Accuracy of occurrence and abundance estimates from insect metabarcoding

Area of Science:

• Ecology
• Genetics
• Bioinformatics

Background:

• DNA metabarcoding is crucial for insect biodiversity assessment using high-throughput sequencing of bulk samples.
• Methodological choices significantly impact the accuracy of metabarcoding data, yet remain under-explored.

Purpose of the Study:

• To assess how lysis methods (non-destructive vs. destructive homogenization) affect insect community recovery.
• To determine the comprehensiveness of metabarcoding in capturing species richness.
• To evaluate the effectiveness of spike-ins in improving abundance estimates and the overall accuracy of species abundance estimations.

Main Methods:

• Evaluated insect metabarcoding accuracy using 4,749 bulk samples from a biodiversity survey.
• Compared mild lysis with destructive homogenization treatments on 856 samples.
• Incorporated biological and synthetic spike-ins to calibrate abundance estimates.
• Validated occurrence and abundance estimates against 15 samples with individually barcoded specimens.

Main Results:

• Lysis methods showed taxon-specific biases: mild lysis favored small/soft-bodied insects, while homogenization favored larger/sclerotized ones.
• Biological spike-ins significantly reduced read number variance per specimen, especially in homogenized samples.
• Synthetic spike-ins were less effective than biological ones.
• Accurate species abundance estimates (+/-1 individual) were achieved for 72.9% of occurrences using Bayesian analysis with spike-in calibration.

Conclusions:

• Reasonably accurate species abundance estimates are achievable with DNA metabarcoding.
• Lysis methods introduce different taxon-specific biases, with homogenization generally improving accuracy.
• Biological spike-ins enhance accuracy more than synthetic ones, aiding robust biodiversity monitoring.