J. & Day, S.J. (1997) Ignimbrites of the Roque Nublo group, Gran Canaria, Canary Islands. Bulletin of Volcanology, 58, 647–654. Pérez-Torrado, F.J., Santana, F., Rodrı́guez-Santana, A. & Melián, A.M. (2002) Reconstrucción paleogeográfica de depósitos volcanosedimentarios Pliocenos en el litoral NE de gran Canaria (Islas Canarias) mediante métodos topográfi- cos. Geogaceta, 32, 43–46. Ree, R.H. & Sanmartı́n, I. (2009) Prospects and challenges for parametric models in historical biogeographical inference. Journal of Biogeography, 36, 1211–1220. Ree, R.H. & Smith, S.A. (2008) Maximum-likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Systematic Biology, 5, 4–14. Rodrı́guez-Sánchez, F., Guzmán, B., Valido, A., Vargas, P. & Arroyo, J. (2009) Late Neogene history of the laurel tree (Laurus L., Lauraceae) based on phylogeographical analyses of Mediterranean and Macaronesian populations. Journal of Biogeography, 36, 1270–1281. Rothe, P. (2008) Kanarische Inseln. Sammlung geologischer Führer, Band 81. Gebrüder Borntraeger, Berlin/Stuttgart. Sanmartı́n, I., van der Mark, P. & Ronquist, F. (2008) Inferring dispersal: a Bayesian approach to phylogeny-based island biogeography, with special reference to the Canary Islands. Journal of Biogeography, 35, 428–449. Schmincke, H.-U. (1967) Mid-Pliocene fossil wood from Gran Canaria. Preliminary note. Cuadernos de Botánica, 2, 19–20. Schmincke, H.-U. (1968) Pliozäne, subtropische Vegetation auf Gran Canaria. Naturwissenschaften, 55, 185–186. Schneider, J.-L., Brunner, C.A. & Kuttner, S. (1998) Epiclastic sedimentation during the upper Miocene–lower Pliocene volcanic hiatus of Gran Canaria: volcanic hiatus of Gran Canaria: evidence from sites 953 and 954. Proceedings of the Ocean Drilling Program, Scientific Results, 157, 293–313. Schneider, J.-L., Pérez Torrado, F.J., Gimeno Torrentec, D., Wassmer, P., del Carmen Cabrera Santana, M. & Carracedo, J.C. (2004) Sedimentary signatures of the entrance of coarse- grained volcaniclastic flows into the sea: the example of the C. L. Anderson et al. 2200 Journal of Biogeography 36, 2189–2201 ª 2009 Blackwell Publishing Ltd breccia units of the Las Palmas Detritic Formation (Mio–Pliocene, Gran Canaria, Eastern Atlantic, Spain). Journal of Volcanology and Geothermal Research, 138, 295– 323. Trusty, J.L., Olmstead, R.G., Santos-Guerra, A., Sá-Fontinha, S. & Francisco-Ortega, J. (2005) Molecular phylogenetics of the Macaronesian-endemic genus Bystropogon (Lamiaceae): palaeo-islands, ecological shifts and interisland coloniza- tions. Molecular Ecology, 14, 1177–1189. Vanderpoorten, A., Rumsey, F.J. & Carine, M.A. (2007) Does Macaronesia exist? Conflicting signal in the bryophyte and pteridophyte floras. American Journal of Botany, 94, 625– 639. Vargas, P. (2007) Are Macaronesian islands refugia of relict plant lineages?: a molecular survey. Phylogeography of southern European refugia (ed. by S. Weiss and N. Ferrand), pp. 297–314. Springer, Dordrecht. Whittaker, R.J., Triantis, K.A. & Ladle, R.J. (2008) A general dynamic theory of oceanic island biogeography. Journal of Biogeography, 35, 977–994. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 A brief review of further Macaronesian fossil floras and potentially fossiliferous strata. Figures S1 and S2 Leaf morphotypes of the Miocene thermophilous flora WNW of Barranco de Mogán/Azulejos. Figure S3 Las Cuevas del Guincho laurisilva vegetation transported by lahar-like flows. Examples of preservation of twigs, wood, and leaf cuticles. Figure S4 Anatomical and taphonomical features of stem morphotypes from Pliocene thermophilous scrub of Barranco de Tirajana. Figure S5 Unidentified infructescence and unidentified fruit/seed from El Hornillo. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such mate- rials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. BIOSKETCHES Cajsa Lisa Anderson is a systematic biologist who has published research relating to several aspects of molecular dating with fossil constraints. She has a post-doctoral position at Real Jardin Botanico, Madrid, and her current research focuses on the integration of methods in historical bioge- ography and dating. Alan Channing is a geologist and palaeobotanist interested in the ecology, physiology and fossilization processes associ- ated with plants in volcanic terrains. Alba B. Zamuner is a wood-anatomist and palaeobotanist interested in fossil floras from the Mesozoic to Tertiary. Editor: José Marı́a Fernández-Palacios Life, death and fossilization on Gran Canaria Journal of Biogeography 36, 2189–2201 2201 ª 2009 Blackwell Publishing Ltd