Why an arrival survival rate is only the first answer
A sandfish juvenile shipment may be described as a success because nearly all the animals were alive when its bags opened. That is an observation at one endpoint, not evidence that the journey had no biological consequences. It does not tell us the true sealed duration, thermal history, water-quality changes, delayed mortality, or whether the animals later resumed normal burrowing in the destination sediment.
Sandfish, Holothuria scabra, are soft-bodied sea cucumbers moved between hatcheries, nurseries and sometimes restocking locations. Small juveniles differ from larger animals in body mass and handling tolerance. The task here is to read a shipment as a sequence of evidence: what was dispatched, what happened in transit, what was directly observed at arrival and what was eventually recovered after release.
A reliable translation of the records preserves species, life stage, origin cohort, bag number, units, observation times and uncertainty. The examples below are constructed for teaching and not observations from a particular farm. Nothing in an illustrative calculation establishes a universal safe transport density, temperature or duration.
Counts and denominators come before percentages
The numerator in arrival survival should represent animals counted or estimated alive at the stated endpoint. The denominator should represent the appropriate starting population. If sorting removed animals before packing, those animals do not belong in the dispatch denominator. If the starting inventory is sampled rather than enumerated, say so clearly because the result is estimated.
Suppose 4,400 juveniles are estimated in a holding vessel, 400 are removed during grading, and 4,000 are packed. If 3,920 are counted alive on arrival, dispatch-to-arrival survival is 3,920 divided by 4,000, or 98%. Relative to the earlier holding inventory, the figure is about 89.1%. Neither number is inherently wrong, but they describe different transitions, and the removed animals cannot be called transport deaths.
A good ledger retains the original inventory, culling or selection count, dispatch count and arrival classifications as separate fields. The scientific reader should be able to reconstruct every percentage without guessing which animals were eligible for it.
What juvenile loading per litre actually measures
Loading expressed as juveniles per litre is a count divided by a reference liquid volume, not a direct measure of physiological safety. Six hundred animals in three litres produce a numerical loading of 200 juveniles/L. The same count in two litres produces 300 juveniles/L. Both calculations are correct but omit important differences in body size, water chemistry, available surface, mixing and journey duration.
A 5-L-labelled bag may contain only 2 L of seawater with a gas space. Dividing animal count by nominal capacity understates loading relative to the actual water. If an additional 0.4 L slurry is added, the denominator may become 2.4 L of combined liquid, depending on the convention being reported. Specify whether liquid amendments are included instead of silently switching measurement bases.
For example, 600 juveniles in 2.0 L initial seawater have a nominal ratio of 300 per litre of initial seawater. After a 0.4 L addition, they occupy approximately 2.4 L and represent 250 per litre of final liquid. These ratios are not inconsistent, only differently defined. State the chosen basis next to the number.
Bag position, water, gas and packing medium are distinct
A bag may contain seawater, a deliberately added preparation and some gas space. A count per bag does not describe all of these compartments. Headspace composition is not the same as dissolved oxygen concentration; movement, pressure, contact area, water temperature and consumption affect gas transfer. An oxygen-filled headspace cannot be translated into a known water oxygen concentration without measurements.
Bag geometry is also an operational variable. A flexible bag laid horizontally and one standing vertically can distribute liquid and animals differently. Handling, closures and unexpected leaks may change conditions during transit. Record orientation and any repacking or leakage events where they matter instead of assuming identical printed bag dimensions guarantee identical conditions.
An algae slurry may be described in millilitres, cells per millilitre or dry mass. These quantities are not interchangeable. If 0.5 L contains 20 million algal cells per millilitre, its nominal input is ten billion cells. That does not mean ten billion cells were consumed by animals. It is an amount supplied, and any consumption claim needs an appropriate measurement.
The sealed journey is longer than the vehicle journey
Transport exposure starts when a bag is sealed and ends when its contents are released from the sealed environment, not necessarily at truck departure or arrival. Sorting, boxing, loading, waiting and unloading all take time. A written claim that a trip lasted four hours may omit biologically relevant intervals if the writer used only driving time.
Suppose a bag closes at 08:10, the vehicle departs at 09:00, arrives at 13:20 and the bag opens at 14:05. Road travel was four hours and twenty minutes, while the sealed interval lasted five hours and fifty-five minutes. The ninety-five-minute difference should not be erased in a technical account.
Always record event timestamps and the time zone when sites may differ. If an opening timestamp is missing, give the interval that can genuinely be defended from the available evidence rather than inventing a precise exposure duration. An uncertainty statement is more useful than a wrong exact figure.
Why two thermometer readings do not describe exposure
A thermometer reading when packing closes and another when the bag opens provide two points, not a full history. A box may be warmed while waiting in sun and later cooled before arrival. The endpoint reading could appear ordinary while the animals had experienced a significant interval of different conditions. Continuous logging is useful when placement, calibration and interval are known.
Imagine a bag-water logger recording 27°C at packing, 31°C during ninety minutes of loading and 28°C at opening. Reporting only ’28°C at arrival’ hides the measured peak. Reporting only the maximum also conceals the duration. An average may conceal a brief excursion, while a maximum alone cannot tell us whether the same level persisted.
If a sensor monitored air outside the bag, label it as box air rather than bag-water temperature. A time-weighted mean requires attention to irregular logging and gaps. The record should separate measured thermal exposure from an unsupported claim that a temperature was harmless or unsafe.
Cooling claims are conditional, not universal
A published sandfish transport study compared durations, media and ambient versus cooler conditions, then observed later outcomes including burying. The experimental populations and arrangements define what the work directly supports. An isolated result should not be rewritten as a blanket instruction to keep any sea cucumber at a particular temperature or to copy a test duration.
Cooling can affect metabolic oxygen demand, yet rapid changes or unsuitable temperatures can create other stress. The practical interpretation therefore depends on the actual species, stage and acclimation conditions. A source saying cooler bags performed better in one comparison does not prove colder is always better.
The SEAFDEC/AQD sandfish hatchery manual provides first-party production procedures that should be consulted in their full context. This explanatory article does not prescribe a live-animal transport protocol. Its purpose is to keep source limitations, measured values and biological claims separate.
Dissolved oxygen requires a mass balance
Dissolved oxygen is commonly measured as milligrams of oxygen per litre of water. A measurement of 6.5 mg/L in 3 L corresponds to about 19.5 mg of dissolved oxygen in that liquid at that instant, assuming the sample represents the whole compartment. This is an inventory, not a rate of oxygen use by the juveniles.
If the same volume later reads 5.0 mg/L, the inventory difference is about 4.5 mg. It would be premature to call all of this animal respiration because headspace exchange, mixing, leakage and microbial processes can affect the balance. A consumption rate requires compatible measurements over time and a credible accounting of sources and sinks.
‘Packed with oxygen’ reports a packaging method; it does not automatically establish that the water remained saturated. Keep headspace composition and water measurements in distinct fields. The correct scientific translation names the compartment and what was actually observed.
pH, carbon dioxide and ammonia have different meanings
In closed or semi-closed systems, respiration and water chemistry can change during transit. A falling pH is not a measurement of total carbon dioxide and does not by itself identify every cause. Carbonate chemistry also depends on alkalinity and other conditions. A report should preserve the measured variable rather than substitute a plausible explanation as fact.
Total ammonia nitrogen and un-ionised ammonia are not identical. Their relationship varies with pH and temperature. The word ammonia is not sufficiently precise when interpreting analytical results with different forms, units and possible biological consequences. A proper worksheet labels the exact analyte and sample time.
Water sampled after dilution in a receiving tank cannot be represented as the chemistry of the sealed transport bag. A sound account names sampling point and method, and avoids inferring stable chemistry merely because most animals arrived alive.
Arrival, release and delayed observation
The bag-opening endpoint is distinct from release into sediment. Acclimation, transfer and receiving-water conditions may influence later outcomes. The environmental transition can occur after the truck and bag procedures have ended. A study that checks only arrival has not measured longer-term release performance.
Suppose 1,000 juveniles are dispatched, 980 are counted alive on arrival and 910 of the same cohort are confirmed alive after five days. Dispatch-to-arrival observed survival is 98%. Dispatch-to-day-five confirmed survival is 91%. The fraction among live arrivals is 910/980, about 92.9%. These measure three different relations.
The decline from 98% to 91% is seven percentage points on the dispatch denominator. Expressed relative to the 980 live arrivals, the loss of seventy is about 7.14%. Write the denominator explicitly whenever the word percent could refer to either comparison.
Missing is not synonymous with dead
Sea cucumbers can bury or move beyond an observer’s view. A juvenile not found in a search is not automatically a confirmed death. Records need distinct categories for confirmed alive, confirmed dead, missing and unassessed individuals. A single observed count will otherwise create a false impression of certainty.
Imagine a bounded chamber receives 300 live juveniles. At follow-up, 275 are confirmed alive, five dead and twenty not located. The observed minimum living fraction is 275/300, or 91.7%. If every missing animal is alive, the possible living fraction reaches 295/300, or 98.3%, under this simplified classification.
These limits are not a formal confidence interval. They show what the categories allow logically. A defensible estimate between them would require a detection model or additional observations. Do not rewrite twenty missing animals as twenty deaths simply to complete a neat table.
Burrowing is behaviour and must be defined
Sandfish burrowing and emergence can vary with time of day, light, temperature, substrate and recent handling. Research on handling has recorded altered emergence behaviour. A juvenile that remains visible for a while is not thereby dead, nor does rapid burial establish complete physiological recovery. The terms ‘normal’ and ‘recovered’ need operational definitions.
If observers count forty animals visible at noon and ten visible after sunset, they cannot infer seventy-five percent mortality. Individuals may have buried, moved, or become harder to find. A useful behavioural comparison follows a standard observation schedule and specifies what qualified as burial or other functional response.
When translated into plain English, the biological record should distinguish an observed behaviour from a diagnosis. ‘Thirty were not visible during the evening survey’ is safer than ‘thirty failed to survive’ without other evidence.
The destination can confound the shipping comparison
The receiving habitat matters: substrate, water exchange, temperature, predator access, enclosure design and observation effort can differ between sites. A comparison of transport bags is not a controlled comparison if each method also sends animals to a different destination. Differences at five days may stem from the site as much as from the journey.
Two groups each deliver 500 living juveniles. In one protected chamber, 470 are confirmed alive after five days; in another open pen, only 390 are recovered. The fractions are 94% and 78%, but the disparity cannot be attributed solely to packing unless the receiving environments and detection methods are sufficiently comparable.
Report treatment and habitat conditions together. Where causation is not identifiable, describe an association or observed difference, then explain which additional comparison would test the suspected cause. Honest limits make the report more useful for decisions.
A complete worked shipment inventory
Six bags begin with 400 juveniles each, and two bags with 350 each. Total dispatch is 6×400 + 2×350 = 3,100. Suppose the corresponding live arrival counts are 392, 397, 390, 395, 398, 391, 342 and 344. Preserve the bag order so any later environmental measurements can be connected correctly.
The first six arrival counts total 2,363 and the final two total 686. Thus 3,049 animals were observed alive. The apparent difference is 51, and the pooled arrival live fraction is 3,049/3,100 = 98.35%. This is a count-based arrival outcome under the assumed complete dispatch enumeration.
Unless the missing fifty-one were independently confirmed dead, call them unaccounted for. Keep the original bag counts rather than publishing only the pooled percentage. Averages may conceal a problem localised to one bag, and the bag is often the relevant treatment unit in an experiment.
Why bags and animals are not always independent replicates
Many juveniles can share the same transport water and thermal conditions in one bag. If four bags are assigned independently to a treatment and each holds fifty animals, there may be four bag-level treatment replicates and two hundred observed animals. Calling the study two hundred independent thermal replicates can underestimate uncertainty.
A mean of bag survival percentages is also not always equal to pooled animal survival. If a small bag of 100 has 100 survivors and a large bag of 900 has 810, the unweighted mean of 100% and 90% is 95%, while pooled survival is 910/1,000 = 91%.
Both numbers are legitimate for specific questions, but they must be labelled. A research translation should state the unit of treatment, unit of observation, weighting basis and uncertainty measure, rather than using the word average without context.
Track the size distribution rather than one average weight
Two groups with the same average body mass can have very different spreads. If one includes many tiny juveniles and several heavy individuals, and another consists of similar-sized animals, their responses may not be comparable. A distribution is often more informative than one mean. Specify whether weight was measured after draining, whether animals had expelled water and whether the process altered their behaviour.
Imagine a group of one hundred juveniles: fifty weigh 1 g each and fifty weigh 5 g each. Mean mass is 3 g and total wet biomass is 300 g. Another hundred animals all weigh 3 g also total 300 g, but their individual dimensions differ. Both have the same nominal biomass per bag, yet identical biology cannot be assumed.
A source saying ‘three-gram juveniles’ should clarify whether that is a mean, median or individual target. When transporting live animals, size range, handling history and count method may influence interpretation. Do not upgrade an average into a statement about every animal.
Salinity comparisons need a genuine common basis
Salinity measured in practical salinity units is a conductivity-related estimate under a defined convention, not automatically the mass of salt in a litre. A packing record of 32 on a salinity instrument and receiving water of 26 points to different water states, but the exact physiological consequence is not established by subtraction alone. Instrument scale and calibration also deserve attention.
The numerical difference is six units on the stated scale. It is not necessarily correct to say that salt mass must fall by exactly six grams per litre, and it would be misleading to promise that a particular change is harmless. Salinity adjustment depends on species and development stage; a report can describe the measured difference without prescribing a transfer method.
Use the source’s measurement convention consistently, and record whether receiving water was sampled before or after additions. A reading after a bag was mixed into a tank cannot be retrospectively presented as the original tank salinity. The location and event order govern the meaning of the number.
Exposure history and dose are not the same claim
A threshold can be used to organise a dataset, but it should not become an invented biological safety standard. A technician may report minutes above a locally chosen caution temperature or below a dissolved oxygen alert value. Unless that threshold has been validated for the animals and situation, the result is a descriptive exposure index rather than an expected mortality probability.
Imagine one bag with a brief high temperature and another with a lower but prolonged increase. Comparing their maxima alone may obscure duration; comparing means alone may obscure the peak. A more informative description records the full trace or meaningful duration summaries, plus the outcome at a known time. Temperature and oxygen can interact, so one index need not explain everything.
If a logger records every fifteen minutes, it may fail to capture a brief excursion between observations. State sampling cadence, gaps and sensor uncertainty. An exact-looking figure printed by software is not evidence of an exact biological exposure.
The temperature-data completeness problem
Four equal-interval readings of 27, 28, 30 and 29°C have an arithmetic average of 28.5°C. If a logger instead records for five minutes, then fails for three hours, then resumes, an average of the surviving values gives a biased view of time exposure unless the data and missingness assumptions are considered. A temperature time series is not just a list of numbers.
To report a time-weighted mean, each valid interval must carry an appropriate duration. Do not silently fill a long missing interval with the last valid reading. That is an imputation, and the conclusions may depend strongly on whether the true bag conditions changed during the gap.
The audit should name logger type, placement, interval, calibration status, fraction of elapsed time covered and the period between packing and opening. A sentence describing uncertainty is preferable to a bold claim that conditions were stable when the measurements cannot establish it.
The arithmetic of waiting after arrival
Transit can include several distinct waiting periods. Suppose a vehicle reaches the receiving facility at 12:55, the transport boxes are signed over at 13:05, and the bags open at 13:20. Road transport has ended, but the animals remain in the original packed environment for twenty-five minutes after arrival. The handover interval is ten minutes and the later wait is fifteen minutes.
If a report counts the journey only until truck arrival, it omits this last exposure. Conversely, calling the entire bag interval road transport inflates driving time. Both mistakes can create confusion when a logistics problem is compared with a biological response.
A small timeline is often the best translation tool: closure, loading, departure, checkpoint, arrival, handover, bag opening, acclimation, release and follow-up. It allows time sums to be checked and identifies where a missing clock entry prevents precise reconstruction.
When a shipment is split or bags are merged
A packing record may begin with eight separate bags but end with fewer receiving groups. If two bags are pooled into one chamber, subsequent survival in that chamber cannot be assigned precisely to one original bag unless animals have reliable individual marks or another justified identification method. This matters when the two bags experienced different temperatures or packing media.
For example, bag A delivers 390 live juveniles and bag B 395. The receiving chamber contains 785 animals after combining them. If 750 are confirmed alive later, the pooled observed retention is 750/785, or about 95.5%. It is not possible from that final count alone to identify how many of the thirty-five unconfirmed animals originated from A or B.
Keep an event-ledger row showing both source IDs and the destination cohort. A phrase such as ‘bag A lost twenty’ would invent attribution without evidence. Good tracking converts a biological report into a traceable set of transitions, including where separate histories cease to be distinguishable.
Why repeated handling affects the measurement
A follow-up survey may require touching, removing or counting animals that would otherwise remain in sediment. Handling can change sea cucumber emergence and response. A study observing animals every day may therefore impose a different disturbance history from a study that looks only once after a week. The measurement process can be part of the treatment.
Suppose one receiving chamber is inspected on days one, two, three and five, and a second only on day five. Different day-five behaviour cannot automatically be attributed to the original transport method. The inspection schedules differ, and observers might change the environment while searching.
Report how often animals were disturbed, whether surveys were visual or physical, how long counting lasted and whether the substrate was replaced or moved. A careful translation should not say all animals were left undisturbed if the records show repeated manipulation. This is a general scientific principle with immediate application to restocking evidence.
Sampling counts from a well-mixed vessel
Very small juveniles may be estimated by taking known-volume subsamples from a larger holding vessel. Suppose five 100-mL samples from a nominal 10-L batch contain 42, 47, 45, 50 and 46 individuals. Their mean is forty-six juveniles per 100 mL. Because ten litres contain one hundred such volumes, the estimated population is about 4,600.
The calculation assumes the vessel was mixed well enough to make the samples representative and that each aliquot was accurately measured. If animals concentrate near the bottom, sampling at one depth may systematically miss them. This is a sampling bias, not a small rounding error.
The estimate should be described as such, not as a census. If arrival numbers use another counting method, apparent survival may reflect differences between estimators. Give the original sample observations, extrapolation factor, and uncertainty or sampling limitations wherever practicable.
Missing animals and detection probability
A survey recovery number is not automatically equal to true survival. Some living juveniles may bury, escape a search area or simply be overlooked. If detection probability is below one, the observed recovered count may understate the living population. However, dividing by an arbitrary guessed detection probability does not cure the problem; the estimate must be supported by an appropriate design.
Imagine a controlled mark-recapture study where the probability of detecting a known living juvenile is estimated at 0.80 under comparable conditions. If ninety-six distinct living juveniles are detected in an eligible sampled group, the simplest illustrative correction is 96/0.80 = 120. This is a model-based estimate under restrictive assumptions, not a guarantee of exactly 120 animals.
The detection estimate, uncertainty, independence assumptions, and potential animal movements all matter. A report should not call this a direct count. Distinguish observed recovery, detection-adjusted abundance and biological survival as three different concepts.
A small sample is not a universal guarantee
A trial with ten juveniles and ten observed survivors yields a sample fraction of 100%, but it does not prove that mortality probability is zero. The sample could be too small to reveal infrequent failures. The same issue appears when a technology provider advertises perfect performance from a short demonstration.
If fifty animals are observed and one is confirmed dead, the observed death proportion is 2%. The point estimate is valid for the stated sample and endpoint, but uncertainty can be substantial, especially with clustering in bags. A larger number of animals in one bag may not provide the same independent evidence as animals distributed across multiple independently treated bags.
Avoid translating ‘no deaths observed’ into ‘mortality is impossible’. Preserve the sample size, timeframe and experimental unit so readers can judge precision. A confidence claim requires a method appropriate to the design, not simply a rounded percentage.
Keeping a chain of custody across languages
Technical forms should preserve cohort identifiers, bag codes, times and measurements across languages. An alphanumeric bag ID is a reference key, not text to translate creatively. If a new receiving label is assigned, the old label must be linked to it. Without that relationship, a later abnormality cannot be tied accurately to its earlier exposure.
Useful fields include origin tank, animal stage, size method, number estimated or counted, packing liquid and its measured volume, bag closure time, box ID, route events, receiving observer, opening time and release area. Add an exception field for leaks, repacking, missing readings and spills. These events often explain apparent inconsistencies.
A bilingual report should not translate ‘estimated’ as ‘counted’, ‘not found’ as ‘dead’ or ‘released’ as ‘survived’. These are differences in evidence status. The correct target-language sentence must preserve what an observer actually knows and what remains inferred.
Transfer exercise: compare arrival fractions
A shipment A dispatches 1,200 juveniles and counts 1,170 alive at arrival. Shipment B dispatches 1,000 and counts 980 alive. A reports additional confirmation of 1,100 live animals on day three; B has no follow-up. Calculate the comparable arrival fractions and explain why the later figures should not be directly ranked.
A’s observed arrival fraction is 1,170/1,200 = 97.5%. B’s is 980/1,000 = 98%. At arrival, B is half a percentage point higher on this simple descriptive measure. A’s day-three confirmed fraction from dispatch is 1,100/1,200 = 91.7%, but B has no day-three denominator-linked observation.
It would be misleading to claim B had 98% long-term success and A only 91.7%. The endpoint timings differ. State the observations, ask for matched follow-up and examine receiving environments before making stronger conclusions.
Transfer exercise: bag-level and pooled averages
One bag with one hundred juveniles has one hundred live arrivals; a second bag with nine hundred has eight hundred and ten. Compute the average of their two percentages and the pooled fraction. Then decide what question each statistic addresses.
The bag percentages are 100% and 90%, so their unweighted mean is 95%. The pooled live count is 910 of 1,000, or 91%. These figures differ because the larger bag contributes far more animals. If the unit of experimental treatment is the bag, bag-level summaries may be pertinent; if the question is the proportion of all dispatched animals alive, pooling is more direct.
A report that states only ‘95% survival’ hides which summary was intended. The repair is to provide the bag starting counts, observed outcomes and weighting method. Precision in aggregation is as important as precision in temperature measurement.
Transfer exercise: identify what died and what disappeared
A chamber starts with eight hundred recorded live arrivals. Five days later the observer confirms six hundred and ninety alive, ten dead and one hundred not found. Calculate the observed lower and upper possible fractions living under the stated categories, and explain the difference between bounds and an estimated probability.
The confirmed living minimum is 690/800 = 86.25%. If every missing juvenile is actually alive, the possible living maximum is 790/800 = 98.75%. These are arithmetic bounds for the simplified ledger. They are not a confidence interval, and they do not establish a biological mechanism.
If thirty missing animals are found alive on the next visit, the confirmed live count rises without any resurrection. The evidence status changed. A defensible report keeps repeated observations rather than treating the first search as omniscient.
Independent review and practical FAQ
Does a 98% live arrival fraction prove the transport method safe? It establishes a defined observation, not an unrestricted safety or welfare conclusion. Consider delayed recovery, behaviour, water quality and a suitable comparison before using a broader term.
Can a published animals-per-bag recommendation be scaled linearly to a larger bag? Not without evidence. Liquid geometry, biomass, gas transfer, duration and animal condition may change. A specific research result or operating manual should remain attached to the conditions in which it was developed.
If two methods show no statistically significant difference, is equivalence proven? No. Small samples and wide uncertainty can mask meaningful differences. Equivalence requires a defined margin and an appropriate analysis.
What if no thermometer reading exists during the trip? Say that thermal exposure was not adequately measured. A good arrival count cannot fill missing physical evidence.
What matters most when writing an accessible translation? Preserve what the source actually measured, state its denominator, and separate direct observation from modelled or inferred conclusions. Clear limits are part of an accurate explanation, not a defect.
A practical check on precision and rounding
Suppose the dispatch count of 3,100 was a true enumeration but an arrival observer could reliably classify only 3,049 living animals, while a small remainder was lost during a spill. The arrival percentage of 98.35% is a valid quotient of observed living count and dispatch count. It is not necessarily an estimated probability of physiological survival, because the status of missing animals is unresolved. Rounding it to 98.4% would alter presentation, not remove the identification uncertainty. The most important audit asks what each number includes, rather than how many decimal digits a spreadsheet displays.
If a source instead estimates both dispatch and arrival populations from subsamples, any ratio compounds two sources of estimation error. The same printed 98.35% would then be far less precise than it looks. A clear explanation identifies direct counts, sample-based estimates, missing groups and the method of aggregation. Whenever a reader can reconstruct those steps, the stated proportion remains useful even if a precise confidence interval is not available. Whenever those steps are hidden, confident language should be reduced accordingly.
A final example of an evidence-preserving report
An operational summary might read: ‘Eight labelled bags carried 3,100 counted Holothuria scabra juveniles. Arrival observers counted 3,049 living animals, an observed live fraction of 98.35% of dispatch. Closure and opening times are retained by bag. Temperature traces were obtained for six bags, while continuous temperature evidence is absent for two. The difference of fifty-one individuals was not fully classified as confirmed deaths.’
That sentence is more useful than a claim that transport was completely safe and successful. It specifies the population and endpoint, acknowledges measurement gaps, and does not promise long-term survival. An independently readable ledger can support later work about behaviour and habitat without rewriting earlier observations.
Related explanations on eduKateSG cover sandfish nursery growth, sea-ranch recapture and sediment monitoring. These are subsequent parts of a biological chain, not proofs that a journey caused a particular long-term result. The test of a good translation is whether another reader can reproduce the arithmetic and identify what is still unknown.
Original sources and related reading
SEAFDEC sandfish hatchery production manual
SEAFDEC sea cucumber programme
FAO AGRIS juvenile transport experiment
FAO AGRIS burrowing and handling study