Skip to main content
. 2020 Feb 27;15(2):e0229689. doi: 10.1371/journal.pone.0229689

Table 3. Freshwater turtle life-history model comparisons.

Comparisons of the Generalized Additive Models created for each life-history trait to estimate capacity for sustainable harvest in freshwater turtles. Models for each trait ordered by decreasing AICc (Akaike information criterion corrected for small sample sizes) values.

Life-history trait Modela Dev. Exp Loglik BIC AICc Δ AICc Wi AICcb
Clutch size
Continuous latitude 77.6 -69.35 207.05 168.09 0.00 1.00
Categorical latitude 75.8 -80.04 237.57 193.80 25.71 0.00
Bioclimate - rain 69.8 -95.48 256.12 218.85 50.77 0.00
Bioclimate - temp 70.5 -98.96 261.27 224.98 56.90 0.00
Clutch frequency
Continuous latitude 17.9 -14.42 81.83 54.94 0.00 0.93
Bioclimate - temp 9.9 -19.25 85.31 61.22 6.28 0.04
Bioclimate - rain 7.0 -21.01 83.69 62.00 9.79 0.03
Categorical latitude 9.3 -19.61 91.34 64.84 12.62 0.01
Age at sexual maturity
Continuous latitude 20.2 -28.82 79.24 68.52 0.00 0.57
Bioclimate - temp 37.7 -19.46 92.49 69.13 0.62 0.42
Bioclimate - rain 31.6 -22.86 99.50 76.08 7.56 0.01
Categorical latitude 32.2 -22.62 104.54 79.42 10.91 0.00
Adult survival
Bioclimate - temp 16.7 14.16 -10.39 -16.31 0.00 0.85
Continuous latitude 6.7 12.09 -6.27 -12.19 4.12 0.11
Bioclimate - rain 1.4 11.09 -4.26 -10.18 6.13 0.04
Categorical latitude 23.8 15.77 7.28 0.73 17.04 0.00

a Models used to predict natural history traits. Each model contained Family as a random effect (smooth term with “re” basis) and body size (log transformed carapace length) as a parametric (not smooth) effect. Continuous latitude included median latitude from all records (S1 Table). Categorical latitude included four latitudinal classes: Temperate (species with latitudinal median and range within temperate zone), “Temp-Trop” (species with latitudinal median within temperate and range overlapping tropical zone/s), “Trop-Temp” (species with latitudinal median within tropical and range overlapping temperate zone/s), Tropical (species with latitudinal median and range within tropical zone). Bioclimate—temp included Mean Temperature of Warmest Quarter (WorldClim: bio10). Bioclimate—rain included Precipitation of Driest Quarter (WorldClim: bio17). Coefficients for individual variables in all models are presented in Table 2.

b Akaike weights (Wi) from largest to smallest. Predictions for each trait were obtained using variables from the 95% confidence subset of models, obtained by first ordering all models in the set by decreasing Akaike weight (Wi), and then sequentially summing the model Wi's in rank order.