Botanically, a tomato is a fruit: a seed-bearing structure that grows from the flowering part of a plant. In 1893, however, the highest court in the land ruled in the case of Nix v. Hedden that the tomato was a vegetable, subject to vegetable import tariffs.

Unfortunately, the vegetal confusion did not end in 1893. Indeed, confusion over botanical categorization has a proud history in America.

Just recently, the US Congress mistook pizza (or, specifically, the tomato paste found on what passes for pizza in school lunchrooms) for a vegetable! And a Fox News anchor apparently had trouble distinguishing between peppers and military-grade pepper spray.

Sheep can do better.

It turns out that many animals possess some basic abilities to distinguish among different types of plants. After all, one wrong bite could mean the difference between a tasty meal and a painful death. Large grazing herbivores, like cows or sheep, however, are faced with a smorgasbord of potential food sources. Having to distinguish among such a large diversity of plants is no small task, and would theoretically place heavy demands on the information processing capabilities of the herbivorous brain. Cecile Ginane and Bertrand Dumont, two French animal behavior researchers, reasoned that such animals could be more efficient at foraging, and the information-processing burden could therefore be reduced, if they were able to assign potential food items to categories.

There are two main types of plants that are found in the fields in which grazers graze: grasses and legumes. In general, legumes are preferred over grasses by grazing herbivores: they are digested faster, contain more nitrogen, and their proteins can be more completely digested. What Gigane and Dumont wanted to know was if sheep (Ovis aries) categorize plants on the basis of botanical grouping. If they learned to avoid a particular grass or legume species, would that aversion generalize to other plants within the same botanical family?

They began by arranging an experimental field that included two grass species and two legume species. The grass species were cocksfoot (Dactylis glomerata) and tall fescue (Festuca arundinacea), and the legumes were alfalfa (Medicago sativa) and sainfoin (Onobrychis viciifolia). They knew, due to prior research, that sheep are able to distinguish these particular within-category plants from each other: that is, they can distinguish cocksfoot from fescue, and alfalfa from sainfoin. Would the sheep categorize these plants at any higher-order levels? In other words, do they form hierarchical categories for plants?

On the first day of the week, one group of lambs was conditioned against tall fescue (grass), and a second group of lambs was conditioned against sainfoin (legume). This was done by pairing the plants with the vomit-inducing lithium chloride. They were given a day to recover, and then on the third day of the week, each group's food preferences were assessed by offering them either cocksfoot (grass) or alfalfa (legume). This pattern was repeated each week for a month, resulting in four testing periods. Combined with baseline preferences, the researchers had data for food preferences at five timepoints.

Initially, all sheep preferred alfalfa (legume) to cocksfoot (grass), which was expected, as legumes are more nutritious than grasses. Therefore, rather than simply measuring the straight preference for each type of plant, the researchers calculated the relative proportions eaten among the different plant types. In other words, the sheep conditioned against sainfoin could still eat more legumes than cocksfoot - but it is possible that the legumes could account for a smaller proportion of their total food intake, following conditioning.

Indeed, as the experiment progressed, the sheep that were conditioned against sainfoin (legume) consumed proportionally less alfalfa (legume) than they did at baseline. That is, the proportion of alfalfa eaten relative to total food consumption decreased for the sheep conditioned against sainfoin. And, the sheep that were conditioned against fescue (grass) ate proportionally more alfalfa (legume) than they did at baseline.

As expected, the total volume in grams (rather than proportion) of alfalfa eaten by either group did not change much over the course of the experiment. However, the sheep did diverge in terms of the volume of cocksfoot that they ate, with those conditioned against fescue consuming less than those conditioned against sainfoin.

Taken together, preferences between alfalfa and cocksfoot differed between each group of sheep, according to the type of plant against which they were conditioned. However, the divergence seemed mainly due to changes in the amount of cocksfoot consumed, rather than the amount of alfalfa consumed. This is probably due to the preference for legumes over grasses in the first place.

Even though the results are not as straightforward as they could have been, this experiment does convincingly suggest that sheep can categorize among the second-order botanical categories of grasses and legumes in addition to the first-order species-level categories. However, the difference between legumes and grasses is not only botanical, but may also be visual (such as leaf arrangement), chemical (such as scent or taste), or tactile (such as texture). In other words, the distinction that sheep make between grasses and legumes could be perceptual rather than conceptual. As always, more research is needed to see just how abstract the categorization capabilities of sheep may be. Still, the ability of sheep to categorize plants at two levels is impressive.

Still no word on whether sheep consider pizza a vegetable.

Ginane C, & Dumont B (2011). Do sheep (Ovis aries) categorize plant species according to botanical family? Animal cognition, 14 (3), 369-76 PMID: 21203791

Images: Sheep via Fescue via Flickr/Tredway. Cocksfoot via Flickr/WixPix. Alfalfa via Flickr/johnnyalive. Sainfoin via Flickr/inklaar.