Measures of reproductive allometry are sensitive to sampling bias

X. Bonnet*, R. Shine, O. Lourdais, G. Naulleau

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

25 Citations (Scopus)


1. In both interspecific and intraspecific comparisons, maternal body size is the strongest predictor of fecundity for many kinds of animals. However, it has not been widely appreciated that the usual empirical descriptors of this relationship (correlation coefficient, slope and P-value of the linear regression between maternal body size and offspring number) are sensitive to a factor that is very labile and subject to methodological bias: the degree of maternal investment, specifically the ratio of litter mass to maternal body mass (relative clutch mass, RCM). 

2. Samples of females used to assess reproductive allometry may often be biased with respect to RCM. For example, RCMs may vary through time within a single population as a consequence of prey availability, or may vary geographically among populations. Also, females with low RCMs may be more difficult to capture, or may be discarded by researchers who do not realize that they are reproductive. 

3. Our analyses on 173 litters of aspic vipers (Vipera aspis Linné) from central western France show that estimates of reproductive allometry are very sensitive to RCM: samples composed of high-RCM females show a positive and highly significant reproductive allometry, whereas samples composed of low RCM females do not. Conclusions also depend on the method of regression analysis used. This result has strong implications for methodology (i.e. selection of samples, choice of analytical methods and timescales of study) in this field of research.

Original languageEnglish
Pages (from-to)39-49
Number of pages11
JournalFunctional Ecology
Issue number1
Publication statusPublished - Feb 2003
Externally publishedYes


  • Reproductive effort
  • Reptiles
  • Snakes


Dive into the research topics of 'Measures of reproductive allometry are sensitive to sampling bias'. Together they form a unique fingerprint.

Cite this