Table 6 Best abundance model results

From: Seed dispersal by Martu peoples promotes the distribution of native plants in arid Australia

 

Diversiflorum

Centrale

Eragrostis

Scaevola

Site distance

0.21 *** [0.11, 0.41]

 

1.63 [0.90, 2.96]

2.05 * [1.05, 4.00]

Water permanence

4.59 *** [2.88, 7.32]

 

0.84 [0.53, 1.32]

 

Site dist × Water perm

  

4.54 *** [1.99, 10.38]

 

Site type Major

   

4.12 ** [1.75, 9.66]

Season Winter

 

0.24 * [0.07, 0.79]

2.09 *** [1.48, 2.95]

 

Fire frequency

 

0.18 ** [0.06, 0.52]

1.40 [1.00, 1.97]

 

Season × TSF

 

0.00 * [0.00, 0.69]

  

Season × Fire frequency

 

6.30 * [1.21, 32.88]

  

TSF

 

32.15 [0.15, 7072.10]

0.01 *** [0.00, 0.05]

0.07 ** [0.01, 0.49]

Soil Carbon

0.50 ** [0.33, 0.78]

 

0.31 *** [0.19, 0.49]

0.21 * [0.06, 0.70]

PCT Sand

0.41 ** [0.21, 0.79]

 

1.31 [0.96, 1.80]

 

PCT Veg cover

  

1.92 *** [1.34, 2.75]

 

NDMI

  

1.60 * [1.11, 2.31]

 

TSF × Site distance

  

9.04 * [1.44, 56.76]

 

N (observations)

109

70

617

177

R2 Fixed effects

37.81%

25.60%

27.50%

33.90%

R2 Fixed + Random

82.75%

92.68%

98.70%

99.90%

  1. Model coefficients (GLM, Poisson, observation level random effects) as the standardized odds ratio (2 sd) for the best presence models (±95% CI). Odds ratios give the relative change in the likelihood of presence for a two-standard deviation increase in the predictor variable. Odds ratios higher than 1 indicate a positive relationship (the predictor variable increases the likelihood of presence); those less than 1 indicate a negative relationship (the predictor variable decreases the likelihood of presence). Interaction (crossed) terms in the model are represented by “X”. As is common in Poisson models, we account for overdispersion through the inclusion of individual-level random effects. Asterisks indicate significant p values (* = <0.05, ** = <0.01, *** = <0.001). Predicted counts for some variables and interactions are presented in Fig. 3.
Back to article page