Skip to main content

Table 1 Results of morphometric analyses of the Pallenopsis patagonica species complex for both data sets (absolute and relative values). Contributions to correctness rate (CR), characters combination for the best LDA performance and p-values for significant differences between geographic regions and sexes are listed

From: Combining morphological and genomic evidence to resolve species diversity and study speciation processes of the Pallenopsis patagonica (Pycnogonida) species complex

Character Absolute Relative
Times the character contributed to an LDA model during character selectiona Increase in correctness rate (mean ± SD)b Character combination for best LDA performance Differences between geographic regions * Differences between sexes * Times the character contributed to an LDA model during character selectiona Increase in correctness rate (mean ± SD)b Character combination for best LDA performance Differences between geographic regions * Differences between sexes *
abdomen W 34 (2.4%) 0.05 ± 0.03   <  0.001 19 (1.23%) 0.09 ± 0.04  
abdomen L 27 (1.9%) 0.21 ± 0.15   <  0.001 2 (0.13%) 0.09 ± 0.01  
eye H 115 (8.11%) 0.22 ± 0.11 x <  0.001 15 (0.97%) 0.1 ± 0.06  
eyes distance 6 (0.42%) 0.06 ± 0.03   22 (1.43%) 0.08 ± 0.03  
ocular tubercle W 15 (1.06%) 0.05 ± 0.03 x 0.004 24 (1.56%) 0.08 ± 0.05  
ocular tubercle H 147 (10.37%) 0.1 ± 0.07 x 102 (6.63%) 0.12 ± 0.05 x 0.005 0.012
ceph. segment 36 (2.54%) 0.09 ± 0.08   <  0.001 24 (1.56%) 0.11 ± 0.06   0.012
cheliphore 1 136 (9.59%) 0.2 ± 0.13 x <  0.001 2 (0.13%) 0.11 ± 0.01   0.048
cheliphore 2 NA NA NA NA NA 5 (0.32%) 0.09 ± 0.03 x
cheliphore 3 5 (0.35%) 0.05 ± 0.02   <  0.001 1 (0.06%) 0.1   0.027
cheliphore 4 8 (0.56%) 0.06 ± 0.03   0.001 23 (1.49%) 0.09 ± 0.04  
palp 41 (2.89%) 0.11 ± 0.1   0.001 198 (12.87%) 0.16 ± 0.08 x
proboscis thick2tip 136 (9.59%) 0.11 ± 0.06   83 (5.39%) 0.15 ± 0.08  
proboscis basis 23 (1.62%) 0.11 ± 0.07   <  0.001 0.027 6 (0.39%) 0.06 ± 0.05  
proboscis thickest 29 (2.05%) 0.17 ± 0.14   <  0.001 7 (0.45%) 0.07 ± 0.05   0.001
proboscis L 2 (0.14%) 0.02 ± 0.01   <  0.001 13 (0.84%) 0.06 ± 0.03   <  0.001
trunk W1 90 (6.35%) 0.11 ± 0.07   <  0.001 41 (2.66%) 0.11 ± 0.05   0.004 0.022
trunk W12 30 (2.12%) 0.09 ± 0.06   0.048 168 (10.92%) 0.17 ± 0.09 x <  0.001
trunk W2 8 (0.56%) 0.07 ± 0.05   0.001 18 (1.17%) 0.07 ± 0.05  
trunk W23 55 (3.88%) 0.08 ± 0.05   81 (5.26%) 0.12 ± 0.06   <  0.001
trunk W3 98 (6.91%) 0.1 ± 0.07   <  0.001 61 (3.96%) 0.1 ± 0.06   0.034
trunk W34 15 (1.06%) 0.06 ± 0.03   73 (4.74%) 0.11 ± 0.05   <  0.001 0.042
trunk W4 5 (0.35%) 0.03 ± 0   <  0.001 23 (1.49%) 0.15 ± 0.09   0.037
trunk H 1 (0.07%) 0.03   20 (1.3%) 0.1 ± 0.04  
trunk L 48 (3.39%) 0.11 ± 0.11   <  0.001 NA NA   NA NA
forehead H 69 (4.87%) 0.07 ± 0.04 x 0.029 45 (2.92%) 0.09 ± 0.04  
WL1 coxa1 17 (1.2%) 0.16 ± 0.08   <  0.001 45 (2.92%) 0.1 ± 0.05  
WL1 coxa2 27 (1.9%) 0.1 ± 0.09   0.001 60 (3.9%) 0.12 ± 0.06   0.012 0.034
WL1 coxa3 4 (0.28%) 0.07 ± 0.03   0.002 3 (0.19%) 0.09 ± 0.05  
WL1 femur 7 (0.49%) 0.04 ± 0.02   <  0.001 0.009 NA NA   NA NA
WL2 coxa1 NA NA   NA NA 15 (0.97%) 0.06 ± 0.03  
WL2 coxa2 20 (1.41%) 0.17 ± 0.12   0.001 66 (4.29%) 0.17 ± 0.11   0.041 0.016
WL2 coxa3 NA NA   NA NA 4 (0.26%) 0.08 ± 0.04  
WL3 coxa1 7 (0.49%) 0.11 ± 0.05   0.001 21 (1.36%) 0.09 ± 0.04   0.023
WL3 coxa3 12 (0.85%) 0.2 ± 0.1   <  0.001 0.008 19 (1.23%) 0.08 ± 0.05   <  0.001
WL4 coxa1 5 (0.35%) 0.18 ± 0.13   <  0.001 1 (0.06%) 0.07  
WL4 coxa2 59 (4.16%) 0.19 ± 0.14   <  0.001 160 (10.4%) 0.16 ± 0.1 x 0.042
WL4 coxa3 2 (0.14%) 0.07 ± 0   <  0.001 0.038 3 (0.19%) 0.08 ± 0.04  
WL4 propodus 8 (0.56%) 0.14 ± 0.16   <  0.001 22 (1.43%) 0.09 ± 0.04  
WL4 tarsus 11 (0.78%) 0.1 ± 0.06   <  0.001 0.025 2 (0.13%) 0.08 ± 0.06  
WL4 tibia2 60 (4.23%) 0.2 ± 0.13   <  0.001 42 (2.73%) 0.11 ± 0.06  
  1. * p-values are only listed for analyses that showed significant differences. aNumber of times the charcter was added and its addition led to a positive increase in cross-validation correctness rate of individual LDA models during repeated character selections (% of total in parentheses). See how the repetitions were organized in Materials and Methods. bAverage increase in cross-validation correctness rate after addition of the character to an LDA model had a positive effect in the character selections