"A healthy fishery" is too vague to decompose, so pin it to four tests a water either passes or fails. A healthy wild fishery replaces itself (recruitment at least matches mortality across a run of years), shows multiple age classes (several year-classes present, not one hatchery cohort ageing out), grows fish to condition (fish reach a size and body condition appropriate to the species and the water's productivity ceiling), and absorbs a bad year (a drought, a scouring freshet or a fire does not flatten the population). Those four tests are the goal node. Everything below is what has to be true for them to pass.

Two of those tests pull against each other, and this is the most misunderstood thing in fisheries: on a fixed food base, numbers and size trade off. High recruitment produces many small fish; constrained recruitment plus abundant food produces few large fish. A water cannot maximise both. Deciding which one a fishery is for is a management choice, not a habitat outcome.

Fishery health is a series of gates, not a score. Each gate is necessary and none substitutes for another.

Gate 5 is different in kind from the first four. Connectivity is not an additional requirement so much as a multiplier on the others: spawning gravel behind a hanging culvert scores zero, and a cold seep a fish cannot reach in August does not count as thermal refuge.

Carrying capacity in a fishery is fixed by the bottleneck, never the mean. A river that averages a comfortable 14 °C but touches 24 °C for nine afternoons in August is a 24 °C river as far as its trout are concerned. Read every habitability variable at its annual worst.

Temperature is the master control, and its drivers stack predictably: elevation and latitude (roughly 0.6 °C cooler per 100 m of climb); source type, which is the single biggest lever — spring-fed and glacial systems are thermally buffered, snowmelt systems are seasonal, and low-elevation surface-runoff systems swing hardest; riparian shade, which dominates on streams narrow enough for a canopy to close and becomes irrelevant on wide rivers; groundwater and hyporheic exchange, which produces the cold seeps fish stack on in a heatwave; sheer water volume as thermal mass; and, on regulated rivers, dam release depth — a bottom-draw (hypolimnetic) release makes a cold tailwater in a valley that would otherwise be too warm, while a surface release does the opposite. In lakes the equivalent structure is stratification, and the summer problem is the squeeze between an epilimnion that has grown too warm and a hypolimnion that has run short of oxygen.

Dissolved oxygen is not an independent variable — it is largely temperature's shadow, and that coupling is vicious. Warmer water holds less oxygen at exactly the moment a fish's metabolic demand for oxygen rises. Riffles and turbulence reaerate; dense algae respire at night and draw oxygen down before dawn; shallow lakes under long ice can winterkill outright.

Water chemistry sets the productivity ceiling — see Gate 2 — and can also cap habitability directly through contaminants. In the Elk Valley, selenium loading from coal mining is the live example of a chemistry variable acting on a fishery independently of temperature or flow.

Flow regime determines how much habitat is actually wet. Base flow at the annual low sets real carrying capacity; freshet magnitude and timing decide whether eggs in the gravel survive; flashiness governs how often habitat is reset; and winter anchor ice and dewatering kill in the season nobody is watching.

Food supply in a fishery is capped by water chemistry before biology gets a vote. Alkalinity and hardness drive primary production: hard, alkaline water over sedimentary geology grows dramatically more algae, therefore more invertebrates, therefore bigger fish, than soft water draining granite. No amount of habitat work raises that ceiling. It is the first thing to establish about an unfamiliar water, because it predicts the size of fish it can possibly hold.

Beneath that ceiling, four things decide how much of the potential is realised. Substrate: clean cobble and gravel with open interstitial space is invertebrate habitat, and sand or silt is close to dead — embeddedness is the metric that matters. Terrestrial subsidy: on small forested streams, ants, beetles and hoppers falling off the bank make up a large share of the summer diet, which makes the riparian strip a food source and not only a shade source. Forage fish: the switch from invertebrates to eating other fish is what produces trophy size, so a piscivore's ceiling is set by its prey population — the gerrard rainbow trout is this principle in its most extreme form. Phenology: food has to arrive when the fish needs the energy, so hatch timing matters as much as hatch volume.

The currency underneath all of it is net energy intake — food delivered, minus the cost of holding position to intercept it. This is why the productive lies are velocity seams beside fast water rather than in it, and it is the mechanism that connects a habitat variable directly to where a cast should go.

Most anglers picture recruitment as spawning, and most recruitment failure is not about spawning. Four conditions have to hold in sequence.

Spawning substrate must be clean gravel sized to the fish, with low fine sediment. Intragravel flow must carry oxygen to the eggs, which is why downwelling sites are chosen and why sedimentation — not sediment on the surface, but fines filling the spaces between stones — is the principal killer of eggs. Incubation stability must hold for the months the eggs are in the gravel, and this is where spawn timing becomes destiny: a spring spawner like the rainbow trout has eggs in the gravel during freshet and can lose an entire year-class to a single scouring event, while a fall spawner incubates through the low, stable winter and faces anchor ice instead. Juvenile rearing habitat — shallow, low-velocity margins, side channels, backwaters and wood — must exist, and it is a different habitat from the one the adults use.

That last condition is where the bottleneck usually sits. Recruitment in most systems is not egg-limited; it is rearing-habitat-limited. Adding eggs to a river short of nursery habitat changes very little, which is why stocking on top of a habitat problem tends to disappoint.

It is also worth knowing where the fish are actually made. In most river systems, self-sustaining populations are produced in tributaries, not in the mainstem the anglers fish. The health of a famous river is frequently a fact about small creeks nobody visits.

The question is never "what kills these fish" but "what kills them, at which life stage, and is that share of mortality something anyone can change". Five sources dominate.

Predation — mergansers, loons, otters, and larger fish — falls hardest on juveniles, and is mostly a habitat-complexity problem: depth, wood and undercut banks are the answer, not predator control. Angling mortality combines harvest with the fraction of released fish that die anyway; on artificial flies in cool water that fraction is small, but it climbs steeply once water passes roughly 20 °C and fights run long, which is the entire logic behind hoot-owl closures. Disease is now a live variable in this region: whirling disease (Myxobolus cerebralis), to which rainbow trout are highly susceptible, was confirmed for the first time in British Columbia at Emerald Lake in Yoho National Park in December 2023, and detected in kootenay lake in December 2024 (CBC). Competition and hybridization act as mortality on the genome as well as the individual — the westslope cutthroat trout loses ground to rainbow trout both by displacement and by interbreeding into cutbow hybrids. Winter removes fish that entered the cold with an energy deficit, which links this gate straight back to Gate 2.

Connectivity is usually discussed as fish passage past dams, which is only one of its four axes. Longitudinal connectivity is movement up and down the channel, blocked by dams, weirs and — far more commonly — undersized or perched road culverts. Lateral connectivity is access to the floodplain, side channels and off-channel ponds where juveniles rear and fish shelter from flood. Vertical connectivity is exchange with the hyporheic zone beneath the streambed, which delivers the cold water and the intragravel oxygen the other gates depend on. Temporal connectivity asks whether the route is open in the season the fish needs it, since a channel that connects only at high water is not connected during a drought.

Species differ enormously in how much connectivity they need. bull trout run over a hundred kilometres between feeding, overwintering and spawning habitat, which makes bull trout presence a statement about the whole watershed rather than about one reach — the single best indicator species in this region.

The reason a handful of variables dominate is structural: they appear in more than one gate. Ranked by how many gates each one touches:

Two consequences follow. First, riparian condition is the highest-leverage restoration target in most systems — it is the only single variable that acts on four gates at once, and unlike geology it can be changed. Second, alkalinity is the one master variable that is purely diagnostic: it tells you what a water can become, and no intervention moves it.

Thermal limits are usually quoted as lethal temperatures, and lethal temperatures are the wrong number for a fishery. Bull trout illustrate the gap precisely: peak growth occurs at 13.2 °C, feeding declines significantly above 16 °C, fish held at 22 °C and above stop feeding entirely, and the upper incipient lethal temperature is 20.9 °C (Selong et al. 2001). A bull trout in 18 °C water is alive and not dying, but it is also barely eating — so the population persists while the fishery quietly hollows out. Manage and read water against the growth threshold, which sits several degrees below the lethal one. Full figures by species: thermal thresholds by species.

The model collapses into five questions, in this order, for any species and any water:

The master-variable list is also a spot-finding list, because fish concentrate wherever several gates are satisfied in the same square metre. Tributary mouths deliver cold water and recruitment at once. Riffle–pool transitions put a fish in slow water beside a conveyor of drift, which is the net-energy equation solved. Wood supplies cover, pool scour and invertebrate substrate simultaneously. Cold seeps and spring inflows become the only fishable water on a hot August afternoon, and their value rises exactly as the rest of the river becomes unfishable. Reading habitat at the system scale and reading water at the cast scale turn out to be the same skill applied at two zoom levels — see reading river flows.