taxonID	type	format	identifier	references	title	description	created	creator	contributor	publisher	audience	source	license	rightsHolder	datasetID
0398BA60F26CFFE848107DF2FA39CB3C.taxon	http://purl.org/dc/dcmitype/StillImage	image/png	https://zenodo.org/record/14794246/files/figure.png	https://doi.org/10.5281/zenodo.14794246	Figure 2. Analysis 1: detection rate of morphological characters (y-axis) versus number of characters included in multinomial logistic regression models (x-axis) in our dataset. A, females. B, males. C, entire dataset.Red dots represent medians; blue boxes represent the 95% detection rate range; upper row of boxes represents true labelling, with all observed detection ranges indicated as blank whiskers; and lower row of boxes represents random labelling, with all observed detection ranges indicated as solid whiskers.Character coding is as in the Materials and Methods.	Figure 2. Analysis 1: detection rate of morphological characters (y-axis) versus number of characters included in multinomial logistic regression models (x-axis) in our dataset. A, females. B, males. C, entire dataset.Red dots represent medians; blue boxes represent the 95% detection rate range; upper row of boxes represents true labelling, with all observed detection ranges indicated as blank whiskers; and lower row of boxes represents random labelling, with all observed detection ranges indicated as solid whiskers.Character coding is as in the Materials and Methods.	2024-03-18	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.		Zenodo	biologists	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.			
0398BA60F26CFFE848107DF2FA39CB3C.taxon	http://purl.org/dc/dcmitype/StillImage	image/png	https://zenodo.org/record/14794248/files/figure.png	https://doi.org/10.5281/zenodo.14794248	Figure 3. Analysis 2: ability of multinomial logistic regression (MLR) models to generalize (ability to classify new specimens): detection rate of morphological characters (y-axis) versus number of characters included in MLR models (x-axis). A, females.B, males. C, entire dataset. Red dots represent medians; blue boxes represent 95% confidence intervals, with all observed detection ranges indicated as solid whiskers; and grey lines indicate the average detection rates of individual molecular clades.Character coding is as in the Materials and Methods.	Figure 3. Analysis 2: ability of multinomial logistic regression (MLR) models to generalize (ability to classify new specimens): detection rate of morphological characters (y-axis) versus number of characters included in MLR models (x-axis). A, females.B, males. C, entire dataset. Red dots represent medians; blue boxes represent 95% confidence intervals, with all observed detection ranges indicated as solid whiskers; and grey lines indicate the average detection rates of individual molecular clades.Character coding is as in the Materials and Methods.	2024-03-18	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.		Zenodo	biologists	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.			
0398BA60F26CFFE848107DF2FA39CB3C.taxon	http://purl.org/dc/dcmitype/StillImage	image/png	https://zenodo.org/record/14794250/files/figure.png	https://doi.org/10.5281/zenodo.14794250	Figure 4. Analysis 3: detection power of morphological characters in genetic lineages of Hansarsia. A, a relative detection power heatmap of individual morphological characters (x-axis; see coding of characters in Supporting Information, Table S1) versus the number of characters included in random sets (in addition to individual characters; y-axis) in females. B, same in males. C, average detection rates (solid lines) and 95% confidence interval (dotted lines) of the 10 strongest (red) and 10 weakest (blue) character sets selected from analysis 2 in females. D, same in males.The relative detection power is defined as a standardized mean difference (difference of means of two subset distributions divided by the standard deviation of the whole set) of detection rate distributions for character subsets that contain a particular character and character subsets that do not contain the same character.Character coding is as in the Materials and Methods.	Figure 4. Analysis 3: detection power of morphological characters in genetic lineages of Hansarsia. A, a relative detection power heatmap of individual morphological characters (x-axis; see coding of characters in Supporting Information, Table S1) versus the number of characters included in random sets (in addition to individual characters; y-axis) in females. B, same in males. C, average detection rates (solid lines) and 95% confidence interval (dotted lines) of the 10 strongest (red) and 10 weakest (blue) character sets selected from analysis 2 in females. D, same in males.The relative detection power is defined as a standardized mean difference (difference of means of two subset distributions divided by the standard deviation of the whole set) of detection rate distributions for character subsets that contain a particular character and character subsets that do not contain the same character.Character coding is as in the Materials and Methods.	2024-03-18	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.		Zenodo	biologists	Vemeshchaka, Alexandem L.;Shatmavin, Alexandem V.;Kulagin, Dmitmy N.;Lunina, Anastasiia A.			
