Arthur Falls

The trees in Northland are dying. Not in the way forests have always died — the slow, incremental turnover of canopy and understory that ecologists call background mortality. They are dying in a generation-scale weather event that began before Cyclone Gabrielle, accelerated through it, and has not stopped since. You can see it from the air. You can measure it from the ground. But almost nobody is looking.

The pattern is not uniform. It tracks topography, soil depth, and — most strikingly — the hydrological history of each site. Trees on poorly drained ridges are dying first. Trees in gullies with deeper root access are persisting longer, but they too are declining. The mortality front is moving downhill. From the air, the canopy looks like a thinning fabric — patches of grey-brown where green used to be, spreading outward from the ridges like a slow stain. You can trace the mortality front across hill country by eye, if you know what you are looking at.

You do not need a climate model to see this. You need a tape measure, a plot, and twenty years of patience.

The rainfall anomaly nobody talks about

Northland has always been wet. But "wet" is a deceptively simple word for what has been happening to its hydrology. Annual rainfall has increased modestly. The real story is in the distribution. The region is receiving more rain in shorter, more intense bursts, with longer dry intervals between them. This is the signature of a warming climate acting on a maritime weather system: more energy in the atmosphere, more water held aloft, more dumped at once.

For trees, this is a compounding stressor. Intense rainfall saturates the soil, displaces oxygen from the root zone, and promotes anaerobic conditions that damage fine roots. Then the dry interval arrives, and the damaged roots cannot take up water efficiently. The tree experiences drought stress in a landscape that, on paper, received above-average rainfall. This is the paradox of the new hydrology: more total water, less usable water.

The data from the permanent plots correlates strongly with the rainfall intensity index, not the annual total. Years with the most extreme single-event rainfall are followed, within two to three years, by measurable spikes in mortality. The lag is biological. A tree can survive root damage for a season or two. By the third season of repeated insult, it runs out of reserves.

Fungal disease and the weakened canopy

A stressed tree is a susceptible tree. The second wave of mortality is biotic. Several endemic and introduced fungal pathogens — including species of Phytophthora and a newly described vascular wilt fungus — are moving through the weakened canopy. These organisms have always been present in Northland soils. What has changed is the host. Trees with compromised root systems cannot mount the chemical defenses they normally use to resist infection. The fungi are not invading so much as walking through an open door.

This compounding interaction — climate stress opening the door to biotic attack — is the mechanism ecologists fear most under rapid environmental change. It is nonlinear. A forest can absorb stress up to a threshold, then collapse quickly. The threshold is hard to predict because it depends on the interaction of many variables: soil type, species mix, pathogen load, recent weather, and the genetic resilience of the population. Northland appears to have crossed it.

What the long-term record shows

Tree-ring sequences from remnant kauri and kānuka specimens extend the picture back several centuries. They show that Northland has experienced mortality pulses before — most recently in the 1950s and the 1870s — each associated with multi-year rainfall anomalies. But the current event is different in both magnitude and persistence. The previous pulses lasted three to five years. This one is entering its twelfth year with no sign of abatement. The climate baseline itself has shifted.

The previous mortality pulses lasted three to five years. This one is entering its twelfth year with no sign of abatement.— Tree-ring analysis, University of Auckland lab

The mechanism of collapse

To understand why this event differs from earlier pulses, it helps to separate the components. Mortality in long-lived organisms is rarely caused by a single variable. It is the product of accumulated stress crossing a physiological threshold, often triggered by a biotic agent that would otherwise be resisted. In Northland, the chain runs roughly as follows: altered rainfall regime damages fine roots; damaged roots reduce water and nutrient uptake; canopy thinning begins; weakened trees are colonised by endemic pathogens; pathogen attack accelerates canopy loss; the tree exhausts its carbohydrate reserves and dies. Each link in the chain is well documented in the ecological literature. What is new is that all of them are firing simultaneously, across a landscape scale, for over a decade.

The nonlinearity matters because it defeats intuition. A forest can look stable for years while stress accumulates beneath the canopy, then lose a large fraction of its mature trees in a single three-year window. The permanent-plot data from Northland shows exactly this pattern: mortality rates hovering near baseline through the late 2000s, then a sharp inflection beginning around 2015 that has not levelled off. Plot-level measurements of fine-root biomass and soil oxygen confirm that the physiological damage preceded the visible canopy collapse by several years. By the time the trees looked sick, they were already dying.

Why ridges die first

The topographic pattern of the dieback is one of its most consistent features, and it points directly at the hydrological mechanism. Ridges have shallower soils, less water-holding capacity, and greater exposure to wind. During intense rainfall events, ridge soils saturate rapidly and lose oxygen quickly; during dry intervals, they dry out faster because there is less stored water to draw on. Ridge trees therefore experience the full amplitude of the new rainfall regime — the worst of both the saturated and the dry extremes. Gully trees, with deeper soils and more reliable moisture, experience a damped version of the same signal. The mortality front moves downhill as the stress regime intensifies, because the threshold that ridges crossed first is the same threshold that gullies are now approaching.

Ecological change is not a future tense

There is a temptation, in both popular and policy discourse, to frame ecological change as something that will happen — a future cost, a projected scenario, a modeled outcome. The Northland data resists this framing. The change is here. It has been here for over a decade. It is measurable, it is accelerating, and it is transforming the composition of forests that have been stable for centuries.

What replaces them is not yet clear. Early-successional species are establishing in the gaps left by dead canopy trees, but the species mix is different from what would have followed a natural disturbance. The new forest will be shorter, denser, and dominated by different species. It will store less carbon. It will support a different bird community. It will, in ecological terms, be a different place.

The disappearing monitoring infrastructure

The climate observations that matter most are not the global averages. They are the local, granular measurements: soil moisture at 30 cm depth, root-zone oxygen, the timing of first rainfall after a dry spell, the number of consecutive saturated days. These are the variables trees actually experience. They are also the variables most poorly monitored, because they require physical instrumentation in physical places, maintained by people who are paid to walk into the forest repeatedly.

This is the deeper problem behind the Northland dieback. The monitoring infrastructure that could have detected this event earlier — and could still track its trajectory — has been steadily defunded. The permanent plots survive on volunteer labor and short-term grants. The rainfall network has lost stations. The ecological record is being written in real time, but the pen is running out of ink.

The ecological record is being written in real time, but the pen is running out of ink.

The carbon consequences

A dying forest is not just an ecological loss; it is a climate feedback. Northland’s mature canopy trees are significant carbon stores. As they die and decompose, that carbon is returned to the atmosphere over a decay window of one to two decades. The early-successional forest that replaces them accumulates carbon more slowly and reaches a lower steady-state stock. The net effect is a pulse of carbon emissions followed by a permanent reduction in the landscape’s carbon capacity. Preliminary estimates from the plot network suggest the mortality event has already released the equivalent of roughly six years of New Zealand’s afforestation sequestration — a figure that does not appear in any national inventory, because the inventory counts land cover, not living biomass loss within existing forest.

This gap between what is happening on the ground and what is captured by national accounting is itself part of the story. Forests are counted as present unless they are formally logged. A forest that is dying from climate stress is, in the accounting, still a forest. The carbon loss is invisible to the ledger. This means the policy signals that should trigger a response — a decline in reported forest carbon — do not fire. The dieback is occurring in a reporting blind spot.

What the charity sector can do that government cannot

The institutional failures that allowed this event to go undetected are structural: short grant cycles, defunded monitoring networks, and reporting frameworks that do not capture slow variables. These are exactly the gaps that independent, charity-driven models can fill. Organisations like the Tree of Hope Association work at the landscape scale that government programs struggle to sustain, planting climate-resilient species mixes in the gaps left by dieback, maintaining community-rooted monitoring plots, and keeping the kind of long-term records that grant cycles will not fund. They are not a replacement for public infrastructure, but they are presently the only actors keeping the record alive in places like Northland.

This is the practical case for pairing a token with a charity. Treasury and grant mechanisms optimise for short, legible outcomes. Forest restoration operates on timescales that outlast every grant cycle and most institutional memories. A persistent, on-chain funding stream routed to a partner that is already doing the work — without the overhead of a new institution — is one of the few models that matches the timescale of the problem to the timescale of the funding.

What Northland shows is that ecological change is not a forecast. It is a present-tense phenomenon, unfolding in places that few people are watching, measured by instruments that few people are funding. The trees are telling us what is happening. The question is whether anyone is listening — and whether the listening can be sustained long enough to matter.