II. MORPHOMETRY OF WINGS OF WORKER BEES OF THE SUBSPECIES APIS MELLIFERA MELLIFERA L. (POLISSYA POPULATION OF ZHYTOMYR REGION)

The uncontrolled spread of the subspecies A. m. carnica, A. m. ligustica, and A. m. caucasica has led to a reduction in the areas of pure "dark forest bees" populations belonging to the subspecies A. m. mellifera in their natural habitats within Ukraine. Due to the need to use dark forest bees in breeding, it became necessary to identify the locations of individual populations of bees belonging to the A. m. mellifera subspecies. The purpose of the work is to create an accessible and at the same time complete methodology for classifying bee wing phenotypes, which would make it possible to determine the probable breed of worker bees, the type and degree of hybridization of the main breed with impurities, and to identify "purebred" bee families by the wing phenotype suitable for further breeding. Material and methods of research: Using discriminant analysis of data, at the first stage of the study, 1500 wings of bee families were classified using 8 features: Ci, Dbi, Disc.sh, Pci, Ri, Сі.3, Сі.2, Сi.2.1, for which there was preliminary information about the possible belonging of the wing phenotype to the subspecies A. m. mellifera or its hybrids. At the second stage, additional 1212 wings of bee colonies were studied, about which there were doubts about their breed. Results of research and discussion: The wings are reliably divided into four clusters, indicating the presence of four sufficiently distinct groups among the studied wings in terms of phenotype. Conclusions and prospects for further research: A classification model has been created that allows for effective discrimination of the wings of working bees of bee colonies in Ukraine, the subspecies A. m. mellifera. Phenotypic values of indices of four Polissia micro-populations of bees, used as reference standards for possible hybridization detection, have been established, which can serve as standards in future research. Four colonies have been found, whose queens produce bees of the A. m. mellifera type of the Polissia population, and three colonies whose queens produce bees of the A. m. macedonica hybrid and can be used for further selection work

Regarding the bee populations in Ukraine, based on morphometric studies using 38 morphological traits according to F. Ruttner [5] and reference specimens of subspecies from the morphometric database (Bundesland Hessen, Oberursel), one-third of the samples from the Polissya region of Ukraine were classified as subspecies A. m. mellifera, one-third as A. m. macedonica, and one-third were identified as hybrids [6]. In case of using only wing morphology traits for classification purposes, all studied bee samples were assigned to three clusters: 1) subspecies of evolutionary lineage C (A. m. carnica, A. m. macedonica, A. m. ligustica); 2) subspecies of lineage O (A. m. caucasica and A. m. anatoliaca) together with samples from Ukraine; 3) subspecies A. m. mellifera. Hybrid samples unexpectedly showed a strong influence of the morphological lineage O. It is impossible to establish from the article which exact wing morphometric characteristics were used for classification. It should also be noted, that the sample of bee families, studied in Ukraine, was limited: only 17 families, obtained approximately from Kyiv, Vinnytsia regions, and Western Ukraine (it is impossible to establish the origin of bee samples from the text of the article). This is a reason to believe that the conclusions, drawn from a small number of families, and a selective regional location require further research to identify bee families, supplement with morphometric data, and reveal the real locations of the dark forest bee population or their hybrids. The authors made the first attempt to study the worker bees of several apiaries in the Zhytomyr region using classical wing morphometrics with the use of five traits (indices) in the work [7]. The results showed that the prevailing phenotypic breeds in this area are the Ukrainian steppe bee, A. m. mellifera (Polissya population), and their hybrids. Of the total of 1423 wings examined, only 116 could be reliably attributed to the A. m. mellifera subspecies, while 272 wings belonged to A. m. mellifera hybrids. It was also found, that the use of five indices and the fact that most of the studied families belonged to one beekeeper did not allow for a reliable determination of hybridization types based on wing phenotype and for making general conclusions about bees in the entire Zhytomyr region. Therefore, researchers faced the task of expanding the area of research and creating a classifier of wings that is accessible and accurate enough to determine the subspecies structure of bees. There was also an urgent need to develop a methodology for determining the type and degree of hybridization based on wing phenotype, without which it would be impossible to interpret the results of morphometric studies of wings.
The purpose of the work -Create an accessible and at the same time complete methodology for classifying the phenotypes of bee wings, by means of which it would be possible to determine the probable breed belonging of worker bees; type and degree of hybridization of the main breed with impurities; identify the "purebred" by the phenotype of the wings of bee colonies suitable for further breeding.
The wings of worker bees were collected from naturally obtained winter cluster, in accordance with the recommendations of ARRIVE guidelines for animal experiments and the UK Animals (Scientific Procedures) Act 1986, and relevant guiding principles, or the EU Directive 2010/63/EU on the protection of animals, used for scientific purposes.  Kevl.6(2022) local bees ** Note: *instrumental insemination; **natural insemination The wing images were processed using the TpsDig software [8]. Index values were calculated using a custom program. In the first stage of the study, 1500 wings from different colonies were classified using eight features (Ci, Dbi, Disc.sh, Pci, Ri, C.2, C.3, C.2.1), as proposed in [9], and based on approximate information about the possible affiliation of the wing phenotype to the subspecies A. m. mel-lifera or its hybrids. Using discriminant analysis and the StatSoft software package [10], the most reliable result was the classification of the wings into four clusters (Tables 2-5, Fig. 2), indicating the presence of four distinct phenotypic groups among the investigated wings. The phenotype of the wings in this study is defined as the set of values for the eight indices mentioned above. The sequence of indices according to the increase of their weights in the classification is as follows: Ri, C, C.3, Ci.2, Disc.sh, Pci, Dbi, Ci.2.1 (Table 2). Therefore, the radi-al Ri and C indices have the smallest influence on the classification of this set of wings from worker bees, while the Dbi and Ci.2.1 indices have the greatest influence. The accuracy (correctness) of classification is satisfactory (95.0 %, Table 3). The arrangement of data (wings) in the space of canonical variables also indicates a qualita-tive discrimination using the obtained classification model (Fig. 2).

Research results
The nature of the distribution of points in canonical variables (Fig.2) indicates that the phenotypes of wings in clusters 2 and 3 are significantly different from each other and do not have common boundaries. Wings belonging to clusters 1 and 4, upon further analysis, for example using Mahalanobis distances, may consist of two subgroups that differ in phenotype. The phenotype of a larger portion of wings will correspond to clusters 1 and 4, while the phenotype of another portion of wings may be similar to cluster 2 or cluster 3. When establishing the tentative affiliation of the wing phenotypes to the four clusters of the classification model (Table 5), literature data for two indices were used: cubital (Ci) and discoidal shift (Disc.sh.) [5,11]. Based on these indices, the wing phenotype of cluster 2 clearly belongs to the subspecies A. m. mellifera. It can be assumed, that cluster 1 also belongs to the subspecies A. m. mellifera. However, it should be noted, that for cluster 1, the value of Disc.sh. is greater than 0, indicating a significant probability of hybridization. Clusters 3 and 4 correspond to local populations of certain hybrids. Taking into account the value of Disc.sh. for cluster 3 (2.5 > 0) and the relatively high value of Ci (2.04), (according to Avatesyan, the value of Ci for the Ukrainian steppe bee is within the range of 2.0-2.3 [11]; according to Polishchuk, it is 2.16-2.62, Disc.sh. (+) for 72 %-94 % of wings) [12], it can be assumed, that the main com-ponent of the bees in cluster 3 is a somewhat hybridized population type of the A. m. macedonica subspecies of the Ukrainian steppe bee.  The same interpretation is also valid for cluster 4, taking into account that the value of Disc.sh.<0 (-0.39) indicates a preference in the phenotype for the subspecies A. m. mellifera, or belonging to the subspecies A. m. caucasica, which is mentioned in the study [6]. This paragraph discusses a study on the morphometrics of the wings of the subspecies A. m. caucasica. The study found that the variation limits and mean values of the indices for seven bee colonies with queens from Georgia were as follows: Si (2.07-2.265), Si (mean) =2.173; Disc.sh. takes values within the range of -0.285 to -1.96, Disc.sh. (mean) =-1.43 [13]. These data are in good agreement with the index values of cluster 4 (Table 4) and with the data of F. Rutner, who reported Si=2.16 [14]. In conclusion, the possibility of the presence of a certain component of the A. m. caucasica subspecies in the genomes of queens from the Polissya region cannot be ruled out. However, a final verdict on this matter can only be made after appropriate genetic research.
The assignment of wing phenotypes from cluster 2 to the subspecies A. m. mellifera, and wing phenotypes from cluster 3 to the hybrid A. m. macedonica, which are phenotypically quite distinct, is consistent with the analysis of the distribution of wings in the canonical variables space, which illustrates a significant difference between clusters 2 and 3 (Fig. 2).
On the second stage of the study, using the obtained classifier, additional 1212 wings from 13 other bee colonies were analyzed, for which there were no previous records or doubts about their belonging to certain breeds, as well as those obtained for processing in the last place (Table 7).  It is necessary to draw attention to the fact that all 502 wings of the Polish standard come directly from the apiary of Polish bee breeder I. Stolyar (the village of Hlybochok, Zhytomyr district), from which queens were purchased by beekeepers at different times. The origin of the bees of the Hybrid.Pol.1, Hybrid.Maced and Hybrid.Pol.2 standards is more diverse, and covers the points of the city Ovruch and district of Zhytomyr. Table 11 Wing index values of worker bees of colonies of the morphometric standard of the local

Discussion
For the arrays of wings, formed by the Hybrid.Pol.1, Polish, and Hybrid.Pol.2 standards, the coefficients of variation of the Disc.sh. index range from 33 % to 50.1 %, which exceeds the "limit of reasonableness," while the coefficients of variation of the other seven indices have entirely acceptable values (<8 %). This fact cannot be explained solely by errors in establishing landmarks on the wing, which determine the value of the discoidal shift index, but indicates that the Disc.sh. index is critically sensitive to minor changes in the genomes of the mothers that determine the wing phenotype within a particular micro-population. This peculiarity can be used advantageously in the future as a criterion for differentiation between ecotypes, populations, and lines. Pol.2 standards, respectively, due to the significant increase in the coefficients of variation of the indices, although these bee colonies, identified in the first approximation according to the established classification, were indicated by Euclidean distances. To assess the breed affiliation of the wing phenotypes of the indicated groups of bee colonies, some of them were analyzed using Mahalanobis distances and the created standard data (Tables 12, 13). Based on the empirical data, accumulated by the authors, and the analysis of the results, presented in Table 12, approximate boundary values of similarity are used, namely: "0-2"high similarity; "2-3.5"significant similarity; "> 3.5"insignificant (or absent) similarity. Four out of five clusters of wings that were tested can be considered significantly similar to the comparison standards we created, except for cluster 3 of bee colony No: 19, for which the Mahalanobis distance is 3.971>3.5.
The data, given in Table 13, allow to obtain information about a more detailed structure of wing phenotypes for each individual cluster. The wings of cluster 1, family No: 10, which are very similar to the Hybrid.Pol.1 standard (Table 12), are distributed somewhat unexpectedly between the standards: Hybrid.Pol.1/Polish = 43.8 %/56.3 %, which actually indicates a greater similarity of the entire cluster to the Polish standard (Table 13). This fact correlates with the significant ambiguity in the distribution of wings between clusters 1 and 2 (Table 3) and, as a result, unsatisfactory accuracy of classifying the wings of cluster 1 (87.9 %). For the other considered bee colonies, No: 27, 19, 22, and 14, the majority of wings are assigned to the standards, determined in Table 12. The presence of a certain portion of wings, assigned to other standards, can be considered as a possible degree of hybridization with a certain breed. For example, for bee colony No. 27, the ratio of wings in cluster 1, Hybrid.Pol.1/Hybrid.Maced = 72/16, which may indicate a slight increase in the influence of the subspecies A. m. macedonica to a phenotype that most likely belongs to the subspecies A. m. mellifera.
A comparison of the average values of the cubital index C of cluster 3 of the classification model (2.04) and the standard Hybrid.Maced (2.096) indicates that the initial assumption that the bees of this cluster belong to the partially hybridized subspecies A. m. macedonica is correct. For 6 bee colonies out of 13 that form the Hy-brid.Maced standard, the value of C>2.14, which corresponds to the typical values for Ukrainian steppe bee populations (Table 10), for the other seven -C<2.076. That is, in a real attempt to classify the wings of "purebred" Ukrainian steppe bees using this classifier, they will be assigned to cluster 3. It should be understood, that this fact does not at all indicate their hybridization. The type and degree of hybridization determine the specific values of the phenotypes and their comparison with the reference data.
In summary, based on the analysis of data, presented in Tables 8-12, it can be concluded, that purebred bees of the Polish population of the subspecies A. m. Mellifera by phenotype, were found only at one apiary location, Stolyar's in the village of Hlybochok, Zhytomyr district. This result provides optimism for the next stages of work, aimed at the selective consolidation of necessary economically beneficial traits (EBTs) in the established micro-population of Polissya bees, namely: forming groups of analogues, evaluating the values of EBTs, and selecting breeding material.
Hybrids of Polish bees of the subspecies A. m. mellifera are present in all other studied apiaries without exception, which correlates with the data [4]. However, we cannot confirm or deny the assumption of the influ-ence of bees of the evolutionary lineage O on the popula-