To project overview

The ecology of the sea in formulae

Annual Review Article 2020

The sea’s ecosystems are made up of many species that interact with one another. Photo: Erik Selander

The seas’ ecosystems affect the earth’s carbon budget and thus the global climate, and are themselves affected by the climate. Professor Thomas Kiørboe, who was awarded the Carlsberg Foundation Research Prize 2019, asks whether it is possible to produce usable forecasts of this complicated interaction using simple models.

By Thomas Kiørboe, professor, PhD, Dr Scient, DTU AQUA, Technical University of Denmark

While weather models, for example, are based on Newton’s laws of motion and a small number of well-tested equations, ecological models are based on a large number of uncertain equations. An ecosystem comprises many species that compete with and eat one another in a complicated network. 

Classical ecological models describe the species’ interactions with one another and with the environment. The problem, however, is that for each species included in the description, the number of equations and parameters increases explosively. It is therefore impossible for classical ecological models to embrace the incredible complexity of ecosystems.

Diatoms are the most important group of phytoplankton in the sea, representing almost onequarter of global primary production. They are surrounded by a hard silicon shell, which protects them from grazers. Photo: Nina Lundholm

Essential features and trade-offs

Our approach to developing models of marine ecosystems takes a different perspective. We use principles of evolutionary biology to understand the dynamics of ecological systems. 

Instead of describing the many species, we describe individuals that – across species – are characterised by a few essential traits (properties) that each have environmentally dependent advantages and disadvantages (trade-offs). 

If we mix individuals with a random set of traits in a model and allow them to interact, only the individuals with the best combinations of traits will survive, according to the principle of “survival of the fittest”. 

The structure of the ecosystem develops, so to speak, from itself (emergence), and the resulting ecosystem will depend on the physical world around it. We use models of this kind to understand and predict the sea’s resources, biogeochemical cycles, and sensitivity to external impacts. 

Two key terms are relevant here: essential traits and trade-offs. “Essential traits” refers to the few properties that are the main determinants of an organism’s fitness, i.e., its ability to gather resources, survive, and reproduce. An example of an essential trait is how the organism gathers resources, whether by means of photosynthesis, eating other organisms, or both. 

Each organism can be awarded a value for each trait – in the example, to what degree it lives by photosynthesis or by eating other organisms. This is, of course, a massive simplification, but we see the grand picture at the expense of details. Just as Picasso can draw a bull with a single stroke of the pen, the art here is to select precisely the few traits that capture the essence of an organism.

Great complexity with a small number of properties

Fig. 1 Single-celled phytoplankton are the sea’s green plants. They vary in size by a factor of more than 1 million: small cells absorb nutrients more efficiently than large cells but, on the other hand, have a greater risk of being eaten by zooplankton.

Is it possible to describe a complex system using a small number of traits? The colour of a computer screen is formed from a mixture of the primary colours: red, green and blue. 

Each colour has 256 intensities, making it possible to obtain 17 million different colours (~species) by combining just three colours (~traits). So, yes, it is possible to achieve great complexity with a small number of traits. 

There are inherent conflicts between an organism’s three fundamental activities: eating, surviving and reproducing. 

No organism can optimise all functions simultaneously, and each essential trait has advantages and disadvantages – these are the trade-offs (see fig. 1-3). The ability to photosynthesise provides a plant with energy, but the plant pays by investing in chloroplasts. 

A fish searching for food simultaneously increases its risk of being eaten by predators. The advantages and disadvantages of the traits thus depend on environmental conditions, both the physical environment (quantity of light) and the biological environment (number of prey and predators).

STAY CURIOUS: Climate change impacts life under the sea

Fig. 2 A copepod creates a feeding current that brings in food particles (phytoplankton) but simultaneously attracts predators. More efficient feeding currents mean greater risk. The red arrows visualise the measured feeding current. Photos: Thomas Kiørboe

Modelling patterns is now possible

It is not possible to study all the organisms in the sea, so the theoretical challenge lies in using observations of a modest number of organisms to deduce general principles for organisms’ interactions with one another and with the environment. 

Through theoretical considerations and experiments using representative organisms, we quantify trade-offs for the essential traits (see fig. 1-3). 

And by understanding the underlying physical mechanisms for interactions, we can generalise the experimental results. 

We have now progressed to the stage where we can model overall patterns in marine ecosystems

We can test our models by comparing observed with predicted distributions of “traits” in the sea. That is, we examine how organisms with specific traits are distributed, and which combinations of traits characterise a given ecosystem. 

We are now able to model large-scale patterns in marine ecosystems. One example is the global size distribution and foraging strategy of zooplankton organisms – two traits that are central to the sea’s ability to remove CO2 from the atmosphere (see below). 

We find good correspondence between what our models predict and what we can observe in nature, but the models provide more complete coverage than the more sporadic observations. 

Fig. 3 Single-celled diatoms are enclosed in a hard silicon shell, which protects them from being grazed. When the diatoms smell zooplankton, the shell thickens in defence (A), but they pay with a lower growth rate (B).

An ecosystem’s “function”, i.e. its ability to produce fish or to remove CO2 from the atmosphere, for example, is a function of the traits (properties) of the organisms that make up the system. The trait-based approach to describe an ecosystem therefore provides direct insight into an ecosystem’s function. 

We can now, with some success, model the sensitivity of marine ecosystems to disturbances (pollution, fishing), the sea’s ability to produce fish, and its capacity to remove CO2 from the atmosphere. 

An example of the last of these is a global model showing how zooplankton’s contribution to reducing atmospheric CO2 has undergone a significant geographic shift over the last 60 years. 

Mechanistically founded, trait-based models of this kind also have the potential to provide robust predictions, e.g. to estimate the ecological consequences of climate scenarios.