3. Identifying the Ecological Correlates of Extinction-Prone Species
33
Table 3.1. Logit regression.
Variable name
Coefficient
Standard error
t ratio
WEAK
5.5424
1.7177
3.2267
CHATHAM
6.4516
2.3182
2.7830
KG
1.3129
0.60606
2.1662
TROPH1
4.8906
1.9156
2.5531
TROPH4
2.6639
1.5320
1.7389
ENDEMIC
2.3008
1.0831
2.1243
Constant
−6.5554
2.2707
−2.8870
Maddala R
2
55.93%
Cragg-Uhler R
2
74.966%
Chow R
2
68.223%
Prediction success rate
91.25%
closely associated with vulnerability to mammalian predators included both size
and flying ability. In the model below, all birds that were either weak fliers or
flightless and that weighed less than 2 kg were classed as WEAK. A more
appropriate measure of the ability to fly would be wing morphometrics and their
relation to body size. However, these data were not available. The restriction to
small birds was necessary to avoid multicollinearity problems with body mass and
flight ability.
Finally, species endemic to restricted areas in New Zealand (ENDEMIC) may
be more vulnerable to the impacts of human settlement than indigenous species
whose natural ranges cover a variety of landscapes and threats.
In all but one instance, a strong statistical relationship was observed between
the independent variables and the dependent variable (current extinction status;
Tables 3.1 and 3.2). Aside from TROPH4 in the probit model, all variables were
statistically significant at a 95% level of confidence. Hence this particular approach has at least successfully identified some of the correlates of extinction risk
in New Zealand birds. Given that the maximum likelihood estimation has a
Table 3.2. Probit regression.
Variable name
Coefficient
Standard error
t ratio
WEAK
2.8575
0.74952
3.8125
CHATHAM
2.8741
0.79795
3.6018
KG
0.65481
0.30414
2.1530
TROPH1
2.2702
0.78822
2.8801
TROPH4
0.99562
0.60489
1.6460
ENDEMIC
1.0092
0.48865
2.0652
Constant
−3.0417
0.82094
−3.7052
Maddala R
2
55.33%
Cragg-Uhler R
2
74.166%
Chow R
2
64.351%
Prediction success rate
87.5%
33
Table 3.1. Logit regression.
Variable name
Coefficient
Standard error
t ratio
WEAK
5.5424
1.7177
3.2267
CHATHAM
6.4516
2.3182
2.7830
KG
1.3129
0.60606
2.1662
TROPH1
4.8906
1.9156
2.5531
TROPH4
2.6639
1.5320
1.7389
ENDEMIC
2.3008
1.0831
2.1243
Constant
−6.5554
2.2707
−2.8870
Maddala R
2
55.93%
Cragg-Uhler R
2
74.966%
Chow R
2
68.223%
Prediction success rate
91.25%
closely associated with vulnerability to mammalian predators included both size
and flying ability. In the model below, all birds that were either weak fliers or
flightless and that weighed less than 2 kg were classed as WEAK. A more
appropriate measure of the ability to fly would be wing morphometrics and their
relation to body size. However, these data were not available. The restriction to
small birds was necessary to avoid multicollinearity problems with body mass and
flight ability.
Finally, species endemic to restricted areas in New Zealand (ENDEMIC) may
be more vulnerable to the impacts of human settlement than indigenous species
whose natural ranges cover a variety of landscapes and threats.
In all but one instance, a strong statistical relationship was observed between
the independent variables and the dependent variable (current extinction status;
Tables 3.1 and 3.2). Aside from TROPH4 in the probit model, all variables were
statistically significant at a 95% level of confidence. Hence this particular approach has at least successfully identified some of the correlates of extinction risk
in New Zealand birds. Given that the maximum likelihood estimation has a
Table 3.2. Probit regression.
Variable name
Coefficient
Standard error
t ratio
WEAK
2.8575
0.74952
3.8125
CHATHAM
2.8741
0.79795
3.6018
KG
0.65481
0.30414
2.1530
TROPH1
2.2702
0.78822
2.8801
TROPH4
0.99562
0.60489
1.6460
ENDEMIC
1.0092
0.48865
2.0652
Constant
−3.0417
0.82094
−3.7052
Maddala R
2
55.33%
Cragg-Uhler R
2
74.166%
Chow R
2
64.351%
Prediction success rate
87.5%