Table 2 Results from binomial generalized linear models predicting the presence of protected areas and the achievement of the 2030 targets (30% coverage on both terrestrial and marine areas) for the 2323 islands globally as a function of 16 factors (see Table 1 and “Methods”).

From: Global correlates of terrestrial and marine coverage by protected areas on islands

  

Presence of protection

30% of protection coverage

 

Terrestrial protected areas

Marine protected areas

Terrestrial protected areas

Marine protected areas

d.f.

AIC weight

F value

AIC weight

F value

AIC weight

F value

AIC weight

F value

Maximum temperature

2

1

11.8***

0.30

 

1

15.7***

1

39.8***

Minimum temperature

2

1

11.7***

0.99

10.8***

1

13.1***

1

11.5***

Maximum precipitation

2

1

9.7***

1

29.8***

0.98

5.8**

1

8.5***

Minimum precipitation

2

0.42

 

0.99

10.7***

0.75

3.0*

0.94

4.8**

Surface area

2

1

13.3***

0.99

19.6***

0.45

 

0.74

3.7*

Distance to the mainland

2

0.96

5.2**

1

18.4***

0.89

4.1*

0.21

 

Maximum altitude

2

0.61

0.6NS

0.40

 

0.96

4.0*

0.20

 

Mean altitude

2

0.45

 

0.48

 

0.82

2.6*

0.15

 

Fractal dimension

2

1

10.7***

1

16.4***

0.42

 

0.99

8.0***

Cropland area

2

0.49

 

0.26

 

0.63

2.1NS

0.14

 

Freshwater area

2

0.26

1.0

0.48

 

0.54

1.3NS

0.38

 

Population density

2

0.20

 

0.30

 

1

22.7***

1

22.8***

Number of languages

2

1

15.0***

1

27.2***

1

15.5***

1

54.0***

Institutional languages

2

1

8.3***

1

14.1***

1

10.5***

1

9.7***

Sovereignty

2

0.38

 

0.92

6.9**

0.57

2.1NS

0.29

 

Human development index

2

1

99.5***

1

113.8***

1

33.6***

1

35.0***

  1. d.f. is the degree of freedom for each factor, AIC weight represents the importance of each factor in the best models and the F value its influence on the predicted variable (NS, not significant, *p < 0.05, **p < 0.01, ***p < 0.001). Only factors retained in the most parsimonious models (according to a backward selection procedure based on AIC) are statistically tested.