A geoduck seed-planting record describes how nursery-grown juvenile clams move from a counted hatchery lot into a defined seabed area, often with temporary predator-protection tubes or netting. To read that record correctly, separate seed received, seed accepted, planting locations prepared, seed actually placed, protective structures installed, later observations and estimated survivors. A planted count is not a survival count, and a successful tube installation is not proof that a juvenile is alive beneath it. Accurate translation preserves every physical unit, date and location boundary.
For a completely fictional example, imagine a farm receives 48,000 graded geoduck seed, rejects 1,200 at receiving inspection and plants 45,000. The remaining 1,800 could have been held for later planting or lost during handling; the ledger must tell us which. If technicians later sample protected units and estimate that 33,750 juveniles remain alive, the apparent survival relative to the planted 45,000 is 75 percent, subject to sampling and detection assumptions. It is not 75 percent of the original shipment unless the denominator is deliberately changed.
The reading sequence is to identify the site and lease boundaries, check habitat exclusions and permits, confirm the seed batch and receiving count, calculate actual planted area, distinguish planting points from protection structures, record the number of juveniles per point, reconcile installed and removed gear, establish a repeatable monitoring design, define alive versus missing versus unexamined, estimate survival and uncertainty, and disclose any environmental or sampling limitations. Never infer that a site is authorized, ecologically suitable or safe to harvest merely because a map displays planting symbols.
The biological and management context
Geoducks are filter-feeding clams whose siphons reach the water while the animal burrows into sediment. In cultivation, hatchery-produced juvenile seed may be planted in prepared intertidal sediments, sometimes inside temporary PVC tubes covered by mesh to reduce early predation. NOAA’s geoduck aquaculture overview describes this production pathway and notes that protection can be used while animals are small. Washington’s WDFW distinguishes farm-raised geoducks from naturally recruited wild stocks and describes predation protection, licensing and the longer grow-out process. A field planting log belongs to aquaculture inventory, not an automatic census of a wild tract.
The term seed refers to juvenile animals appropriate for a particular farm’s planting specification, not plant seeds or mature shellfish. “Outplanting” is the transfer and placement of nursery stock into the intended grow-out environment. “Recovery” may describe animals found in a later sample, whereas “survival” is a biological condition inferred from evidence. “Planting density” may mean individuals per square metre of eligible area, individuals per planting hole or individuals per tube. Those definitions must be given before anyone can compare two farms or two experimental treatments.
A site map is more than a decorative image. Boundaries may include lease limits, property or permit zones, shoreline reference, depth intervals, eelgrass avoidance areas and unsuitable substrate. The planted footprint may be smaller than the overall leased parcel. If an operation uses an 8,000-square-metre lease but plants seed in only 5,200 square metres, dividing planted animals by the whole lease gives a lower density than the actual planted-footprint density. Neither figure is inherently wrong if labeled properly; the error is silently substituting one denominator for another.
Area coordinates require an explicit reference system. Latitude and longitude in decimal degrees cannot simply be treated as metre coordinates, and a map’s visual scale may not hold for arbitrary measurements. A grid recorded in metres under a stated local or projected coordinate system can support area calculations when the mapping procedure is correct. A translation that drops the coordinate reference, truncates precision or swaps northing with easting can move a plotted planting point. Sensitive sites need verified field survey data, not approximate reconstruction from a paragraph.
The seed inventory chain
A hatchery shipment may list number packed, size-grade distribution, batch identity and dispatch time. The farm’s receiving log may separately record number arrived, quantity accepted, quality holds and losses observed. If the shipping manifest says 50,000 seed while receiving reports 49,300, the 700 difference is not necessarily confirmed mortality. It can arise through counting disagreement, damaged packaging, discarded material or animals held separately. Reconciliation needs evidence. A report should describe the status of the discrepancy rather than adding seven hundred to a “died in transit” column without inspection.
A seed lot can be split across zones. Suppose 24,000 juveniles are allocated to North Plot, 18,000 to South Plot and 6,000 retained at the nursery. The allocation totals 48,000, but only 42,000 are scheduled for field planting. If 41,500 are actually placed because the tide window closes, planting efficiency relative to planned field allocation is 41,500 divided by 42,000, or about 98.81 percent. That operational fraction says nothing yet about later biological survival. The retained and unplanted seed should remain in separately labeled inventories.
Planting windows are affected by tides, access, weather, staff and the animals’ condition. A field crew may complete fewer sites than planned. Time stamps on each section matter because a later photograph may be taken after some protective gear was removed. When a report says “all plots planted on Tuesday,” verify whether the work log and map show completed positions, planned positions or both. An attractive map containing every intended point is not the same as an as-built record of the number of juveniles placed in each point.
In one hypothetical as-built ledger, six rows each describe a plot block of 200 planting points, for 1,200 points in total. If each point received four juvenile geoducks, the documented planted number is 4,800 animals. However, if the third block had only 160 points completed, the correct total is 1,160 points and 4,640 animals. Counting planned points instead of completed points overstates planting by 160 animals. The reader should multiply recorded seed per point by actual completed points, checking for any variable stocking within blocks.
Protection tubes, mesh and biological claims
A predator-protection tube is an installed structure, not an animal. Several juveniles may be placed inside one tube, or the design may use another arrangement, depending on the farm and approved method. A count of 2,000 tubes cannot be relabeled as 2,000 geoducks without a one-animal-per-tube rule supported by the planting ledger. Some plots use netting or protective systems with different geometry. The report should record the device type, installed count, physical location, intended removal condition and any damage or replacement observed later.
Temporary gear creates a separate environmental and maintenance inventory. Tubes can be installed, dislodged, buried, damaged, recovered, replaced or removed as juveniles grow. “Tube present” indicates equipment status; it does not verify juvenile survival. “Tube missing” can indicate a lost device, an intentional removal or an unrecorded change, not necessarily a dead clam. Equipment inspection and biological sampling should therefore have separate fields. Conflating the two destroys the ability to distinguish a gear-management issue from a biological survival issue.
Predation protection is context-specific. Washington’s WDFW describes protective tubes or nets as common during early vulnerability to predators such as crabs and other marine animals. The success of protection depends on local species, sediment, installation and maintenance. Observing fewer losses in protected plots does not automatically establish causality if the protected sites also had better habitat. A well-designed comparison uses a documented sampling frame and, when feasible, replicated treatments that avoid confounding. Translate correlation cautiously rather than selling a particular device as guaranteed protection.
The presence of eelgrass or other sensitive habitat may affect where geoduck aquaculture can be located and which operations are permitted. NOAA describes habitat considerations for farm equipment, while local approval decisions depend on their own authorities and site-specific conditions. The responsible wording is that geoduck farms must follow applicable permits and habitat protections, not that every bare patch on a satellite image is legal to plant. A planting map can document the operator’s plan, but only the authoritative permit and field assessment establish the permitted footprint.
From an as-built map to actual planted density
Planting density is a ratio with a numerator and a spatial denominator. If a crew places 24,000 juvenile geoducks across an eligible planted footprint of 4,000 square metres, the nominal density is six animals per square metre. That value is not the same as six per tube unless the planting structure and number of animals per tube happen to support such a conversion. If some squares of the footprint were deliberately left unplanted, density within occupied locations may be higher. The report should state whether it uses total approved area, actual prepared area, occupied bed or planting holes.
Area boundaries can be revised following field inspection. For a fictional example, a mapped planting polygon originally measures 3,600 square metres but a later survey identifies 600 square metres of excluded habitat. If 18,000 animals were placed only within the remaining 3,000 square metres, the correct eligible-footprint planting density is six per square metre; dividing by the initial 3,600 produces five. The biological planting did not change when the map was corrected. The reported ratio changed because its spatial denominator changed, so both versions need clear boundary dates.
A planting grid may be uneven. Suppose one section has two seed per point on 400 completed points, another has four per point on 300 and a third has six per point on 200. Total placed is 800 plus 1,200 plus 1,200, or 3,200 juveniles across 900 points. The point-weighted mean is about 3.56 juveniles per completed point. Averaging two, four and six without weighting gives four, which misrepresents the actual planting. Weighting by completed point counts is essential whenever zone sizes or treatment allocations differ.
Spatial precision is not the same as reporting many decimal places. A GPS observation with several digits may still be uncertain because of tide conditions, equipment, reference systems or underwater placement. If an as-built map reports each tube at a centimeter-precise point but the field survey cannot support that accuracy, do not promise exact centimeter positions. For a useful planting summary, the important evidence is the credible boundary, date, actual completed locations and documented mapping method. Avoid combining coordinate pairs from incompatible reference systems without transformation.
Design a field survival check before interpreting it
Survival is the fraction of animals that remain alive after a stated interval, relative to a clearly defined starting cohort. But buried geoducks are not necessarily visible during a quick site inspection. A monitoring method may use sampled tube interiors, excavation under authorized protocols, marked reference units or another scientifically justified approach. The report must distinguish animals observed alive, animals confirmed dead, animals missing and units that were not inspected. If an animal cannot be located, its fate may be uncertain rather than confirmed mortality.
Time interval is part of the survival statement. A reported 80 percent survival at one month and 65 percent at twelve months should not be compared as if they cover the same exposure period. Nor should a survival calculation substitute planned planting for the actual planting denominator when some seed remained at the nursery. A useful title names the interval, such as apparent six-month survival among planted juveniles. That wording lets the reader infer the biological question without imagining survival from fertilisation all the way to adult harvest.
Imagine a fictional survey inspecting a random sample of 120 planting positions, each recorded as having four juveniles at planting. Initial animals represented by the sample are therefore 480. The monitoring team recovers 360 clearly live juveniles. Observed live recovery is 360 divided by 480, or 75 percent. Whether this estimates biological survival depends on recovery efficiency and representative position selection. If juveniles escaped the immediate sample area but remain alive, the apparent recovery fraction may understate actual survival.
A missing tube does not automatically imply four missing animals. The protective structure could have been removed intentionally after burrowing, dislodged by waves or replaced under a different identifier. A proper monitoring dataset links the original point to maintenance records so that missing gear can be distinguished from an uninspected biological position. A report that substitutes tube retention for animal survival measures equipment persistence, not necessarily living biomass. The two measures are useful when correctly named, and misleading when merged.
Detection probability matters especially when monitoring is nondestructive. Suppose an independent test under comparable conditions finds that a method locates an identifiable juvenile in ninety percent of known occupied positions. That result could inform an explicitly modeled detection adjustment, with uncertainty and assumptions. It should not be imposed blindly on another beach. The species’ burrowing behavior, substrate, tide level, staff and stage of growth can change observation conditions. If no validated detection estimate exists, label the result as observed recovery rather than pretending that all unseen animals died.
Sampling design and uncertainty
Sampling only tidy, easily accessed rows can bias a survival estimate if those rows also have better sediment or receive more maintenance. A randomized or otherwise defensible design gives each relevant portion of the planted population an appropriate chance of selection. Stratification may be useful when the farm has distinct depth zones, substrates or planting dates. The statistical estimator must match that design. Dividing all recovered animals by all sampled points can produce a deceptive overall estimate if the sampled point counts and zone areas differ substantially.
Suppose Zone A has 1,000 planted positions and an estimated live recovery of 80 percent, while Zone B has 3,000 positions and an estimated live recovery of 60 percent. The position-weighted overall rate is (1,000 times 0.80 plus 3,000 times 0.60) divided by 4,000, or 65 percent. The simple unweighted mean of 80 and 60 is 70 percent and does not describe the combined planting. If the number of seed per point differs by zone, the analysis should weight by initial animals rather than points to calculate an animal-level survival fraction.
Uncertainty can arise from which positions were sampled, how accurately the original seed number was recorded, whether individuals were detected, and how boundaries were mapped. A confidence interval for a proportion addresses some sampling uncertainty under a stated design; it does not automatically incorporate all these other sources. A phrase such as “estimated survival 65 percent” is more faithful than “exactly 65 percent lived” when the figure came from sample expansion. A useful report includes sample size and methods before an apparently precise result.
Repeated inspections of the same sample positions produce longitudinal data, not independent new animals. If fifty tubes are checked at month one, month three and month six, the record contains three visits to fifty units, not 150 different experimental units. Treating the visits as independent can understate uncertainty and make comparisons misleading. Link records by persistent position ID and explain replacements, losses or inaccessible visits. A time-series model should recognize this dependence rather than inflate the sample merely because the spreadsheet contains many rows.
Survival and growth need different measurements. A juvenile may remain alive yet grow slowly, while an initially smaller individual may grow quickly. A monitoring report could record live recovery, shell dimension and condition separately. If growing animals are measured only when easily found, the length sample may overrepresent larger or shallower individuals. Describing the mean length of recovered geoducks as the mean length of all planted geoducks requires evidence about detection and sampling. Keep the qualified population attached to each metric.
Protection performance versus habitat differences
A field trial might compare protected planting points with unprotected controls. If survival differs, ask whether treatment assignment was random, whether sites had similar substrate and exposure, and whether enough independent plot units existed. Twenty tubes clustered in one protected corner and twenty controls in a distant exposed area do not prove the tubes caused the difference. Treatment and location are confounded. If the design uses entire plots as treatment units, the number of independent plots matters more for inference than the total juveniles housed inside them.
A worked fictional comparison has three protected plots with live-recovery estimates of 72, 76 and 74 percent, and three comparable controls with 51, 55 and 53 percent. The mean protected proportion is 74 percent and mean control proportion is 53 percent, a difference of 21 percentage points. This describes the imaginary plot-level data. It is not automatically a statistically significant causal effect, since variances, assignment, detection and plot independence require analysis. Reporting a percentage-point difference accurately is different from saying protection increased survival by 21 percent relatively.
The relative difference in that same fictional example is (74 minus 53) divided by 53, or about 39.6 percent higher mean recovery in protected plots relative to controls. That is mathematically different from a 21-percentage-point absolute difference. A translated press release that replaces points with percent can understate or overstate results. A full interpretation should also tell readers the period of monitoring, initial seed numbers, sample method and whether the effect was consistent across replicate sites. A treatment result without these qualifications is incomplete.
Gear can alter the local environment as well as discourage predators. Tubes and mesh may influence sediment flow, small-animal behavior, access or habitat use. A comparison should not presume that ecological effects are zero solely because animals survived. NOAA discusses environmental considerations for geoduck farms, including nearshore habitats. A planting record is not a full environmental impact assessment, and an equipment count is not evidence of environmental safety. Describe what monitoring actually measured and leave wider causal claims to appropriately designed studies.
Gear recovery is a different performance measure
After early protection is no longer required under the approved method, an operator may remove tubes or mesh. A gear ledger should record items installed, added, replaced, recovered, damaged and unresolved. The number removed can exceed the original installed count if replacements were also installed and later collected. It can be smaller without implying that all missing gear remains in the environment; some items may have documented earlier removals. A responsible report reconciles unique gear IDs or standardized counts across dates rather than declaring an ecological outcome from a single aggregate.
Suppose a fictional farm installed 1,000 tubes, replaced 80 damaged ones during the first year and later recovered 1,055 tubes over all removal events. The total devices introduced was 1,080, so twenty remain unresolved if all counts refer to compatible units and no further transfers occurred. Dividing 1,055 by the original 1,000 would give a nonsensical recovery rate above 100 percent. The correct denominator includes replacement devices. This example shows why an inventory of physical equipment needs its own event history, separate from juvenile survival.
Equipment removal does not mean the surviving geoduck crop has been harvested. The clams may remain buried for years after early protection is removed. Likewise, a missing juvenile during a tube-removal inspection is not necessarily a confirmed mortality because the individual may have moved deeper. NOAA’s general geoduck overview describes long grow-out periods, while farm plans and permits determine specific schedules. Avoid treating a widely quoted number of years as a guarantee of when every planted lot will be ready for commercial sale.
A complete fictional site calculation
Imagine a clearly hypothetical farm with a licensed parcel of 9,000 square metres. Environmental and engineering mapping identifies 5,500 square metres as eligible for a particular planting operation. Technicians complete 2,000 planting points, assigning three seed to each point. Documented planting is 6,000 animals. If all points lie within the eligible 5,500-square-metre footprint, mean planting density over that eligible area is about 1.091 animals per square metre. Over the entire parcel it would be about 0.667. Both ratios are arithmetically defined, but only the first describes density within the mapped eligible footprint.
Suppose 500 of those planting points occupy a small 600-square-metre intensive demonstration block. It contains 1,500 seed, or 2.5 per square metre of block area. The remaining 1,500 points carry 4,500 seed over 4,900 eligible square metres, approximately 0.918 per square metre. Simply averaging 2.5 and 0.918 gives about 1.709, which does not describe the combined site because the zones have different areas. The correct total density is 6,000 divided by 5,500, approximately 1.091. This area-weighted interpretation matters whenever a map contains heterogeneous planting plots.
Six months later, inspectors sample 100 demonstration-block planting points and recover 222 living geoducks, while 150 other-block points yield 330. Each point originally received three, so observed sample recoveries are 222 of 300, or 74 percent, and 330 of 450, or about 73.33 percent. If positions were selected representatively and every living animal was detected, weighted extrapolation could estimate about 1,110 live juveniles in the demonstration block and 3,300 in the other block, a combined 4,410. These are model-dependent estimates, not a complete count of buried animals.
The combined apparent survival estimate in this fictional site is 4,410 divided by 6,000, or 73.5 percent. The simple average of 74 and 73.33 would be approximately 73.67 percent; its small difference is still a denominator problem. A different site with more unequal treatment sizes could show a much larger distortion. Before presenting a percentage, the reader should reconstruct how plot-level rates were weighted and whether unequal initial stocking per point requires weighting by animals rather than by points.
Imagine an auditor discovers that 200 planting points in the ledger were marked planned rather than completed, though the aggregate 2,000-point total included them. The correct actual planted count would become 1,800 times three, or 5,400 juveniles if no other discrepancies exist. Any survival fraction using 6,000 as its denominator would need reanalysis. Yet no biological mortality has been documented by this correction. The missing 600 are animals that were never shown to have been planted, not evidence of juvenile loss after field placement.
Long-term interpretation and economic limits
Growth and survival during early months do not guarantee an adult harvest quantity. Juveniles experience later mortality, changing conditions, variable growth and potential operational decisions about site use. A forecast of adult tonnage requires a defensible survival trajectory, size or weight distribution and harvest assumption for the relevant crop. Reporting an early recovery rate as if it were the final commercial yield would skip most of the biological pathway. The correct sentence is that early monitoring observed or estimated a defined fraction alive at a stated date.
A planting trial can also have a control purpose rather than a production purpose. Researchers may deliberately leave some positions unprotected or test alternative devices, but they need ethical and permitted protocols. Results can inform decisions about local protection, yet an apparent effect may depend on habitat, predator community or exposure. If a trial reports a 15-percentage-point difference, translate it as a difference between measured proportions under those conditions, not a promise that every future farm will gain fifteen percent surviving animals.
Field monitoring can change the animals’ environment. Excavation to verify survival may disturb the very positions that researchers plan to inspect again. If the same locations are resampled, note whether the procedure is destructive and whether it alters later outcomes. A longitudinal series based on untouched reference plots is not methodologically equivalent to repeated excavation of the same juveniles. In a public explanation, these details clarify why some apparent time trends require more careful interpretation than a simple line graph suggests.
Common errors and practical repairs
Error one: describing planned seed as planted seed. Repair it by checking as-built points, actual placement and retained inventory. Error two: equating tube count with animal count. Repair it by retrieving the documented number of juveniles per structure and any variable allocation. Error three: using the total lease as the planting-density denominator. Repair it by reporting the relevant eligible and occupied footprints. Error four: interpreting every missing tube as confirmed juvenile death. Repair it with gear-maintenance and biological monitoring records.
Error five: treating the number of juveniles recovered by a sampling method as a full census when detectability is incomplete. Repair it by labeling observed recovery and describing any validated adjustment. Error six: reporting a simple average of plot rates despite unequal planted numbers. Repair it with appropriate weighting and a stated population. Error seven: announcing a statistically causal protection effect from geographically confounded plots. Repair it by discussing assignment, replication and plausible alternative explanations. Error eight: promising environmental approval from a map alone. Repair it by checking the applicable authority and permit.
Error nine: converting a sampled six-month apparent survival into final adult harvest weight without a grow-out model. Repair it by restricting the statement to the measured time interval. Error ten: calling a density change after map revision a biological mortality event. Repair it by showing both boundary versions and the original planting count. Error eleven: confusing equipment-removal dates with crop-harvest dates. Repair it by maintaining separate operational timelines. Error twelve: assuming every site should use the same planting specifications. Repair it by anchoring every number to a particular plan, species, area and measurement.
Independent transfer questions and answers
Problem one: A farm plants 3,600 juveniles at four per completed point. How many points were completed? Answer: 900. If the planted footprint is 1,800 square metres, density is two juveniles per square metre of that footprint. Problem two: A map covers 2,400 square metres but excludes 400 square metres. If the same 3,600 were planted in the remaining area, eligible-footprint density is 1.8 animals per square metre, not 1.5. The difference comes entirely from the chosen area denominator.
Problem three: Three zones receive 600, 900 and 1,500 seed. A sample-based monitor estimates apparent six-month live recovery at 80, 70 and 60 percent, respectively. Estimated living animals are 480, 630 and 900, totaling 2,010. The overall apparent survival is 2,010 divided by 3,000, or 67 percent. The unweighted average of the three percentages is 70 percent, which is wrong for the combined animal-level fraction. The weighting follows original planted animals, not the number of zones.
Problem four: Equipment records show 500 initial tubes, 30 additional replacements and 510 recovered tubes over all removal events. What remains unresolved? Answer: twenty devices, assuming counts refer to the same scope and none were otherwise transferred. Is the recovery percentage 102 percent? No. The correct introduced-device denominator is 530, giving about 96.23 percent recovered. This remains an equipment inventory ratio, not an animal survival percentage. Separate the two even when both are collected on the same field visit.
Problem five: Inspectors find 72 living juveniles in 100 originally one-animal planting points but the monitoring protocol may miss some buried animals. Can they say survival was exactly 72 percent? No. They can report an observed recovery fraction of 72 percent and then explain detection limitations. Problem six: A site map contains areas of eelgrass excluded from planting. May a writer claim those excluded areas were barren and unnecessary habitat? No. Exclusion usually represents a habitat or management condition, not evidence of ecological worthlessness.
Problem seven: Two treatment groups have apparent recovery of 70 and 50 percent, each over the same interval. What is their difference? Twenty percentage points. What is their relative difference against the fifty-percent group? Forty percent greater. These phrases are not interchangeable. Problem eight: 5,000 seed are allocated to a site, 4,500 placed and 500 returned to a nursery. What is the planting completion fraction? Ninety percent of the allocation. Does that reveal survival? No. The unplanted 500 are not part of the field survival denominator.
Frequently asked questions
Does a geoduck planted in a predator tube remain in the tube until harvest? Not necessarily; early gear can be removed while the animal remains buried and grows. Does a missing tube prove the juvenile died? No. Is seed received the same as seed planted? Only if records show complete placement without exclusions, holds or returns. Can a planting map prove a farm is authorized? No; permits and relevant approvals are separate documents. Should farm planting density use total lease area or occupied footprint? It depends on the question, and the denominator must be specified.
Can a single inspection establish a long-term survival curve? No. Can a high tube recovery rate demonstrate high clam survival? No; gear and animal inventories measure different things. Can a sample-based apparent survival percentage be exact? No; sampling and detection uncertainty may apply. Are protected plots necessarily comparable with unprotected controls? Only if the experimental design and habitat conditions make the comparison defensible. Does early survival mean the crop will achieve its projected adult biomass? No; later growth, mortality and management conditions remain.
Make the evidence understandable
A short, truthful summary of a geoduck planting report can say where seed were placed, how many juveniles were documented as planted, which protective systems were used, what portion of the planted area was monitored and how many animals were observed or estimated alive at the stated interval. Then it can identify unresolved equipment, biological or sampling questions. This is more useful than a promotional claim that a farm achieved perfect planting or a guaranteed harvest, because it connects the numbers to decisions a reader can check.
Vocabulary is part of accuracy: “planned” is not “placed,” “located” is not “alive,” “observed” is not “estimated,” and “approved area” is not “total land.” Readers can practise those everyday English contrasts using the verified eduKateSG adult vocabulary guide. That link is a language-learning resource, not a source of geoduck farming rules. For the biological and regulatory context, consult the public agencies and specialist sources linked below.
The final test is whether the report can be reconstructed backward. Start from an estimated surviving population, trace it to field sampling and detection assumptions, trace the denominator to actual completed planting points and seed per point, and trace the location to an as-built map and authorized eligible footprint. Keep temporary gear in its own inventory. This method prevents a chain of minor ambiguities from becoming a major false claim about production, environmental effects or legal access.
Primary sources and methodological references
NOAA Fisheries — Geoduck Aquaculture explains hatchery seed production, intertidal planting, temporary predator-protection methods and habitat considerations.
Washington Department of Fish and Wildlife — Geoduck Fisheries distinguishes wild-stock management and commercial geoduck aquaculture and describes predator protection and permitting.
Washington Department of Natural Resources — Dive Team outlines tract boundaries, habitat conditions, sampling and harvest records in state operations.
Google Search Central — Helpful, people-first content emphasizes original, accurate and useful explanation.