Quiz time – in the map below, can you name the two species whose distributions are shown in red and blue dots? You can see that the occurrences of A are associated with warmer temperatures, whereas the occurrences of B are associated with colder temperatures and higher latitudes.

Whatever you guessed, you were probably wrong. Species A represents all the franchises of the American chain, Dunkin’ Donuts, while Species B represents franchises of the Canadian donut chain, Tim Hortons. Data are for the year 2010. Sometimes economic diversity can look a lot like ecological diversity.

I became interested in this topic on a few long road trips where the same ‘species’ would appear over and over again – endless seas of Burger Kings, Denny’s restaurants every few miles, multiple Starbucks on the same city block. Do these businesses obey the same macroecological rules that biological species do?

I recently finished up a study of this question. It was just published in the open-access journal PLoS One as “Separating Macroecological Pattern and Process: Comparing Ecological, Economic, and Geological Systems“. We bring together datasets for North American trees, birds, and minerals, then contrast these patterns with data for several hundred chain businesses as well as a recent United States Census of business types. Below (Lima, Peru) and above (Canóvanas, Puerto Rico) you can see a lot of these American chains have extended their distributions far from home.

Macroecology is generally concerned with the distribution of diversity: how common are different sets of species across space and time? One central ‘first order’ metric is the species-abundance distribution, which describes the number of species that have a certain number of individuals. Most ecological communities show ‘hollow curves’ where most species are rare (low # of individuals) and only a few species are common (high # of individuals). Another key ‘first order’ metric is the decay of similarity with distance, which states that for two ecological communities separated by a given distance, the fraction of species shared between those communities decays as distance increases. This makes intuitive sense: the pine forests of Arizona are more similar to the pine-juniper forests of New Mexico than the hardwood forests of Georgia.

We show that using these and other macroecological metrics, economic and ecological systems look remarkably similar. Abundance distributions and several other first-order patterns all follow the same patterns. Only by examining the less-considered spatial scale dependence of these patterns (‘second-order’ metrics) can we distinguish economic and ecological systems.

These results suggest that first-order metrics are statistically inevitable when objects are partitioned into categories over space, and only second-order metrics let us test theories that are uniquely relevant to ecology.

One of the most striking second-order patterns is that at larger spatial scales, economic diversity maintains much higher similarity at large distances. Working with our chain business dataset, we constructed test communities of the approximate size of a county (30 x 30 miles), then computed a Jaccard similarity coefficient between all possible pairs of communities. We found that even at 4000 km distances, more than 80% of the same chain businesses are found in each community. This ‘everything is everywhere’ finding provides quantitative support for the feeling of anonymity and homogeneity that I have felt on many of my travels through my country.

The commonality of first-order patterns does suggest that macroecology can provide some strong constraints on possible economic diversity patterns. For example, the ‘hollow curve’ abundance distribution suggests that more businesses will always be rare than are common, consistent with the continued existence of a few large and influential companies in every system, and contrary to our hopes of building communities centered around all small or local businesses. However, the constraint is only on the existence of a ‘hollow curve’ distribution, and not on its parameters – so shifts in the mean and variance of this distribution are certainly still possible.

I think the key thing this study leaves unaddressed is how strong macroecological constraints are on economic systems that are rapidly changing: undergoing growth or collapse (as in the Panama City neighborhood above), or shifts from rural to suburban or urban land use (as in the Miami neighborhood below).

We’re following up on all of these questions, and I hope that macroecology will soon have much more to say about non-ecological systems.