Skip to main content
. 2021 Apr 5;9:e11187. doi: 10.7717/peerj.11187

Table 2. Model comparisons for the distance-duration relationships.

Comparison of non-linear, segmented linear and simple linear models of the relationships between feeder distance (d, in km) and the duration of the waggle phase (tw), the return phase (tr) and the whole dance circuit (tc) (each in s) based on the data obtained in this study. The models are ordered according to their ΔAICc values (the “best” models are highlighted in bold).

Dance component Model Is is biased?§ Formula r# ΔAICc
Waggle phase non-linear no tw=0.1993+2.00180.6717(1e0.6717d) 0.946 0
segmented no tw={0.2917+1.4282d,d1.03281.0767+0.6683d,d>1.0328 0.947 1.5
linear yes tw=0.4475+1.1152d 0.935 7.7
Return phase linear no tr=1.3712+0.5238d 0.582 0
non-linear no tr=1.2686+0.88760.6035(1e0.6035d) 0.587 1.9
segmented no tr={1.2303+0.8613d,d0.71.5725+0.3725d,d>0.7 0.596 3.4
Dance circuit non-linear no tc=1.467+2.8930.6519(1e0.6519d) 0.904 0
segmented no tc={1.4926+2.4201d,d0.72.2844+1.2889d,d>0.7 0.906 1.4
linear yes tc=1.8187+1.6390d 0.894 2.6

Notes:

§

Does the model provide biased predictions, i. e. do the predictions systematically deviate from the actual observations at any point over the range of feeder distances? (Decision aided by the inspection of plots of fitted values versus residuals)

#

Pearson’s r of a correlation between fitted and observed values, a measure of how well the models fit the data