Sea urchin size surveys are easiest to understand when every number is tied to an actual animal, a precise way of measuring the hard test, and a defined area of seabed. A report may show a large mean test diameter, many juveniles, or a fall in density, yet these patterns can arise from different sampling methods rather than biological change. This guide explains how to read the measurements, recalculate the denominators and communicate what the survey can legitimately show.
The test is the rigid body of the urchin beneath its spines. Test diameter is not spine-to-spine width, and an animal count is not automatically a density until survey area is known. We work through clearly invented teaching datasets, show the arithmetic for abundance and size groups, and distinguish biological signals from observer, habitat, visibility and selection effects. Values are examples for learning, not observations from a named field site.
Quick reading route: identify the species and survey question; find the measurement protocol; check units and size-bin endpoints; reconstruct belt-transect or quadrat area; treat zeros and missing survey records differently; decide how sites and depths should be weighted; compare matched years; then state uncertainties and alternative explanations before describing a population trend.
What the test measures, and what it does not
The test consists of fused skeletal plates forming the urchin body; the long spines are external appendages and must not be included in a stated test-diameter measurement. A caliper applied across intact spines gives an entirely different dimension. Some surveys measure the maximum horizontal test diameter, others adopt a specified axis or a photographic approximation. A translator should not quietly turn all of them into body width. Record whether live animals were handled, whether the test was irregular, and whether the measuring device contacted the skeleton directly.
Suppose an animal is 63 mm across its test while the visible spine tips span 118 mm. The first number belongs in a test-size frequency chart; the second may describe the occupied space or a defensive silhouette. Averaging the two would yield 90.5 mm, a number that measures neither feature. If an old report uses the vague word diameter, describe the ambiguity rather than silently treating it as test diameter.
Repair the wording by saying ‘maximum test diameter, spines excluded, measured to the nearest millimetre’ only when the protocol supports those details. In a translated field sheet, keep the measurement name, the unit and the physical landmark together. If the method is missing, write ‘reported diameter; inclusion of spines not specified.’ That preserves the limitation for future analysts.
Species identification comes before a combined histogram
Different sea urchin taxa can have different test shapes, adult dimensions, habitats and recruitment histories. A size distribution combining two species may become bimodal simply because the sample contains two different body plans. Even within a single genus, a field label such as ‘black urchin’ may refer to more than one species in different places. When species identity is uncertain, a safe record uses a genus-level or unidentified category rather than inventing precision.
Imagine 40 animals from taxon A averaging 45 mm and 60 from taxon B averaging 80 mm. The pooled mean equals (40 times 45 plus 60 times 80) divided by 100, or 66 mm. Neither taxon has a mean of 66 mm. A chart without a species legend could mislead a manager into thinking the common animal changed size, when only the species mixture changed. The correct comparison needs separate species counts and size statistics.
Translate scientific names without replacing them by a local common name that may be ambiguous. Keep a specimen or photograph reference where the study permits it, and annotate uncertain identifications. The ecological question determines whether combined density across taxa is meaningful, but species-specific growth, maturity and survival claims need species-specific evidence.
Millimetres, centimetres and rounded calipers
A tenfold scale error is easy to introduce when a figure labelled centimetres is transcribed into a spreadsheet expecting millimetres. A reading of 6.2 cm is 62 mm, not 620 mm and not 6.2 mm. Precision is a separate matter: a caliper rounded to the nearest millimetre cannot certify hundredths of a millimetre merely because a computer displays two decimal places. Instrument resolution, observer variation and biological variation must not be conflated.
Suppose five animals are written as 5.8, 6.0, 6.3, 6.5 and 6.9 cm. Their corresponding millimetre values are 58, 60, 63, 65 and 69, with a mean of 63 mm. If the transcriber changes the unit header without converting the values, the reported mean becomes 6.3 mm and the biological interpretation collapses. A simple range check against the original field sheet would catch the problem.
Keep raw entries and converted entries in different columns. State any rounding convention in the methods and avoid repeatedly rounding intermediate arithmetic. When older data mix units, mark the source unit for every observation and make the conversion reproducible; a clean-looking column is not proof that all observations were recorded on the same scale.
Size bins are intervals with explicit boundaries
A size-frequency chart groups individual diameters into intervals. Labels such as 0–20, 21–40 and 41–60 mm are straightforward if diameter was rounded to whole millimetres. For measurements with decimals, the analyst must specify whether bins are half-open, such as 20 to less than 40 mm, or use another unambiguous convention. Boundary choices determine which column receives an animal exactly at 40.0 mm.
Consider four sizes: 19.9, 20.0, 39.9 and 40.0 mm. With intervals [0,20), [20,40) and [40,60), the counts are one, two and one. If a second analyst treats both ends as inclusive and records 20.0 in the first and second bins, the total becomes five for only four animals. That is a boundary error, not a biological increase.
Read the axis note and source code where available. Preserve an explicit lower-bound-inclusive, upper-bound-exclusive rule in any bilingual data dictionary. A reader should be able to reconstruct the count from raw lengths without guessing where endpoint values went. Resist interpreting apparent cohort peaks when bin definitions changed between surveys.
A frequency count differs from a proportion
Frequency is the number of animals within a category. Relative frequency divides that number by the appropriate total and reports a proportion or percentage. The distinction matters when sample sizes differ. Twenty small urchins from a census of 40 are half the observations; twenty from a census of 200 are only one tenth. Describing both as ‘twenty juveniles’ obscures the distributional difference while a claim of recruitment may depend on how animals were selected.
Suppose survey A measured 50 animals: 15 below 40 mm, 20 from 40 to under 60 mm, and 15 at least 60 mm. The relative frequencies are 30%, 40% and 30%. Survey B measured 100 animals with 20, 50 and 30 in the same bins, giving 20%, 50% and 30%. The small-size count increased from 15 to 20, but its share fell.
Translate ‘number measured’ separately from ‘number observed’. A deliberately capped size subsample of 50 animals should not be used as though it were the total field abundance. Where samples were taken conditional on detection or accessibility, relative frequencies describe the observed sample, not necessarily every animal living in the habitat.
A belt transect becomes density only with area
An observer who counts animals in a known rectangular belt is sampling an area, not just a line. If belt length is L metres and searched width is W metres, nominal planar area is L times W square metres. Density is the number of detected animals divided by that surveyed area. The result is a detected density under a stated method unless detectability has been estimated and corrected. A line length alone cannot be used as the density denominator.
In a training example, four 30 m by 2 m belts each cover 60 square metres. Together they survey 240 square metres. If 96 animals are counted, the observed density is 96/240 = 0.40 animals per square metre. Dividing 96 by 120 m of line would instead give 0.80 animals per metre, a different physical unit and a different question. The width must never disappear during translation.
The NOAA survey description for Pacific reef habitats uses 2 m by 30 m swaths and records sea urchin counts and sizes within segments. That is an example of a defined method, not a universal protocol for every shore or study. State both geometry and detection conditions instead of reporting a bare ‘urchins per transect’ as though transects always have identical area.
Quadrats provide a different area ledger
A quadrat represents an area with defined sides, often examined for every visible animal of the target category. Results become comparable only when quadrat size and the number actually completed are reported. One square metre and five square metres are not interchangeable units of observation. Partial or obstructed quadrats require a recorded rule; otherwise the denominator overstates the searched area and understates density.
Imagine eight 1 m by 1 m quadrats with counts 0, 1, 2, 1, 3, 2, 0 and 3. The total is 12 across 8 square metres, or 1.5 observed animals per square metre. A second survey uses eight 0.5 m by 0.5 m quadrats and counts six animals. It sampled only 2 square metres, so its observed density is 3.0 per square metre, not 0.75.
Keep counts and sampled area in the same row or linked record. The average count per quadrat can be useful but cannot be translated into area density without knowing quadrat dimensions. When the study uses unusual plot shapes, derive the effective area from the original method rather than substituting a standard square.
Zero, missing and not surveyed are separate records
A genuine zero states that an observer looked within a valid sampling unit and detected no target animals. A missing value states that the observation was not recorded or cannot be used. ‘Not surveyed’ says the planned unit was not examined. These are distinct situations. Turning blanks into zeros artificially lowers estimated abundance; deleting genuine zeros exaggerates it. Both mistakes are common in data export and translation.
Suppose a table lists ten intended 1-square-metre quadrats. Six are valid with counts 0, 1, 0, 3, 2 and 0, while four were not reached because of conditions. Valid density is 6 animals over 6 square metres, or 1.0. If the four unsurveyed plots become zero, the reported density drops to 0.6, although nothing biological was observed in those plots.
Use separate status fields such as complete, incomplete, missed and invalid, while retaining the measured count only where relevant. A report should state the achieved sample size alongside the planned one. The difference is an effort and coverage limitation; it is not evidence that the inaccessible habitat was empty.
Random placement, convenient placement and patchiness
Sea urchins often aggregate in crevices, on particular substrates or near food, so where a transect is placed matters. Haphazard placement, random selection and permanent fixed plots are different designs. A survey near a convenient entry point may describe that accessible area well but may not estimate an entire coastline without assumptions. A translation that calls every chosen plot ‘random’ can transform an observational convenience sample into false statistical authority.
Consider two microhabitats of equal area: exposed rock with an observed density of 0.2 animals per square metre, and creviced rock with 2.0. A crew deliberately takes nine exposed-rock quadrats and one creviced-rock quadrat. The sample-weighted mean is 0.38, but an equal-area habitat average is 1.1. The difference reflects allocation, not new animal behaviour.
Report the selection mechanism and habitat coverage. The analyst can apply design weights only if a defensible sampling frame or area proportions exist. Do not repair non-random sampling by merely increasing decimal precision or adding a confidence interval. Precision around a biased estimate remains bias.
Depth strata and area weighting
An offshore reef may be separated into shallow, middle and deeper strata, each with different area and urchin distributions. A simple mean of stratum densities answers the question ‘what is the average of these three stratum estimates?’ It may not answer ‘what is the density across the entire mapped habitat?’ The latter requires an area-weighted combination when the area of each stratum differs and the design supports that approach.
Suppose shallow ground covers 10% of mapped habitat at 3 animals per square metre; middle ground covers 30% at 1.5; deep ground covers 60% at 0.5. Estimated habitat-wide density is 0.10 times 3 plus 0.30 times 1.5 plus 0.60 times 0.5, or 1.05 animals per square metre. The unweighted average, (3+1.5+0.5)/3, is about 1.67 and overstates the area-weighted value.
Display stratum area, measured effort and estimated density separately. Make the target population clear: all mapped seabed, only suitable habitat, or only surveyed patches. If mapped area is uncertain, describe that uncertainty and do not present a weighted value as an exact census.
Detection is not identical to presence
Spines, body size, turbidity and refuge behaviour all influence whether a diver or camera can see an urchin. One animal concealed in a crevice counts as present biologically but may count as undetected observationally. Observed density is therefore a property of both animals and survey technique. Formal detection corrections require a justified model or repeated observations; they cannot be guessed from a convenient multiplier.
An illustrative model takes 80 detected animals in 200 square metres, yielding 0.4 observed animals per square metre. If an independent and appropriate detection study supported probability 0.8 under matched conditions, a simple estimate would be 80/(200 times 0.8) = 0.5 per square metre. Without that evidence, reporting 0.5 as ‘true density’ would be an invented correction.
Describe torch use, survey speed, crevice inspection, substrate and water visibility. NOAA’s documented transect methods sometimes specify divers inspecting crevices with flashlights precisely because visibility is part of counting effort. Keep detection assumptions beside any derived number, especially across years when observers or equipment changed.
Observer calibration and measurement repeatability
Two competent observers may differ slightly on where the widest test diameter lies or whether a partly buried animal is inside a plot. The correct response is not to erase disagreement but to estimate repeatability and document the protocol. Calibration sessions with the same set of animals or photographs can reveal systematic offsets and inconsistent classification at category boundaries.
Suppose observer A records the same five tests as 50, 61, 44, 72 and 55 mm, while observer B records 51, 60, 45, 73 and 54. Differences B minus A are +1, -1, +1, +1 and -1 mm, averaging +0.2 mm. That small average does not mean there is zero measurement error; individual differences still range by a millimetre. A second exercise should compare spread and large-animal performance.
Retain observer ID and calibration date. If a historical series has a change of survey staff, mention it as a plausible discontinuity rather than instantly declaring a population shift. Sensible quality control flags implausible measurements and compares repeated readings before any correction is applied.
Mean, median and the long upper tail
Mean diameter sums observed diameters and divides by the number measured; the median is the middle ordered value. When a few very large animals occur among many smaller ones, the mean can drift upward while the median remains anchored to the typical individual. Neither statistic by itself describes how many recruits or harvest-sized animals are present.
Consider measured tests of 20, 22, 24, 25 and 89 mm. The mean is 36 mm because their sum is 180 mm across five animals, while the median is 24 mm. A report saying ‘average size is 36 mm’ might sound as though most animals are roughly that large; the raw observations reveal the opposite. A box plot or the complete distribution conveys the pattern better.
Report the sample size and distributional spread, and label which measure of central tendency was used. A translated abstract that changes ‘median’ to ‘mean’ invalidates the published interpretation, even if the numbers happen to look similar. The size metric must travel unchanged across languages.
Variation is not the same as uncertainty in the mean
A standard deviation describes variation among measured individuals; a standard error describes estimated sampling uncertainty in the sample mean under particular assumptions. They do not answer the same question. The formula standard error = sample standard deviation divided by the square root of sample size is not automatically valid when animals were clustered within only a few transects and treated as independent replicates.
Suppose 100 independent measurements have sample standard deviation 12 mm. The naive standard error is 12/10 = 1.2 mm. But if all 100 animals came from two adjacent patches, the effective independent spatial sample size may be much smaller. Adding more animals within the same patch refines local description but cannot establish the coast-wide mean with the confidence of 100 independent sites.
State the real sampling unit: individual, quadrat, transect or site. A confidence interval should reflect the study design and hierarchical structure where feasible. Never translate an individual-level standard deviation as a measurement tolerance, and never use the sign ‘±’ without saying which uncertainty measure follows it.
Missing small animals and size-selective detection
Very small urchins may hide more successfully, blend with substrate or be excluded by a field minimum. If the smallest visible group is undercounted, an apparent absence of new recruitment may be a detection artefact. Conversely, a new macro-photography method can reveal many juveniles in a later year and make recruitment look stronger than it is under a comparable protocol.
Imagine 100 animals with diameter at least 40 mm are recorded in both years. In year one, a rapid diver survey also finds five below 40 mm; in year two, a careful crevice search finds 30. The raw juvenile share changes from 5/105, or 4.8%, to 30/130, or 23.1%. Without standardized detection effort, this difference cannot be confidently assigned to a spawning or settlement event.
Document minimum size detectable or recorded, search detail, lighting and photograph scale. If methods cannot be reconciled, compare only a defensible common size range and say so. That gives up some information but avoids a false ecological story.
Sampling fixed numbers creates a different denominator
Some programmes measure a fixed number of individuals at each site, even if many more are counted. This supports comparisons of size distributions under the chosen collection protocol, not abundance from size-subset counts. If a site contains 1,000 observed animals and the investigator measures 50, a histogram total of 50 is not the population total and must not appear in a density equation.
Suppose site A has 300 counted urchins and 50 measured; site B has 30 counted and all 30 measured. A size table has 80 records, but the combined observed abundance is 330. Dividing the size table by total surveyed area will estimate density from a deliberately selective subset and could understate it more at the crowded site. Keep census, subsample and measured-complete flags apart.
Any synthesis should report the size-sampling rule and how the chosen animals were selected. Randomly selecting 50 may permit a distribution estimate for that site; grabbing the easiest 50 may not. Unequal site sample sizes and unequal site areas also require separate weighting decisions.
An observed large-size class is not automatically mature
Test diameter is related to life history but does not independently demonstrate gonadal maturity, successful spawning or marketable roe. Species, season, food environment and physiological condition can change those relationships. A size class chosen for management may be a legal or operational threshold rather than a universally valid maturity boundary. This distinction matters in translations of fisheries assessments.
If a chart shows 75 animals above a locally specified 60-mm threshold among 250 measured, the fraction above that threshold is 30%. It is not correct to write ‘30% of all animals are mature’ without evidence linking size to maturity for the relevant species and sampling period. The expression ‘large-size group’ or ‘above the defined management size’ is safer.
For gonad index, measured roe grades and harvestable tissue see the separate eduKateSG article on gonad index. Keep the diameter histogram as a size observation and let reproductive studies answer reproductive questions. A precise boundary in a chart is not proof of a biological step change at that diameter.
Cross-sectional size does not by itself measure growth
Growth is change in an individual or cohort through time. A cross-sectional histogram contains sizes at one moment. Comparing two such histograms can suggest change but cannot separate growth, recruitment, death, movement or selective sampling on its own. Repeated measurement of identified animals, validated cohort models or other longitudinal evidence is needed before a growth rate deserves that name.
Suppose the median test diameter rises from 48 mm in spring to 54 mm in autumn. The numerical difference is 6 mm over the interval, but it is not necessarily each animal’s gain. If new small animals were missed later or larger individuals moved into the plots, the same medians could emerge without the claimed growth. An identified cohort averaging 48 then 54 mm under a valid design would answer a different question.
Use wording such as ‘median measured size increased’ for repeated cross-sections. Only translate it as ‘urchins grew 6 mm’ when the study actually followed the animals or supports a defined cohort inference. Keep seasonal timing and measurement methods explicit.
Length–weight models are empirical, not identities
A fitted relationship between test diameter and wet mass is often summarized as W = aD^b, where the coefficients and units depend on the species, sample, conditions and fitting method. These models can be useful for estimating biomass but do not mean weight is mathematically fixed by diameter for every animal. A log-linear fit and a fit on the original scale also have different error properties.
For a purely hypothetical model W = 0.0008 times D cubed when D is in millimetres and W in grams, a 50-mm animal predicts 100 g, because 50 cubed is 125,000. A 60-mm animal predicts 172.8 g. Increasing diameter 20% has raised predicted mass 72.8% under this assumed cubic model. This is a model illustration, not a species-specific calibration.
Name the fitted equation, sample range, species and units. Never convert diameter from mm to cm without transforming the coefficient. A model extrapolated beyond the measured range can give impressively precise but biologically unsupported biomass values.
Biomass needs abundance and a defensible mass estimate
To estimate the standing wet biomass of urchins in an area, an analyst must combine an abundance estimate with average individual mass or a size-to-mass function. This is a different calculation from density or mean test diameter. Variation in body condition and the mass of retained water can influence live mass, so the exact mass protocol matters.
Consider a site with estimated observed density 1.5 animals per square metre over 1,000 square metres of suitable habitat. This corresponds to 1,500 observed-equivalent animals under the design. If representative mean wet mass is 80 g, a simple estimate is 120,000 g or 120 kg. It is not 120 kg of edible roe; most of that mass is not roe. If density and mass are biased, biomass inherits both problems.
Keep ‘whole wet mass’, ‘gonad mass’ and ‘edible graded product’ distinct. If a relationship predicts mass from diameter, expose prediction uncertainty and whether the sample represents smaller animals. A report using extrapolated total area should call the result an estimate, not a counted harvest.
Size-selective harvest changes the observed distribution
Removing larger individuals can shift an observed histogram toward smaller animals without any failure of growth or recruitment. Likewise, closing a fishery or concentrating surveys away from harvested areas may change the observed proportion of large animals. The meaning of a new distribution depends on fishing history, access, substrate and collection method.
Imagine a reef with 40 large and 60 small observed animals. After a selective removal of 20 large animals, the remaining observed distribution is 20 large and 60 small, so the large-size share changes from 40% to 25%. Total animals also fall from 100 to 80. Reporting only the change in mean diameter would hide the abundance change and the mechanism behind it.
Look for harvest dates, minimum legal size, gear, collecting pressure and closures. Do not label a size decrease as environmental deterioration if selective harvest provides an alternative explanation. Equally, do not assume harvesting caused a shift without matched evidence. The data support competing hypotheses until the design discriminates among them.
Site and year comparisons need matched methods
Annual monitoring becomes persuasive when plot locations, depth bands, measurement instruments and search procedures are comparable. A new survey may look more precise simply because it uses better cameras, not because the animals are fundamentally different. Permanently marked transects allow repeated-local comparisons but do not automatically represent places that were never surveyed.
Suppose year one records 70 animals along 200 square metres of belt transects, giving 0.35 per square metre. Year two records 90 animals but surveys 300 square metres, giving 0.30 per square metre. The count increased by 20 while observed density declined by about 14.3%. Neither statement alone gives a full picture, and a year-to-year claim should acknowledge that the sampled spatial footprint changed.
Report observed counts, achieved surveyed area, habitat allocation and survey timing beside every density point. When methods vary, use comparable subsets or a design-aware model and say what assumptions it adds. A tidy line graph without methods can mislead readers more effectively than an untidy but honest table.
Uncertainty intervals and rare patches
An urchin population may be concentrated in a few patches, making a standard normal confidence interval unreliable when most transects are empty and a few contain many animals. The ecological variation is real. A responsible analyst may summarize the distribution of transect-level densities, use an appropriate design-based interval or bootstrap independent sampling units, and avoid a false impression of symmetric error.
Suppose four equal-area transects contain 0, 0, 2 and 38 animals. The mean is 10 animals per transect, but three quarters of transects contain at most two. An individual-based average conceals the patch structure. Applying a simple margin of error to the 40 pooled animals while ignoring among-transect heterogeneity would make the estimate look unjustifiably stable.
Explain the chosen uncertainty method and its sampling unit. Wide intervals can be an honest result rather than a weakness to hide. When only four independent transects exist, readers need that small replication count and the raw spread, not a table crowded with unsupported decimals.
Photogrammetry changes both scale and visibility
Photographs can reduce handling and preserve visual records, but diameter cannot be read in physical units without scale, camera geometry or a calibrated reconstruction. Perspective can make a body farther from the camera look smaller, and spines can hide the margin of the test. Software segmentation may consistently place an edge that differs from direct caliper placement.
Suppose a known 100-mm scale bar occupies 250 pixels and a test occupies 150 pixels in the same measurement plane. The estimated test diameter is 150/250 times 100, or 60 mm. If the scale bar is closer to the camera than the urchin in a perspective image, this proportional calculation may be biased. The arithmetic can be correct while the geometric assumption fails.
Retain image ID, scale-bar location, camera angle and whether a correction was used. Compare representative photographed animals with direct measurements where possible. A translation should describe ‘image-derived diameter’ if that is what was measured, rather than silently recasting it as a physical caliper reading.
A data dictionary keeps the meaning of every field
Most errors arise when individual numbers are separated from column definitions. A useful dictionary specifies sample_id, species, site, stratum, date, observer, sample_method, survey_length_m, survey_width_m, area_m2, test_diameter_mm, size_bin, counted_total, measured_total and record_status. A field also needs allowed missing codes and any derived formula.
For example, a row with area_m2 = 60 and count = 24 yields detected_density_m2 = 0.4. If the same row has measured_total = 12, those 12 size records must not replace the 24 count records. A bilingual file might distinguish ‘observed’, ‘measured’ and ‘estimated’. Translating every one as ‘collected’ would introduce an unsupported handling claim and corrupt later reporting.
Write terms once, define them in plain language and link each derived field to its numerator and denominator. Keep a small provenance column identifying the source sheet and processing version. A later analyst should be able to reverse the calculations without relying on a vanished email explanation.
Worked synthesis: reconstructing a multi-site report
Consider a fictional survey with three equal-width 30-by-2-m transects at site North and two such transects at site South. North has counts of 12, 18 and 0; South has counts of 25 and 15. The areas are 180 and 120 square metres, respectively. North observed density is 30/180 = 0.1667 per square metre; South is 40/120 = 0.3333. The pooled density across actual surveyed area is 70/300 = 0.2333, rather than the simple unweighted mean of 0.25.
Now suppose North measured only 20 of its 30 animals and South measured 20 of its 40. Of the measured animals, 8 in North and 4 in South are below 40 mm. The measured-sample small-size shares are 40% and 20%, respectively. They do not establish the true sitewide juvenile fractions unless the measured subsets are representative. The combined 12/40 = 30% describes the pooled measured sample, not a coastwide random census.
A faithful explanation reads: ‘Seventy urchins were observed over 300 square metres; site-specific detected densities differed. Small-animal frequencies come from a limited size subsample, and their representativeness was not established.’ This is less dramatic than ‘recruitment doubled in the north’, but it is supported by the actual numbers.
Independent transfer problems and checked answers
Problem one: three valid 2 m by 25 m belts contain 10, 15 and 20 observed urchins. Find total surveyed area and detected density. Answer: each belt covers 50 square metres, total 150; the count is 45, and 45/150 = 0.30 animals per square metre. A denominator of three belts yields 15 per belt, a valid separate metric but not area density.
Problem two: a size sample has 10 animals below 40 mm, 30 from 40 to under 60 and 20 at least 60. Find the three relative frequencies. Answer: total measured is 60; proportions are 16.7%, 50.0% and 33.3%, subject to rounding. These are proportions of measured animals, not of all animals observed unless the sampling protocol establishes that equivalence.
Problem three: two strata occupy 25% and 75% of habitat and have estimated densities 2.0 and 0.4. Calculate area-weighted density. Answer: 0.25 times 2.0 plus 0.75 times 0.4 = 0.8 per square metre. An unweighted mean of 1.2 assigns equal habitat weight when the areas are unequal. Problem four: a blank record means bad weather prevented surveying. Answer: it is missing effort, not a zero count.
Frequently asked questions and decision rules
Does a larger mean diameter prove healthier animals? No. Mean size responds to recruitment, mortality, selective harvest and the mix of species or sites, while individual health requires additional evidence. Does a high number of measured juveniles prove recruitment increased? Not unless observers used comparable detection methods, size bins and spatial coverage. Does 0 per square metre mean extinction? No; it means zero detected in the specified area and conditions.
Can the number of urchins observed be called total population size? Only for an explicitly complete census of a bounded population. Most field transects sample part of a larger habitat and require justified extrapolation. Should an ecological abstract report a single significant digit? It should preserve measured precision and uncertainty, not round so heavily that two distinct values become falsely equal. Should an uncertain species name be translated into a definitive name? No.
The practical rule is to retain definitions alongside numbers: test versus spines; measured versus counted; observed versus inferred; site versus habitat; random versus convenient sample; area density versus per-transect count; and change in observed distribution versus change in living population. A good explanation does not remove the caveats that give the numerical result its actual meaning.
Sources and method references
Original methodological sources: NOAA NCEI archive description of 2 m × 30 m reef swath surveys, including sea urchin size counts: https://www.ncei.noaa.gov/archive/archive-management-system/OAS/bin/prd/jquery/accession/details/277329 ; NOAA Pacific coral reef survey data dictionary with test-size bins and transect fields: https://www.fisheries.noaa.gov/inport/item/73796 ; Edmunds PJ and colleagues, long-term Diadema monitoring and size methods in PNAS: https://doi.org/10.1073/pnas.2218901120 . These are method examples, not universal measurement laws.
Useful eduKateSG routes
Related reading on eduKateSG: How to Read Sea Urchin Basket Stocking Density, Surface Coverage and Growth: https://edukatesg.com/2026/11/01/how-to-translate-easily-sea-urchin-basket-stocking-density-coverage-growth/ ; How to Read Sea Urchin Gonad Index, Roe Colour Grades and Marketable Yield: https://edukatesg.com/2026/11/01/how-to-translate-easily-sea-urchin-gonad-index-roe-colour-marketable-yield/ ; How English Works | Vocabulary and Meaning: https://edukatesg.com/how-english-works-vocabulary-and-meaning/ .