Original Finding
The Seed Coverage Curve Flattens After Two Foods
Hulled sunflower reaches 43 of our 73 feeder-using species by itself. Adding insect food takes the pair to 65. Every food after that adds single digits, which puts a number on how little the third, fourth and fifth bag are buying you.
The short answer
- Sunflower in a hulled form (hearts or chips) reaches 43 species, 59% of the feeder-using birds in our matrix, on its own.
- Adding insect food, which at a feeder means mealworms and suet, takes the total to 65 species (89%).
- The third food adds 4 species. The fourth adds 3. The curve is essentially flat past two.
- This is a coverage argument, not a "buy one bag" argument. Specialist foods exist precisely to reach species the staples miss, and safflower and nyjer earn their place for reasons coverage does not measure.
Method at a glance
- Question
- How much species reach does each additional food actually buy?
- Data
- Feeder bird trait matrix, 73 feeder-using species
- Method
- Greedy set cover over normalized food tags
- Measure
- Distinct species reached, unweighted
- Reproducible
- Yes, from the published CSV
- Type
- Analysis of compiled data, not a feeding trial
The result#
| Add | Food | New species added | Cumulative | Coverage |
|---|---|---|---|---|
| 1 | Sunflower hearts or chips | +43 | 43 | 59% |
| 2 | Live or dried insects | +22 | 65 | 89% |
| 3 | Cracked corn | +4 | 69 | 94% |
| 4 | Acorns and tree nuts | +3 | 72 | 99% |
| 5 | Fruit and jelly | +1 | 73 | 100% |
The shape is the finding. Going from no feeder to one bag of hulled sunflower is worth 43 species. Going from four foods to five is worth 1. Most seed advice presents a long list of foods as if each were pulling comparable weight, and on coverage grounds they are not close.
Each food on its own#
Set cover measures marginal contribution, which is what you want when deciding what to buy next but which understates foods that arrive late. Here is raw reach, ignoring overlap entirely.
Why hulled sunflower wins#
It is the same reason our black oil sunflower guide gives, now with a number attached. Sunflower is high in fat, and in hulled form it removes the one barrier that excludes small-billed birds from sunflower: the shell. A bird that cannot crack a hull can still eat a heart. So hulled sunflower inherits the whole sunflower audience and adds the soft-billed birds on top.
Hearts cost noticeably more per pound than seed in the shell, and this analysis says nothing about whether they are worth it. That is a cost-per-usable-gram question, because you are not paying to ship hulls you throw away. Our hull waste protocol is the way to measure it for a specific bag, and the seed cost calculator does the arithmetic.
One important caveat on hearts, which our food guides cover: without a shell they spoil faster, they turn to mush when wet, and they need a feeder that drains and a schedule that keeps them fresh. Coverage is not the only axis.
Why insect food is the big second step#
Because it reaches a group that seed cannot reach at all. Warblers, wrens, bluebirds, creepers and most woodpeckers are not seed birds in any meaningful sense, and no amount of sunflower brings them in. Suet and mealworms are the entry point to roughly a third of the feeder community.
This is the same structural point as the feeder analysis, where the suet cage ranks second: the best second purchase is the one with the least overlap with your first. Seed variety and food variety are different things, and buying a second kind of seed is usually the lower-value move.
The case for the specialist foods#
Coverage is one axis and a narrow one. The specialist foods on this site earn their place on axes this analysis cannot see, and it would be a misreading to conclude otherwise.
Safflower reaches fewer species by design. That is the point: cardinals and finches eat it while squirrels, grackles and starlings mostly refuse it, which makes it a pest management tool that happens to be a seed.
Nyjer is a narrow food with a devoted audience. If goldfinches and siskins are what you want, its coverage number is irrelevant to you.
Nectar looks small here and reaches a group nothing else touches at all.
Cracked corn ranks third on marginal gain and comes with real drawbacks: it molds fast, it draws large flocks and mammals, and it is the food most likely to create the problems in our problems section.
What this does not show#
Binary, unweighted, compiled. Same three limits as every analysis run on this matrix: "eats this" does not distinguish staple from occasional, every species counts once regardless of abundance, and the underlying facts are our profiles rather than direct observation.
Coverage ignores cost, spoilage, waste and pest pressure, all of which are real reasons to choose a food. See the food guides for those.
Sunflower forms are collapsed. We tag hulled sunflower separately from black oil in the shell, but a species that eats one nearly always eats the other. If you merge them the curve gets slightly steeper, not flatter.
A real preference trial would answer a different and better question. Coverage says a species will eat something. It does not say what it would choose. That is what the paired-choice protocol is for.
What to do with it#
Starting out: one bag of hulled sunflower or black oil sunflower, and a suet cage. That is 89% of the feeder-using species in the matrix for two purchases, and it is a substantially better use of the same money than three kinds of seed.
After that, stop buying for coverage and start buying for a target. Pick the species you actually want from the attracting guides and buy the food that species needs, which is how the specialist foods are meant to be chosen in the first place.
And avoid the cheap bagged mixes, which is the one piece of standard advice this analysis reinforces rather than complicates. Mixes are padded with milo, wheat and red millet that most feeder birds throw on the ground, so their effective coverage is the coverage of whatever small fraction of the bag is sunflower. Our hull waste and cost protocol measures that directly.