1Describing a community

Species composition is the list of species present. Two ponds could both hold 300 animals and yet share no species at all.

Species diversity has two parts:

• Species richness: the number of different species.
• Relative abundance (evenness): how equal the numbers of each species are.

A wood with ten tree species is richer than one with four. But if one of the ten species makes up 95% of the trees, a walker would hardly notice the other nine; the wood is rich but uneven, and its diversity is lower than richness alone suggests.

2Simpson's diversity index

The formula sheet gives

$$\text{Diversity Index} = 1 - \sum \left(\frac{n}{N}\right)^2$$

where $n$ is the number of organisms of one species and $N$ the total number of organisms of all species. For each species, find its share of the total, square it, add the squares, and subtract the sum from 1. The index runs from 0 (one species only) towards 1 (many species in equal numbers). Higher means more diverse.

Worked example (invented). Pond invertebrates in a net sample:

$$\text{Diversity Index} = 1 - 0.325 = 0.675$$

Why evenness matters. Two invented meadows each hold 100 plants of the same four species.

Same richness, very different diversity. Meadow B is dominated by clover, so its index is low.

Simpson's index rises with both more species and more even numbers; to compare sites, the higher index is the more diverse community.

The commonest slips: forgetting to subtract from 1 (which gives a number that falls as diversity rises), squaring the counts instead of the proportions, and dividing by the number of species instead of the number of organisms.

3Interactions between populations

Every interaction between two species can be labelled by its effect on each: + (benefits), − (harmed) or 0 (unaffected).

Parasitism, mutualism and commensalism are kinds of symbiosis: two species living in close, long-term association. Parasites usually harm their host without killing it quickly, unlike a predator.

All of these drive population dynamics, the rise and fall of numbers over time. The CED names four models to know.

Predator–prey cycles. When prey are plentiful, predators eat well and breed, so predator numbers rise. More predators eat more prey, so prey numbers fall. With less food, predator numbers then fall, which lets prey recover, and the cycle repeats. The predator's peaks come after the prey's, because it takes time for extra food to become extra predators.

In Figure 1, find a prey peak, then the predator peak just after it: the gap between them is the lag.

Cooperation. Individuals of one species, or of two species in a mutualism, can raise each other's success: ants that defend an acacia tree from herbivores in return for food and shelter.

Trophic cascades. A change at the top of a food chain cascades down. If predators that eat sea urchins are removed, urchins multiply and graze the kelp forest bare; restore the predators and the kelp returns. The effect skips a level: predators help the producers by eating the herbivores.

Niche partitioning. A species' niche is its role in the community: what it eats, where and when it feeds, what conditions it tolerates. Two species that need exactly the same resources compete strongly. Species that coexist often divide the resource: one bat species feeds on insects above the treetops at dusk, another among the branches later at night. Partitioning reduces competition and lets more species live in the same community.

4Error bars: are two means really different?

Community studies compare means: the mean number of beetles in plots with and without a competitor, for example. Means from samples vary by chance, so the question is whether a difference is bigger than chance would produce. Error bars answer it.

The formula sheet gives the standard error of the mean, $SE_{\bar{x}} = \frac{s}{\sqrt{n}}$, where $s$ is the sample standard deviation and $n$ the sample size. Bars drawn at $\bar{x} \pm 2SE_{\bar{x}}$ cover roughly a 95% confidence interval for the true mean.

Worked example (invented). Ground beetles were counted in 16 plots where a competing beetle species had been removed, and 16 control plots.

The two ranges do not overlap, so the difference is likely statistically significant: removing the competitor probably did increase the beetle numbers. If the ranges had overlapped, you could not conclude that the means differ.

Two cautions. Overlapping bars do not prove the means are the same; they mean the data do not show a difference. And separate bars support a difference, they do not prove its cause; for that you need a controlled design.