WHAT A FISHING MAP MEANS BY “STRUCTURE”

The basics, fast

Most fishing apps draw structure on a map and leave you to assume the map knows the seafloor. It does not. It knows a list. Here is ours, stated plainly enough that you can decide how much weight to give it.

• 374 points, ten kinds, three countries. 157 artificial reefs, 76 FADs, 56 seamounts, 50 wrecks, 9 banks, 7 submarine canyons, 6 guyots, 5 ridges, 4 knolls, 4 rubble grounds. United States 200, Australia 153, New Zealand 21.
• The depths split perfectly along one line. Natural features: 87 points, zero with a depth. Deployed structures: 287 points, 258 with one. Wrecks are 50 of 50.
• In Australian water this list is almost entirely man-made — 135 of 153 points, 88.2%. In New Zealand water it is the mirror image: 16 of 21 are natural, 76.2%. Same file, two completely different kinds of knowledge.
• Two objects share the word “structure” and should not. Australian artificial reefs sit in 3–46 m (median 28 m). Australian FADs sit in 7–306 m (median 64 m). One is bottom you work over; the other is a mooring you pass.
• And the number that governs everything below: the median distance from one of our 326 Australian fishing spots to the nearest structure in this list is 38.7 km.

What this guide will not tell you

Worth stating before anything else, because it shapes what is and is not below.

• No marks. Every coordinate discussed here is already published by the agency that deployed the structure or by a public seafloor gazetteer. Nothing here came off anyone’s sounder, and nothing here is a spot.
• No protected-species sites. The gazetteer this page describes is deliberately conservation-cleared: a place known mainly as aggregation habitat for a protected species is not in it and will not be added. Naming such ground is the harm, whatever the intent.
• No bag limits, size limits or closed seasons. Those change, and a number printed in an article is a number the article eventually gets wrong.
• No claim that structure is where the fish are. Structure is one term in one model for one species, described below with its own weight and its own floor. It is not a promise.
• No claim that this list is the seafloor. That is the entire point of the guide.

The split that runs through the whole file

Start with the oddity, because it looks like a data-quality problem and is not one.

Take the ten kinds and ask a single question of each: how many of these points tell you how deep they are?

Eighty-seven natural features, not one depth between them. Two hundred and eighty-seven deployed structures, 258 depths. The line is exact, and once you see why, it stops being a gap and becomes the most informative thing in the file.

A deployed structure has a deployment record. Somebody lowered it off a barge at a surveyed position on a known day, and the depth at that position is part of the paperwork — it is how the agency proves the thing is where it said. So the number travels with the coordinates, and you inherit it.

A seamount does not have “a” depth. It has a summit, a base, and several thousand metres of flank in between, and which of those you mean depends entirely on what you are doing. A named seafloor feature in a gazetteer is a name and a position, because that is what a gazetteer is for. The absence is honest: there is no single number to carry.

So the rule to take away is not “our natural-feature data is worse”. It is that a depth in this list is a signature of human deployment. If a structure tells you how deep it is, somebody put it there.

Australia is a deployed coastline — in this data

The country split is where that observation stops being trivia.

Australia’s 153 break down as 74 artificial reefs, 58 FADs, 6 guyots, 6 seamounts, 3 rubble grounds, 3 banks, 2 submarine canyons and 1 knoll. New Zealand’s 21 are 6 seamounts, 5 ridges, 5 wrecks, 3 knolls and 2 banks.

It would be a mistake to read that as a statement about the two seafloors. The Tasman does not stop having seamounts at the Australian EEZ. What the table actually measures is which structures somebody chose to publish coordinates for — and those are two different questions with two different answers.

Deployed structures are published on purpose. A state fisheries agency sinks an artificial reef or moors a FAD precisely so that recreational anglers go and fish it; announcing the position is the point of the spend. Natural features get into a gazetteer only if a survey named them, and naming is driven by hydrography and research priorities, not by whether anyone fishes there.

So an Australian structure map is closer to a map of fisheries policy than a map of the bottom. That is not a criticism of the data — it is the most useful single sentence about it, because it tells you exactly where the list will be rich (near cities, where programmes are funded) and exactly where it will be empty (everywhere else), regardless of what the seafloor is doing.

Two different objects, one word

Before going near the map, a distinction that the word “structure” hides and that the depths make obvious.

Of the 106 Australian deployed structures carrying a depth, the quartiles run 3 m, 25 m, 30 m, 56 m, 306 m. That spread is not noise. It is two populations stacked on top of each other.

The artificial reefs stop at 46 m. The FADs run to 306 m — nearly seven times deeper than the deepest reef. These are not variations on a theme; they are different pieces of equipment aimed at different fish, and the depth column is what separates them at a glance.

It matters because “nearest structure” does not distinguish them. A nearest-structure calculation will happily hand a bottom-fishing plan a 300 m FAD mooring, or hand a pelagic troll a 6 m inshore reef module. If you are using any app’s structure layer — ours included — read the kind and the depth before you read the distance.

One more line the latitudes draw: every Australian FAD in this list sits between 12.1°S and 28.0°S. Not one is south of the northern rivers of New South Wales. FAD programmes are a warm-water instrument, and below that line the deployed structure you will find is reef, not buoy — the artificial reefs run from 12.1°S all the way to 38.3°S.

What is actually near the places people fish from

Now the map, anchored to seven fixed points so you can check every number. Distances are great-circle from the coordinates printed in the first column.

Victoria is the best-served water in the country by a distance — 17 structures inside 25 km of the Heads, against Sydney’s 2. That is the Port Phillip artificial-reef programme showing up as data: a cluster of numbered modules placed in the bay, each with its depth on the record.

And two of Australia’s major fishing cities have nothing at all. Adelaide’s nearest known structure is 643.7 km away; Hobart’s is a guyot 255.3 km offshore with no depth. Within 100 km of either, this list holds zero points.

That is worth sitting with, because neither city is short of fishing. South Australia’s gulfs and Tasmania’s south-east are among the most productive water on the continent. What they are short of is published offshore structure — no FAD programme, no large artificial-reef deployment in the gazetteer we draw from. Our own earlier guides to those two waters were built on the tide and on shore access precisely because that is what the data there supports.

If you fish out of Adelaide or Hobart and our structure layer looks empty, it is empty. That is the honest answer, and it is better than drawing something plausible.

What that does to the model — stated against ourselves

This is the section that costs us something to write, so it is the one worth reading.

Our kingfish habitat score uses distance-to-structure as its strongest locating term. That choice comes from the research rather than from taste: adult yellowtail kingfish hold to seafloor topography hard, with roughly 80% of captures within 1 km of it. The model turns that into an exponential decay with a scale of 8 km, and gives adults a structure floor of 0.20 — so a place far from anything known is damped, not zeroed. Juveniles get a floor of 0.85, because juveniles demonstrably do not hold to structure, and a term that is right for adults must be nearly switched off for them.

Good geometry. Now measure it against the list it runs on. For each of the 326 Australian fishing spots we publish, take the distance to the nearest structure point in the gazetteer:

The quartiles: 0.0 km at the closest, 10.8 km at the lower quartile, 38.7 km median, 180.0 km at the upper quartile, 832.2 km at the worst. And for 299 of the 326, the nearest thing the list offers is something people built — 197 artificial reefs, 61 FADs, 41 rubble grounds — against 27 natural features.

Read those two facts together and the conclusion is unavoidable. For roughly four out of five Australian spots, the nearest structure we know about is beyond the distance at which the term still varies. The model’s strongest locating signal is sitting on or near its floor across most of the coastline, which means that over most of Australia it is not localising anything. It is a constant.

That is not a bug in the arithmetic. The decay is doing exactly what it was asked to. It is a statement about the input: a term can only discriminate where the data underneath it discriminates, and a list of 153 published points cannot resolve a coastline of tens of thousands of kilometres.

The concrete version, from our own testing: a proven, repeatedly-fished peak scored 0.33 because the nearest structure the map knew was 14.7 km away. Add that one mark and the same place scores 1.00. Nothing about the water changed. The map learned one thing.

What actually fixes it

Two answers, and only one of them scales.

Public bathymetry does not fix it. This is the same limit we ran into writing about Port Phillip’s artificial reefs: a 250 m national grid cell averages a two-metre reef out of existence, and the little pinnacles anglers actually fish are smaller than the cells that would have to show them. Buying resolution is not an option at national scale, and interpolating relief that the survey never saw would be inventing structure — which is worse than having none, because it looks the same on a screen.

Your own marks do fix it, for you. A logged catch position is a structure point of exactly the kind the model wants and the public data cannot supply. In our app those coordinates stay on your device: owner mode shows you your own marks on your own map, they are never uploaded, and nobody else ever sees them. The rule we hold ourselves to is that spot secrecy is the promise that matters — a catch can be shared without the ground being given away.

There is a community version of the same idea, and it is switched off until it can be done without betraying anyone: marks snapped to a coarse grid, a cell becoming a structure point only once several separate anglers have independently fished it, so no individual’s position is recoverable from the result. Until that clears its own bar, the honest state of affairs is the one described above.

How to use the layer as it actually is

Practical, given everything above.

• Read the kind before the distance. “Nearest structure” can mean a 6 m reef module or a 306 m FAD mooring. They are not substitutes and the distance alone will not tell you which you got.
• Use the depth where it exists — it is the deployed half. If you want bottom structure in Australian water, filter to artificial reefs and expect 3–46 m. If you want a pelagic mooring, that is the FAD set, median 64 m, and only north of about 28°S.
• Treat a blank as a blank. Around Adelaide and Hobart the list is genuinely empty inside 100 km. An empty structure layer there is information about the data, not about the fishing.
• Expect the structure term to be flat where you fish. Four in five of our Australian spots sit beyond the 8 km decay. If a score looks oddly uniform across a stretch of coast, this is usually why.
• If you know a peak, log it. That is the only route by which the map gets the resolution the model was designed around, and on our side of it the coordinates never leave your phone.

What we model here, and what we do not

The standing coverage section, because a guide that describes a model should say where the model stops.

The structure term described above belongs to the kingfish score specifically. It is an adult prior — the same research that supports it also shows juveniles ignoring structure, which is why the juvenile floor sits at 0.85 and the term is nearly inert for them. Thermal suitability is treated as a filter rather than a locator, and deliberately asymmetrically, because activity keeps rising into warm water well past the point a symmetric curve would start penalising.

What this page does not claim: that distance-to-structure is the right first term for any other species we score; that the 374 points are a survey of anything; or that a structure with a depth is a structure worth fishing. The depth tells you what somebody deployed and how deep they put it. Whether it holds fish on the day you get there is the part no dataset in this article speaks to.