Ecology has a calendar. It is set by when plants flower, when bats roost, when birds breed and when newts return to ponds. It does not move.
Consent programmes have a calendar too. It is set by board approvals, funding decisions, statutory deadlines and, from 2 November 2026, by the point at which a development consent order application becomes subject to mandatory biodiversity net gain.
These two calendars are not aligned, and on nationally significant infrastructure projects they are frequently in open conflict. That conflict is about to become a compliance issue rather than a scheduling inconvenience.
Habitat condition assessment for the statutory biodiversity metric depends on being able to identify what is growing. In practice that means the growing season. Practitioners generally work to a window running from mid to late April through the end of August for most habitat types, with heathland and wetland assessment often pushed later into June to September. Urban habitats and individual trees can be assessed more flexibly.
Grassland is the awkward case, and grassland is everywhere on a linear scheme. Distinguishing a modified grassland from an other neutral grassland, and assigning a condition score to it, is difficult outside the flowering season. Get it wrong and the habitat is undervalued, which sounds like a conservative error until you realise it changes the number of units you have to deliver.
Protected species surveys layer their own windows on top:
Note what that list does. It stacks almost everything into the same five months, and it places bat hibernation surveys in direct opposition to everything else. A site that needs both a full habitat condition assessment and hibernation roost data cannot get them in the same season.
Surveys carried out outside the optimal window are not automatically invalid, but they are less reliable, and a negative result from a sub-optimal survey should not be read as absence. That distinction is well understood by ecologists. It is less well understood by programme managers, and it is exactly the sort of thing that gets probed in examination.
Three features of the NSIP regime sharpen the problem.
A draft habitat management and monitoring plan is required at application. The ecology cannot be a work in progress at submission. Survey work, habitat design and management planning all have to be substantially advanced before the application goes in. That pulls the survey requirement forward by a season or more compared to a scheme where detail follows consent.
The baseline covers a worst-case envelope, not a final design. Where locations and construction areas are not yet fixed, guidance directs applicants to assess the realistic worst-case extent of habitat loss using maximum design and construction parameters. You are surveying the envelope. The envelope is bigger than the scheme.
The corridors are long. A 4,500-hectare linear asset cannot be walked in a fortnight. Traditional survey effort scales roughly with area, and area on an NSIP is not modest.
Put those together and a single missed season is not a delay of weeks. It is a delay of a year.
There is a second timing problem that receives less attention: survey data ages.
CIEEM's advice on the lifespan of ecological reports gives a usable rule of thumb. Data less than twelve months old is likely to be valid in most cases. Between twelve and eighteen months it is still likely to be valid, with caveats for mobile species and for sites that have gained new features. Between eighteen months and three years, a professional ecologist needs to visit the site and review whether the data still holds. Beyond three years, the report is unlikely to be valid and most surveys will need to be redone.
NSIP consent programmes routinely run longer than three years from first ecological survey to decision. On a long programme it is entirely possible to survey a corridor, watch the design evolve, and arrive at submission with a baseline that has quietly expired.
The trap is that this is invisible until someone asks. Then it is a re-survey, in a season that may be six months away, on a programme that has no float.
There is no way to make plants flower in February. But there are ways to stop the calendar from setting the whole programme.
Separate the wide view from the deep view. Most of the survey effort on a large corridor goes into establishing what habitat is where. A much smaller proportion goes into the condition and species work that genuinely requires a specific window. Splitting those two questions lets the first one happen year-round and reserves scarce field time for the second.
Use remote habitat mapping to target field effort. A consistent habitat classification across the whole order limits, available in any month, tells you where the interesting ground is before you commit a survey team to it. It also gives you the evidence base for the exclusion argument the BNG boundary plan requires, because to justify leaving unimpacted habitat out of the baseline you need to have looked at it.
Front-load the envelope. If the baseline has to cover a worst-case envelope anyway, mapping the full envelope early costs less than mapping it twice as the design shifts.
Track data age deliberately. Put survey dates on the risk register alongside consent milestones. An eighteen-month marker on ecological data is a cheaper conversation than a three-year one.
Design the programme around the constraint, not against it. The teams that handle this well treat the survey calendar as a fixed input, in the same category as a grid connection date, and build backwards from it.
Satellite and aerial habitat mapping removes the seasonal constraint from one part of the problem. Habitat extent and classification can be produced at any time of year, across very large areas, consistently, and repeatedly.
It does not remove the constraint from the rest. Condition assessment involves judgements about sward structure, species composition and management that are made on the ground, in season, by a competent person. Protected species surveys require licensed ecologists in specific windows. Neither of those is going away, and neither should.
What changes is the sequence. Instead of sending teams out to find out what is there, you know roughly what is there and send teams out to confirm it and score it. The same field budget covers more ground, and it covers the ground that matters.
This is why the honest version of remote sensing includes a confidence layer. Accuracy tells you what we think is there. Confidence tells you whether you should act on it without checking. On an application that will be examined, that distinction is the whole point, and we cover it in more depth alongside the questions an examining authority is likely to ask.
If you have an NSIP heading for an application after 2 November 2026, the practical question is which season your baseline depends on and whether you still have it. The full picture of what the November change requires is set out in BNG for NSIPs: what changes on 2 November 2026, and if the answer is that you are already tight against the window, what is realistically deliverable before November is the more useful read.
If you want a view of habitat across your order limits without waiting for April, talk to us about a baseline scope. We will be straightforward about what we can tell you now and what still needs a season.