Estimation of Individual Growth Trajectories When Repeated Measures Are Missing

Area of Science:

• Ecology
• Population Biology
• Quantitative Biology

Background:

• Individual variation in growth is influenced by genetics, environment, and health, impacting population dynamics.
• Estimating growth in wild populations is challenging due to missing data and observation errors.
• Linear mixed models (LMMs) can underestimate individual differences when data is incomplete.

Purpose of the Study:

• To develop a flexible and robust method for modeling animal growth trajectories.
• To compare the performance of state-space models (SSMs) against LMMs for growth estimation with missing data.
• To apply the developed method to quantify growth variation in a wild Soay sheep population.

Main Methods:

• Simulated datasets with varying levels of missing repeated measures and observation error.
• Fitted state-space models (SSMs) using the R package 'growmod'.
• Compared SSMs with linear mixed models (LMMs) for bias and computational efficiency.

Main Results:

• SSMs showed significantly less bias than LMMs when substantial data was missing (up to 87.5%).
• The SSM approach was computationally faster than traditional Markov chain Monte Carlo methods.
• Growth in Soay sheep decreased with age, population density, adverse weather, and reproductive status.

Conclusions:

• SSMs provide a reliable and efficient method for estimating individual growth trajectories in wild populations, even with imperfect data.
• This approach enhances the ability to link individual attributes and environmental factors to growth variation.
• The findings have significant implications for understanding population dynamics and wildlife management.