Ecology Explained

Here are a few of my favourite ideas in ecology, told through the studies that put them to the test. Why did only half a lake turn green? How could a salamander make an insecticide more harmful to tadpoles it couldn’t reach? Choose a question, compare what happened, and see how the idea connects to my work.

Keystone speciesWhat makes a species a keystone?A keystone species shapes its community far more than you would expect from how common it is.

A rocky shore without sea stars

On the Washington coast, Robert T. Paine removed ochre sea stars from a patch of shore. What would happen to the other species sharing that rock?

Compare illustrations of the shore with sea stars and after their removal

Sea stars ate mussels, leaving patches of rock where barnacles and algae could live.

Over time, the mussel bed spread. Other species lost their foothold, and fewer species shared the rock.

Paine’s original experiment (1966)

Other keystone species

Sea otters. In Alaska’s Aleutian Islands, otters eat sea urchins. Where otters were absent, heavy grazing by urchins left little kelp standing.
Estes and Palmisano (1974)

Beavers. In New York’s Adirondacks, beaver dams created wetlands. Some plants grew there that researchers found nowhere else along the streams they surveyed.
Wright and colleagues (2002)

My water-louse recovery study
Multiple stressorsCan a predator make a pesticide more harmful?Tadpoles can pick up signs of a predator in the water, even when it cannot reach them.

The salamander behind the mesh

Rick Relyea and Nathan Mills raised gray treefrog tadpoles in tanks with an empty cage or a caged salamander larva. Compare the two groups that received the same low dose of the insecticide carbaryl in their first experiment.

Can a predator make a pesticide more harmful?

With the cage empty, 40% of the tadpoles survived at this dose.

With a salamander in the cage, 3% survived. It could release chemical cues into the water but could not catch the tadpoles outside.

Relyea and Mills (2001)

Survival stayed high in the control tanks without the insecticide. Together, those comparisons show that the effect of carbaryl on survival depended on whether the tadpoles could sense a predator.

How this connects to my work

I asked a similar question about feeding. My colleagues and I measured how much leaf water lice ate near a caged dragonfly larva and farther away, then tested whether a fungicide changed that pattern.

Would water lice feed near a predator?
Ecosystem managementWhich nutrient should we keep out of a lake?When a lake turns green with algae, deciding what to change means finding out what is feeding the bloom.

A lake divided by a curtain

Lake 226 and its dividing curtainThe two adjoining basins have clear water. A curtain spans the narrow neck between them.Curtain
Lake 226, Ontario · Schematic
Compare the two basins
Schindler (1974)

In Ontario, David Schindler and his colleagues stretched a curtain across the narrow neck of Lake 226. They added equal amounts of nitrogen and carbon to each basin. Only one basin also received phosphorus.

The two basins shared the same weather. Their different responses gave managers a reason to reduce the phosphorus entering lakes like this one.

How this connects to my work

I’m interested in what we should measure when water quality changes. In one experiment, I tested whether water lice in clean water still ate less if their leaves had previously soaked in a fungicide.

What did the leaves carry into clean water?
Ecological forecastingHow far ahead can we trust a forecast?At a drinking-water reservoir in Virginia, a sensor could check last week’s prediction against today’s water temperature.

Predicting the temperature of Falling Creek Reservoir

For 475 days, Quinn Thomas and his colleagues issued daily forecasts and checked them against later readings. Here is how the forecast error changed at one metre below the surface.

How far ahead can we trust a forecast?

Across the seven-day forecasts, the root mean square error was 1.13 °C. This measure gives larger misses more weight.

At the same depth, the root mean square error for the sixteen-day forecasts was 1.62 °C.

Thomas and colleagues (2020)

The longer forecast bought nine extra days of warning at the cost of a larger error. Whether that trade-off is worthwhile depends on the decision someone needs to make.

How this connects to my work

In my current forecasting work, I ask whether an animal’s size today is enough to predict how much it will eat next week, or whether a past heatwave leaves something that a size measurement misses.

My work on environmental history and prediction