Bill morphology and biometrics of three sibling woodpecker species from sympatric populations

ABSTRACT

Capsule Different methods of comparing bill morphology between woodpecker species show different, but not contradictory results.

Aims Differences among similar, closely related species which co-occur are still highly debated. In such a scenario, species should diverge morphologically to reduce competition. We studied this phenomenon, in three closely related woodpecker species that co-occur in eastern Slovakia: Great Spotted Woodpecker Dendrocopos major, Syrian Woodpecker Dendrocopos syriacus and White-backed Woodpecker Dendrocopos leucotos. The Great Spotted Woodpecker has the widest niche and lives sympatrically with both Syrian and White-backed Woodpeckers, while the distributions of the latter two species do not overlap because of their use of different habitats. We predicted that Great Spotted Woodpeckers should differ morphologically from both sympatric species: Syrian and White-backed Woodpeckers.

Methods Comparisons were made between species, based on dorsal and lateral views of maxilla using both geometric morphometric techniques and traditional measurements.

Results We confirmed our hypothesis, and found significant differences in bill shape between Great Spotted Woodpecker and both congeners. This is in contrast to the observed higher similarity of Great Spotted Woodpecker and Syrian Woodpecker in traditional body measurements. However, surprisingly, Syrian Woodpecker and White-backed Woodpecker do not differ significantly in bill shape.

Conclusions This work indicates that geometric morphometry appears to be a promising tool for the investigation of how interspecific competition influences the shape of the bill between co-occurring species.

The study of community assembly emphasizes competition-mediated ecological interactions (Meiri et al. 2014). Competition for resources is regarded as one of the major causes of phenotypic differentiation in closely related species (Schluter 2000). Such divergence in both phenotype and available resources in sympatric congeneric species is caused and/or maintained by ecological character displacement. This phenomenon has recently been considered to be one of the key drivers generating biodiversity (Pfennig & Pfennig 2012, Stuart & Losos 2013). However, the status of ecological character displacement is still the subject of debate, and not all examples of this phenomenon are considered to be fully compelling (Schluter 2000, Dayan & Simberloff 2005, Stuart & Losos 2013, Dufour et al. 2015). This is because, for most cases, interspecific competition has not been indubitably documented and confounding mechanisms have not been unequivocally ruled out. Therefore, most documented cases of ecological character displacement are also consistent with other evolutionary and ecological processes (Stuart & Losos 2013). Ecological competition takes place not only between species, but mainly within them (Doebeli 2011). Sexual size dimorphism can enable males and females of the same species to specialize on different sized food items and therefore minimize intraspecific competition. However, interspecific competition is thought to limit sexual dimorphism, since larger competitors in the community will prevent the larger sex from evolving to a larger size, and smaller species may prevent the smaller sex from becoming even smaller (Meiri et al. 2014). Strong intraspecific competition can select for character divergence between individuals within a population (Bolnick et al. 2003, Scott et al. 2003).

Food is usually considered to be a major factor limiting population sizes (MacArthur & Wilson 1967). When the sexes compete over food, theory suggests that sexual dimorphism can evolve via frequency-dependent natural selection, enabling males and females to specialize on different types of food (Slatkin 1984, Bolnick & Doebeli 2003). Therefore, strong intersexual competition can lead to selection for accentuated sexual differences. Sexual size dimorphism is often assumed to be more likely to evolve where intraspecific competition is fierce, but where interspecific competition, which can prevent the sexes from diverging, is weak (Greenberg & Olsen 2010, Cooper et al. 2011, Luther & Greenberg 2011). In the case of birds, bill characteristics are crucial for effective foraging in a given niche (Grant 1999, Benkman 2003, Grant & Grant 2006, Berns & Adams 2010). Changes in bill traits together with other features, such as different food preferences and foraging behaviour, can give a competitive advantage over species that have overlapping food and habitat preferences. European woodpeckers (Picidae) are a good example of a group of species living in sympatry (Angelstam & Mikusiński 1994, Kosiński & Kempa 2007) that form an ideal system for the study of species and sexual shape dimorphism. Three closely related species of woodpeckers: Great Spotted Woodpecker Dendrocopos major, Syrian Woodpecker Dendrocopos syriacus, and White-backed Woodpecker Dendrocopos leucotos, co-exist in large parts of their ranges in central and southeastern Europe (Winkler & Christie 2002). The Great Spotted Woodpecker is the most numerous woodpecker in the Palearctic and Europe (Michalek & Miettinen 2003), with a range that covers almost the whole of Europe and much of Asia. The Great Spotted Woodpecker is the most opportunistic of the European woodpeckers because it is able to occupy many different habitats, from pure coniferous forests to deciduous forests (Michalek & Miettinen 2003, Mazgajski & Rejt 2006), and has a large tolerance to fragmentation of habitats (Myczko et al. 2014, Michalczuk & Michalczuk 2016a). The Syrian Woodpecker is an expansive species, which began to extend its distribution through the Balkans toward northwest Europe around 1890 (Nowak 1971); it now also inhabits Central Europe and has been a regular breeder there since the 1950s (Ferianc 1950, Ciosek & Tomiałojć 1982, Danko et al. 2002, Michalczuk 2014). The Syrian Woodpecker occurs sympatrically with the Great Spotted Woodpecker in the majority of the former’s range and both species can use many different types of habitat (Michalczuk & Michalczuk 2016a, Kajtoch & Figarski 2017). However, the Syrian Woodpecker is more common in more open and synanthropic areas than the Great Spotted Woodpecker, especially in orchards and vineyards (Danko et al. 2002, Michalczuk & Michalczuk 2006, 2016a, 2016b, Ciach & Fröhlich 2013). Furthermore, the Syrian Woodpecker does not occur in boreal/hemiboreal areas or coniferous forests (Angelstam & Mikusiński 1994). In comparison to the two previous species, the White-backed Woodpecker is declining in northern Europe (Carlson 2000) but generally stable in the remainder of its range (Pasinelli 2006). Most of the range of this species overlaps with that of the Great Spotted Woodpecker, especially in Asia (Winkler & Christie 2002). It is the most specialized of the three studied species and is associated with old-growth deciduous or mixed forests (Wesołowski 1995, Carlson 2000). For this species, one of the most important factors is a large amount of dead wood in the habitat (Czeszczewik & Walankiewicz 2006). The Great Spotted and Syrian Woodpeckers are very closely related (Fuchs & Pons 2015) and hybridize relatively often (Cramp 1985, Michalek & Miettinen 2003, Michalczuk et al. 2014), but Great Spotted and White-backed Woodpecker hybrids are also known (Cramp 1985, Michalek & Miettinen 2003). Surprisingly, Syrian Woodpeckers is the least well known of the European woodpeckers (Pasinelli 2006) with a lack of morphological data from Central Europe (Cramp 1985).

Little is known about the differences in morphometry between these three species in co-occurring populations. Such information may be crucial, not only for understanding the niche separation between these very closely related species, but also to provide more general conclusions for causes of success under mutual competition. Earlier classical studies, for example, on Darwin finches, suggested that external measurements, especially bill size and shape, as well as body size, are important for understanding niche utilization by similar species (Boag & Grant 1981, 1984, Price 1987). Two other traits that should be related to foraging behaviour are wing length and tail length. Changes in size and/or shape of the wing should affect the capability of the animal to fly and manoeuvre (James 1982, Fitzpatrick 1999, Hromada et al. 2003a). Tail feathers are used to store kinetic energy during excavating behaviour (Winkler & Christie 2002).

An excellent, but often overlooked, source of ecological data is museum collections. Saris Museum in Bardejov, Slovakia holds one of the most prominent collections in the Carpathian region, including birds, accompanied with excellent secondary data (Hromada et al. 2003b, Kuczyński et al. 2003, Roselaar 2003, Hromada et al. 2015, Kaczmarski et al. 2017). Great Spotted, Syrian and White-backed Woodpeckers are well represented in this collection.

In this study, we predicted that Great Spotted Woodpeckers should differ morphologically from both sympatric species (Syrian and White-backed Woodpeckers) due to interspecific competition. However, bill measurements traditionally used for comparisons between species, such as bill length, do not sufficiently demonstrate differences in bill structure. It is possible to compare other traits only when the bill possesses specific and characteristic measurable points. But, like most bird beaks, woodpecker bills do not have such features. Therefore, we have here tested the suitability of geometric morphometrics which allows a comparison of bill shapes in the absence of the specific landmarks needed for comparison using traditional methods.

Methods

Research area

Woodpeckers from sympatric populations were collected in eastern Slovakia, mainly in its northern part, in the transition between the Eastern and Western Carpathians. The landscape in the region is hilly and diverse, with many valleys entering the mountains from the south. Elevation varies between 100 and 1100 m above sea level. Managed fir-beech forests with small relics of primeval forests dominate in the mountains, while the valleys are characterized by farmland, with a patchy mosaic of arable fields, pasture, orchards and gardens, interspersed with small woods, and scrub (Hromada et al. 2003b, Hromada et al. 2015). The Great Spotted Woodpecker lives here in sympatry with either the White-backed Woodpecker (in mixed and coniferous forests) or the Syrian Woodpecker (in valleys, lowlands, urbanized areas, and orchards). However, the White-backed and Syrian Woodpecker niches in this area do not overlap and can be described as allotopic (Rivas 1964). The Syrian Woodpecker expanded its range from the southeast into the studied area at the beginning of the 1950s (Danko et al. 2002, Michalczuk 2014).

Woodpecker specimens

The analysed material was based on the extensive collection of Great Spotted, Syrian and White-backed Woodpeckers collected from 1957 to 1974, and now held in the Saris museum in Bardejov, Slovakia (Hromada et al. 2003b, Roselaar 2003, Hromada et al. 2015) and additionally in the East Slovakian museum, Košice, Slovakia. Birds were collected throughout the year except for July, although in August and September only a few specimens were obtained. To assess bill shape morphology we used only sexually mature specimens of the three species. The samples used for geometric morphometric comparisons consisted of 34 specimens of White-backed Woodpeckers (20 females and 14 males), 89 specimens of Great Spotted Woodpeckers (43 females and 46 males), and 33 specimens of Syrian Woodpeckers (13 females and 20 males). For the data on traditional measurements we used only museum specimens from the Saris museum. These measurements had been made on freshly obtained birds before preparation and conservation, using the same methodology, by one person (Tibor Weisz), who was not aware of the future use of the data (Hromada et al. 2003b, Hromada et al. 2015). The six traditional measurements were as follows: body mass to an accuracy of 0.05 g, lengths of wing, body, and tail to an accuracy of 0.5 mm, and bill length (skull to bill tip) and tarsus lengths to an accuracy of 0.05 mm. Measurement methods followed Svensson (1992): wing length was the flattened wing, tail length was taken on the underside using a rule, and tarsus length was taken as the distance between the sole side of the opened foot, abutted on a calliper at a right angle and measured to the proximal point of the tarsometatarsus. Some measurements were missing for some specimens because of damage during collection, and the numbers of specimens actually used are shown in the tables.

Data collection for geometric morphometry

Each bill was photographed in two different projections: dorsal (from above) and lateral (from the side) using a Sony Cyber-shot DSC-WX60 digital camera. The photographs were taken with a 4.5 mm focal length. The bill was photographed against a white background in a laboratory, using graph paper as a scale. The camera was mounted on a tripod and photographs were taken from the same distance and position for each bill. We used tpsUtil software (Rohlf 2012) to build a tps file for each bill from the image, separately for the two projections. Afterwards, the tps file was opened in tpsDig2 software (Rohlf 2010) where every image was scaled to the graph paper. Then, on every image in the same projection, three landmarks were distributed at local minima and maxima of the bill edge (Figure 1). Next, two curves with five equally spaced locations (semi-landmarks) were digitized on bill edges between landmarks (Figure 1). Semi-landmarks are landmarks digitized on the curve or surface with positions along the curvature (Mitteröecker & Gunz 2009). On the dorsal view, a landmark on the end of the bill and two landmarks on the left and right side of the bill near the head were digitized. The first curve was made on the right-hand side of the bill from the head to the end of the bill, and the second curve on the left-hand side. On the lateral view, landmarks were located at the end of the bill and at the most upper and lower location near the head. The first curve was then made on the superior (upper) edge of the maxilla (Figure 1(b)), and the second on the inferior (lower) edge. Before taking the photographs, the feathers were arranged to expose the greatest extent of the bill. This ensured that the first landmarks were digitized near the skull. Both curves were made in the same direction, from the head to the end of the bill. On the lateral view of the bill, landmarks and semi-landmarks were distributed only on the maxilla, because the mandible was partially covered by the upper part of the bill.

Figure 1. Image of a woodpecker bill with semi-landmarks shown. A – Dorsal view and B – Lateral view. Red points are landmarks, white points are semi-landmarks digitized along the curves.

The bills of woodpeckers do not have many characteristic landmarks which could be useful in geometric morphometrics. Thus, in our study we also used semi-landmarks to analyse shape differences within and between species. A landmark superimposition was then undertaken using a Procrustes Generalized Analysis (GPA) with adjustment to each semi-landmark during each iteration of the GPA optimization. Slide of semi-landmarks along curves was carried out to reduce the bending energy between each specimen. This part of the analysis was calculated using TPSRelw (Rohlf 2017). Next, all semi-landmarks were converted to landmarks in order to carry out geometric morphometric analysis in MorphoJ. This procedure allowed a comparison of shapes through translation (averaging the position of landmarks), scaling to one size, and rotation (minimizing the summed squared distances between landmarks; Zelditch et al. 2012). Centroid size was used as a measure of geometric scale and calculated as the square root of the summed squared distances of each landmark from the centre of the landmark configuration (Zelditch et al. 2012). To estimate the technical error measurement, the landmarks and semi-landmarks were distributed by two independent observers (ZO and AMK) twice each on 30 randomly selected specimens.

Geometric morphometric analysis

A Procrustes analysis of variance (anova) test was carried out to analyse the level of the measurement error associated with semi-landmark digitization. The Procrustes anova used in geometric morphometric analysis is a variant of the anova used in traditional statistical analysis. This method compares variation in shape within groups to variation in shape between groups, and is applied to test the hypothesis that samples do not differ in their means (Zeldith et al. 2012). A multivariate regression was subsequently carried out for each view (anterior and lateral) separately to analyse how the size of the bill (centroid size) influenced its shape (Procrustes coordinates). This type of association is called allometry and can be a crucial component of shape variation (Outomuro & Johansson 2017). In the case of a significant multivariate regression, geometric morphometric analysis would be carried out taking the allometry component into account. We used principal components analysis (PCA) to describe major trends in shape variation within the entire sample of birds. To focus on the differences between sexes within the three species a canonical variate analysis (CVA) was then carried out. To test the significance of shape differences we computed tests based on 10,000 randomizations in CVA. The significance level of the CVA was Bonferroni-adjusted (Sokal & Rohlf 1995) to 0.05/15 = 0.0033 owing to the large number (15) of sex/species comparisons on each view. The semi-landmarks were established using tpsRelw while all geometric morphometric analyses were carried out using MorphoJ software (version 1.06d, Klingenberg 2011).

Linear measurements

We compared interspecific and intersexual morphometric differences using anova based on three species groups and two sex groups and their interaction followed by post hoc Tukey multiple comparisons. All six traditional variables passed homogeneity of variance tests but two failed Kolmogorov Smirnoff normality tests. Consequently, as a precaution, all multiple comparisons were derived from bootstrapping based on 1000 samples. Pearson correlations were calculated between morphometric measurements, separately for each species. All statistical tests were two-tailed and the significance threshold was 0.05.

Results

The suitability of geometric morphometry for comparison of bill shapes

In the Procrustes anova, the Procrustes coordinates of each sample were the dependent variable, in turn the observer who digitized landmarks and semi-landmarks on bills was the independent variable. Variation among samples (mean square = 3.8 × 10⁻⁴) was much higher than that between the double distribution of semi-landmarks in both observers (mean square = 8.3 × 10⁻⁷). Moreover, Procrustes anova did not show significant differences (P > 0.05) in semi-landmark distribution between the two observers; hence, high intra- and inter-observer agreement was achieved. The multivariate regression which analyses association between bill size (centroid size as the independent variable) and bill shape (coordinates as the dependent variable) was not significant for either analysed view (dorsal view: per cent predicted 0.853%, P = 0.3075; lateral view: per cent predicted 0.186%, P = 0.6984) thus, subsequent geometric morphometric analysis was carried out without taking into account the allometric component.

Bill shape differences

The PCA for bill shape showed that the female and male Great Spotted Woodpeckers were most variable in the two projections. A total of 88.06% of the variability in the dorsal view and 85.69% in the lateral view could be explained by the first component (PC1). PC1 in the dorsal view represented the width of the proximal part of the bill; in the lateral view it represented the extent of curve (Figures 2 and 3). PC2 explained 9.43% of the variability for the dorsal view and 10.85% for the lateral view. PC2 represented the width and height in the central part of the bill (Figures 2 and 3).

Figure 2. Results of PCA of bill shapes for the dorsal view. Grey lines represent bill shape for the principal component.

Figure 3. Results of PCA of bill shapes for the lateral view. Grey lines represent bill shape for the principal component.

The CVA showed that White-backed and Syrian Woodpecker bill shapes differed significantly from bills of Great Spotted Woodpeckers in both views (lateral and dorsal), the results are presented in Figures 4 and 5, and Tables 1 and 2. In general, White-backed and Syrian Woodpeckers were not distinct in bill shape in the two views. Only male White-backed Woodpeckers were characterized by significantly different bill shape from female Syrian Woodpeckers in the anterior view. CV1 explained 81.95% of the variation in the dorsal view and 79.67% in the lateral view. In both views, CV1 described bill width over its entire length and the extent of curve of the bill. In the analysed material, CV2 explained 9.51% of the variation in the dorsal view and 10.59% in the lateral view. CV2 represented the width and height in the central part of the bill. Both sexes of White-backed and Syrian Woodpecker were characterized by negative values of CV1 indicating a relatively narrower and less curved bill than female and male Great Spotted Woodpeckers which were characterized by positive values of CV1. In turn, positive values of CV2 characterized a relatively wider bill near the head. Significant differences between sexes were noted only for White-backed Woodpeckers in the anterior view (Table 1). Male White-backed Woodpeckers generally had a relatively narrower bill near the head.

Figure 4. Results of the CVA of bill shapes for the dorsal view. Grey lines represent bill shape for the canonical variate.

Figure 5. Results of the CVA of bill shapes for the lateral view. Grey lines represent bill shape for the canonical variate.

Table 1. P-values from permutation tests in CVA. Analysis carried out on the dorsal view of the bill.

	GSW F	GSW M	WBW F	WBW M	SW F
GSW M	0.0308
WBW F	0.0001*	0.0001*
WBW M	0.0001*	0.0001*	0.0023*
SW F	0.0001*	0.0001*	0.1589	0.0002*
SW M	0.0001*	0.0001*	0.0276	0.0191	0.0556

Notes: GSW: Great Spotted Woodpecker; WBW: White-backed Woodpecker; SW: Syrian Woodpecker; F: female; M: male.

*Significant differences at P < 0.05 after Bonferroni correction (equivalent to P < 0.003).

Table 2. P-values from permutation tests in CVA. Analysis carried out on the lateral view of the bill.

	GSW F	GSW M	WBW F	WBW M	SW F
GSW M	0.6108
WBW F	0.0001*	0.0001*
WBW M	0.0001*	0.0001*	0.4408
SW F	0.0012*	0.0001*	0.8158	0.2879
SW M	0.0001*	0.0001*	0.6729	0.5621	0.5719

Notes: GSW: Great Spotted Woodpecker; WBW: White-backed Woodpecker; SW: Syrian Woodpecker; F: female; M: male.

*Significant differences at P < 0.05 after Bonferroni correction (equivalent to P < 0.003).

Body trait differences

A summary of traditional morphometric measurements analysed by species, sex and species*sex interaction revealed significant differences between groups in all body traits (Table 3(a–f)). In each species, mean bill length was significantly longer in males than females (Table 3(a)). The longest bills in both males and females were recorded for White-backed Woodpeckers and were significantly longer than all other analysed groups. The shortest mean bill length was recorded for female Great Spotted Woodpeckers but was not significantly different from female Syrian Woodpeckers. Bills of male Syrian Woodpeckers were significantly longer than male Great Spotted Woodpeckers. Sex differences in body mass were significant only for the White-backed Woodpecker, where males were heavier than females (Table 3(b)). Both sexes of the White-backed Woodpecker were heavier than birds from all other groups. Male Great Spotted Woodpeckers were significantly heavier than male Syrian Woodpeckers. Statistically significant sex differences in wing length only occurred in the White-backed Woodpecker (Table 3(c)). Both sexes of the White-backed Woodpecker had significantly longer wings than all other groups of birds. Both sexes of the Syrian Woodpecker had significantly shorter wings than all other groups. We did not find any statistically significant sex differences in mean tail length of the analysed groups (Table 3(d)). Only the Syrian Woodpecker had significantly shorter tails than all other analysed groups. Tarsus length was significantly greater only in male White-backed Woodpeckers compared to other groups (Table 3(e)). We did not find any statistically significant sex differences between mean body length in analysed groups except for Great Spotted Woodpeckers (Table 3(f)). Only the Syrian Woodpecker had a significantly shorter mean body length than all other groups and male White-backed Woodpeckers had significantly longer mean body length than both sexes of Great Spotted Woodpecker.

Table 3. Summary of analyses of traditional body trait differences in White-backed, Syrian and Great Spotted Woodpeckers. Means that do not share the same superscript letter are significantly different (P < 0.05).

(a) Bill length (mm)	mean ± se (n)
	Female	Male
White-backed	34.1^b ± 0.3 (15)	36.7^a ± 0.3 (14)
Syrian	27.9^de ± 0.6 (8)	29.9^c ± 0.2 (12)
Great Spotted	26.5^e ± 0.2 (38)	27.5^d ± 0.2 (40)
Significance of effects: species: F2,121 = 434.91, P < 0.001, sex: F1,121 = 48.26 P < 0.001, species × sex interaction: F2,121 = 4.79 P = 0.010.
(b) Body mass (g)	Mean ± se (n)
	Female	Male
White-backed	94.3^b ± 1.0 (15)	101.0^a ± 1.0 (14)
Syrian	77.0^cd ± 2.3 (8)	78.2^d ± 1.3 (12)
Great Spotted	80.8^cd ± 1.2 (37)	81.7^c ± 0.8 (40)
Significance of effects: species: F2,120 = 106.69, P < 0.001, sex: F1,120 = 6.28, P = 0.014, species × sex interaction: F2,120 = 2.92 P = 0.058.
(c) Wing length (mm)	Mean ± se (n)
	Female	Male
White-backed	141.8^b ± 0.7 (15)	144.3^a ± 0.4 (14)
Syrian	132.0^d ± 1.1 (8)	132.5^d ± 1.0 (12)
Great Spotted	139.2^c ± 0.5 (38)	137.8^c ± 0.6 (40)
Significance of effects: species: F2,121 = 66.82, P < 0.001, sex: F1,121 = 0.56, P = 0.456, species × sex interaction: F1,121 = 4.28, P = 0.016.
(d) Tail length (mm)	Mean ± se (n)
	Female	Male
White-backed	95.6^a ± 0.9 (15)	96.6^a ± 0.9 (14)
Syrian	89.5^b ± 1.0 (8)	88.4^b ± 0.7 (12)
Great Spotted	97.4^a ± 0.5 (38)	96.3^a ± 1.0 (40)
Significance of effects: species: F2,121 = 25.32, P < 0.001, sex: F1,121 = 0.18, P = 0.670, species × sex interaction: F2,121 = 0.60, P = 0.551.
(e) Tarsus length (mm)	Mean ± se (n)
	Female	Male
White-backed	23.9^b ± 0.3 (15)	25.2^a ± 0.4 (14)
Syrian	23.0^bc ± 0.8 (8)	22.8^c ± 0.3 (12)
Great Spotted	23.6^b ± 0.2 (38)	23.9^b ± 0.2 (40)
Significance of effects: species: F2,121 = 9.59, P < 0.001, sex: F1,121 = 2.90, P = 0.091, species × sex interaction: F2,121 = 2.54, P = 0.083.
(f) Body length (mm)	Mean ± se (n)
	Female	Male
White-backed	228.1^ab ± 1.3 (15)	231.8^a ± 1.6 (14)
Syrian	209.9^d ± 2.7 (8)	212.9^d ± 1.5 (12)
Great Spotted	225.5^b ± 0.9 (38)	221.9^c ± 1.0 (40)
Significance of effects: species: F2,121 = 54.20, P < 0.001, sex: F1,121 = 0.60, P = 0.439, species × sex interaction: F2,121 = 4.94, P = 0.009.

Discussion

The traditional body trait measurements showed a higher similarity between the Great Spotted and Syrian Woodpeckers than either did with White-backed Woodpecker. However, as predicted, the bill shape analysis indicated a specific shape of the Great Spotted Woodpecker bill which differed significantly not only from that of the White-backed Woodpecker but also from that of the closer related Syrian Woodpecker. Surprisingly, despite the significant difference in the bill length of Syrian and White-backed Woodpeckers, the variation of bill shapes in these two species was very similar. The lack of differences between the bill shapes of Syrian and White-backed Woodpeckers (if we compare the same sexes) may be a result of the lack of competition between the two, because of the spatial isolation in the environment of these two species which inhabit completely different habitats (Wesołowski 1995, Danko et al. 2002, Czeszczewik & Walankiewicz 2006, Michalczuk & Michalczuk 2016b). Moreover, it is surprising in these different habitats that the optimum shapes of the bills are similar (Figures 2–5).

Because of a lack of direct data about the existing competition of the studied species in Slovakia in terms of dietary analysis, we cannot confirm unequivocally that differences in bill morphology in co-occurring populations are a direct effect of character displacement. However, our findings can at least suggest the existence of niche differentiation in the three studied woodpecker species. For investigation of co-evolutionary morphological response to interspecific competition, Dayan & Simberloff (1998) suggested a careful consideration both of guild composition and proper morphological traits. Yet, both White-backed and Syrian Woodpeckers co-occur widely with Great Spotted Woodpecker, which is a generalist and uses habitats preferred by both other species (Cramp 1985, Wesołowski 1995, Michalczuk & Michalczuk 2016a, 2016b). The foraging behaviour of Great Spotted and Syrian Woodpeckers is similar. The most important differences are a higher frequency of gleaning and probing, and less excavating, in Syrian Woodpecker than in the Great Spotted Woodpecker (Winkler 1973). All three examined woodpecker species mainly forage on arthropods, especially insect larvae, and all species include plant material in their diet (Cramp 1985, Michalek & Miettinen 2003). Great Spotted Woodpeckers show a greater tendency than Syrian Woodpeckers to seek deep wood-boring insects, and White-backed Woodpeckers – in comparison to Great Spotted Woodpeckers – forage more frequently pecking deep into wood (Hogstad 1978, Cramp 1985). A major component of the White-backed Woodpecker diet is large wood-boring insect larvae, which occur much less frequently in the diet of Great Spotted and Syrian Woodpeckers (Cramp 1985, Michalek & Miettinen 2003). There are strong differences between winter foraging niches of the three species (Hogstad 1971, Cramp 1985). Great Spotted Woodpeckers forage on conifer seeds, mainly Scots Pine Pinus sylvestris (Hogstad 1971, Myczko & Benkman 2011). This can explain their specific bill shape which could be an adaptation for processing conifer cones.

Thus, it is clear that there are considerable dietary niche overlaps in the interspecific pairs, White-backed/Great Spotted Woodpeckers and Syrian/Great Spotted Woodpeckers, which can lead to competition and accentuation of trait divergence (Gray et al. 2005, Stuart & Losos 2013) in studied sympatric species. Moreover, all three species are closely related (Fuchs & Pons 2015) and more closely related species compete more fiercely for resources (Wiens & Rotenberry 1981). We did not confirm any significant shape differences between White-backed and Syrian Woodpeckers in comparisons of the same sexes in the dorsal view of the maxilla, and independent of sex in the lateral view. This means that their bill shapes are morphologically similar (Figures 2–5, Tables 1 and 2). Syrian Woodpeckers prefer more open areas greatly modified by humans (Mošanský & Mošanský 1999, Michalczuk & Michalczuk 2016b) in comparison to the semi-natural forests preferred by White-backed Woodpeckers (Czeszczewik & Walankiewicz 2006). This difference in habitat preferences results in strict spatial isolation between the two species and a lack of direct competition for resources. Despite a lack of any difference in bill shape between White-backed and Syrian Woodpeckers, both species differed significantly in bill length. The bills of White-backed Woodpeckers were significantly longer, which may simply reflect the difference in body size between the two species. For both sexes, mean dimensions of almost all analysed body traits were significantly larger in White-backed Woodpeckers than in the other analysed species (Table 3(a–f)). Only female White-backed Woodpecker wing length and body length were similar to those of female Great Spotted Woodpeckers. White-backed Woodpecker tail length was similar to tail length in the Great Spotted Woodpecker, and female White-backed Woodpecker tarsus length was similar to all other groups. On the other hand, longer bills in White-backed Woodpeckers may be a consequence of their foraging behaviour involving excavation of prey deeper within the tree (Cramp 1985), therefore, longer bill and greater body mass can be beneficial. The comparisons of body mass within the same sexes between sympatrically occurring Great Spotted and Syrian Woodpeckers showed a lack of significant differences, thus between these two species we should expect the strongest interspecific concurrence.

Sex differences in foraging niches are well known in woodpeckers. Most studies on foraging niches of sympatric European woodpeckers reveal many differences in their foraging behaviour (Jenni 1983, Török 1990). For example, Hogstad (1978) reported differences between sexes in foraging niches for White-backed and Great Spotted Woodpeckers and suggested dominance by males in optimum feeding sites. Moreover, Osiejuk (1998) reported many intersexual differences in foraging patterns in Great Spotted Woodpeckers, especially in winters of poor Scots Pine cone crops. However, in winters with a rich crop of Scots Pine cones, the foraging behaviour of both sexes became very similar (Osiejuk 1994, 1998). It is obvious to expect that such differences in foraging niche should also encourage sexual dimorphism; particularly in bill size and shape. We confirmed significant sex differences in bill length of all three woodpecker species (Table 3(a)), but following Bonferroni correction the only significant difference between the sexes in shape of bills was in the dorsal view of White-backed Woodpeckers (Tables 1 and 2).

We have demonstrated that different methods of comparing bill morphology between species show different, but not contradictory results. The traditional measurement of bill length to compare morphology shows different results than the landmark-based geometric morphometric analysis described here. This apparent discrepancy is a consequence of woodpecker bills, like those of most birds, not having characteristic points enabling repeatable measurements for describing bill shape. It follows that using the results of these two techniques together allows us to more thoroughly compare differences in bill morphology between groups of birds.

Acknowledgements

We thank T. Jászay, the curator of the Natural History collection of Saris Museum in Bardejov for help during investigations and P. Tryjanowski for useful comments on this paper and for statistical advice.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

MH and ZO were supported by grant OPV ITMS: 26110230119.