Variations in how to display the trophic interactions of an ecosystem—the food web, the regulating pressures between organisms of a particular space—have existed since the roots of the idea in community ecology. Nature—or, in particular, the relationships between organisms—has long been described as a web by many. Take, for example, the now-classic food web of Bear Island, the southernmost island of the Norwegian Svalbard archipelago, by Victor Summerhayes and Charles Elton in their seminal 1923 paper. In their illustration below, organisms are grouped into functional clusters, and arrows depict one organism being consumed by another, outlining a basis from which these systemic illustrations would build upon for decades.
But this depiction of life wasn’t quite so scientifically until Charles Darwin published his On the Origin of Species in 1859, in which he wrote,
I am tempted to give one more instance showing how plants and animals remote in the scale of nature, are bound together by a web of complex relations. I shall hereafter have occasion to show that the exotic Lobelia fulgens is never visited in my garden by insects, and consequently, from its peculiar structure, never sets a seed. Nearly all our orchidaceous plants absolutely require the visits of insects to remove their pollen-masses and thus to fertilise them. I find from experiments that humblebees are almost indispensable to the fertilisation of the heartsease (Viola tricolor), for other bees do not visit this flower […] Hence we may infer as highly probable that, if the whole genus of humble-bees became extinct or very rare in England, the heartsease and red clover would become very rare, or wholly disappear. The number of humblebees in any district depends in a great measure upon the number of field mice, which destroy their combs and nests; and Col. Newman, who has long attended to the habits of humble-bees, believes that “more than two-thirds of them are thus destroyed all over England.” Now the number of mice is largely dependent, as every one knows, on the number of cats; and Col. Newman says, “Near villages and small towns I have found the nests of humblebees more numerous than elsewhere, which I attribute to the number of cats that destroy the mice.” Hence it is quite credible that the presence of a feline animal in large numbers in a district might determine, through the intervention first of mice and then of bees, the frequency of certain flowers in that district!
Darwin also described this web of complex relations as an entangled bank, “clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth,” and he reflected on how each of these forms, different they may be, are dependent on one another in direct and indirect pathways while at the same time producing the kinds of ecological pressures that lead to adaptation and the rise—or elimination—of species.
On the Origin of Species and Darwin’s subsequent writings were foremost about natural selection, the non-random process by which biological traits become more or less common in a population. But Darwin’s idea can be, and was, applied the same to how organisms themselves become more or less common in an ecosystem—the question became the following: what are those forces and relationships that allow some species to foster in a particular environment, while others only play a minor role? Community ecologists divide these species into two initial groups: autotrophs, which produce organic matter from inorganic substance to compose the primary level of an ecosystem (the first trophic level), and the heterotrophs (1+ trophic levels), which cannot fix carbon and thus must consume organisms from the initial or subsequent trophic bases. In terms of the relationships between the autotrophs and heterotrophs, food web ecology polarizes the dynamic between which group regulates the other. In bottom-up control, also known as donor control, the autotrophs (i.e., plants and algae) produce the primary resources, thereby maintaing the nutrient availability that organisms at higher trophic levels are dependent upon. Contrasting bottom-up control is the top-down dynamic, also known as a trophic cascade, in which predation and herbivory are the factors that dominate whether a species will thrive in an environment. More typically, however, analyses in community ecology strive for some middle ground between the two.
While the organization of a food web is inherently more complex—incorporating the movement of energy and mineral nutrients; dividing the heterotrophs into carnivores, herbivores, decomposers; delineating even further between competitor species, generalist consumers, specialist predators, etc—this basic model holds steady. Because this organization is predominantly a visual one—spatial and temporal patterns and variations are a large part of community dynamics in ecological research, as outlined in a recent review by Massol et al.—an interesting aspect of studying trophic interactions lies in the possibilities for visualizing data. The first graphical interpretation of a food web was published in 1880 by Lorenzo Camerano, in a manuscript entitled “On the equilibrium of living beings by means of reciprocal destruction.” Frustrated by his time’s scientific divide over which animals are useful or beneficial to agriculture, Camerano wrote the following.
To have an exact and clear idea, I repeat, of the relations between, for example, insectivorous birds and phytophagous insects, and between these and plants, these groups cannot be studied separately. Rather it is necessary to study each in relation to all other animals to see the general laws governing the equilibrium of animal and plant species.
While Camerano’s reasoning for the paper, published in the Accademia delle Scienze di Torino, was particular and applied towards the practical, his ultimate goal in the article and its resulting visualization was of general and theoretical concern.
Compare that template of trophic relationships among organism groups—including primary producers, predators, ectoparasites, endoparasites, and carnivores—with a particular illustration of an ecological community of individual species with regard to enemies of a specific phylogenetic order, Coleoptera.
In the former, more-theoretical visualization, Camerano sides with the theory of bottom-up control—vegetation is the base layer of the trophic sphere, with interactions between higher trophic levels expanding in outward rings. In the latter, allying with top-down ideas of control, complex relations are derived not from primary production, but rather from multiple pressures of predation, forming direct and indirect influences between heterotrophs. It is the latter depiction—with its emphasis on predator–prey relationships and linear–nonlinear associations between species—that has most influenced how many visualizations of trophic dynamics are drawn. In Summerhayes and Elton’s depiction of Bear Island, the first illustration in this article, the ecologists organized the biota with priority on that relationship between predator and prey—protozoa eat the bacteria, polar bears eat seals. Even where nutrients and zooplankton entered the situation, their connection to the rest of the system was given within the framework of consumption dynamics—bacteria consume nitrogen, worms eat dead plants, algae take in mineral salts.
And nor is this choice of visualization a poor or incorrect one. However, an interesting note lies in the shift between how to organize a food web in the 19th and early 20th centuries and what kind of information should and can be included in more sophisticated images now. When one adds in the more quantitative turn in food web analysis—such as with ecological pyramids and trophic dynamics, both of which incorporate the second law of thermodynamics to consider how energy is conserved or lost as it passes between organisms—the complexity inherent in the early food webs above exponentially increases in magnitude.
The question in producing a food web is, then, how to produce a visualization of that complexity without overwhelming our perception of reality. Sacrifice quantitative rigor, and the illustration instead appears as illusion, an over-simplification. Introduce too much information, and the traceable dissolves into surplus. Such are the balancing acts Edward Tufte outlines in his 1983 The Visual Display of Quantitative Information. In the book, the statistician argues that good information design reveals the greatest number of ideas in the shortest time, with the least ink, in the smallest amount space, akin to what Italo Calvino described as lightness in literary prose—both must, in the end, tell not only a good story, but a proper one, without sacrificing either extreme of this information equilibrium.
As some of the first illustrations of food webs came from aquatic systems, bodies of water, as simultaneously self-contained microcosms and hubs in ecological networks, are good places to look for growing trends in data visualizations. An interesting and successful update on the classic food web comes from a paper by Kevin Caillouët et al. in a 2008 paper on enclosed aquatic communities in New Orleans, Louisiana. After Hurricane Katrina flooded 80% of the city in August of 2005, floodwater from Lake Pontchartrain to the north inundated the urban terrain for over three weeks. By completely covering most regions of the city in brackish water, flooding resulted in the production of novel aquatic habitats in lowlands, depressions, and artificial containers, to name a few. Much as a forest fire produces a blank template from which previously occupied niches can be reclaimed and monocultures overthrown, Hurricane Katrina produced a blank template from which new food webs—new trophic interactions, in new magnitudes and organizations—could be drawn. In particular, on the back of research suggesting correlations between flooding and increased mosquito populations, Caillouët et al. sought to examine how these new aquatic communities would work for or against local mosquito populations, and how this in turn could influence the human risk of mosquito-borne diseases such as that caused by West Nile virus. In particular, the team examined the quasi-contained ecosystem of abandoned swimming pools, the largest of artificial containers and one of the most prone habitats to support mosquito larval development.
By sampling pools in two flooded neighborhoods of New Orleans near the 17th Street Canal levee breach, the researchers were looking at this artificial environment as a mesocosm of sorts. Unlike the simplified microcosm, a controlled, laboratory-based ecosystem used to simulate and predict the behavior of its natural counterpart, a mesocosm instead brings the natural under controlled conditions, providing a link between observational field study and the artificiality of the laboratory. These mesocosms, which were being colonized by various invertebrates, fish, and mosquito species, allowed the team to ask the following questions about the composition of these new and temporary environments: if mosquito larvae were using the pools for development; if fish moved in from Lake Pontchartrain and were able to establish populations; if predatory invertebrates migrated to and colonized pools; and, lastly, if and to what extend these latter organism groups regulated mosquito populations. To explore how these questions translate into science, Caillouët et al. relied on a conceptual model, the first step in transforming idea into design, the prerequisite for collecting and interpreting data.
The conceptual model (A) of Caillouët et al. is not all too different, as a theoretical insight, from the initial plate by Camerano in 1880. Just as Camerano endeavored to clearly illustrate the relationships that draw together insectivorous birds and phytophagous insects, and between these and plants, the modern model food web illustrates the general relations between four main groups of organisms: the top predators, the specialist predators, the competitor species, and the prey. Much as Summerhayes and Elton used arrows in their 1923 depiction of an aquatic food web to illustrate the direction of the relationship between two or more organisms, the above model demonstrates a direct (and negative) effect of one group on another with solid lines, while dotted lines indicate an indirect (and therefore positive) effect.
These effects, as well as the overall model, are based upon the ecological theory that specialist predators are more efficient at reducing prey than their generalist counterparts, the top predators. However, because these generalists and competitor species prey upon the same food source, it would be a logical hypothesis to predict that these two groups would indirectly benefit prey by reducing the direct effect of the specialist. As this predator–prey web applies to the swimming pool mesocosm, the researchers applied their aquatic system to the model as follows: mosquito larvae as the prey, Western mosquitofish as the specialist predators, odonates (dragonflies and damselflies) as competitor species, and large predacious insects (giant water bugs and waterscorpions) as the generalists. Revealing the actual composition of this community, the proportion of each group and each species to the overall collection of organisms in each mesocosm, was a simple enough task. Out of 92 abandoned swimming pools sampled, 64% contained mosquito larvae, of which 86% belonged to the species Culiseta inornata and 8% to Culex quinquefasciatus, the main vector of West Nile virus in the Mississippi delta. As per the generalist predators, 16 families of predatory invertebrates were found, and dragonfly nymphs, the competitor species to the mosquito larvae in this community, were found in roughly half the pools. The various fish species, the specialists of the ecosystem, were found in 47% of pools, with Western mosquitofish constituting 75% of the overall abundance.
However, using a statistical technique known as path analysis, Caillouët et al. were able to not only produce a food web more akin to the second plate of Camerano—one applied to an actual group of organisms—but also tease apart the direction and magnitude of those relationships, combining the classic web visualization with the complexity of the quantitative turn.
The second part of Caillouët’s model (B) demonstrates the goodness of fit between the food web as a concept and the strength of those ecological relationships in reality. As predicted by ecological theory, the Western mosquitofish in New Orleans abandoned swimming pools did have a large, direct, and negative effect on the presence of mosquito larvae—likewise, the magnitude of this effect, as revealed by the path analysis, was shown to be dampened by the presence of those predatory invertebrates and competitor odonates. Unlike the initial model food web, the uniform solid lines of direct relationships here vary in wideness—the thicker the line, the greater the direct effect of the given species on the other, given by a negative value as well. In the case of the generalists and competitors on mosquito specialists, then, both combined had a negative effect of roughly 0.5. As this relates to the mosquito larvae themselves, the indirect influence of these two groups of organisms, as revealed by statistical techniques, was a combined negative 0.48. While this implies the two were able to reduce Western mosquitofish abundance and effectiveness, the primary specialist species still exerted a far larger negative effect of 0.74, which significantly outcasts those indirect influences, depicted with the dashed lines.
Much like Camerano’s reasoning for his paper in the Accademia delle Scienze di Torino—of practical importance to resolving the question of which animals are useful or beneficial to agriculture—the findings of Caillouët et al. were both applied data and of broader theoretical concern. As the study found that Western mosquitofish were still effective colonizers of abandoned swimming pools as well as regulators of mosquito larvae, this species—native to Lake Pontchartrain—was introduced in several months time, before the larvae matured into adulthood, into the thousands of pools throughout New Orleans. In terms of human health concerns, the use of the fish as a native biocontrol measure against diseases like West Nile virus was one much-needed application—a study by the same group that year demonstrated a greater-than two-fold increase in risk for West Nile neuroinvasive disease after Hurricane Katrina.
Yet in terms of theoretical contributions, Caillouët et al. make and reiterate the broader point that trophic relationships should be evaluated prior to making any manipulation to ecological communities for human-related benefits. And as with other advances in food web ecology in the past, the real value of the research can be taken from the conversion of raw data into visualization. The beauty in this visualization lies in its simplicity. One could make the argument that the numbers in part B, which indicate the specific magnitude of the relationship, aren’t even necessary. Instead, they only reiterate the actual visual component of the figure, that of the thickness of the solid line. In this way, the numbers are what Edward Tuftle referred to as “chartjunk,” the useless, non-informative, or—in this case—information-obscuring elements of quantitative information displays. And yet it is a shortcoming that can be overlooked, for otherwise the depiction might seem almost too simple, too streamlined, lacking the rigor that it inherently required to produce. Perhaps then, as applies to the modern web of trophic dynamics in this mesocosm, Darwin’s image of a web of complex relationships just slightly misses the mark. Perhaps contrary to Calvino’s lightness and Camerano’s radii of lines, we in fact need more weight added to these visuals, a quality that webs, in being light, suspended in air, lack.