The Canary Islands are characterised by an outstanding level of biodiversity with a high number of endemic taxa. The Canarian avifauna is no exception and six extant avian species are recognised as endemic. However, we have a limited understanding of the genetic structure of these taxa, which makes it difficult to identify conservation priorities based on the existence of unique lineages. we analysed the diversification process and demography of the Canary Islands Chiffchaff Phylloscopus canariensis using two mitochondrial genes (cytochrome b and cytochrome oxidase subunit I). Although the species colonised the Canary Islands during the early Pleistocene, our results suggest a recent process of differentiation, which ran in parallel to the Last Glacial Maximum. Chiffchaffs from Gran Canaria emerge as a unique lineage since none of the haplotypes found on this island were recorded anywhere else in the archipelago. Our findings suggest a process of divergence according to the gradual colonisation and subsequent isolation of nearby islands. Demographic results show a stable trend of Chiffchaff populations until the colonisation of Gran Canaria (ca. 15,000 years ago), where the species significantly increased its effective population size. Nowadays, the effective population size of the Canary Islands Chiffchaff is stable, which highlights the ability of Chiffchaffs to adapt to local disturbances related to human activities. Overall, our results provide a scenario of incipient differentiation of the Canary Islands Chiffchaff and, importantly, underscore once more the role of Gran Canaria in driving speciation processes in this archipelago.—Illera, J.C., Ramírez, A., Rodríguez, L., Polypathellis, K. & Pérez-Tris, J. (2020). Maternal genetic structure reveals an incipient differentiation in the Canary Islands Chiffchaff Phylloscopus canariensis. Ardeola, 67: 401-414.
The avifauna of the Canary Islands has fascinated naturalists and biologists since the nineteenth century due to the diversity of breeding species and, especially, to their conspicuous phenotypic differentiation in relation to their mainland counterparts. This interest resulted in the description of a plethora of new endemic species and subspecies during the nineteenth and twentieth centuries (Martín & Lorenzo, 2001). Taxonomic descriptions were traditionally performed by visual inspection of individuals, with subsequent identification of diagnostic traits. However, this approach was not associated with rigorous statistical analyses, which raises doubts about whether these populations are idiosyncratically distinct (Illera et al., 2016). In recent years, the routine use of molecular tools has provided insights into our understanding of the genetic diversity, colonisation pathways and taxonomic identity of many taxa in the Canary Islands (Juan et al., 2000; Stervander et al., 2015; Kangas et al., 2018; Senfeld et al., 2020). Some of these findings have revealed strikingly high genetic differentiation between nearby populations, which is consistent with subspecific classification (Dietzen et al., 2003, 2015; Päckert et al., 2006). These isolated and genetically distinct populations probably went unnoticed because they showed similar morphological traits (Kvist et al., 2005; Illera et al., 2018). Overall, these findings highlight the need for a rigorous assessment of avian biodiversity in the Canary Islands to identify overlooked cryptic diversity, which should be officially recognised under Spanish and Canarian conservation laws. The final aim should be to support the management of all biological resources according to their conservation status.
Over the last two decades, knowledge of the evolutionary history of many avian taxa in the Canary Islands has noticeably increased thanks to a combination of phylogenetic or phylogenomic data and fossil information (reviewed in Illera et al., 2012, 2016; Van Doren et al., 2017). However, we have a limited understanding at the population level of the genetic structure of most endemic bird species that occur in the Canary Islands. This knowledge gap precludes recognising new cryptic lineages, which should be a critical factor in conservation management (Illera et al., 2007, 2016).
The Canary Islands Chiffchaff Phylloscopus canariensis is one of six endemic avian species that inhabits the Canary Islands. It is a small (<10g) sedentary passerine (González-Melián, 2018) that occurs on the central and western islands, and is one of the most common land species that breeds in the Canary Islands (Martín & Lorenzo, 2001; González-Melián, 2007). Although the species is absent from the eastern islands (i.e., Fuerteventura and Lanzarote), a recognised (extinct) endemic subspecies (P. c. exsul) was breeding on these two islands probably until the first quarter of the twentieth century (Figure 1). Chiffchaffs show wide ecological plasticity in with respect to climatic conditions. Thus, the species inhabits vegetated areas ranging from sea level to alpine habitats (>2,400m a.s.l.), being absent only from sparsely vegetated areas (Martín & Lorenzo, 2001).
Recent phylogenetic analyses have indicated that the Canary Islands Chiffchaff is related to the Mountain Chiffchaff P. sindianus, a species with a Caucasian distribution (Tietze et al., 2015, but see Alström et al., 2018). This result is remarkable because the distributions of other chiffchaff species are geographically closer to the Canary Islands than that of the Mountain Chiffchaff. Thus, the Common Chiffchaff (P. collybita) and the Iberian Chiffchaff (P. ibericus) are common breeding and/or wintering species in northern Africa and the Iberian Peninsula (Gordo et al., 2017). However, according to the limited nodal support found in both phylogenies, the phylogenetic relationships within this clade do not yet appear to have been resolved (Tietze et al., 2015; Alström et al., 2018). In fact, until the beginning of the twenty-first century, the Canary Islands Chiffchaff was considered to be a subspecies of the Common Chiffchaff (Martín & Lorenzo, 2001). However, significant differentiation found in mitochondrial sequences, and acoustic and morphological data between taxa of the Chiffchaff complex (i.e., P. ibericus, P. canariensis and P. sindianus) supported a new taxonomical classification in which the Canary Islands Chiffchaff emerged as a species in its own right (Helbig et al., 1996; Raković et al., 2019).
The chiffchaffs colonised the Canary Islands during the early Pleistocene, 2.28 million years ago (Valente et al., 2017). According to its taxonomic classification, the Canary Islands Chiffchaff is a polytypic species with two subspecies (P. c. canariensis and P. c. exsul), which means that individuals of each subspecies experienced reproductive isolation from one another. However, it is also plausible to expect further differentiation, especially between the central and western islands, as has been found in other avian species. Thus, distinctive lineages have emerged for a wide number of taxa on Gran Canaria, including European Robins Erithacus rubecula, Common Chaffinches Fringilla coelebs, Blue Chaffinches F. polatzeki and Canarian Blue Tits Cyanistes teneriffae (Pestano et al., 2000; Dietzen et al., 2003; Kvist et al., 2005; Suárez et al., 2009).
Here, we provide a comprehensive phylogeographic framework based on two mitochondrial genes: cytochrome b (cyt-b) and cytochrome oxidase subunit I (COI). Our approach had three aims. The first goal was to scrutinise the genetic structure of the Canary Islands Chiffchaff across the central and western islands, with the specific aim of estimating the genetic variation and relationships among islands. Based on the taxonomic hierarchy recognised to date (there is only one extant subspecies), we predicted that there is contemporary gene flow among current populations and an absence of significant genetic structure. The second goal was to determine the pathway of colonisation and phylogenetic relationships of the Canarian populations. Here we expected to find a very recent differentiation process, in which genetic differentiation increases with geographical distances among populations. Finally, our third objective was to analyse the demographic evolution of the Canary Islands Chiffchaffs, with the final aim of determining the role of human arrival on its effective population size. Because the Canary Islands Chiffchaff is a common avian species inhabiting woodland habitats, we predicted that the reduction of forest habitats, that began immediately after humans arrived in the archipelago (especially after the European conquest), should have resulted in a significant reduction of the effective population size.
Material and methods
Individuals (n = 86) were caught at various localities, using mist nets, between 2003 and 2010. we ringed all individuals with unique numbered aluminium rings. we collected blood samples (≈ 25µl) by brachial or jugular venipuncture using insulin syringes (0.5mm), and diluted them in 1000µl of pure ethanol in screw-cap microfuge tubes. Finally, we released all birds at the site of capture.
We extracted DNA using an ammonium acetate method (dx.doi.org/10.17504/proto cols.io.knycvfw), for working with a final concentration of 10-50ng/µl. we amplified a 1018 base pair (bp) fragment of the mitochondrial cyt-b gene using primers L14995 (5′ CTCCCAGCCCCATCCAACATCTCA GCATGATGAAACTTCG 3′) and H16065 (5′ CTAAGAAGGGTGGAGTCTTCAGTT TTTGGTTTACAAGAC 3′) (Helbig et al., 1995). PCR reactions were set up in a final volume of 10µL including 5µL of GoTaq® Green Master Mix 2× (Promega Corporation, Madison, wI, USA), 0.5µL (10mM) of primers L14995 and H16065 and 1µL of DNA. Reactions were performed on a G-Storm GS2 thermal cycler (Somerton Biotechnology Centre, Somerset, UK) under the following conditions: initial denaturation at 94°C for 4 minutes followed by 35 denaturation cycles at 94°C for 30s, annealing at 50°C for 45s, and extension at 72°C for 60s, and a final extension at 72°C for 10min. PCR products were checked by electrophoresis in 1.5% agarose gels stained with GelRedTM nucleic acid gel stain (Biotium, Inc., Hayward, CA, USA). The samples were purified using the E.Z.N.A. Gel Extraction Kit OMEGA. Sequencing reactions were performed from both ends using the Perkin Elmer BigDye v. 3.1 (Applied Biosystems, Carlsbad, CA, USA) terminator reaction mix in a volume of 10µL using 1µL of PCR product and L14995 and H16065 primers. Sequencing reaction conditions were: initial denaturation at 94°C for 2 minutes followed by 25 denaturation cycles at 94°C for 30s, annealing at 50°C for 30s, and extension at 60°C for 2 minutes, followed by a final extension at 60°C for 1 minute. The final product was sequenced on an ABI PRISM® 3130xl Genetic Analyzer.
We also amplified a 615 bp fragment of the mitochondrial COI gene using the primers PasserF1 (5′ CCAACCACAAAGA CATCGGAACC 3′) and PasserR1 (5′ GTA AACTTCTGGGTGACCAAAGAATC 3′) (Lohman et al., 2009). PCR reactions were performed using the following conditions: initial denaturation at 94°C for 3 minutes followed by 35 denaturation cycles at 94°C for 1 minute, annealing at 58°C for 1 minute, extension at 72°C for 1.5 minutes, and a final extension at 72°C for 10 minutes. Sequencing reactions were performed with the same reagents and conditions that were used with the cyt-b gene.
We aligned sequences by hand using BioEdit version 7.0.9 (Hall, 1999) and Chromas version 2.5.1. ( https://technelysium.com.au/wp/chromas/) programmes. we used the DnaSP programme (Version 5.10.01; Librado & Rozas, 2009) to calculate the genetic diversity estimated per population (island), in terms of the number of haplotypes, haplotype and nucleotide diversities and theta (2Neµ). we assessed genetic distances between populations using pairwise FST values and tested their statistical significance by performing 10,000 permutations using the ARLEQUIN programme (version 126.96.36.199; Excoffier & Lischer, 2010). we also evaluated isolation by distance by testing whether pairwise genetic distances were correlated with geographic distances between islands, for which we performed a Mantel test using the vegan package (Oksanen et al., 2018) in the R programme (Version 3.5.1; R Core Team, 2018). we used Rousset's approach, which compares the FST pairwise genetic distance matrix (FST / (1 – FST)) with log-transformed geographic distance (Rousset, 1997).
We inferred the network of relationships among mitochondrial sequences (both cyt-b and COI) using the TCS programme (Version 1.21; Clement et al., 2000), which implements the probabilistic method of statistical parsimony. we used a 5% limit, and ambiguity characters were treated as missing data. The graph file obtained in TCS was then used to classify haplotypes and display the final network using the TCS Beautifier programme (tcsBU) (Santos et al., 2016).
We estimated population divergence times using the BEAST programme (Version 1.8.1; Drummond et al., 2012). we only included the sequences of cyt-b, which is the most commonly used mitochondrial gene for dating divergence times in avian taxa (Valente et al., 2017, 2020). we inferred the most appropriate nucleotide substitution model (HKY+ I) according to the JModelTest programme (Version 2.1.4; Darriba et al., 2012). we used a strict molecular clock, with a mean of 0.01 and standard deviation of 0.0075 substitutions per site per million years (Illera et al., 2008). we ran two independent MCMC analyses of 50,000,000 generations, with a burn-in of 5,000,000 steps. Finally, we evaluated the convergence of the chains with Tracer (Version 1.7.1; Rambaut et al., 2018). Recently, two studies have revealed problems with the joint estimation of tree topology provided by BEAST (Emerson & Hickerson, 2015; Emerson et al., 2015). For this reason, we performed another Bayesian phylogenetic inference using the Mr. Bayes programme (Version 3.1.2; Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003). we ran Mr. Bayes for 10,000,000 iterations, and we sampled trees and model parameters every 100 generations. we set up a burn-in of 2,500,000 generations, and performed two independent runs to ensure stability in the posterior probabilities. we also assessed the convergence of the MCMCs obtained with Mr. Bayes using Tracer.
Genetic diversity estimates from cytochrome b and COI genes for the Canary Islands Chiffchaff Phylloscopus canariensis. N: Sample size. Hap: Number of haplotypes. HD: Haplotype diversity and standard deviation (SD). ND: Nucleotide diversity and standard deviation (SD). θ: Theta (2Neµ), where Ne is the effective population size and µ the mutation rate.
[Estimas de diversidad genética del mosquitero canario Phylloscopus canariensis obtenidas con los genes mitocondriales citocromo b y citocromo oxidasa subunidad I. N: Tamaño de muestra. Hap: Número de haplotipos. HD: Diversidad haplotípica y desviación estándar (SD). ND: Diversidad nucleotídica y desviación estándar (SD). θ: Theta (2Neµ), donde Ne es el tamaño efectivo poblacional y µ es la tasa de mutación.]
We evaluated the changes in the effective population size (Ne) of the Canary Islands Chiffchaffs using the Bayesian skyline plot implemented in BEAST (Drummond et al., 2012). This plot is a coalescent demographic reconstruction, which is effective for detecting population expansions, contractions and bottleneck events.
Cytochrome b pairwise FST values obtained in the Canary Islands Chiffchaff Phylloscopus canariensis. P values obtained after 10,000 permutations are indicated with asterisks. * = P < 0.05; * * = P < 0.01; * * * = P << 0.001.
[Valores FST de las comparaciones por pares entre las diferentes poblaciones de mosquitero canario Phylloscopus canariensis. El valor de significación P se obtiene tras 10.000 permutaciones y se indica con asterisco. * = P < 0,05; ** = P < 0,01; *** = P << 0,001.]
In total, we obtained 170 sequences from cyt-b (n = 84) and COI (n = 86) genes (15-18 sequences per population, Table 1). Unique haplotypes have been deposited in the NCBI gene bank database (MT237934-52; MT241 268-241300). Cytochrome b showed higher genetic diversity, with three times as many haplotypes (n = 33) as COI (n = 11). Tenerife and La Gomera exhibited the highest values of genetic diversity both for haplotype and nucleotide diversity, while the most peripheral populations (El Hierro and Gran Canaria) provided the lowest estimates (Table 1).
Haplotype networks for cyt-b and COI were concordant (Figure 2 and Supplementary Material, Appendix 1, Figure A1 (Ra_11_Supplementary Electronic Material.pdf)), depicting a limited differentiation in the Canary Islands Chiffchaffs. Gran Canaria was the only island for which no haplotypes were found to be shared with other populations. The parsimony network based on cyt-b showed the La Gomera population connecting the two most frequent haplotypes, which were found in individuals from three and four islands.
All pairwise FST values were significant (P < 0.05), and all pairwise FST values including Gran Canaria reached the highest differentiation levels with high statistical support (P < 0.001) (Table 2). Finally, pairwise FST and geographical distances were positively and significantly correlated (Mantel statistic r = 0.761, P = 0.033) (Figure 3).
Bayesian analyses confirmed the incipient isolation of the Gran Canaria population, with all individuals grouping together. However, two individuals from Tenerife also grouped with the Gran Canarian birds within the same pool with high nodal support (Posterior probability = 0.97). The remaining individuals pooled in different groups but without showing a clear differentiation among islands ( Supplementary material appendix 1, Figure A2 (Ra_11_Supplementary Electronic Material.pdf)).
The differentiation process of the extant Canary Islands Chiffchaff populations started recently, at the end of Pleistocene (29,900 years; 22,900-37,000 years ago, lower and upper 95% highest posterior density values, respectively). Gran Canaria began its genetic divergence from the remaining populations shortly after the extant chiffchaffs started to differentiate (14,900 years; 9,681-20,040 years ago).
Our Bayesian Skyline reconstruction showed a population size history in which the Canary Islands Chiffchaffs experienced a demographic expansion event coincident with the divergence of Gran Canaria (ca. 15,000 years ago). This expansion finished approximately 5,000 years ago, and nowadays the species shows a stable trend (Figure 4). Assuming a generation time of two years for the Canary Islands Chiffchaff, we estimated an effective population size (Ne) of 200,736 individuals (61,130-525,048 birds, lower and upper 95% highest posterior density values, respectively).
Although chiffchaffs colonised the Canary Islands during the early Pleistocene (Valente et al., 2017), our results showed an incipient differentiation process that ran in parallel with the Last Glacial Maximum. The ongoing process of differentiation was inferred from three independent analyses: 1) haplotype networks; 2) significant pairwise FST values found among all populations; and 3) the Bayesian inference obtained with BEAST and Mr. Bayes. Overall, our findings suggest a recent differentiation mirroring a process of incomplete lineage sorting with a limited contemporary gene flow among extant populations. The fact that two individuals from Tenerife appeared nested with the Gran Canaria cluster, and the limited gene flow supported by FST values, could represent the retention of an ancient polymorphism from the ancestor of the Canary Islands Chiffchaff. It is interesting that acoustic information published to date also suggests a similar pattern of differentiation. For example, Henning et al. (1994) found conspicuous differences in some acoustic traits; specifically, singing and alarm calls (i.e., contact calls) were found to differ among some populations. They distinguished two main groups: one group pooled individuals from Gran Canaria and Tenerife, and the other group pooled individuals from the remaining populations. In addition, to a lesser extent, these authors also found differences between all Canarian islands. In addition, Naguib et al. (2001) studying chiffchaff calls in Tenerife and La Gomera found significant differences between these islands. Taken together, these acoustic findings suggest a context of limited dispersal events between nearby islands.
There was no evident relationship between the Last Glacial Maximum and divergence events, but a similar pattern of an early Pleistocene colonisation event and recent differentiation has also been found in Berthelot's Pipit Anthus berthelotii in Macaronesia (Illera et al., 2007; Spurgin et al., 2014). we have a limited understanding about how Quaternary climatic oscillations have created or extirpated genetic diversity through the contraction, expansion and extinction of Canarian biota in the archipelago (García-Verdugo et al., 2019). However, it is plausible to speculate that the climate changed towards drier conditions during the late Pleistocene, although with episodes of wetter periods (Alonso-Zarza & Silva, 2002; weigelt et al., 2016), which favoured the expansion of some avian taxa in Macaronesia.
The divergence process found in Gran Canaria, where the population appears to be isolated from the remaining Canary Islands Chiffchaff populations, is remarkable. This genetic finding highlights once more the role of Gran Canaria in providing distinctive lineages of Canarian avifauna. The level of genetic differentiation is lower than that observed in other Canarian species, such as Robins, Canarian Blue Tits, and Common and Blue Chaffinches (Pestano et al., 2000; Dietzen et al., 2003; Stervander et al., 2015; Illera et al., 2018). However, it is similar to the level of differentiation found in the Southern Grey Shrike Lanius meridionalis (Padilla et al., 2015), and slightly higher than that documented in the Stone-Curlew Burhinus oedicnemus (Mori et al., 2017) and Spectacled warbler Sylvia conspicillata (Illera et al., 2014). why does Gran Canaria frequently emerge as a unique lineage within Canarian birds? Gran Canaria is the third oldest island in the archipelago (14 my), the third largest island (1,560km2), and supports a wide variety of habitats, which could further favour diversification processes. Therefore, a combination of age, habitat diversity, and isolation from other islands with similar ecosystems (such as Tenerife, which is approximately 60km away), could explain the recurrent differentiation patterns observed.
The incipient differentiation process recorded with the mitochondrial sequences limits our ability to unravel the pathway of chiffchaff colonisation within the archipelago. Our findings show a positive correlation between geographical and genetic distances, suggesting gradual genetic differentiation with increasing distances. However, the incomplete lineage sorting recorded, and the absence of sequences from the extinct subspecies from the eastern islands, precludes identifying the ancestral population. A future study using high genomic coverage and wide population sampling, including the extinct taxa of the eastern islands, is now needed to infer which population acted as a source and served as the cradle of diversification for the Canary Islands Chiffchaff.
Finally, our demographic analyses suggests that the effective population size of the Canary Islands Chiffchaff has evolved according to its diversification history in the Canary Islands. we did not record any bottleneck event in the Canary Islands Chiffchaff, which could have been attributable to the arrival of aboriginal people (ca. 2,000 years ago) or Europeans (ca. 500 years ago) (Alcover et al., 2009). Instead, we found a stable trend in the effective population size from the onset of its diversification in the central islands (Tenerife and La Gomera) and western islands (La Palma and El Hierro) islands until the colonisation of Gran Canaria (ca. 15,000 years ago), where the effective population size of the Canary Islands Chiffchaff increased remarkably (Figure 4). This last result seems to be a logical consequence of occupying a large island such as Gran Canaria. A new equilibrium was reached 5,000 years ago, with no evidence of a drastic reduction of the effective population size thereafter. Overall, this result suggests a limited role of human arrival (either aboriginal or European) in determining the population size of the Canary Islands Chiffchaff, which could be explained by the considerable ability of chiffchaffs to adapt to local disturbances.
The regional government of the Canary Islands provided the permits for sampling the birds. J.C. Illera was funded by the Spanish Ministry of Science, Innovation and Universities (Ref.: PGC2018-097575-B-I00) and by a GRUPIN research grant from the Regional Government of Asturias (Ref.: IDI/2018/000151); J. Pérez-Tris was funded by the Spanish Ministry of Science, Innovation and Universities (CGL 2017-82117-P); and A. Ramírez was funded by the Ministry of Education and Science (Ref.: EX2006-0733). Leticia Viesca helped in the laboratory. Alice Cibois and two anonymous reviewers provided helpful comments on the manuscript. Elisa Bauer helped with the translation of the German text, and Allison Perry polished the language. Juan Varela provided the illustration of the Canary Islands Chiffchaff.
JCI designed and supervised the study. JPT, AR and JCI performed the fieldwork. JCI, LR and KP carried out the molecular work. JCI performed statistical analyses. JCI drafted the manuscript and all authors commented and approved the final version of the manuscript.
Supplementary Electronic Material
Additional supporting information may be found in the online version of this article. See the volume 67(2) on www.ardeola.org
Figure A1. Parsimony network of the Canary Islands Chiffchaff Phylloscopus canariensis performed with the TCS programme, based on COI sequences.
Figure A2. Bayesian consensus radial tree for the Canary Islands Chiffchaff Phylloscopus canariensis inferred with Mr. Bayes, based on cyt-b sequences.