Sample size considerations for species co-occurrence models.

Area of Science:

• Ecology
• Statistical Modeling

Background:

• Multispecies occupancy models are crucial for understanding species interactions.
• Convergence and estimation issues commonly arise with limited sample sizes.

Purpose of the Study:

• To evaluate a new model's ability to estimate co-occurrence under various sample sizes and interaction strengths.
• To assess the impact of model complexity (number of species and covariates) on estimation accuracy.

Main Methods:

• Simulation study examining a recently developed multispecies occupancy model.
• Utilized both standard and penalized likelihood approaches.
• Varied sample size, detection probability, interaction strength, and model complexity.

Main Results:

• Model performance is highly sensitive to sample size, detection probability, and interaction strength.
• High bias in co-occurrence estimates occurs with <100 sites (high detection) or 400-1000 sites (low detection).
• Strong co-occurrence is detectable with >200 sites (high detection), but weak co-occurrence is rarely detected even with large datasets.

Conclusions:

• Reliable inference of species co-occurrence necessitates significantly larger datasets than currently common.
• Occupancy patterns are generally robust to sample size, but co-occurrence inference is not.
• The model can quantify strong co-occurrence and predict occupancy in larger datasets, but caution is advised for small datasets or weak interactions.