Page 7 - Voda_Dapporto_alii_2016

www.nature.com/scientificreports/

probabilities following local extinctions. In contrast to very isolated oceanic islands, the Western Mediterranean
islands maintained relatively strong biotic relationships with the neighbouring mainland, especially for organisms
with high dispersal ability, such as butterflies 24,28 . Accordingly, our analyses of faunistic diversity and popula-
tion genetic differentiation supported that contemporary island-source dynamics, filtered by the main ecolog-
ical characteristics of islands (predominantly area and isolation), are the main determinants shaping butterfly
communities in the circum-Sicilian islands. Accordingly, a direct comparison between dispersal tendency and
genetic variation resulted in a significant negative correlation (Fig. 2c), and about half of the species in these
communities are taxa with high dispersal ability (P. rapae, P. brassicae, C. croceus, Leptotes pirithous, Lampides
29
boeticus, Vanessa atalanta and V. cardui) and/or with low genetic diversification (Fig. 5). Island populations of
30
these species likely include both local residents and recent immigrants . Besides these widespread species, island
communities include rare species (i.e. species that occur on only few islands), most of which showed considerable
population differentiation over the study area. Out of 32 species, 10 occur only on one or two islands and 18 on
less than half of them. Although their overall number is similar to that of the widespread species, they represent a
relatively small part of each island’s diversity (on average 27.8%). These rare species did not show a nested pattern
and generated inter-island faunistic dissimilarity mostly because of turnover.
The non-nested pattern of rare species revealed that basic rules of ecological filtering and colonization-
extinction dynamics were not sufficient to explain occupancy patterns and that a larger array of determinants is
involved, including: i) dispersal from local source populations (e.g. P. baton to Aeolian Islands, widespread north
African species to Lampedusa), ii) interaction with sister species and lineages (as revealed by mutual exclusion,
see below), iii) human impact and iv) Pleistocene connections. We confirmed the existence of marked differences
in terms of butterfly diversity between north Africa and southern Europe 24,25,31,32 , and individual islands showed
different degrees of similarity to the two sources, evidently based on their relative distance and paleogeographic
connections. The richness analysis attributed to Malta, Levanzo and Lampedusa a set of species that supposedly
colonized them during the LGM, likely through the land-bridges that connected these islands to mainland at that
moment. This hypothesis is further supported by the finding that all the rare species with regional genetic diver-
sification occurring on these islands are genetically very similar to the populations inhabiting the areas to which
these islands were connected during the LGM.
Another important finding is the pervasive occurrence of chequered distributions 31,32 . This is evident not only
for certain pairs of cryptic species (Pontia edusa-P. daplidice and P. icarus-P. celina), but also at the intraspecific
level. Many species displaying over 1% intraspecific genetic divergence (C. alceae, P. machaon, L. phlaeas, Aricia
spp., L. megera, P. aegeria, C. pamphilus) show chequered distributions not only across the wide channel separat-
ing Sicily from north Africa, but also across the narrow (3 km) Messina strait between Sicily and southern Italy.
Chequered distributions challenge the hypothesis of neutrality for species and lineages, but the relative impor-
tance of the potential mechanisms involved is still unknown. A combination of factors, such as reproductive
interference, reduced dispersal, density-dependent phenomena and differences in climatic niches is probably at
the basis of the observed patterns .
32
Island communities and the need for specifically tailored conservation measures. All the islands
investigated host a significantly high fraction of dispersive species having widespread distributions in the study
area and displaying low levels of genetic differentiation. Their long-term presence on islands is expected because
any local extinction event would soon be followed by recolonization involving a population genetically similar
to the original one. This is obvious for those species that reach Europe every spring and are unable to overwinter
30
there [for V. cardui], but also applies to many others with permanent populations on islands and with high dis-
persal capacities (e.g. P. rapae, P. brassicae, C. croceus, L. pirithous, L. boeticus, V. atalanta). Any actions aiming to
preserve the local island populations of these widespread species should be cautiously considered since it could
represent a waste of economic resources in detriment of other, more important, conservation endeavours. On
the contrary, rare species should be carefully monitored and conservation efforts immediately directed towards
declining populations . Their non-nested distribution patterns, with a prevalence of unpredictable species
33
replacements between islands, and their regionally structured genetic diversity, are strong indications that these
island populations have unique histories 1,6,12 . Moreover, the rare species generated the faunistic identity and the
particular genetic structure of each island community, thus providing an unbalanced contribution to diversity .
31
Hipparchia leighebi is a taxon endemic to the Aeolian islands with still debated status. Our results show that
it represents at least a slightly diverged (0.6%) COI lineage endemic to this group of islands, where no other
lineage of Hipparchia was detected. A COI haplotype of M. jurtina was the only one detected on Vulcano and
the most common on Lipari, where it coexists with typical Sicilian haplotypes, and was not found elsewhere .
34
Similarly, Lampedusa hosts a substantially diverged lineage (0.8%) of the north African clade of P. machaon
(Fig. 3, Supplementary Note). If population declines are detected for these entities, they should be considered as
priority taxa for conservation in the islands where they occur. Many single individuals with unique haplotypes on
islands have also been found, but their uniqueness probably represents the effect of random sampling in species
with high genetic variation. This was the case for a single specimen of L. boeticus collected on Gozo island, having
a highly diverged COI sequence close to a genetic lineage previously detected in Madagascar (Supplementary
Note).
Priority for conservation should also be given to populations that apparently represent glacial relicts and that,
in the current ecological settings, would unlikely recolonize a specific island. In fact, these species are the most
threatened on the studied islands (Supplementary Note), possibly because they lack metapopulation dynamics
that maintain population numbers (rescue effect) and are subjected to inbreeding depression. The most striking
example is that of the Maltese islands, where L. phlaeas and A. agestis are believed to be extinct, while M. jurtina,
C. pamphilus and P. aegeria are rapidly declining due to human impact , [Paul Sammut personal observations],
35

Scientific RepoRts | 6:28828 | DOI: 10.1038/srep28828 7