Morphometric versus genetic variation in the Eurasian Reed Warbler Acrocephalus scirpaceus in Italy

ABSTRACT Capsule The Eurasian Reed Warbler Acrocephalus scirpaceus from breeding sites in Italy has little genetic variability in the mitochondrial cytochrome b gene but shows morphological variation suggesting isolation of some populations. Aims To investigate the morphometric and genetic trait variation of Eurasian Reed Warbler populations at a large temporal and spatial scale. Methods We analysed morphometric traits and the mitochondrial cytochrome b (cyt b) gene of samples collected over 25 years, from populations at seven sites across the Italian peninsula and islands. Results While we found no genetic differences, we detected significantly smaller body sizes in individuals breeding in Sicily and Sardinia compared to those from mainland Italy. There were also significant body size differences between two Sicilian populations: one larger breeding on the southern coast and the other smaller at an inner highland site. Conclusion The morphometric data, together with the site fidelity of individuals, are consistent with there being isolation of some populations. Specifically, our results are consistent with an example of polyphenism, morphological plasticity, in the populations of the main Italian islands, Sicily and Sardinia.

The Eurasian Reed Warbler is a philopatric (Pollo et al. 2018) and monogamous (Leisler & Catchpole 1992) species, two traits which probably drive population isolation and, in turn, accumulation of differences between populations over time.
Consequently, many subspecies have been described across the Palearctic region and Africa, although, to date, there is no complete agreement between studies investigating the variation.The European populations are considered stable (BirdLife International 2017) under the current definitions of the subspecies but the phylogeny of these is still subject to debate.According to del Hoyo et al. (2006) and Kennerley & Pearson (2010), populations living in sub-Saharan Africa should be separated from the Eurasian Reed Warbler and be identified as the African Reed Warbler Acrocephalus baeticatus (online Table S1).This is in contrast to Olsson et al. (2016) who considers all included taxa as belonging to the Eurasian Reed Warbler.Recently, Olsson et al. (2016) confirmed that the Iberian and Moroccan populations form a taxonomically distinct clade, A. scirpaceus ambiguus, following reports from Thévenot et al. (2003) and Amezian et al. (2010) who described some wintering individuals in Morocco as belonging to this clade.Whether all populations from North Africa, thus from Morocco to Tunisia and north Libya, belong to A. s. ambiguus is yet unknown (Figure 1).Hering et al. (2016) described Libyan specimens, from the Siwa Oasis complex, as another subspecies of Eurasian Reed Warbler A. scirpaceus ammon, which is sedentary in the Egypt/Libya border region and characterized by small size.Finally, populations in northern Libya and Tunisia have not been clearly ascribed as belonging to any of the known subspecies.
There are significant morphometric differences between the Continental European-Mediterranean subspecies of the Eurasian Reed Warbler between geographic sites, with A. s. scirpaceus from northern and eastern European populations being characterized by longer wings (Wilson et al. 2001, Procházka et al. 2011) and A. s. ambiguus and ammon characterized by smaller biometrics (online Tables S1 and S2).Using ten polymorphic microsatellite loci and extensive sampling, Procházka et al. (2011) found three genetically different populations in Europe, with one of them in the Iberian Peninsula.Overall, they found small but significant genetic differentiation among populations across Europe.Furthermore, they interpreted their data as showing high levels of gene flow, suggestive of a possibly recent divergence of European populations after a rapid range expansion.
A recent investigation on Eurasian Reed Warblers in Italy concluded that Italian samples showed a high genetic similarity with the nominate subspecies of European Reed Warbler A. s. scirpaceus (Ilahiane et al. 2020).We question, however, whether populations breeding on the main Mediterranean islands (Sicily and Sardinia), might show morphometric and/or genetic differences to those breeding on the Italian mainland peninsula.The Mediterranean flyway is used by a minority of migratory breeders who cross the Mediterranean by stopping over on numerous small islands (e.g.Pantelleria, Ustica; Massa et al. 2015) and reaching Sicily from Tunisia and Libya, an area where South-Saharan birds stop to breed.We are however unaware whether, first, this North African population has different morphometric and genetic features in comparison with the other known subspecies and, second, some individuals cross the Mediterranean to breed on the main Mediterranean islands.
Here, we analysed data collected over 25 years from breeding and migratory individuals across seven breeding and stop-over sites.Our main aim was to highlight potential differences among populations by investigating geographical variation in morphometric and genetic traits of reed warblers.

Study areas
We sampled at seven sites distributed across Italy and its main and minor islands as follows: Mirandola, Mortizzuolo, Modena, Italy (44°51 ′ 40 ′′ N 11°06 ′ 25 ′′ E).Included in the Nature 2000 Network (SPA ITA 4040014), it is a protected area with marshland formed by several fresh-water ponds restored with European Union (EU) grants in the 1990s.The vegetation along the bank and open waters is mainly dominated by Phragmites australis, where many pairs of Eurasian Reed Warblers breed.Ringing activity has been conducted on the site since 1995 (Giannella & Gemmato 2003).
Molentargius, Saline Natural Park, Sardinia, Cagliari (39°12 ′ 28 ′′ N 9°08 ′ 59 ′′ E).The area is a Special Protection Area (SPA; ITB044022), a Special Area of Conservation (SAC; ITB040022) and a Regional Natural Park.Its habitats include freshwater ponds, saltwater and a sandy agricultural plain.The presence of areas with different salinity favours a rich variety of plant and animal species.There are over 200 ha of Phragmites australis reedbed, which form an important site for stopover, breeding and wintering of passerines, including a large population of Eurasian Reed Warblers.Ringing activity has been carried out discontinuously since 1992.
Persano Oasis, Campagna and Serre, Salerno, Campania, Italy (40°36 ′ 12 ′′ N 15°08 ′ 46 ′′ E).The Oasis of Persano lies in the high part of the Plain of Sele River.It is a wetland of international importance, and is included in the Ramsar Convention, within Zone Spéciale de Conservation (ZSC) and SPA IT8050021 'Medio corso del fiume Sele-Persano' and also part of the Nature Reserve 'Foce Sele-Tanagro'.It extends over 110 ha, of which 70 ha flood.Ringing activity has been conducted since 2000.
Gela Lake, Sicily, Italy (37°01 ′ 17 ′′ N 14°20 ′ 30 ′′ E).Since 2000, the area has been protected as a nature reserve and an SPA.This is a retrodunal humid area, approximately 120 ha wide.Generally, it contains freshwater, but in some summers the water may become brackish.The vegetation along the bank is dominated by Phragmites communis and much less by Typha.Also, Arundo donax and Tamarix sp. are present, the latter with thick formations.The surrounding areas are heavily cultivated, especially with greenhouses (Triolo et al. 2011).Ringing activity allowed sample collection between 2001 and 2014.
Vendicari, Sicily, Italy (37°47 ′ 30 ′′ N 15°05 ′ 29 ′′ E).Since 1984 the area has been a nature reserve (Oasi faunistica di Vendicari) and also an SPA.The area is part of a complex of retrodunal marshes with salt waters, approximately 150 ha wide with most vegetation along the banks dominated by Phragmites communis.Ringing activity has been carried out since 2003.
Pergusa Lake, Enna, Sicily, Italy (37°30 ′ 50 ′′ N 14°1 8 ′ 18 ′′ E).The site in the centre of Sicily was established as Special Nature Reserve in 1995 (SPA).Pergusa is an endorheic lake 667 m above sea level (ASL) with a perimeter of 4.5 km and is surrounded by hills between 750 and 911 m ASL.The vegetation along the bank is dominated by Phragmites australis, where many pairs of Eurasian Reed Warblers breed regularly.Due to a severe drought drastically reducing the lake size, the warbler population risked extinction at the end of the 1900s.At the beginning of the 2000s, the lake filled again resulting in increased reed vegetation and in turn, a renewed population of reed warblers.Ringing activity allowed sample collection between 2003 and 2019.

Ringing and biometrics
We analysed data collected during the ringing activities carried out between 1993 and 2018 at the study sites.Specifically, we determined inter-annual fidelity to the breeding site and mist-net transects when we recaptured individuals (both adults and juveniles) ringed in June-July in Mirandola, Molentargius, Persano, Gela and Pergusa lakes.Birds were captured using four-shelf mist-nets then the following biometrics were recorded: wing chord (mm), length of Primary 8 (P8, primaries numbered in descending order, mm), tarsus length (mm) and body mass (g).All lengths were measured to the nearest 0.5 mm, except for tarsus which was measured to 0.01 mm, following Spencer (1972).The wing and tarsus measurements are taken during bird ringing and are considered to have high repeatability both within and between measurers (Alatalo & Lundberg 1986) which, in our case, also means within and between ringing sites.Despite there being no global consensus on the best practice to measure the tarsus (Caravaggi et al. 2021), in Italy all the ringers adopt the method standardized for contributing to the EURING scheme.Each individual bird was aged using the EURING age codes, namely '3' for birds in their first calendar year, as indicated by a partial moult limited to a small part of the body plumage at most, or age '4' for birds in at least their second calendar year, as indicated by postbreeding moult which was either partial, to be completed in the African wintering grounds, or, in very rare cases, complete (R. Ientile pers.obs.).Males and females were sexed only during the breeding season, when a cloacal protuberance or brood patch, respectively, are present.To determine biometrics of breeding individuals, our sample size included sexed adults captured only in May at the Persano Oasis and in June-July in all other sites.We excluded individuals captured in other months to avoid analysing migratory individuals, which could already be present in August.At Pergusa and Gela lakes, measurements were taken across months from breeding and nonbreeding periods.On Ustica Island, measurements were from migratory unsexed and unaged individuals during the migration period (April-May).These data were compared to the unified dataset from the other six populations including both males and females.

Molecular analysis
DNA was extracted from feathers, using the procedure described in Tagliavia et al. (2016).We proceeded by comparing three samples from Pergusa (Sicily), and three from Molentargius (Sardinia) with corresponding sequences from selected Mediterranean areas found in GenBank.
We amplified the mitochondrial cytochrome b (cyt b) gene, as indicated by Ilahiane et al. (2020) and following the protocol described in Olsson et al. (2016) with some modifications.The locus was chosen based on the existing literature on the studied species, whose sequences are available in public databases (i.e.GenBank).This allows for robust comparison of the obtained sequence data with those from previous studies, which include samples from many different geographical areas.
Polymerase chain reactions (PCR) were performed in a final volume of 50 µl using Phusion High-Fidelity DNA Polymerase (Thermo Scientific) in HF buffer, following the manufacturer's directions for reaction and cycling condition settings.The amplicons were verified by agarose gel electrophoresis on gel agarose using GelRed (Biotium) as DNA dye.PCR products were precipitated with 2.5M ammonium acetate plus 2.5 volumes of ethanol and then suspended into molecular biology-grade double distilled water and sequenced by Macrogen Europe (the Netherlands).High-quality sequences were manually checked and edited, then trimmed and aligned using either Clustal Omega or UGENE.Bioinformatic analyses were restricted to the sequence region corresponding to that of sequences available in GenBank.In particular, after multiple sequence alignments, sequences were clustered by the neighbour-joining (NJ) method using the software default settings, and a distance matrix was generated.

Statistical analysis
Morphometric data were normalized by means of a Box-Cox transformation prior to each analysis.Morphometric differences among populations were analysed using a one-way analysis of variance (ANOVA) where each of the biometric measures was treated as a dependent variable and study site as a factor with seven levels.To assess differences in the two Sicilian sites of Gela and Pergusa in the different months, a general linear model (GLM) was applied, in which the wing length was treated as a dependent variable and site and month as fixed factors.All significant differences were further analysed with Tukey post hoc tests.We used Minitab software for all statistical analyses.We conducted a temporal analysis (from June to September, ANOVA with repeated measures followed by F statistics for significance assessment) of morphometric traits recorded in Pergusa and Gela to detect any possible difference between breeding and migratory individuals.

Biometrics
Body mass was highly variable throughout the seasons, so was not included in the analysis.Mean wing length and P8 length did not differ significantly between breeding males and females in the Italian peninsula (Table S3, Figures 2 and 3), but they were significantly different from Sicilian and Sardinian ones.Wing length of females from Molentargius (Sardinia) and Pergusa (Sicily) was significantly smaller than females from other Sicilian sites (Figure 3), although differences in the P8 length were not significant (Figure 4).There were no significant geographical differences in the tarsus length (Figure 4).Wing length measurements of unsexed individuals captured on Ustica Island were larger than male and female individuals captured at Sicilian and Sardinian breeding sites but did not differ significantly from those of peninsular Italy (Persano; one-way ANOVA, F 5, 2289 = 93.55,P < 0.01).
At Gela Lake (Sicily), mean values of P8 length were fairly consistent during July but gradually increased in August and September, with the exception of a very low value, even lower than in July, recorded in the second period of August.Variation in P8 length for 10-day periods was significant for pairwise comparisons (one-way ANOVA: F 6, 1095 = 11.91,P < 0.01; Tukey's HSD test, the majority P < 0.05).The temporal analysis of wing length at Pergusa and Gela (Sicily) over the breeding and migration months showed significant differences between the two sites (F 1,810 = 67.44,P < 0.01) and between months (F 6,810 = 4.87, P < 0.01).The site of Gela in August-September holds many migrating individuals coming from central Europe, as is evident from the significant increase in the mean wing length over the two previous months.At Pergusa Lake (only 52 km away from Gela), however, the mean wing length did not change significantly with month from June through to September, and showed only a small increase when migrant birds stopped at the site (September), suggesting that the site is not important for stopover on migration (Figure 5).

Genetic analyses
Sequence analyses showed only negligible differences among Sicilian and Sardinian samples, as well as between them and others coming from the distribution area of the nominate subspecies of the Eurasian Reed Warbler A. s. scirpaceus.Moreover, no new or unique haplotypes were detected (Figure 6, Table S4).Such a scenario made the genetic data set unsuitable for further phylogenetic analyses.

Discussion
Our genetic data confirmed the findings of Ilahiane et al. (2020), that there are no genetic differences among our study populations of Eurasian Reed Warblers in Italy, and they are similar to other European populations.Note that our genetic results refer to the analysis of a single mitochondrial locus.While this choice was to enable comparison with other studies (Ilahiane et al. 2021, Ollson et al. 2005), the possibility does remain that analyzing further loci, including nuclear loci, might result in a different outcome.However, to be properly informative, such an analysis would require the availability of sequences of other loci in public databases but these are currently unavailable.In fact, analysing cyt-b was the only way for a full and robust comparison of the sequence data obtained in our study with those from previous ones, which included samples from many different geographical areas.
We did, however, find significant differences in morphometric traits between Italian breeding populations of Eurasian Reed Warblers.Sicilian and Sardinian breeding populations were smaller than those breeding on the Italian mainland.Migratory individuals stopping over the small island of Ustica in spring were larger than the other Sicilian breeders but more similar than breeders on the Italian mainland.
Interestingly, Sicilian and Sardinian breeding populations showed biometrics similar to those of the subspecies A. s. ambiguus, which breeds in the Iberian Peninsula and Morocco.Nevertheless, the subspecies ambiguus has a more rounded wing than A. s. scirpaceus, completes the moult before migrating to sub-Saharan wintering areas, and in some cases may overwinter in Morocco, while Sicilian and Sardinian populations moult after their autumnal migration to the sub-Saharan wintering areas.Only exceptionally, in Sardinia and Sicily, a complete postbreeding moult in Eurasian Reed Warblers has been recorded (Nissardi & Zucca 2001, R. Ientile pers. obs.)During our long-term study, we did not find any individual with plumage traits consistent with the subspecies A. s. fuscus: slightly greyer, and more olive- brown than nominate, less tinged rufous above and slighter whiter, less creamy-buff, on the sides of the breast and on the flanks; in worn plumage, paler than nominate.We may conclude, therefore, that Sicilian and Sardinian breeding populations belong to the subspecies scirpaceus; however, this fuscus has been found in central Europe during migration (Arbabi et al. 2014).These results are consistent with breeding Eurasian Reed Warblers in Sicily and Sardinia displaying polyphenism, a morphological plasticity that can be attributed to external influences and therefore, which takes place in the absence of genetic differences (Fusco & Minelli 2010, Pigliucci 2010).Very likely it is dependent on adult philopatry (4.7%-19.8%),which is a prerequisite of an isolation process.
We do not know if different populations of Eurasian Reed Warblers have different wintering areas (we obtained only one winter recovery in the Central African Republic of one breeding individual from Pergusa), but morphometric differences reported in this paper demonstrate that there is an isolation in progress, likely due to absence or scarcity of gene flow between different populations.Statistical differences detected between populations breeding at Gela and Pergusa, only 52 km away from each other, show that inside the same regional area, there is little or no gene flow between populations.Additionally, Kralj et al. (2010) found differences in the wing length and indications of limited gene flow between coastal and inland breeding populations of Eurasian Reed Warblers in Croatia.In particular, Pergusa is located at 667 m ASL and is scarcely used by autumn migrants, while Gela is on the coast and gets an annual influx of migrants in the late summer and autumn.Furthermore, in the 1990s the lake of Pergusa completely dried up, and the population of Eurasian Reed Warblers was close to local extinction.Most likely a bottleneck has occurred and when, in the first years of 2000s, the water filled the lake again, the reed vegetation could support a new population of birds starting from the few surviving individuals, which likely were smaller in size than the average breeders in Sicily.This is only a hypothesis, probably reinforced by the site fidelity of the Pergusa population (5.4% of ringed adults recovered in the following years).
Previously, Cramp (1992)   does not have high levels of gene flow, but they express a phenotypic plasticity, in the sense of a 'traitand environment-specific property of the genotype which may or may not be advantageous and may or may not be the result of adaptive evolution' (Pigliucci 2010).
Sedentary passerine species are known to vary clinically in size from north to south, but also most trans-Saharan species have differentiated populations, recognizable as geographic subspecies.Following Vaurie (1959), at least 33/44 (75%) passerine species migrating from tropical Africa to Europe were subdivided into subspecies; this means that they were recognized as populations living in different parts of the geographic breeding range that generally do not meet.However, there are many cases of overlapping distributions and the consequent gene flow between different populations, such as the case of the wagtails Motacilla (Harris et al. 2018).We do not know any case of trans-Saharan passerines where their clinal variation is described.However, in most cases, subspecies are only the extremes of a gradual latitudinal variation, with larger size specimens in the north and smaller specimens in the south, such as in the case of the Northern Wheatear Oenanthe oenanthe (Svensson 1992).
Interestingly, however, Eurasian Reed Warblers in Italy, even if variable, do not vary clinically, but populations elsewhere in Europe are biometrically similar.Overall, small-sized Eurasian Reed Warblers living in Sicily and Sardinia could be the result of a long isolation, due to scarce gene flow and site fidelity, indicating that most likely they are not the extremes of a clinal variation, but the product of isolated populations.S4.
Finally, our results show that even if genetic studies are useful to understand differences between populations of several bird species, the study of phenotypic traits is always indispensable.Biodiversity is not only related to the number of species but also their variability in space and the evolutionary processes that contribute to that variability.In other words, investigations on phenotypic traits still remain a crucial tool for biodiversity conservation although further investigations should aim to verify whether the use of a single gene provides a robust protocol for this type of genetic data investigation.

Figure 1 .
Figure 1.Distribution of main populations of Eurasian Reed Warbler subspecies (light grey = breeding areas, dark grey = wintering area).

Figure 2 .
Figure2.Wing length ± SD (mm) of male and female Eurasian Reed Warblers and significant differences among populations from the Italian peninsula, Sardinia and Sicily.Letters on columns indicate the statistical differences among the different populations (one-way ANOVA followed by Tukey post hoc test; males: F 5,693 = 24.48,P < 0.01; females: F 5,793 = 29.18,P < 0.01).

Figure 3 .
Figure3.Length of 8th primary ± SD (mm) of male and female Eurasian Reed Warblers and significant differences among populations from the Italian peninsula, Sardinia and Sicily.Letters on columns indicate the statistical differences among the different populations (one-way ANOVA followed by Tukey post hoc test; males: F 5,749 = 37.45, P < 0.01; females: F 5,844 = 21.86,P < 0.01).
and Kennerley & Pearson (2010) highlighted the geographical variation in the measurements of Eurasian Reed Warblers in the Palaearctic region.Present data show that, in contrast to the proposal by Procházka et al. (2011), the species

Figure 4 .
Figure 4. Tarsus length ± SD (mm) of male and female Eurasian Reed Warblers and significant differences among populations from Italian peninsula, Sardinia and Sicily.Letters on columns indicate the statistical differences among the different populations (one-way ANOVA followed by Tukey post hoc test; males: F 5,546 = 4.98, P < 0.01; females: F 5,577 = 8.85, P < 0.01).

Figure 5 .
Figure 5. Mean wing length ± SD (mm) of Eurasian Reed Warblers from June to September in two Sicilian breeding sites, Gela and Pergusa lakes.Letters on columns indicate the statistical differences among the different populations and months (one-way ANOVA followed by Tukey post hoc test; males: F 6,810 = 4.87, P < 0.01).

Figure 6 .
Figure 6.Neighbour-joining tree with bootstrap values created from the sequences obtained by mitochondrial cytochrome b (cyt b) gene in Sicily and Sardinia compared with corresponding sequences from selected Mediterranean areas found in GenBank.Details on Percent Identity Matrix and GeneBank Access Numbers are provided in online TableS4.