Sándor Csősz1,2, Kadri Kiran1, Celal Karaman1, Albena Lapeva-Gjonova3. 1. Evolutionary Ecology Research Group, Institute of Ecology and Botany, Centre for Ecological Research, Alkotmány út 2-4, Vácrátót H-2163, Hungary Evolutionary Ecology Research Group, Institute of Ecology and Botany Budapest Hungary. 2. MTA-ELTE-MTM Ecology Research Group, Pázmány Péter sétány 1/C, Budapest H-1117, Hungary MTA-ELTE-MTM Ecology Research Group Budapest Hungary. 3. Sofia University, Faculty of Biology, Department of Zoology and Anthropology, Bulgaria Sofia University Sofia Bulgaria.
Abstract
In this paper, we provide numeric morphology-based evidence that the dark-colored Poneracoarctatavar.lucida Emery, 1898, formerly considered a synonym of P.coarctata (Latreille, 1802), is conspecific with the lighter-colored Poneratestacea Emery, 1895. Species hypotheses are developed via NC-PART clustering, combined with Partitioning Algorithm based on Recursive Thresholding (PART), and via PCA combined with gap statistics. We obtained our results from an extensive dataset from the 10 continuous morphometric traits measured on 165 workers belonging to 73 nest samples. Linear discriminant analysis (LDA) confirmed the grouping of hypotheses generated by exploratory analyses with 100% classification success when all ten morphometric traits were involved. The Anatolian Turkish black and the predominantly European yellow samples, did not separate based on their morphometric characteristics. These two color variations broadly overlap in their geographic range in Anatolian Turkey. The investigated type series of Poneracoarctatavar.lucida Emery, 1898 (collected from Kazakhstan) fell within the P.testacea cluster instead of P.coarctata and is also classified with high certainty as P.testacea by confirmatory LDA. Therefore, we propose the synonymy of Poneracoarctatavar.lucida Emery, 1898 with Poneratestacea Emery, 1895. As no other morphological differences than color patterns were detected between the "black" and "pale" P.testacea samples, we hold that these populations constitute geographically occurring color variations of the same species. Finally, our quantitative morphology-based results show that relying on color patterns is not a robust approach in identifying European Ponera samples, particularly in the east, but using multivariate analyses of morphometric traits is advised instead. Sándor Csősz, Kadri Kiran, Celal Karaman, Albena Lapeva-Gjonova.
In this paper, we provide numeric morphology-based evidence that the dark-colored Poneracoarctatavar.lucida Emery, 1898, formerly considered a synonym of P.coarctata (Latreille, 1802), is conspecific with the lighter-colored Poneratestacea Emery, 1895. Species hypotheses are developed via NC-PART clustering, combined with Partitioning Algorithm based on Recursive Thresholding (PART), and via PCA combined with gap statistics. We obtained our results from an extensive dataset from the 10 continuous morphometric traits measured on 165 workers belonging to 73 nest samples. Linear discriminant analysis (LDA) confirmed the grouping of hypotheses generated by exploratory analyses with 100% classification success when all ten morphometric traits were involved. The Anatolian Turkish black and the predominantly European yellow samples, did not separate based on their morphometric characteristics. These two color variations broadly overlap in their geographic range in Anatolian Turkey. The investigated type series of Poneracoarctatavar.lucida Emery, 1898 (collected from Kazakhstan) fell within the P.testacea cluster instead of P.coarctata and is also classified with high certainty as P.testacea by confirmatory LDA. Therefore, we propose the synonymy of Poneracoarctatavar.lucida Emery, 1898 with Poneratestacea Emery, 1895. As no other morphological differences than color patterns were detected between the "black" and "pale" P.testacea samples, we hold that these populations constitute geographically occurring color variations of the same species. Finally, our quantitative morphology-based results show that relying on color patterns is not a robust approach in identifying European Ponera samples, particularly in the east, but using multivariate analyses of morphometric traits is advised instead. Sándor Csősz, Kadri Kiran, Celal Karaman, Albena Lapeva-Gjonova.
Entities:
Keywords:
Biogeography; exploratory analyses; gap statistic; morphometry; species delimitation
The taxonomy of the European representatives of the tiny hypogeic genus has for several decades been apparently unambiguous. The two European species belonging to this genus, (Latreille, 1802) and Emery, 1895 constitute one of the continent’s most easily distinguishable species pairs (Csősz and Seifert 2003; Scupola 2006; Czechowski and Radchenko 2010; Attewell et al. 2011). Conspicuous size and color features help to tell these species apart. is larger, generally black, and has a higher petiole, while its typically lighter yellow congener, , is significantly smaller, having a low and thick petiolar node (Csősz and Seifert 2003). In addition, these species differ in habitat preferences; prefers more xerothermous biotopes than its sister species, . Based on several environmental variables measured in 25 study plots in Central Europe (Seifert 2017), sites can be characterized as having higher maximum calibrated soil temperatures and lower soil moisture values. These previously considered European species extend their distribution to Turkey (Kiran and Karaman 2021), and has been known to occur north of the Black Sea coastline and reach the Caucasus range (Kiran and Karaman 2020). However, the most recent investigations revealed that a third morph, with a mixed combination of traits, also appears to occur in this region. This form is similar to in its shiny black or dark brown color. At the same time, the lower petiole node, dense pilosity, and multivariate analyses on their morphometric data place representatives of this particular group in . Therefore, the normal morphological approaches that help to separate these species in the rest of the European part of the Western Palaearctic region seemed to fail in accurately identifying them in Turkey.This problem prompted us to examine the possibility of whether a third species appeared in this region and, if so, to describe it appropriately. Ant taxonomists have often considered color traits unreliable as species-level traits due to the high intraspecific color polymorphism (Seifert 2003a,b, 2019; Seifert et al. 2017) that may reportedly lead to taxonomic errors. Therefore, we analyzed an extensive set of continuous morphometric data via robust multivariate statistic procedures, NC-clustering and principal component analysis (PCA) to infer species boundaries. Both these approaches were used in combination with Gap statistic algorithms that estimate the number of clusters in the data and assign observations (i.e. specimens or samples) into partitions. We tested the validity of the recognized pattern via confirmatory linear discriminant analysis (LDA).We compared the morphologically recognized clusters with the color patterns of the samples. As a result, the third morph found in Turkey broadly overlaps with the cluster of European via the complex morphometric protocol. The Anatolian Turkish black and the predominantly European yellow samples did not separate on the basis of their morphometric characteristics, and these two color variations broadly overlap in their geographic range in Anatolian Turkey. The investigated type series of Emery, 1898 (collected from Azerbaijan) fell within the cluster instead of and is also classified with high certainty as by confirmatory LDA. Therefore, we propose the synonymy of with . As no other morphological differences than color patterns were detected between the “black” and “pale” samples, we hold that these populations constitute geographically occurring color variations of the same species. Thus, making a subspecies of is proven unnecessary.
Material and methods
In this research 10 continuous morphometric traits were measured on 165 workers belonging to 73 nest samples. The material is deposited in the following institutions: Entomological Museum of Trakya University (EMTU), Hungarian Natural History Museum, Budapest, Hungary (HNHM), Muséum d’histoire naturelle, Genève (MHNG), Natural History Museum, Genoa (MSNG), private collection of Sándor Csősz (PCSC). The full list of material investigated is listed below:
Material examined
Type material
Lectotype: “Bonifacio, leg. REVEL 1872” (1 w, MCSN); paralectotypes (3 workers) Rapallo / Liguria / Mai 1891” (1 w, MHNG); and two other specimens (on one pin) labelled by Emery “Gallia merid.” and with a blue label “Cotypus” (2 ww, MHNG)Syntypes: “Lenkoran / (next line) Korb / next label Syntypus var. Emery, 1898” (3 ww), MCSN) [1st and 3rd measured, 2nd not measurable due to glue obstructing the view]
Pigmentation scaling was performed via a subjective evaluation of body coloration ranging from whitish yellow (score 1) to black (score 5). The specimens were illuminated via Photonic Optics 2-arms Illuminator with neutral white color temperature, 5900 K (equivalent to halogen, 4000 K). Specimens with light pigmentation (score 2 and score 3) were classified as “pale morph” (Fig. 1.), darker (score 4 and score 5) specimens were considered “black morph” (Fig. 2). Very light, whitish-yellow (score 1) specimens have not been found.
Figure 1.
Light-colored (score 2) worker from Hungary. Specimen: CASENT0906719, from www.antweb.org.
Figure 2.
A syntype worker of representing a dark-colored (score 4) worker from Azerbaijan. Specimen: CASENT0903905, from www.antweb.org.
Light-colored (score 2) worker from Hungary. Specimen: CASENT0906719, from www.antweb.org.A syntype worker of representing a dark-colored (score 4) worker from Azerbaijan. Specimen: CASENT0903905, from www.antweb.org.
Protocol for morphometric character recording
Morphometric characters are defined as in Seifert (2018). All measurements were made in an Olympus SZX9 stereomicroscope at a magnification of 150× for each character. Morphometric data are given in µm throughout the paper. All worker individuals were measured by SC. Definitions of morphometric characters are listed below (for details see Csősz and Seifert 2003: 202–204):CL Cephalic length;CW Cephalic width;FL Maximum width of frontal lobes;FR Minimum distance between frontal carinae;ML Mesosoma length;PEH The maximum height of petiole;PEL Petiole length;PH Height of petiole node;PW Petiole width;SL Scape length.
Statistical framework on morphometric data–hypothesis formation and testing
Exploratory analyses through NC-PART clustering
The statistical procedure has been done on worker caste only. The prior species hypothesis was generated based on workers through the combined application of NC clustering (Seifert et al. 2014) and Partitioning Algorithm based on Recursive Thresholding (PART) (Nilsen et al. 2013; see also Csősz and Fisher 2016). The optimal number of clusters and the partitioning of samples are accepted as the preliminary species hypothesis in every case in which the two clustering methods, ‘hclust’ and ‘kmeans’ through PART, have yielded the same conclusion.
Exploratory analyses through sPCA in combination with Gap statistics
Structure in morphometric data was also displayed in a scatterplot via a principal component analysis (sPCA; Baur and Leuenberger 2011). The sPCA does not return an estimate on the number of clusters, hence an iterative gap statistic algorithm (function ‘gap’) implemented in package clusterGenomics (Tibshirani et al. 2001; Nilsen et al. 2013), was employed to determine the optimal number of clusters within data and also assigned observations (i.e. specimens, or samples) into partitions.
Hypothesis testing by confirmatory analyses
The validity of the prior species hypothesis imposed by the two exploratory processes was tested via a cross-validated linear discriminant analysis (CV-LDA), and the best fitting simple ratio is found via multivariate ratio analysis (MRA). Statistical analyses have been done in R (R Core Team 2019).
Results
Finding biodiversity patterns through numeric morphology
Altogether two clusters were revealed to be the most parsimonious solution by both NC-PART clustering (Fig. 3.) and by the gap statistic based on PC scores (Figs 4, 5.). The grouping hypotheses generated by hypothesis-free exploratory analyses is confirmed by LDA with 100% classification success when all 10 morphometric traits were involved. This pattern is also supported by the examination of external morphological traits (e.g. shape of petiolar node, and density of pilosity on the first gaster tegite). appears to exhibit uniform color patterns throughout the whole distributional area, but has a rather bimodal coloration; the western (European) population has a pale brown, or yellow appearance, whilst the eastern, Anatolian samples are black. This geography driven shift does not appear in other morphological features and the multivariate analyses of continuous traits do not support separateness.
Figure 3.
Dendrogram solution for the Western Palaearctic representatives of . Sample information in the dendrogram follows this format: abbreviated country code, locality name, and/or a special collection code followed by final species hypothesis separated by underscore. Two columns of rectangles represent results of partitioning resulted by method PART using two cluster methods ‘hclust’ and ‘kmeans’.
Figure 4.
Gap statistic for dataset of Western Palaearctic samples. Two-cluster solution is highly supported by the elbow at 2 components by the dispersion curve (left) and by the peak at cluster number four by the gap curve (right). Number of clusters in the data (X axis), the total within-cluster dispersion for each evaluated partition (Y axis for the left plot) and the vector of length Kmax giving the Gap statistic for each evaluated partition (Y axis for the right plot) is illustrated.
Figure 5.
Ordination biplot for shape principal component analysis (sPCA) based on species identity. Color codes represent: : blue circles, : orange asterisks, lectotype: orange triangle, syntypes: dark brown rectangles.
Dendrogram solution for the Western Palaearctic representatives of . Sample information in the dendrogram follows this format: abbreviated country code, locality name, and/or a special collection code followed by final species hypothesis separated by underscore. Two columns of rectangles represent results of partitioning resulted by method PART using two cluster methods ‘hclust’ and ‘kmeans’.Gap statistic for dataset of Western Palaearctic samples. Two-cluster solution is highly supported by the elbow at 2 components by the dispersion curve (left) and by the peak at cluster number four by the gap curve (right). Number of clusters in the data (X axis), the total within-cluster dispersion for each evaluated partition (Y axis for the left plot) and the vector of length Kmax giving the Gap statistic for each evaluated partition (Y axis for the right plot) is illustrated.Ordination biplot for shape principal component analysis (sPCA) based on species identity. Color codes represent: : blue circles, : orange asterisks, lectotype: orange triangle, syntypes: dark brown rectangles.
Taxonomic acts
We hold does not split into separate subspecies and the “pale morph” and the “black morph” of belong to the same species. Therefore, we synonymize Emery, 1898, which was formerly considered a junior synonym of (see Taylor 1967), with Emery, 1895.
The spatial distribution pattern
Geographic distribution of and (including the “pale morph” and the “black morph”) broadly overlap in Europe and in Anatolian Turkey, where occupies significantly (p = 0.046) lower altitudes (519 m [5 m, 743 m]) than (900 m [0 m, 1900 m]). The two color variations of does not show significant differences (p = 0.92) in their vertical distribution (“black morph” (n = 14) 896 m [442 m, 1791 m], “pale morph” (n = 10) 919 m [0 m, 1900 m]).
Species delimitation
The distinctive morphology of these species allows for considerable reduction of morphological characters, so that workers of the two taxa and can be separated based on the combination of three continuous morphometric traits (CW, PEL, and PH; Fig. 6) with 99.4% (a single misclassified case out of the total 165 individuals).
Figure 6.
The best morphometric ratio (petiole length / petiole height; PEL/PH) is illustrated on the head length (CL). Scatterplots of the most discriminating ratio on the head length between workers of Western Palaearctic representatives of ; : black circles, : red triangles. The thin dashed line illustrates best separation.
The best morphometric ratio (petiole length / petiole height; PEL/PH) is illustrated on the head length (CL). Scatterplots of the most discriminating ratio on the head length between workers of Western Palaearctic representatives of ; : black circles, : red triangles. The thin dashed line illustrates best separation.Coefficients of linear discriminants help to place and identify samples via placing workers in the discriminant space using the linear discriminant function (LD) as follows (morphometric traits are in micrometers):LD = 0.0449*CW − 0.0893 * PEL + 0.0750 * PH − 21.4786(n = 93) = +1.99 (+0.22, +5.02)(n = 72) = −2.57 (−5.05, +0.27) (including both “pale morph” and “black morph”)A simple morphometric ratio of two petiole characters (petiole length / height of petiolar node, PEL/PH) appears an excellent numeric key to tell these species apart but is slightly affected by allometry. Therefore, a graphical display of this ratio on the head length (CL) is also provided as an easy-to-use asset aiding routine determinations (Fig. 6). This key yields a 98.18% classification success (n = 165).
Discussion
The yellow and black morphs do not differ via multivariate analyses of continuous morphometric traits based on an extensive material collected in a wide geographic range. Furthermore, no other shape characteristics support their separation. The investigated type series of (the “black” morph) fell within the cluster instead of in both exploratory analyses (NC-clustering and PCA) and is also classified with high certainty as by confirmatory LDA. Therefore, we propose synonymy of with instead of .Although the yellowish phenotype of is dominant in the Western Palaearctic (western Turkey and Europe), the black morph is prevalent in Turkey; distribution of these color variations broadly overlaps in Anatolia (Fig. 7). The sympatric pattern might indicate a subspecific rank. However, due to the lack of morphological differences other than the color of morphs, we hold that these populations constitute geographically occurring color variations of the same species, making a subspecies of unnecessary.
Figure 7.
Geographic map of species in Europe and Turkey. Color codes for species are as follows: : red circles, “black morphs”: blue triangles, “yellow morphs”: yellow rectangles.
Geographic map of species in Europe and Turkey. Color codes for species are as follows: : red circles, “black morphs”: blue triangles, “yellow morphs”: yellow rectangles.In conclusion, relying solely on color patterns is not a robust approach in identifying European samples, but using multivariate analyses of morphometric traits or the presented numeric key is advised to distinguish these species. Straightforward pigmentation patterns, such as variation of light versus dark, red versus brown, or red versus black on the whole body or specific body parts, are frequently unreliable taxonomic characters in insects and other animals. A minor point mutation may change pigmentation profoundly, whereas complex morphological structures are less easily changed. Many textbook examples demonstrate relatively simple pigmentation genetics (e.g. Lus 1932; Barrion and Saxena 1987; Majerus 1998; Andrés and Cordero 1999; Majumdar et al. 2008). Taxonomic failures or pitfalls resulting from pigmentation in ants have been repeatedly reported. Seifert (1997, 2018, 2020) described color dimorphism as superimposed by an allometric change in the group and species. Intraspecific color polymorphism superimposed by geographic clines has also been shown for species of the group (Seifert 2003a). Furthermore, extreme intraspecific color polymorphism is reported in several species of (Seifert 2003b; Seifert et al. 2017) and in (Seifert 2019) that reportedly leads to taxonomic errors.
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.