Detecting the "invisible fraction" bias in resurrection experiments

Area of Science:

• Evolutionary biology
• Ecology
• Conservation biology

Background:

• The resurrection approach revives ancestral plant generations from dormant seeds to study evolution in response to environmental changes.
• Project Baseline collected seeds from 61 North American plant species for future resurrection experiments.

Purpose of the Study:

• To investigate the potential bias, termed the "invisible fraction" problem, in resurrection experiments due to non-random seed mortality during storage.
• To identify conditions under which this bias can significantly impact estimates of evolutionary change.

Main Methods:

• The study theoretically analyzes how non-random seed mortality, influenced by genetically correlated seed and postemergence traits, can bias ancestral phenotype estimates.
• It proposes using family structure within seed collections to diagnose "invisible fraction" bias by correlating family survival rates with trait means.

Main Results:

• Low seed survival combined with selective mortality and strong genetic correlations between seed and postemergence traits can lead to substantial bias.
• A correlation between family mean survival and family mean of a trait indicates non-random mortality and a biased sample mean.

Conclusions:

• The "invisible fraction" bias is a significant concern for resurrection studies with low seed survival.
• The family-structured sampling in Project Baseline allows for the detection of this bias, and future statistical methods may enable its correction.