Reading path: Underwater images and evidence →
Decode the axes before the shape
A figure can make a pattern easy to see, but only after you know what is plotted. Read both axes, including units, scale, and labels. “Temperature” might be degrees at a specific depth; “abundance” might mean organisms per volume, per image, or per transect. Those quantities are not interchangeable. Check whether an axis begins at zero or uses a restricted range that visually magnifies a difference.
Identify what each point, bar, line, or color represents. A dot might be one sample, a site average, or a model estimate. Lines may connect observations through time, show a fitted relationship, or simply guide the eye. The caption usually explains these conventions. If it does not, consult the methods before treating the visual as self-explanatory.
Read the sample labels and uncertainty
Look for labels showing sites, dates, depths, treatments, or organism groups. A legend may distinguish several locations that would otherwise look like one broad dataset. Check the sample size when provided and ask what it counts: individual organisms, sampling events, or independent sites. Many plotted marks do not always mean many independent observations.
Error bars and shaded bands communicate uncertainty or variation, but their meaning depends on the caption. They may show standard deviation, standard error, a confidence interval, or another range. These have different interpretations. Overlapping intervals do not by themselves prove that groups are the same, and separated intervals do not automatically establish a meaningful ecological effect. Use the analysis reported in the text for formal comparisons.
Connect the visual to the claim
Once the figure is decoded, describe the pattern narrowly. Suppose a plot shows median animal counts at three depths over four survey dates. A careful summary might say that counts were higher at the middle depth on most plotted dates, while noting that the pattern was not identical every time. It would be too strong to claim that the species prefers that depth everywhere unless the sampling and analysis support that wider inference.
Check whether the figure actually displays the variable named in the paper's headline claim. A map of sampling locations establishes where observations were collected, not what caused the observed distribution. A time series can show a change in measured conditions; it may not show the ecological response unless that response is also measured. Figures are evidence within a design, not standalone proof of every sentence in the discussion.
Build a useful figure summary
A concise figure note has three parts: variables, visible pattern, and limitation. For example: “The panel plots measured water temperature against the number of detected drifters per tow; counts are generally higher in the cooler samples; sampling covers two sites during one week.” This wording conveys what can be seen and signals the bounds of the evidence without dismissing the result.
Use captions as an entry point, then return to the main text for how measurements were made and how the authors analyzed them. If the plot appears to contradict the prose, check the scales, panels, groups, and units before concluding there is an error. Slow, literal reading of a figure often reveals whether a paper's central summary is well matched to its underlying data.
Have a correction or a source to suggest? Send a note →