Print

Farrell, Catherine Cassidy

Biology

2008

MS

We found published exponents were not correlated with sample size, average mass, or log (difference in mass). For mammals and birds, the allometric exponents were tightly clustered with means of 0.72 and 0.73 respectively. The reptile data spanned a wider range, but had a mean of 0.74. Likelihood analysis suggests that b = 3/4 is significantly more probable than b = 2/3. We built a linear regression simulation with experimental error in mass and showed that such measurement error systematically lowers estimates of the allometric exponent. Measurement error probably contributes to the observation that published allometric exponents often fall short of b= 3/4 as predicted by theoretical models.