A sandfish larval tank can look full of life while its daily spreadsheet makes little sense. One column may record auricularia larvae per litre, another the number seen in a microscope subsample, and a third the microalgae cells added as feed. All three may be labelled count, yet their sampling units, biological meaning and possible errors differ fundamentally.
This article explains how to translate Holothuria scabra larval rearing observations into meaningful estimates of stage composition, tank population, food concentration and apparent survival. The route moves from first principles to numerical examples, then to diagnostic questions and independent exercises. Values introduced for arithmetic are hypothetical, never alleged hatchery measurements or instructions to change a facility’s protocol.
The most important distinction is between what a sample directly establishes and what a larger-scale estimate assumes. A counted subsample is evidence about that sample. Estimating an entire tank adds mixing, volume, sampling and classification assumptions. A change in calculated density may also reflect water exchange or animal movement rather than genuine mortality.
How to read the full larval-rearing chain
From embryos to feeding auricularia
The mechanism has to be stated before the number is interpreted. Sandfish develop from fertilised eggs through embryonic states to auricularia larvae, then doliolaria and pentactula stages before juveniles. Auricularia is a swimming, feeding phase; it is not a generic word for every early sea cucumber. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
Return to the notebook or microscope sheet, not just the chart caption. Log the observed stage, the time since spawning or fertilisation, the sample identifier and the staging criteria. Mixed stages should appear as separate categories, with uncertain classifications marked uncertain instead of forced into an attractive uniform group. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
A numerical example makes the denominator visible. A sample of 200 individually classified organisms may contain 132 auriculariae, 48 earlier stages and 20 indeterminate forms. The observed auricularia fraction is 132/200, or 66 percent, only for that examined sample. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
There is a plausible competing explanation that the neat summary obscures. A photograph at one magnification may not reveal the features that distinguish adjacent stages. Delayed development, damaged specimens and poor focus can make a stage fraction seem more precise than the biological evidence allows. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The repair is a change in documentation and measurement logic, not merely a better sentence. Have assessors use a shared staging guide, blinded rechecks when possible, and the actual developmental structures visible under the microscope. Retain images and a category for unresolved specimens rather than guessing. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
The language in a technical translation carries the measurement model. Translate larva, post-larva, juvenile, swimming form and feeding stage according to morphology, not according to how old the tank is. Age in days is context; it is not a substitute for observed identity. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
Now change one assumption to test understanding. Invent a mixed-stage sample in which 20 percent are indeterminate. Show how excluding those animals changes the denominator and why both the all-specimen and interpretable-specimen proportions should be reported. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Obtaining a representative tank sample
An important first-principles distinction gives the rest of the result its meaning. A tank census usually begins with a finite water subsample, not a direct count of all swimming animals. Whether the subsample represents the whole population depends on distribution, mixing, vertical movement and the timing of collection. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
A reliable log separates the actual observation from later calculations. Document vessel volume, sampling depth, distance from aeration or inlets, sampling time, mixing before withdrawal, aliquot size, and whether the sample was returned. Pooling subsamples must not erase their distinct positions. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
This is where a dimensional check can prevent a convincing but impossible result. If four separate 50-millilitre samples contain 11, 13, 10 and 12 larvae, the mean is 11.5 larvae per 50 millilitres, equivalent to 230 larvae per litre. The spread across samples reveals variation, not simply rounding noise. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
The principal interpretation hazard is the jump from a sample to a stronger causal claim. Sampling only at the tank surface may oversample or undersample organisms if the larvae are not vertically uniform. Sampling after a disturbance can also mix densities that normally differ across the tank. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The most useful next step is to make this inference independently checkable. Specify a repeatable sampling grid and timings, and retain raw aliquot counts. For sparse organisms increase subsample volume rather than pretending a single two-larva sample gives a stable tank concentration. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
Precise wording prevents a biological category or physical unit from silently changing. Translate representative sample as a claim needing evidence. Random, systematic, pooled and convenience samples are not interchangeable labels; each describes how the observations were selected. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
The following transfer question shows why memorising one formula is insufficient. Compare two 50-millilitre aliquots, one with 2 larvae and another with 18. A simple average is possible, but explain why it deserves an uncertainty discussion before scaling to millions of millilitres. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Scaling sample counts to total tank population
The mechanism has to be stated before the number is interpreted. Population estimate equals observed concentration multiplied by effective working volume, provided the sampled distribution represents that volume. Keeping millilitres, litres and cubic metres distinct prevents thousandfold errors. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
Return to the notebook or microscope sheet, not just the chart caption. Read tank dimensions, water-level calibration, displacement by equipment, and any recirculation volume separately. The relevant volume is where larvae actually reside during sampling, not automatically the manufacturer’s maximum capacity. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
A numerical example makes the denominator visible. At 230 larvae per litre and 2,500 litres measured working volume, the scaled estimate is 575,000 larvae. That is a model-based estimate, not an independently verified census of 575,000 individual organisms. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
There is a plausible competing explanation that the neat summary obscures. An estimate changes if water depth falls or a screened compartment is excluded. A concentration measured after a partial drain cannot be multiplied by the previous morning’s volume without reconciliation. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The repair is a change in documentation and measurement logic, not merely a better sentence. Write the equation in words and units before using a calculator; show concentration times effective volume, and attach the timestamp of both measurements. Preserve a plausible range when counts vary across subsamples. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
The language in a technical translation carries the measurement model. Translate larvae per millilitre into larvae per litre by multiplying by 1,000; converting litre-based density to cubic-metre-based density multiplies by another 1,000. The result changes numerically, not biologically. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
Now change one assumption to test understanding. If the same 230 larvae per litre occur in 1,200 litres rather than 2,500 litres, calculate 276,000 estimated larvae. The concentration is identical, but the estimated tank population is very different. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Microalgae as feed, not a larval population
An important first-principles distinction gives the rest of the result its meaning. SEAFDEC/AQD identifies Chaetoceros calcitrans as food for feeding sandfish auriculariae. The algal cells are not larvae, and their concentration is an environmental input rather than a count of sea cucumber offspring. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
A reliable log separates the actual observation from later calculations. Record algal species or culture identifier, culture age, quality observation, cell density, any dilution, delivered volume, feeding time and whether the cell target refers to an initial dose or a maintained concentration. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
This is where a dimensional check can prevent a convincing but impossible result. Suppose a culture contains 12 million algal cells per millilitre and an operator adds 5 litres. Five litres are 5,000 millilitres, giving 60 billion added cells before any losses, consumption or nonuniform distribution. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
The principal interpretation hazard is the jump from a sample to a stronger causal claim. A feeding sheet that states 30,000 cells per millilitre without defining the tank volume or measurement time cannot reconstruct the amount introduced. Counting fragments as cells can also skew a microscope tally. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The most useful next step is to make this inference independently checkable. Keep counts from the algal stock and from the larval tank on separate lines. Label measured versus target concentration, and record whether a value was calculated from a dose or actually sampled. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
Precise wording prevents a biological category or physical unit from silently changing. Translate cell density, culture concentration and feed ration with exact units. In particular, a cells-per-millilitre target is a concentration, while litres of culture delivered is a volume, and these are not synonyms. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
The following transfer question shows why memorising one formula is insufficient. Two concentrated cultures have the same delivered volume but one has half the cell density. Calculate their different delivered cell totals and explain why a litre of culture is not a consistent feeding unit. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Feed-dose arithmetic and dilution
The mechanism has to be stated before the number is interpreted. Turning a culture stock into an intended tank concentration is a material-balance calculation. The simple product of stock concentration and added volume must be related to the water volume and any preexisting algae. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
Return to the notebook or microscope sheet, not just the chart caption. Track the culture’s measured cells per millilitre, the stock volume added, working tank volume, background cell counts and expected loss processes. The calculation is only a planning estimate when consumption and filtration are active. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
A numerical example makes the denominator visible. For 2,000 litres of water, a hypothetical target increase of 30,000 cells per millilitre represents 60 billion added cells. From a stock of 12 million cells per millilitre, that requires 5,000 millilitres, or 5 litres, before losses. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
There is a plausible competing explanation that the neat summary obscures. If a tank already holds 10,000 cells per millilitre of the same counted organism, adding the full target dose overshoots the incremental target. Unidentified plankton and species mixtures complicate this further. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The repair is a change in documentation and measurement logic, not merely a better sentence. Distinguish target final concentration from target incremental addition; write down which formula was used. Check whether culture volume itself materially changes the working volume, especially in small test vessels. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
The language in a technical translation carries the measurement model. Translate increase to, increase by, maintain at and dose with as four different instructions. Losing these prepositions can turn an accurate formula into an incorrect procedure. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
Now change one assumption to test understanding. A smaller tank of 500 litres requires one quarter of the cells for the same intended concentration increase, all else equal. Calculate 15 billion cells and the corresponding 1.25 litres of the stated stock. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Water exchange and changing denominators
An important first-principles distinction gives the rest of the result its meaning. Larval rearing logs often report percent water exchange. This represents a volume operation, not automatically the proportion of larvae lost, the proportion of feed removed, or the share of original water remaining. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
A reliable log separates the actual observation from later calculations. SEAFDEC/AQD describes periodic 20–30 percent exchanges during a specific sandfish larval-rearing practice. The written method also matters: screened retention, top-up sequence, mixing and sampling time all change what an exchange means. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
This is where a dimensional check can prevent a convincing but impossible result. In a 1,500-litre tank, a nominal 20 percent exchange corresponds to 300 litres replaced. Under ideal mixing, three sequential 20 percent replacements leave 0.8 cubed, or 51.2 percent of the initial water fraction, not 40 percent. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
The principal interpretation hazard is the jump from a sample to a stronger causal claim. Larvae retained by screens may remain in the tank while dissolved material is diluted. Screen failure, siphoning and mechanical damage create additional pathways that cannot be inferred from exchange percentage alone. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The most useful next step is to make this inference independently checkable. Reconcile withdrawal, replacement and measured water level; record screen condition, observed entrainment, timing of feeding and sampling. Estimate losses only from actual evidence instead of assigning every population decrease to water exchange. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
Precise wording prevents a biological category or physical unit from silently changing. Translate exchanged, discharged, topped up, recirculated and completely renewed precisely. If a log uses turnover, identify whether it means volumetric exchange or system flow per unit time. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
The following transfer question shows why memorising one formula is insufficient. Given 900 litres operating volume and two 150-litre replacements, calculate each event’s fraction of the tank. Then explain why adding those percentages is not generally the same as the fraction of initial water removed under mixing. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Apparent survival versus true biological survival
The mechanism has to be stated before the number is interpreted. A population estimated by subsampling on two dates can support an apparent survival ratio, but mortality is not the only reason counts decline. Transfers, release, entrainment, sampling bias and stage-specific detectability all matter. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
Return to the notebook or microscope sheet, not just the chart caption. Record baseline stocked population, later estimates, removals, tank mergers, filter losses, mortality observations and sampling method. Also identify whether the later population includes all developmental stages or only those easy to recognise. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
A numerical example makes the denominator visible. If 200,000 embryos were estimated at baseline and 160,000 larvae later estimated, the simple apparent proportion is 80 percent. This is not necessarily a controlled estimate of biological survival for a closed unchanged cohort. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
There is a plausible competing explanation that the neat summary obscures. A tank can appear to gain larvae when mixing improves or counting technique changes. A ratio above 100 percent should trigger an audit, not an assertion that larvae multiplied during that developmental stage. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The repair is a change in documentation and measurement logic, not merely a better sentence. Provide a transparent flow ledger and sensitivity analysis. If the later count could plausibly vary by 20 percent due to sampling, communicate the resulting range rather than a false single-decimal precision. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
The language in a technical translation carries the measurement model. Translate apparent survival, recovery, retention, hatch percentage and settlement yield as different statistics. A cohort percentage has meaning only when its start and endpoint populations are explicitly defined. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
Now change one assumption to test understanding. A cohort starts with 400,000 estimated larvae, transfers 60,000 out, and retains 280,000 by the next survey. Explain why 280,000/400,000 understates retention in the original tank if out-transfers are ignored. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Uncertainty from aliquots and observers
An important first-principles distinction gives the rest of the result its meaning. Microscope counts are measurements subject to sampling variation and classification disagreement. A careful reader distinguishes variance among aliquots from systematic sampling bias and from uncertainty in effective tank volume. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
A reliable log separates the actual observation from later calculations. Retain individual count values and calculate dispersion before averaging. Log who classified stages, the field of view or chamber definition, cleaning and calibration procedures, and any inter-observer disagreement. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
This is where a dimensional check can prevent a convincing but impossible result. For a set of four 50-millilitre aliquots with counts 11, 13, 10 and 12, the mean is 11.5 and sample standard deviation is about 1.29 counts. This describes those aliquots, not every source of tank-level uncertainty. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
The principal interpretation hazard is the jump from a sample to a stronger causal claim. Very small counts behave differently from large ones, and nonuniform distribution may violate the assumptions of a simple Poisson model. More aliquots can address random error but not a systematic sampling location bias. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The most useful next step is to make this inference independently checkable. Use repeat sampling from distinct locations, show ranges, and refrain from declaring treatment effects when intervals overlap or the sampling design cannot separate day effects from method changes. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
Precise wording prevents a biological category or physical unit from silently changing. Translate precision as repeatability and accuracy as closeness to truth. A result can be highly repeatable but biased if the same unrepresentative sampling location is used every day. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
The following transfer question shows why memorising one formula is insufficient. Ask a colleague to classify twenty saved microscope images without access to the first observer’s labels. Report agreement and disagreements rather than silently overwriting the original record. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Diagnosing a declining feed or larval count
The mechanism has to be stated before the number is interpreted. When algal cell concentration drops, the immediate cause may be larval feeding, filtration, dilution, cell death or measurement drift. When larval counts drop, additional biological and sampling explanations arise; the two curves should not be conflated. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
Return to the notebook or microscope sheet, not just the chart caption. Compare time-aligned records of feed addition, tank volume, water exchange, larval stage, temperature, salinity, microalgae cell samples and mortality observations. Draw a timeline rather than pairing unrelated daily averages. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
A numerical example makes the denominator visible. Suppose tank algae concentration falls from 30,000 to 12,000 cells per millilitre across an interval while 25 percent water is exchanged. Even perfectly mixed replacement with algae-free water would predict a fall to 22,500 before other processes. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
There is a plausible competing explanation that the neat summary obscures. Calling the entire decline consumption would therefore overstate feeding. Conversely, assuming every decline is dilution would hide possible growth, grazing and measurement differences. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The repair is a change in documentation and measurement logic, not merely a better sentence. Reconstruct a simple mass balance first, add plausible flux terms, then name which inputs are unknown. If competing explanations remain, propose the specific next measurement that could discriminate between them. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
The language in a technical translation carries the measurement model. Translate correlation, change, consumption and inferred uptake carefully. A time-series association is not proof of a unique process without experimental or mass-balance evidence. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
Now change one assumption to test understanding. For the same scenario, compute the observed 18,000-cells-per-millilitre decrease and the 7,500-cells-per-millilitre change expected from one ideal 25 percent dilution. State the unexplained difference with caveats. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
From laboratory notes to an honest published statement
An important first-principles distinction gives the rest of the result its meaning. The end of the analysis is a narrative that can survive independent scrutiny. It should tell the reader what was counted, how it was scaled, which conditions changed, what uncertainty remains and what the evidence does not prove. Keep the lifecycle and population boundary explicit throughout the report. Stages are observational categories, and their definitions control which counts can be compared as an animal develops rather than being born again.
A reliable log separates the actual observation from later calculations. Maintain a chain from archived observations to final charts: original sample labels, calculation workbook version, excluded images, specimen identification, tank-volume source, and decisions about indeterminate developmental stages. Always state the timestamp, sample selection, units and instrument or microscopy method. Without those, a number that looks like a field measurement may really be an estimated value or a target from an operating procedure.
This is where a dimensional check can prevent a convincing but impossible result. If a report states 575,000 larvae in a tank, rewrite it as an estimated 575,000 larvae from a mean sampled density of 230 per litre in 2,500 litres, subject to spatial sampling and volume uncertainty. Round only after the units cancel correctly, then interpret the computed quantity in words. This is especially important when data pass through several volume conversions, density factors and population estimates.
The principal interpretation hazard is the jump from a sample to a stronger causal claim. Beware polished charts that omit units or align two results from different observation times. A persuasive graphic can still make an invalid biological comparison when it hides changing denominators. An association can be real without supporting the proposed cause. Reconstruct the timing of sampling and interventions before making an environmental or biological diagnosis.
The most useful next step is to make this inference independently checkable. Show the raw-to-summary route and flag limitations near the claim they constrain. Keep a mechanism-oriented explanation accessible to a learner without pretending the observed farm was a controlled laboratory experiment. Treat missing data as missing instead of assigning a convenient assumed value. Keep the original observations intact, annotate transformations, and tell reviewers exactly where another measurement could change the conclusion.
Precise wording prevents a biological category or physical unit from silently changing. In bilingual documentation, keep organism names, concentrations, stage names and units stable across both languages. When the original source is ambiguous, preserve the ambiguity and add a request for clarification. Translations between specialist languages and everyday English should preserve uncertainty, sample-versus-census distinctions and the domain meaning of rates. A readable phrase that changes the statistic is not faithful.
The following transfer question shows why memorising one formula is insufficient. Complete a short audit: identify one concentration estimate, one inferred population, one unknown loss pathway and one action that would improve measurement on the next observation day. Give the answer together with the assumption that made it possible. If a changed sampling design or missing volume would make the answer unstable, that is itself a useful result.
Fully checked integrated example: a hypothetical two-day tank record
A hypothetical 2,000-litre tank has 150,000 estimated swimming larvae on day one, equivalent to 75 larvae per litre. On day two, five independent 100-millilitre aliquots yield 8, 7, 9, 6 and 10 larvae. The average is 8 larvae per 100 millilitres, or 80 per litre. At an unchanged measured volume of 2,000 litres the estimate is 160,000 larvae. This apparent increase does not establish reproduction: the original and new estimates both have sampling uncertainty, and mixing may have changed.
At the same time, a feed culture of 10 million cells per millilitre supplies 3 litres of algae. The delivery is 30 billion cells. Distributed across 2 million millilitres of tank water, that represents a nominal 15,000 additional cells per millilitre before grazing, removal and growth. A researcher cannot label that nominal number a measured post-feeding concentration without sampling the tank.
Suppose a 20 percent water exchange immediately follows this dosing and the exchanged liquid was well mixed, with replacement seawater containing no target alga. The idealised post-exchange concentration attributable to the dose would be 12,000 cells per millilitre. Screen retention might keep larvae while water leaves, so the same simple 20 percent factor must not automatically be applied to larval abundance. These two quantities occupy different ledgers.
Independent practice: solve before reading the answer
Question 1: Four 100-millilitre samples contain 5, 6, 7 and 10 auriculariae. For a well-mixed tank of 1,500 litres, what is the estimated count? Answer: the mean is 7 per 100 millilitres or 70 per litre, so 105,000 larvae. The spatial representativeness assumption must accompany the result.
Question 2: A culture has 8 million algal cells per millilitre and 2.5 litres are added. How many cells were delivered? Answer: 2,500 millilitres times 8,000,000 equals 20 billion cells. It does not reveal the resulting standing concentration without tank volume and losses.
Question 3: A tank starts with 500,000 estimated larvae, transfers 100,000 out and contains 280,000 at survey. What is the apparent fraction retained or accounted for? Answer: 280,000/500,000 is 56 percent still in that tank; including the documented 100,000 transfer gives 76 percent accounted for, subject to measurement uncertainties.
Question 4: A 30 percent idealised replacement is followed by a 20 percent replacement. What fraction of original well-mixed water remains? Answer: 0.70 times 0.80 equals 0.56, or 56 percent. Simply subtracting 50 percentage points ignores that the second exchange acts on the new mixture.
FAQ and limits
Can larval counts alone identify the cause of a hatchery problem? No. Samples show abundance under a specified method; diagnosing causes requires comparison with stage, water quality, feed, movement and equipment records.
Is microalgal concentration the same as feeding rate? No. Concentration describes standing cells in a volume. Feeding rate additionally specifies time, additions, losses and the biological consumers.
Why not report precise survival to two decimals? A precise printed ratio is not equivalent to a precise estimate. Sampling error, stage classification, transfer losses and missing data may dominate rounding.
Can methods for different sea cucumber species be combined? Only with justified species-specific evidence. Sandfish Holothuria scabra husbandry cannot simply be inferred from Apostichopus japonicus or other taxa.
What should be carried into an English translation? Species, developmental stage, sample unit, volume, timing, raw observation, calculation method and the actual strength of the inference.
First-party sources and verified eduKate reading
References: SEAFDEC/AQD sea cucumber culture overview; SEAFDEC/AQD sandfish hatchery manual; WorldFish research on sea cucumber larval morphology; Google Search Central people-first guidance. Worked data in this article are hypothetical for teaching, not reported field results.
More reading: How to Translate Easily index; English vocabulary in context; sandfish broodstock and spawning records.