Ecology Field IA (IB Biology and ESS): Sampling and Data

Field investigations make some of the best IAs and some of the worst. The good ones are original, personal and full of real data. The bad ones die on the day of collection because the site was wrong, the sampling had no logic, or the data came back in five different notebooks. Most of what follows comes from coaching field IAs this autumn. If you want to see a finished report first, read the annotated example of a full field IA. The fieldwork advice is the same for both subjects, but the marking isn't: Biology uses the four-criterion scientific investigation, while the ESS individual investigation has six criteria, including one on a strategy for an environmental issue. If you take ESS, read the ESS IA guide alongside this one.

1. Choose a site that won't sink your data

The first question isn't "what's interesting" but "where can I actually collect clean data". Sites I've seen students propose and then drop:

Mowed parks and lawns: the vegetation reflects the gardener's schedule, not the variable you're studying.
Road verges that get cut, farmland that gets sprayed and ploughed, stone-lined riverbanks with no natural gradient.
Private or closed land you can't access on the day.

What you want: natural or semi-natural, publicly accessible, safe. Nature reserves and woodland edges are usually better than parks. Scout the site in advance and take photos. Check that the gradient you want to study actually exists there.

And a practical one: you have to be able to reach it, more than once, and nobody should fall in a river.

2. Transect or random sampling?

Expect the examiner, or your teacher, to ask why your sampling is laid out the way it is. Have the answer written into your method.

Use a transect when you're studying a gradient: distance from water, from a footpath, from a road, from a forest edge. The transect runs along the gradient, and that's your justification.
Use random sampling when you're comparing areas without a gradient (two meadows, a trampled vs untrampled zone). Generate random coordinates with a random-number app, so you're not unconsciously choosing "nice" spots.

A phone compass and GPS coordinates are enough to lay a transect along a bearing and come back to the same line later.

3. Find your confounders before the examiner does

With one student designing a lichen survey (lichen cover vs distance from a motorway), we listed out loud everything else that could change lichen cover along the transect: houses nearby, a second road, tree density, tree species, trunk height, which side of the tree faces the road. Then we controlled what we could:

One tree species only. A city tree register helped find enough trees of the same species.
Same trunk height for every quadrat.
Same side of the trunk.
A photo of every tree and quadrat, as evidence and for re-checking cover later.

What you can't control, you record and discuss. Using official air-quality monitoring data from a nearby station as secondary context strengthened that design a lot.

Background research predicts the shape of your result. For a plant that prefers moist but well-drained soil, both drought and waterlogging stunt it, so you should expect a curve with an optimum, not a straight line. Soil type then becomes a confounder to note. Knowing this before collecting makes the analysis and the evaluation much stronger.

4. Measure properly

Soil moisture by mass. Simple, cheap and defensible:

Take each soil sample into a sealed, numbered tube or bag, in collection order.
Weigh it (wet mass).
Dry it in an oven until the mass stops changing.
Weigh it again (dry mass).
Moisture content (%) = (wet mass − dry mass) ÷ wet mass × 100. Some methods divide by dry mass instead; state which one you used.

Plant height: photograph each plant with a ruler next to it. You get the measurement and a record you can check later.

Percentage cover: use a gridded quadrat and count the squares covered. Decide your rule for half-covered squares before you start, and state it.

Write down what you couldn't measure (pH probe broke, rain on day two). That goes straight into your evaluation.

5. Plan for missing data

Something will go wrong in the field. Decide in advance what you'll do, and write it into the method. For example: "If fewer than three specimens are present at a sampling point, the nearest specimens within 1 m are measured and the actual sample size is recorded." A gap then becomes a stated method decision, not a hole in your data.

6. One master spreadsheet

The messiest IA data I've seen came in separate tables per sample, without IDs, written up days later. Do this instead:

One master sheet, filled in on the day or the same evening.
Hierarchical IDs for every measurement: site A, transect 1, point 3, replicate 2 → A1.3.2.
Label sites A, B, C so your write-up can refer to them simply.
Check replicates as you go. One pair of soil samples in a student's data read 94% and 56% moisture at the same point. Spotting that on the day means you can resample. Spotting it at home means it becomes a limitation.
Look for gaps in your range. If you have points at 0, 10, 40 and 50 m and nothing in between, go back for 20 and 30.

7. Analysis: pick the right test

Always: a clear graph (scatter for a gradient, bar chart with error bars for group comparisons), means and standard deviations, and the trend described with numbers.
Gradient or relationship between two continuous variables: a correlation (Spearman's rank or Pearson), with the strength and significance reported.
Comparing two groups or sites: a t-test (or Mann–Whitney U if the data isn't roughly normal).
Comparing three or more groups: ANOVA (or Kruskal–Wallis).
State your null and alternative hypotheses before the test.

How to choose between these tests, and how to report the result, is in the guide to statistical tests for the IA. Google Sheets and Excel can do correlations and t-tests. Make sure you can explain what the test does and what your p-value means, not just paste it.

Small samples are an honest limitation, not a disaster. With only three plants per point, averages and standard deviations are still meaningful. Say clearly what the sample size allows you to conclude and what it doesn't.

8. The two rules that matter most

Describe the method you actually used. If you measured distance by pacing instead of with a tape, say so and discuss the uncertainty. Done honestly, that earns evaluation marks.
If the data contradicts your hypothesis, it's a result. Report it and discuss why. Changing or inventing data is academic misconduct and puts the whole diploma at risk.

Before collection day: a kit card

The night before, write a checklist of everything you need: quadrat, tape or GPS app, labelled tubes, marker, scale or ruler, phone charged, data sheet with IDs ready, permission confirmed, backup date. Then do a two-minute "defend my design" talk to yourself or a friend: why this site, why this layout, what could go wrong? It's surprising how often that exposes a problem in time to fix it.

The free IA planner Notion template includes the kit card, the missing-data protocol and a "defend your design" section.