Residual Data Reveals Greater Efficacy than Procrustes Data in Differentiation and Relationships Between the Populations of Apis florea

Background: Honeybees have an important role in the pollination of plants.Apis florea Fabricius, 1787, has been described over the last two centuries.A. florea is distributed in Vietnam, southeastern China, India, Nepal, southern Thailand, Sri Lanka, Sudan, southern Iran, Pakistan, Saudi Arabia, and Oman.The climatic zones of A. florea change from tropical rainforest in the east to savannah in the west.Furthermore, westwards, the climatic zones change from tropical to subtropical steppe and semi-desert.Morphometric data were used in addition to analyses of molecular data to study the honeybees.Morphometric analyses are flexible tools employed by users in different areas of the world.The geometric morphometric method uses landmarks that can be located precisely on the vein junctions of wings.The geometric morphometric method has been used to compare the populations of A. florea.In addition, traditional or standard morphometric methods have been used to study the populations of A. florea in various areas of the world.The landmark configurations eliminate the effects of position, orientation, and size of shapes.Data on shapes are converted into Procrustes coordinates using the geometric morphometric method.Regression analysis between centroid size and Procrustes coordinates creates new data on the residuals, which can be applied in other analyses.No published research has utilized residual coordinates (residual data) to compare the populations of A. florea.Therefore, the current research aimed to compare the performance efficacy of residual and Procrustes coordinates in differentiation and relationships among the populations of A. florea in various areas of the world.Methods: Samples of the A. florea honeybee species were prepared from their distributed areas.The right forewings of honeybees were used to study the relationships among the various populations.Eighty samples were selected from each area.A tps file was built by importing the right forewing pictures into TpsUtil V. 1.64 software.Then, the tps file was loaded into tpsDig V. 2.18.Twenty landmarks were digitized in the vein junctions of the forewings.This was followed by the raw data obtained from landmarks, loaded into the MorphoJ software V. 1.06d, and converted into Procrustes coordinates for use in future analyses, followed by analyzing the Procrustes coordinates.Mahalanobis distances and canonical variates were obtained using permutation tests.The regression between Procrustes data and centroid sizes was calculated, and an allometry test was performed afterward.The residual coordinates (residual data) were obtained after the removal of the size effect (size correction) from shape variables.The residual and Procrustes data were imported into PAST software v.3.19, and the populations of A. florea were compared using Canonical Variate Analysis (CVA).Moreover, clusters were drawn with the residual and Procrustes data using SAS v.8 software.Results: Multivariate (MANOVA) and pair-wise analyses of residual and Procrustes data were tested for the populations of A. florea.The Procrustes and residual data of the populations showed statistically significant differences using MANOVA (p < 0.001).Moreover, the tested pairwise comparisons indicated that all populations were significantly different in Procrustes and residual data (p < 0.001).The landmark configurations of forewings were superimposed, and variations were obtained between the populations.The highest variation was found in the vein junction of R and Rs, landmark 19 (S2 = 0.0000622).The lowest variation was observed in the vein junction of Cu and 1m-cu, landmark 8 (S2 = 0.0000109).The populations of various areas were compared using Canonical Variate Analysis (CVA).In the CVA of Procrustes data (Procrustes coordinates), the first and second components included 75.94% of all variation (CV1 = 28.74% and CV2 = 47.20%).In addition, in the CVA of residual data, the first and second components included 83.06