Putting biogeography’s cart back behind taxonomy’s horse: a response to Triantis et al.



In a recent paper, two of us discussed diversity patterns and diversification processes in the Azores flora. Triantis et al. (2012, Journal of Biogeography, 39, 1179–1184) challenged our hypothesis that palaeoclimatic differences had an effect on diversification rates and suggested that area, island age and isolation explain diversity patterns. They did not, however, fully address the results from our subsequent paper, in which we showed that diversity patterns evident from phylogeographic studies differ markedly from those suggested by checklists. Checklists are working hypotheses and we suggest that the discrepancies evident between molecular data and checklists may be indicative of deficiencies in our taxonomic understanding of the Azores flora. Patterns of molecular and morphological diversity need to be better understood, and the discrepancies between checklists and molecular data accounted for, before we can establish the relative importance of factors such as palaeoclimate, area, island age or isolation in generating endemic diversity patterns in the Azores flora.

In a letter to Joseph Hooker dated Christmas Day, 1844, Charles Darwin (Darwin Correspondence Project, 2012, letter 803, p. 3) commented on a recently published enumeration of the Azores flora (Watson, 1843) and noted: ‘Watson’s Paper on [the] Azores has surprised me much; do you not think it odd, the fewness of peculiar species…?’

More than 160 years later, the patterns of Azorean plant diversity that (briefly) puzzled Darwin remain to be fully explained. Our two recent papers (Carine & Schaefer, 2010, hereafter C&S; and Schaefer et al., 2011a, hereafter SEA), sought to document and explain those patterns. Understanding the evolution of the distinctive but poorly studied Azores flora has the potential to inform our appreciation of the processes generating island plant diversity patterns more generally and we therefore welcome the interest in the topic that our papers have stimulated, notably the correspondence article by Triantis et al. (2012, hereafter TEA). The reply of TEA does, however, contain misconceptions and omissions relating to the findings of C&S and fails to take into account fully the results and conclusions of SEA. Here, we address these issues and highlight what we believe to be the key challenge to understanding patterns of diversity in the endemic flora of the Azores.

The pattern

C&S described patterns of diversity in the Azores flora based on the taxonomic concepts used at that time and that are still reflected in the currently available checklist data for the Azores (Borges et al., 2010). Analysing those data, C&S showed that the Azores is unusual not only in the apparently low number of endemic species (as pointed out by Darwin) but also more specifically in the low number of single-island endemic taxa (SIEs) and the resulting left-skewed PKI curve (presence in k-island, i.e. the distribution of endemic taxa per number of islands of occupancy). Thus, in the Azores, endemic taxa that are apparently distributed across the entire archipelago constitute a much larger proportion of the endemic flora than do single-island endemics. In contrast, in the eight other oceanic archipelagos investigated by C&S [Canary Islands, Madeira, Cape Verdes, Hawaii, Galápagos, Marquesas, Society Islands, Ogasawara (Bonin) Islands], single-island endemics account for a substantially greater proportion of the endemic flora than do endemics that are widespread across the archipelago.

In addition, C&S used distribution data in conjunction with ecological data to identify genera within which either inter-island allopatry or ecological speciation may have occurred. Contrary to the suggestion of TEA, this analysis was performed on the complete endemic angiosperm flora with the nine lineages identified (Table 1 in C&S) constituting all lineages that contained two or more endemic taxa. C&S found instances of putative allopatric diversification to be even more limited than putative instances of ecological diversification. As C&S (p. 82) noted: ‘The limited incidence of adaptive, ecological speciation in the Azores... is consistent with the limited ecological diversity of the islands. However, the limited incidence of inter-island allopatry… is surprising when inter-island distances in the archipelago are considered’. This was the pattern that they sought to explain.

SEA subsequently collected population-level genetic data from a range of Azorean endemic plant lineages to further test the patterns observed. In marked contrast to the pattern suggested by checklist data, molecular data indicated that allopatric diversification has been common in the archipelago’s endemic flora. Molecular diversity in most lineages examined showed a geographic signal and the data suggest that, at least at the molecular level, SIEs are more common than previously documented, with even young islands having endemic entities. SEA sampled c. 11% of the endemic flora. We have since continued to sample additional endemic vascular plant lineages and have found further instances of previously overlooked diversity in Smilax, Ranunculus and Centaurium (H. Schaefer, unpublished data), and also in Viburnum and Leontodon (M. Moura et al., unpublished data) but so far not in Bellis or Rumex (H. Schaefer, unpublished data). Overall, with nearly 20% of Azorean endemic lineages now sampled from two or more islands, the patterns documented by SEA are still evident and still differ markedly from those suggested by checklist data. We are continuing to expand this dataset.

TEA argued that whilst such phylogeographic data are important ‘...when such evidence is only available for a limited number of lineages, extrapolating the patterns observed to the level of the whole biota may be misleading’. They then went on to ‘...use existing knowledge and the available checklists for plants, arthropods and land snails to address why there are so few Azorean endemic species and why they are so widespread’ [our italics]. We would suggest that any explanation for diversity patterns in the Azorean endemic flora needs to account not only for the pattern suggested by checklist data (i.e. that which TEA sought to explain) but also the markedly different pattern observed with molecular data which suggest many more restricted endemics.

The explanation?

Darwin offered no explanation for the low levels of endemic diversity observed in the Azores flora. Subsequent explanations, reviewed by C&S, have included the ages of the islands, the ages of the endemic lineages and the low levels of habitat diversity. C&S acknowledged that the low level of habitat diversity in the Azores has contributed to the limited amount of endemic diversity observed in the Azores flora (cf. the suggestion of TEA) and focused on trying to explain why allopatric speciation in the Azores was so limited. C&S focused in particular on the much lower incidence of allopatric speciation observed in the Azorean endemic flora in comparison with the Canarian endemic flora. The palaeoclimate hypothesis was proposed to explain that pattern. Subsequently, Cardoso et al. (2010) proposed another hypothesis, suggesting that human-mediated extinctions coupled with high dispersal ability offered a plausible explanation for the similar patterns observed in the Azorean endemic spider fauna.

TEA present data (Table 1 in TEA) that contradict the data presented by C&S. As TEA note, their data correspond to ‘conditions at sea level and the actual climate would have varied according to the interaction of each island’. Whether global models such as those used by TEA are sufficiently robust to predict palaeoclimatic patterns in the topographically complex Macaronesian islands is beyond our area of expertise. However, it has been suggested that island plants are extremely sensitive to climate change (Bramwell, 2011) and whilst the response of plants to future climate change is difficult to predict via correlative-style species distribution models, a recent study suggested that even relatively small rises in global temperature could have a profound impact on the extent of laurisilva on Tenerife, leading to fragmentation and a marked reduction in area of occupancy (Martín-Esquivel, 2010; although cf. Sperling et al., 2004). Whilst there remains no consensus on timings (cf. Rognon & Coudé-Gaussen, 1996; Ortiz et al., 2006; Meco, 2008), changes to the wind regimes in the Canaries have certainly occurred and, over a longer time-scale, fossil plant deposits, for example those attributed to Tetraclinis articulata Mast., suggest changes to the climate of at least one of the islands in the Canarian archipelago (Gran Canaria, Anderson et al., 2009; see also Fernández-Palacios et al. (2011) for further discussion on palaeoclimate in the region). Consistent with the hypothesis of C&S, Connor et al. (2012) recently found that late Quaternary palaeoclimatic variations detected using geochemical proxies were not clearly reflected as changes in past forest composition or structure in the Azores. However, palaeoecological data for the Macaronesian islands are limited (Forjaz et al., 1970; van Leeuwen et al., 2005; de Nascimento et al., 2009; Connor et al., 2012) and more data are needed to understand the impact of past climate change on the biota of the archipelagos.

C&S (p. 86) concluded that: ‘If climate has contributed to the diversification of the flora and recent climatic history has played a major role in shaping endemic diversity patterns across Macaronesia, then palaeoclimate needs to be integrated into explanations of diversity patterns across the region’. We would suggest that until the nature and extent of palaeoclimatic change in the region has been robustly established, it is appropriate that the potential impact of past climate on the patterns of diversity observed in Macaronesia is taken into account. This was acknowledged by Whittaker et al. (2008) and Fernández-Palacios et al. (2011) but it appears to have been discounted by TEA who, after rejecting the palaeoclimate scenario presented by C&S, argue that the low levels of endemic plant diversity documented in Azorean checklists are explained by the ages of the islands, their small size, the low levels of habitat diversity and the inter-archipelagic missing step.

Taxonomy – a key challenge

The hypotheses of C&S, TEA and Cardoso et al. (2010), together with earlier hypotheses reviewed by C&S, all seek to explain the paucity of endemics and the widespread distributions of Azorean endemic taxa. None, at least as currently framed, are able to explain also the markedly different patterns of molecular diversity documented by SEA.

The diversity of island plants is considered by some to be relatively well known from a taxonomic perspective (e.g. Joppa et al., 2011). However, as Caujapé-Castells et al. (2010) note, the terrain of many islands is a powerful impediment to botanical exploration and, consequently, large gaps in knowledge of the status and distribution of endemic plants remain in most archipelagos. We concur with TEA that additional effort is needed to increase our taxonomic knowledge of the Macaronesian flora. We wish we also shared TEA’s confidence in the robustness of the diversity patterns derived from current checklists. As SEA acknowledge, checklists have played an extremely important role in the development of biogeographic theories. However, they are imperfect working hypotheses that continue to develop. Both C&S and TEA document the increase in endemic diversity recognized in the Azores flora in recent years. TEA contrast the endemic element of the flora in the 2005 and 2010 checklists for the Azores and suggest a relatively trivial increase in the proportion of endemics. However, whilst this is true when the entire flora is considered, the figures are markedly skewed by the size of the introduced flora, which comprises over 75% of the total flora of the Azores. In relative terms, there has been a 6% increase in the size of the endemic flora in just 5 years and without a particularly focused taxonomic effort.

We have contributed to the compilation of those checklists (Schaefer, 2003; Silva et al., 2010) and the results of SEA surprised us. They caused us to reflect on the strikingly different patterns we observed and to consider two important questions. First, do the geographically restricted and genetically distinct entities identified by SEA correspond to morphologically distinct taxa? Second, if they do, why have they been overlooked?

As SEA noted, molecular patterns observed in Angelica lignescens J. P. Reduron and Danton correlate with previously documented morphological differences. Taxonomic revisions of A. lignescens and a number of other groups are now underway and preliminary results suggest that genetic diversity correlates with morphological differences in several groups. We therefore disagree with TEA’s statement that ‘it is clear that the proportions of SIEs will remain well below those of the other archipelagos for plants...’. Rather, we expect the rate of description of new endemic taxa in the flora to increase. Given the challenges of terrain highlighted by Caujapé-Castells et al. (2010), new species are likely to be discovered in the Canaries and other archipelagos as well. However, whereas in the Azores an increase in SIEs will markedly change the shape of the PKI curve and the incidence of intra-archipelago diversification events, these would not be fundamentally altered by the discovery of new species in the Canaries, where the existing patterns would rather become more pronounced. Significantly, it would appear that such a ‘Linnean shortfall’ is unlikely to be restricted to Azorean plants. TEA present data on endemism in land snails, plants and arthropods for the Macaronesian archipelagos (Table 2 in TEA). For arthropods, they report 1077 native species in the Azores, of which 9.3% are considered SIEs. However, we note that in the Hymenoptera alone, more than 200 Azorean ‘morphospecies’ are awaiting formal description (or identification as introduced species) and none of them have been listed in checklists to date (P. A. V. Borges, Universidade dos Açores, Azorean Biodiversity Group CITA-A, pers. comm., 2010).

Whilst new plant species are collected in the field, most are ‘discovered’ in the herbarium, where the critical and detailed comparison of specimens can take place. It has been proposed that more than half of the plant species world-wide that are yet to be discovered may already have been collected and are awaiting discovery in herbaria, with the lack of taxonomists an important rate-limiting factor in their discovery (Bebber et al., 2010). In the case of the Azores, a database of specimens of endemic plants is being compiled by one of us (K.J.). This includes specimens from 12 herbaria located in the Azores and mainland Portugal and from herbaria in Europe and the United States that hold significant collections of Azorean endemic plants. In total, 2652 collections (excluding duplicates) have so far been examined. Most taxa show a biased collection effort, with some islands well represented and others not in the herbaria sampled. Moreover, in many cases, specimens have been consecutively collected over the years at the same easily accessible and well-known places. By way of example, Scabiosa nitens Roem. and Schult. occurs on eight of the Azorean islands and it is locally common (Schaefer, 2003). Forty specimens of this species have been included in our database, of which 10 were collected from one island (Flores); Pico and Santa Maria are each represented by just two collections and Faial by one. Is S. nitens really a single widespread taxon or does it contain overlooked taxa? The current sample of specimens – that are also dispersed in different herbaria – would appear to be inadequate to answer this question. Addressing this aspect of the Linnean shortfall, through both extensive gap filling collections, particularly from less accessible areas, and effectively mobilizing the highly dispersed collections that already exist (e.g. through digitization), should be considered a key priority for researchers working on Azorean endemic plants. This is also crucial for management and conservation of the endemic flora, where the best intentions and concepts will fail when endemic species are overlooked or exotics are misidentified as threatened endemics, as in the case of the Azorean Marsilea population (Schaefer et al., 2011b).

TEA note that ‘...patterns of diversification on the Macaronesian archipelagos, particularly on the Canary Islands and the Azores, have been extensively discussed over the past few years...’. This is certainly the case, albeit with a strong bias towards the Canaries. Notably, however, prior to the publication of SEA, phylogeographic data for Azorean plants were almost totally lacking and even in the Canaries, Caujapé-Castells (2011) highlights significant sampling deficiencies in the molecular data available to date. The phylogeographic data now available for Azorean plants do not conform to the patterns that people have previously sought to explain. We suggest that these new data highlight deficiencies in our taxonomic understanding of the Azores endemic flora. Until patterns of molecular and morphological diversity in the Azores flora are better understood and integrated into a revised taxonomic framework, any discussion of the importance of palaeoclimate (C&S), area, island age, isolation or intra-archipelago missing steps (TEA), or indeed dispersal ability and extinction (Cardoso et al., 2010) is putting the biogeographic cart before the taxonomic horse.

Editor: José María Fernández-Palacios