## Sampling a populations abundance

Often in ecology we conduct studies to measure variables of a population (or community). These are not necessarily 'experiments' because we often are not manipulating variables or creating treatments - we are just measuring what is there. Before we sample a population, we must first decide:

• who the population members are;

• where the population boundaries are;

• how to count individuals.

In some cases these answers are straightforward, whereas in others they are more complex. To sample, we must be able to identify every member of a species and differentiate it from other similar species. This is difficult when individuals change in appearance with age (developmental plasticity) or with their environment (environmental plasticity). In some cases we can only identify plants to genera; for example, if we sample during a time when the plant is not flowering. Secondly, unlike mammals where every individual is a separate entity, plants can exist as clonal organisms, and therefore a researcher will have to decide whether to sample a ramet or genet (see Chapter 5). The third question of population boundaries was addressed in Chapter 2, where we also discussed measures of abundance.

The fourth decision of sampling methodology is the subject of the rest of this section. If we want to describe a population, we cannot count or measure every individual. Therefore, we measure some of the individuals and use that to represent the population as a whole. The samples we take should be both random and representative of the population we are sampling. Fortunately, randomly collected samples are usually representative (Underwood, 1997). One method to collect random samples is to generate random coordinates within a habitat and then sample at those points. The next step is to decide whether to sample using a plot or plotless sampling technique. Your decision will be based on the species and habitat type as well as on the type of data you wish to collect.

### Plot sampling

Plot sampling often involves quadrats. These are physical sampling units that are placed over the vegetation and act as boundaries for the sample. The optimal number, size and

Fig. 10.4. Cumulative estimate (running mean) of a population variable estimate (e.g. height) as the number of quadrats sampled increases. Below 10 quadrats the estimates vary widely, but begin to level off after 15.

Fig. 10.4. Cumulative estimate (running mean) of a population variable estimate (e.g. height) as the number of quadrats sampled increases. Below 10 quadrats the estimates vary widely, but begin to level off after 15.

shape of quadrat will depend on the species being studied, the statistics to be carried out on the data, and the financial and physical resources of the researcher (Underwood, 1997; Zar, 1999; Quinn and Keough, 2002). Sampling many smaller quadrats is generally better than sampling fewer larger quadrats (Kershaw, 1973). This is because the accuracy of data typically increases as the number of quadrats increases. However, after a certain point increasing the number of quadrats will not improve the estimate of the variable, and will only increase the time and cost of the study. To determine whether enough quadrats have been sampled to get an accu rate estimate of a variable, we can compare how the estimate of a variable (e.g. height) changes as the number of quadrats used increases (Fig. 10.4). The point where the curve levels out indicates minimum quadrat number. This 'running mean' approach is used commonly in ecology.

The size of the quadrat will depend primarily on the vegetation type. Understorey vegetation is often sampled with 1-m2 quadrats, understorey trees and shrubs with 10-m2 quadrats and canopy trees with 100-m2 quadrats; however, these are just guidelines. The quadrat size is a compromise between the size of individual plants and

12 metres

0 0