How DNA Unlocks the Secrets of What Animals Really Eat
The hidden world of wildlife diets is finally revealing its secrets, one DNA sequence at a time.
Imagine trying to understand human culture by only studying what people eat at their largest Thanksgiving meal. For decades, ecologists faced a similar challenge when studying animal diets—until now. Dietary DNA metabarcoding has revolutionized our ability to see exactly what animals consume by identifying genetic traces of plants and prey in their feces. But this powerful tool comes with its own controversy: how should scientists separate true dietary signals from genetic background noise?
This article explores how standard approaches to filtering data may be dramatically altering our perception of nature's dinner table—and why a seemingly technical decision in the laboratory could reshape fundamental ecological theories.
Dietary DNA metabarcoding enables researchers to identify and characterize trophic interactions with unprecedented taxonomic precision. The process works by collecting fecal samples, extracting all DNA present, amplifying specific genetic barcodes, and using high-throughput sequencing to identify all species present 1 9 .
Think of it as reading nature's grocery list from a biological receipt left in animal scat.
This technique has revealed astonishing dietary details that traditional observation methods missed. For instance, it has documented previously unknown fish species in pinniped diets, identified jellyfish in white shark meals when morphological analysis found none, and detected fragile prey items that are completely digested beyond visual recognition 9 .
Early in its development, researchers noticed that sequencing errors, contamination, and technical artifacts often appeared as low-abundance sequences in their data 1 . An apparently simple solution emerged: remove all sequences below an arbitrary abundance threshold, typically around 1% of Relative Read Abundance (RRA), and base ecological analyses solely on the presence or absence of the remaining food taxa 1 4 .
This approach became so widespread that it's now often perceived as necessary—but evidence of its sufficiency is lacking, and the practice risks introducing more errors than it eliminates 1 .
The central problem with arbitrarily removing low-abundance sequences is that this process may erroneously eliminate true dietary items in ways that dramatically impact ecological interpretations 1 4 . Consider what might be lost:
Nutritionally important food sources that are consumed infrequently
Evidence of dietary breadth that defines generalist species
Seasonal or opportunistic foraging behaviors
Weak trophic links that may be crucial to ecosystem resilience
Computer simulations demonstrate that abundance-filtering differentially impacts dietary profiles based on how specialized or generalized an animal's diet actually is 1 . When researchers applied different RRA thresholds (0%-5%) to simulated dietary data, they found that removing low-abundance sequences immediately caused a 75-80% loss of dietary richness in specialized feeders 1 .
The most common filtering strategies essentially make it impossible to distinguish between true dietary specialists and generalists that consume many foods in small quantities 1 . As one research team concluded, "Although the practice of removing low-abundance sequences may continue to be a useful strategy to address research questions that focus on a subset of relatively abundant foods, its continued widespread use risks generating misleading perceptions about the structure of trophic networks" 1 .
To understand how this methodological decision plays out in real-world ecology, let's examine groundbreaking research conducted in Yellowstone National Park 1 8 .
Researchers collected fecal samples from bison and bighorn sheep during both winter and summer seasons 8
Using the P6 loop of the chloroplast trnL(UAA) intron—a genetic region ideal for identifying plants—they sequenced DNA from 35 samples 8
After basic quality control, the team applied a series of relative read abundance thresholds from 0% to 5% to the same dataset 1 8
They examined how each threshold affected measurements of dietary richness within and between species, across seasons, and at population levels 8
The Yellowstone experiment yielded striking demonstrations of how arbitrary thresholds alter ecological interpretation. The tables below summarize key findings from this research.
| RRA Threshold | Impact on Richness |
|---|---|
| 0% (no filtering) | Maximum richness detected |
| 1% | 75-80% richness loss in specialists |
| 3% | Further richness reduction |
| 5% | Only most abundant items remain |
| Research Finding | 0% vs 1% Threshold |
|---|---|
| Dietary overlap | Moderate vs minimal overlap |
| Seasonal shifts | Gradual vs abrupt changes |
| Individual variation | Considerable vs reduced variation |
The data showed that these thresholds didn't just subtly adjust numbers—they fundamentally altered apparent feeding strategies and ecological relationships 1 . As the researchers noted, these filtering strategies "can markedly alter the composition of dietary profiles in ways that scale-up to obscure ecological interpretations about dietary generalism, specialism, and composition" 1 .
Filtering at 1% RRA threshold caused immediate 75-80% loss of dietary richness in specialized feeders, dramatically altering ecological interpretation.
What does it take to conduct dietary DNA metabarcoding research? Here's a look at the essential tools and reagents:
| Research Tool | Function | Importance |
|---|---|---|
| Chloroplast trnL(UAA) intron P6 loop | Plant identification barcode | Short, variable genetic region ideal for degraded DNA in feces 8 |
| Blocking primers | Inhibit host DNA amplification | Increase detection of prey DNA; concentrations must be optimized 3 |
| DNeasy PowerWater Sterivex Kit | DNA extraction from filters | Efficiently extracts DNA from environmental samples 7 |
| ZBJ-ArtF1c/ZBJ-ArtR2c primers | Invertebrate DNA amplification | Targets 157bp COI fragment for prey identification 7 |
| mlCOIintF-XT/jgHCO2198 primers | Alternative invertebrate markers | Targets 313bp COI fragment for broader taxonomic coverage 7 |
| Illumina MiSeq platform | High-throughput sequencing | Generates thousands of sequences per sample simultaneously 1 7 |
If arbitrary filtering causes such problems, what should researchers do instead? The scientific community is developing more sophisticated approaches:
Solutions that distinguish errors from true signals based on sequence patterns rather than just abundance 1
Designs that include controls and mock communities to quantify error rates directly 3
Approaches that account for technical artifacts without completely removing rare sequences 1
As we move beyond simplistic filtering, dietary metabarcoding continues to reveal astonishing ecological insights. The technique has been successfully applied to everything from Alpine salamanders (revealing an extraordinary 177 prey taxa) 7 to sea turtles 9 and even human diets through the innovative "FoodSeq" approach 5 .
The precautionary principle in dietary DNA metabarcoding reminds us that seemingly minor methodological decisions can have profound consequences for how we understand nature. What we initially dismissed as "noise" might actually be the whisper of ecologically important but rare dietary items—the nutritional equivalent of crucial but overlooked ingredients in a complex recipe.
As we refine these molecular tools, we're not just improving technical methods—we're fundamentally rethinking how animals navigate their nutritional world. We're discovering that individual variation matters more than we thought, that seasonal transitions involve more nuanced dietary changes than we realized, and that nature's food webs are connected by more threads—both thick and thin—than we ever imagined.
and it turns out the menu is far more diverse, dynamic, and fascinating than we ever expected.
For further exploration of this topic, see the original research in Molecular Ecology (Littleford-Colquhoun et al. 2022) and related studies at Kartzinel Lab, Brown University .