Skip to main content

Table 8 Percent of correctly discriminated species of the subgenus Boophilus , using the sets of descriptive covariates

From: A global set of Fourier-transformed remotely sensed covariates for the description of abiotic niche in epidemiological studies of tick vector species

Records reported as

AUC

annulatus

australis

decoloratus

geigyi

microplus

1. Discriminant analysis with 7 coefficients of LST, 7 coefficients of NDVI, and 5 coefficients of LAI. Correct determinations: 82.4%

annulatus

0.955

69.94

0.00

2.76

25.15

2.15

australis

0.977

0.00

93.39

1.17

0.00

5.43

decoloratus

0.905

2.21

0.18

79.85

9.26

8.50

geigyi

0.986

1.41

0.00

1.41

97.18

0.00

microplus

0.924

2.53

1.86

16.39

1.31

77.91

2. Discriminant analysis with 7 coefficients of LST, and 7 coefficients of NDVI. Correct determinations: 72.9%

annulatus

0.959

46.79

0.00

3.21

48.08

1.92

australis

0.995

0.02

94.45

1.44

0.00

4.08

decoloratus

0.922

4.94

0.09

73.84

8.94

12.19

geigyi

0.989

3.47

0.00

1.39

93.75

1.39

microplus

0.946

5.36

1.33

12.06

0.73

80.52

3. Discriminant analysis with 12 months of remotely sensed LST and NDVI. Correct determinations: 62.3%

annulatus

0.931

32.69

1.28

4.49

57.05

4.49

australis

0.991

0.20

96.83

1.22

0.00

1.75

decoloratus

0.889

4.27

1.91

69.93

10.77

13.12

geigyi

0.979

6.25

0.69

0.00

92.36

0.69

microplus

0.919

4.68

5.93

16.69

2.62

70.08

4. Discriminant analysis with monthly remotely sensed LST and NDVI, after removal of months with high collinearity. Only values for January, March, May and October were included for LST. Data for February, March and July were removed from NDVI. Correct determinations: 56.7%

annulatus

0.912

34.62

1.28

0.64

57.05

6.41

australis

0.947

0.49

90.83

3.20

0.00

5.48

decoloratus

0.761

5.12

12.06

52.89

14.10

15.84

geigyi

0.971

9.72

1.39

0.69

88.19

0.00

microplus

0.826

8.23

15.12

18.87

2.46

55.32

5. Discriminant analysis with 12 months of gridded interpolated temperature and rainfall (Worldclim dataset). Correct determinations: 69.7%

annulatus

0.872

42.31

1.92

3.85

44.23

7.69

australis

0.996

0.00

99.67

0.13

0.00

0.20

decoloratus

0.923

4.23

1.33

78.29

7.96

8.19

geigyi

0.985

5.56

0.69

2.78

85.42

5.56

microplus

0.949

2.66

7.78

14.88

0.81

73.87

6. Discriminant analysis with monthly gridded interpolated temperature and rainfall (Worldclim dataset) after removal of the months with high collinearity. Only data of temperatures of January, April, June and September were included. Data of rainfall of February, August and December were removed. Correct determinations: 65.1%

annulatus

0.889

44.23

4.49

0.64

45.51

5.13

australis

0.982

0.00

97.63

2.35

0.00

0.02

decoloratus

0.851

8.41

3.65

72.15

10.63

5.16

geigyi

0.965

15.97

0.00

0.00

76.39

7.64

microplus

0.879

6.45

16.57

14.88

3.06

59.03

7. Discriminant analysis with “bioclim variables” derived from monthly gridded interpolated temperature and rainfall (Worldclim dataset). Correct determinations: 57.9%

annulatus

0.941

28.21

1.92

8.97

46.15

14.74

australis

0.990

0.00

97.45

0.00

0.00

2.55

decoloratus

0.876

3.91

2.58

74.64

9.34

9.52

geigyi

0.968

1.39

2.08

3.47

83.33

9.72

microplus

0.901

1.49

11.09

19.60

1.98

65.85

8. Discriminant analysis with “bioclim variables” derived from monthly gridded interpolated temperature and rainfall (Worldclim dataset) after removal of variables with high collinearity. Correct determinations: 57.4%

annulatus

0.930

29.49

1.92

7.69

53.21

7.69

australis

0.991

0.00

95.12

0.00

0.09

4.79

decoloratus

0.890

3.51

2.49

73.84

10.85

9.30

geigyi

0.979

1.39

0.69

1.39

86.81

9.72

microplus

0.920

2.30

15.52

19.68

5.85

56.65

  1. For some of these datasets of descriptive covariates, the analysis was repeated with every variable included (e.g., the 12 months of average temperature) and after the highly correlated variables were removed. A discriminant analysis was conducted, and its reliability evaluated by the percent of records correctly predicted and the area under the curve (AUC). The AUC is a general measure of model performance and does not consider individual results of true positives for each species. Therefore, some models may perform better for a particular species while having a general low AUC. The percent of correctly determined records of each species is also included.