Biometric differences between several populations of Cordulegaster boltonii (Odonata: Cordulegastridae) in Ibero-Maghrebian area

Biometric data of the exuviae of female larvae of the dragonfl y Cordulegaster boltonii collected in Portugal, Spain and Morocco were analysed to determine whether the size of three exuvial structures measured differed depending on the geographic localities of the populations. Based on the results recorded for the 16 populations studied, head width was negatively correlated with latitude and the greatest length of the gonapophysis was recorded for the Iberian populations at the centre of this peninsula. Multivariate cluster analysis revealed a clear separation of the Moroccan population. A second cluster separated the southernmost population (Sierra Nevada) from the remaining Iberian populations. Four population groups were distinguished: those located in watercourses in the north and central area of the Iberian Peninsula, those in Iberian watercourses in the East and Middle South, the Sierra Nevada and North Morocco. Some of these results coincide with the results of genetic studies of other authors.


INTRODUCTION
Morpho-biometric characters of Odonata have been used to confi rm interspecifi c differences recorded by genetic analysis (Lee & Lin, 2012;Kohli et al., 2013). These characters enable species to be differentiated during the larval phase when other characters are not consistent or diffi cult to observe (Landwer & Sites, 2006), can be used to analyse the parallel evolution of species in the same genus (Stoks et al., 2005) and to relate genome size with body size (Ardila-García & Gregory, 2009). Similarly, they have been utilized to separate specimens or populations of the same species, either by analysing adults (e.g., Cabuga et al., 2017;Corso, 2019) or exuviae (Casanueva et al., 2017a).
During the larval stages odonates are preyed upon by other species, mainly fi sh (Morin, 1984;McPeek, 1990). Many species defend themselves by means of abdominal spines against predators (Arnqvist & Johansson, 1998;Johansson Eur. J. Entomol. 117: 260-264, 2020 doi: 10.14411/eje.2020.028 NOTE the 9th sternite (SP9) and length of the 9th sternite (S9), which is SP9/S9, and (c) length of gonapophysis (LG). The fi rst is a morphological character and therefore not quantifi able, while the other two are easily quantifi able. The last two were used in this study. An additional quantifi able character was also utilized; head width (He; i.e. the maximum distance between the lateral margins of the compound eyes), which does not vary during the emergence period (Casanueva et al., 2017c). These three variables are very poorly correlated, so collinearity is avoided (LG-SP9/S9: r = 0.30; LG-He: r = -0.03; He-SP9/S9: r = -0.12). The LG and He variables were adjusted to normality using the Shapiro-Wilk test, which then met the homogeneity of variance based on the Bartlett's test. Therefore, their average values were compared using an analysis of variance ANOVA. The SP9/S9 ratio values do not adjust to normality, so it was necessary to use the Kruskal-Wallis non-parametric analysis. In all cases the minimum signifi cance value was P < 0.05. A multivariate cluster analysis was used to group the variables by achieving the maximum homogeneity in each group and greatest difference between groups. For this, the nearest neighbour, similarity index or Euclidean distance was used as an algorithm. Clusters were joined based on the smallest distance between two groups. With a bootstrapping of N = 10000, the percentage support was calculated for each of the nodes.

RESULTS
Measurements of 590 exuviae collected from sixteen Iberian and Maghrebi populations were analysed (Table 1). There were signifi cant differences between populations for the SP9/S9 ratio (Kruskal-Wallis test, H = 212.18, P = 0.000), LG (F 15, 571 = 65.66, P = 0.000) and He (F 15, 571 = 54.06, P = 0.000). The SP9/S9 ratio did not clearly separate between the populations sampled, but the highest values were recorded for the northern and central populations and the lowest for the southernmost Iberian Peninsula (Monachil, located at Sierra Nevada) and Moroccan populations (Fig. 2). The populations at the centre of the Iberian Peninsula (Tormes and Alberche) had a longer gonapophysis than the other populations and the shortest was recorded for the Monachil population (Fig. 3). The head of the exuviae from Morocco was the widest (Fig. 4) followed by those from the Middle South and East

MATERIAL AND METHODS
During 2016 and 2017 exuviae of fi nal instar larvae of C. boltonii were collected from fourteen mountain rivers on the Iberian Peninsula and one in Morocco ( Fig. 1, Table 1). In addition, exuviae that were collected from 1989-1992 from the population in the Bejarano River were also used. All samples were collected in May, June and July, which for this species is the period when most adults emerge (Ferreras-Romero & Corbet, 1999;Casanueva et al., 2017b). In the laboratory, the exuviae were measured, always by the same person (FC), using a Nikon SMZ800 binocular microscope with an eyepiece micrometre. Since males do not have a gonapophysis, only female exuviae were used.
Based on the work of Verschuren (1989), Boudot & Jacquemin (1995) proposes three criteria of exuviae for differentiating populations of C. boltonii: (a) form of the setae on the posterior edge of the 9th sternite, (b) ratio of the length of the lateral spine on   Iberian Peninsula populations (Bejarano and Júcar). The head width was negatively correlated with latitude. Multivariate cluster analysis (Fig. 5) clearly separates the Moroccan population (Amsemlil) from the others. A second cluster separates, with a difference of 70%, the southernmost population (Monachil), which hereafter is called cluster B, from the other clusters, which are referred to hereinafter as clusters C and D. Clusters C and D differ by 33%. Cluster D includes the central and northern Iberian populations within which there are two groups that differ by 70%: cluster D1 including the seven populations in the centre and one from the north of Portugal, and cluster D2 including the four northernmost populations studied.
Since this analysis indicates that the Moroccan and Monachil populations differ from all others, they were not included in the non-parametric ANOVA used to determine whether clusters C, D1 and D2 differ from each other (Table 2). These analyses indicate that the variables He, GL and SP9/S9 differ signifi cantly (P = 0.000), both when the three clusters are compared separately and when C is compared with D1 and D2 combined (Table 2). However, when clusters D1 and D2 were compared with each other, the SP9/S9 ratio did not differ signifi cantly (P = 0.249). This grouping coincides only partially with the one proposed by Boudot & Jacquemin (1995) who analysed exuviae of C. boltonii from the Iberian Peninsula and Morocco. These authors suggest that the South Iberian populations have a mixture of the characteristics of the populations in North Africa and those at the centre of the Iberian Peninsula and those of the northern Iberian populations differ from those at the centre of the peninsular. On the other hand, Froufe et al. (2013) conclude that, of the populations they studied, only that in North Africa (C. boltonii algirica) is genetically different from the populations in Europe. This same difference also appears when the biometric characteristics of the exuviae were studied.    According to Froufe et al. (2013), it seems likely that North Africa was colonized twice, once by a common ancestor of Cordulegaster princeps (Jacquemin & Boudot, 1999) and C. boltonii, and more recently by C. boltonii. However, adults in Czech populations of C. boltonii only make short trips (< 10 km) (Hančíková, 2014). If the same occurs on the Iberian Peninsula and in Morocco, it is possible that the populations of C. boltonii on both sides of the Mediterranean Sea do not mix with each other, because they are separated by a sea. This isolation could have given rise to a separate subspecies in North Africa (C. b. algirica). Froufe et al. (2013) state that over a long term an allopatric speciation process could even be taking place. The results presented here show that the Iberian populations are clearly different from that located in North Africa (Morocco) (Fig. 5) and therefore support the hypothesis that they are separate population groups, without reproductive contact that allows gene fl ow. Furthermore, the population at Monachil (Sierra Nevada), the most southern and possibly also the most isolated of those studied on the Iberian Peninsula, is the most different from the other peninsular populations. In any case, in order to determine if this is true, it is necessary to study in more detail the genetical and anatomical characteristics of the populations of C. boltonii in North Africa and Sierra Nevada.

DISCUSSION
The Central System of populations on the Iberian Peninsula are clearly a distinct group. The eastern populations resemble those in the North. This indicates that they may have had a common origin, probably located in the North of the Iberian Peninsula. From there they would have spread south along two routes, one following the mountains in the North (populations 2, 3, 4, and 5) (Fig. 1) and the other the mountains in the centre, with both routes meeting in the mountains in northern Portugal (population 1).
In summary, genetic studies (Froufe et al., 2013) along with biometric analyses can be used for grouping or segregating different C. boltonii populations. This method could also be used for studying populations in other areas where this species is present, and even those of other species of Cordulegaster morphologically similar to C. boltonii (Corso, 2019).