Follow the data, stupid …

A perennial problem with ecology is that it is a discipline that is far better at describing problems than it is at solving them. The Water Framework Directive (WFD) encapsulates this: after nineteen years, we have a pretty good idea of the condition of Europe’s waters but have made very little progress in restoring the half that do not yet achieve good ecological status.

The reason for this is, I suspect, because describing the problem is a task that lies squarely within the remit of a scientist whilst finding solutions requires interactions that go beyond the boundaries of science, encountering vested interests along the way.   The agricultural sector’s enthusiasm for the environment is tempered by their desire to maximise yield and earn a living from the land, politicians are wary of regulations that may deter business or raise prices for the consumer and all of us are too wedded to the luxuries that the modern world offers.

The WFD can be seen as an embodiment of the social contract, articulated by philosophers such as Thomas Hobbes whereby individuals forego some rights in order to transcend the state of nature (“… nasty, brutish and short.”) and give us access to the benefits of an ordered society.  In this case, we all consent to forego some freedoms in return for a share in the benefits that a healthy aquatic environment will bring to all of us.   “Freedom” might seem like a weighty word in this context but anyone who has watched their sewerage charges creep steadily upwards over the past twenty years should recognise this as the price we pay for the freedom to flush away life’s less desirable by-products.

The problem is defining the point at which we hand over our natural rights to a higher authority.   We understand this when driving: an urban speed limit of 30 miles per hour reflects the point at which the risk we pose to other road users are deemed societally unacceptable and our right to drive as fast as we wish has to be curtailed.  If we can translate that principle into environmental governance then we can set “speed limits” for the major pressures that impact on the aquatic environment.   How do we get from an ecologist’s understanding of a “healthy” river (“good ecological status”, in WFD parlance) to the “speed limit” for nutrients, widely recognised as one of the major pressures affecting both freshwater and marine systems?

That’s been the focus of some work I’ve been doing under the auspices of the European Commission’s Joint Research Centre, one strand of which has just been published in Science of the Total Environment.  This paper looked at the threshold concentrations for nutrients (phosphorus and nitrogen) used by EU countries, noting the very wide range of values chosen as the national “speed limit”.   The situation is complicated because, just as is the case for roads, different types of rivers require different limits and we had to look for variation between countries amidst an array of variation within countries.   What emerged, however, was a clear relationship between the threshold values and the method used to set the standard.  Those that had applied statistical or modelling techniques to national data generally had tighter thresholds than those that relied upon “expert judgement”.  I’ve included the two figures from this paper that make this point.


Range of good/moderate lake phosphorus (a) and nitrogen (b) threshold values grouped by method used to determine the value. Different letters indicate groups that are statistically different (p ≤ 0.05).   Fig. 7 from Poikane et al. (2019).

“Expert judgement” is one of those slippery terms that often creeps into official reports.   There needs to be space within a decision-making process for an experienced professional to see through the limitations of available evidence and present a reasoned alternative.  However, “expert judgement” too often becomes a shorthand for cutting corners and, in this case, grabbing numbers from the published literature that seem vaguely plausible.  There is also a darker side because, having unhitched decision-making from the evidence, “expert judgement” can become a euphemism for the “art of the possible”.  I have seen this occur during discussions around setting and revising river phosphorus standards in the UK: the regulators themselves are under pressure to balance environmental protection with economic development and tight standards can potentially limit what can be done in a catchment.

Another of our recent papers (this one’s not open-access, I’m afraid) shows that setting standards using empirical models is far from straightforward and we also recognise that standard setting is just one part of a longer process of nutrient management.   However, setting inappropriate standards simply as an expedience seems completely barmy, as you are never going to attain your desired ecological benefits.   The cynical view might be that, as the process of environmental change is invariably greater than the electoral cycle, there is limited accountability associated with such decisions, compared with more immediate political capital kudos from bringing investment and jobs to a region.


Range of good/moderate river phosphorus (a) and nitrogen (b) threshold values grouped method used to determine the value. Different letters indicate groups that are statistically different (p ≤ 0.05).   Fig. 8 from Poikane et al. (2019).

All of our work has shown that, in most cases, the relationship between biology and nutrients is weak and, for this reason, large datasets are needed if robust inferences are to be drawn.  This leads to one further consequence of our work: setting environmental standards may only be possible if countries pool their data in order to produce big enough datasets with which to work.  This is particularly the case for smaller countries within the EU, but also applies to water body types that may be relatively infrequent in one country but are more widespread elsewhere.   I had recent experience of this when working on the Romanian stretches of the Danube: they simply did not have a wide enough gradient of conditions in their own territory, and we had to incorporate their data into a larger dataset in order to see the big picture (see “Beyond the Tower of Babel …”).    Writing about the benefits of international collaboration as the Brexit deadline looms obviously has a certain irony, but it needs to be said.  Far from being the distant and unaccountable law maker of Brexiteer mythology, in this field the European Commission has been quietly encouraging Member States to share experience and promote best practice.  One can only speculate about the future of the UK environment once free of Brussels oversight.


Philips, G., Teixeira, H., Poikane, S., Salas, F. & Kelly, M.G. (2019).   Establishing nutrient thresholds in the face of uncertainty and multiple stressors: a comparison of approaches using simulated data sets.   Science of the Total Environment684: 425-433.

Poikane, S., Kelly, M.G., Salas Herrero, F., Pitt, J.-A., Jarvie, H.P., Claussen, U., Leujak, W., Solheim, A.L., Teixera, H. & Phillips, G. (2019).  Nutrient criteria for surface waters under the European Water Framework Directive: Current state-of-the-art, challenges and future outlook.  Science of the Total Environment 695.  

Note on figures:

The methods used by Member States to derive nutrient thresholds are described in more detail in Poikane et al. (2019).   In brief, the approaches are:

1 – regression between nutrient and biological response;

2 – modelling;

3 – distribution of nutrient concentrations in water bodies classified (using ecological criteria) as high, good or moderate status;

4 – distribution of nutrient concentrations in all water bodies using an arbitrary percentile;

5 – expert judgement.  This includes values taken from the literature or from older European Directives. For example, for nitrate, the common use of the value 5.65 mg-N L−1 in freshwaters is likely to be derived from the guideline value of 25 mg L−1 of nitrate in the Nitrates Directive (91/676/EEC) or now repealed Drinking Water Directive (80/778/EC).

6 – The so-called OSPAR Comprehensive Procedure is used widely in coastal and transitional waters. In this, a water body is considered to be an ‘Eutrophication Problem Area’ if actual status deviates 50% or more from reference conditions.

7 – insufficient information.


Tales from prehistory


Microscopes and Monsters has been quiet for a couple of weeks, as I have been on holiday, part of which was spent “off grid” at the Green Man Festival in Wales.  From there, we headed to London for a Proms concert (two music festivals in a week!) via the Cotswolds and the ancient landscapes of Salisbury Plain.

My first visit to Stonehenge was 50 years ago, at which time you could pull off the A303 and wander amongst the columns unconstrained by fences and barriers.   Now the visitors are guided to a visitor centre two kilometres from the site, and offered a shuttle bus after being relieved of £20.  Or, if you prefer, you can walk across Salisbury Plain to the monument.  On a sunny afternoon in August this becomes part of the experience as there are ancient burial mounds (some pre-dating Stonehenge) both alongside the path and dotted around the horizon.   Stonehenge itself gradually rose up ahead of us, and we experienced a little of what the ancients must have felt as they approached Stonehenge along the processional way.

The last time that I was here was a stop off between field work on the nearby River Wylye and a meeting in Reading.   At the time I was engaged with two separate projects concerned with the health of chalk streams, which are characteristic of this part of southern England.   The approach we used elsewhere in the country was to compare what we found in samples we collected with what we expected to find if that site was in a pristine condition.

There was, at the time, a vigorous debate about how this “reference condition” should be defined.   This debate had a theoretical component (epitomised by Brian Moss’ paper in the reference list) but also a more pragmatic element (encapsulated by the other paper).  This was necessary because an ultra-strict, but theoretically sound, approach might not yield enough data from which a robust prediction of the “expected” ecology could be derived.   In essence, we searched out remote regions of the UK where population density was low and agriculture was not intensive and used these to derive our understanding of what to expect in the more densely-populated regions of the country.


Part of Avebury stone circle, Wiltshire, August 2019.  The photograph at the top of the post shows Stonehenge.

This worked quite well (although Brian Moss, predictably, had his own pithy thoughts on the approach).  However,  we simply could not find any sites that fulfilled our criteria of low population density and a low intensity of farming in those parts of lowland Britain where the underlying geology was Jurassic limestone, Cretaceous chalk or another formation that resulted in very hard water.  Our estimates of ecological health in such regions depend, as a result, on extrapolation and judgement rather than evidence.   That is all well and good for an academic journal but, in the case of the River Wylye, Wessex Water were being asked to spend hundreds of thousands of pounds to upgrade sewage works and, rightly, felt that they needed something more in order to explain the consequent price rises to their customers.

The OS maps of the region around my sampling points on the Wylye were dotted with symbols marking ancient monuments (long and round barrows, in particular) and the huge, mysterious religious sites of Stonehenge and Avebury lie just to the east.  Together, they point to continuous occupation of the area for over four thousand years, which means that it is hardly a surprise that we found no sites that met our criteria for a “pristine” stream.   The chalk streams of southern England are famous and rightly regarded as a threatened habitat, but they are not natural.  It is better to think of them as aquatic equivalents to hay meadows or hedgerows: ecologically-rich habitats that have been created by human activity, rather than as a result of “natural” ecological processes.

That means that it we need to diverge from a strict definition of “reference conditions” in order to set a baseline for ecological expectations in such circumstances.  For macrophytes – the larger aquatic plants – there is an expectation that the flora in this baseline state will be rich; however,  this assumption does not work for the microscopic algae in chalk streams.  We also found that river stretches where the macrophytes are thriving and, apparently, healthy, often have diatoms that suggest nutrient enrichment.  That is a puzzle for which we think we may have a solution, and which I will write about in a future post.


Silbury Hill, part of the Avebury World Heritage Site.  It is a Neolithic site whose original purpose is unknown though, to a visitor from north-east England, it looks remarkably like a slag heap.

We use low population density and absence of intensive agriculture as a proxy for “natural” in the uplands but need to treat this assumption with care too.  There might be fewer grand Neolithic monuments in the north of England or Scotland but signs of ancient habitation are there if you care to look (see “More reflections from the dawn of time”).   The moorland where these streams rise is, itself, an artificial habitat, created when early agriculturalists removed the natural tree cover.  Modern streams in these areas are, therefore, exposed to more light than in their primeval states and that will have important consequences for the plant life that lives within them.  They may be the best we have, but are hardly “natural”.

Two factors, both highly pragmatic, brought this debate to a close.   The first was realisation that, whatever the rights and wrongs of purist versus practical standpoints, most of our rivers are very degraded and alterations in the approach used to define the “expected” condition would be unlikely to change this broad scale picture.   About sixty-five per cent of our rivers fail to achieve good ecological status despite the flaws in the reference concept.  The second factor was simply that, since the financial crisis in the 2008-2010, the UK environment agencies have had too few resources to improve the reference concept.   As any such “improvement” will almost certainly make the true state of UK rivers look even worse than it does at the moment, a more cynical argument is that few of the bureaucrats involved in the process have any great enthusiasm for the task anyway.


Moss, B. (2008).  The Water Framework Directive: total environment or political compromise.  Science of the Total Environment400: 32-41.

Pardo, I., Gómez-Rodríguez, C., Wasson, J.G., Owen, R., van de Bund, W., Kelly, M., Bennett, C., Birk, S., Buffagni, A., Erba, S., Mengin, N., Murray-Bligh, J. & Ofenböeck, G. (2012).  The European reference condition concept: A scientific and technical approach to identify minimally-impacted river ecosystems.  Science of the Total Environment420: 33-42.


The presence of absence in Castle Eden Dene


Some of my strongest impressions of Castle Eden Burn after last week’s visit concerned not what I found in the stream, but what was not there.  I mentioned in my previous post that I had not seen the mosses that I associated with streams in northern England in Castle Eden Burn, but there were other species, too, that I had expected to see but had not noticed.   Once I have noticed that something is absent, this absence becomes present.  I have noticed the presence of absence.  Woohoo: I’ve shoehorned Jean-Paul Sartre’s Being and Nothingness into a blog about ecology.

When I got back home I had read a chapter about the FBA’s study of the Winterbourne in Dorset, an intermittent stream flowing off the chalk downland, and noticed that they had recorded plants there that I knew from north-east English rivers, but which I could not remember seeing in Castle Eden Burn.  Was this because I had not searched the stream environs thoroughly, or is this a real difference between intermittent streams on chalk and on Magnesian limestone?

I went back this weekend to try to answer these questions, taking Heather with me, as her skills with the higher plants far exceed mine, and walked as much of the stream bed as we could, starting near the remains of a footbridge at NZ 424 389, and making our way downstream to Denemouth, where Castle Eden Burn joins the North Sea.  If my original intention was to better understand the burn by traversing space within the Dene, my first lesson concerned time: a week with some heavy rainfall separated my two visits and it was clear straight away that the Burn had been flowing during the week, with a fine layer of silt and mud spread across much of the surface, making parts of it slippery to walk upon.  There were standing pools of water at several points in the upper part of the Burn too.   Within a week the stream had come and gone, offering scant opportunities for any water-loving organism to establish.

We made our way along the Burn through the delicious silence of the forest.  The banksides were richly vegetated: masses of opposite-leaved golden saxifrage plus the mosses I described last time and many others, along with plenty of harts-tongue fern (Asplenium solopendrium).   Then, with a very clear demarcation, there was the stony stream bed with very little vegetation at all.    We looked hard for three plants, in particular, that I associated with the damp margins of streams, and which I had expected to see here: Verronica beccabunga (brooklime or water speedwell), Rorippa nasturtium-aquaticum (water cress) and Mentha aquatica (water mint).  None seemed to be present in any of the stretches we visited apart from a single sorry looking brooklime in the freshwater marsh at Denemouth .

What we did find, a little further downstream, was a pebble and gravel-dominated stretch with a straggly array of plants, all bent over in the direction of flow.   These included broad-leaved dock (Rumex obtusifolius), nettles (Urtica diocia), a few shoots of Himalayan balsam (Impatiens glandulifera) and some grasses.    Were I not standing on a dry stream bed I would have assumed that this was a bare piece of ground being colonised by typical ruderal species.  And that, I think, offers some insights into the ecology of Castle Eden Burn.   This is not a stream that occasionally dries out: it is a long-thin terrestrial habitat that is occasionally flushed through by water.   Welcome to north-east England’s premier wadi.


Rumex obtusifolius and other ruderal vegetation on the stream bed of Castle Eden Burn, August 2019.

This hypothesis really needs corroboration by a hydrologist, but the graph I showed in “Out of my depth …” shows that, despite flow being generally low,  episodes of high flow are scattered throughout the year, and I suspect that these keep the substratum mobile and, more important, stop organic matter accumulating to give amphibious plants an opportunity to establish.   The water table, too, I guess, is too far below the stream bed in between the spates to make it easy for plants to stay hydrated.   This is one of the main differences between Castle Eden Burn and the southern chalk streams, which are characterised by very stable flow regimes

From the point at which Blunt’s Burn enters Castle Eden Burn (NZ 436 396) there does seem to be permanent flow down to the sea.  Still, however, there was very little in-stream vegetation.  That was in contrast to the forest around us, which was floristically-rich (Heather has written more about this on her blog) and, on this warm summer morning, positively humming with bees and aflutter with butterflies.

A large embankment takes the busy A1086 over the Dene, the Burn passing through a long culvert at this point, after which there is a viaduct taking the coastal railway line across before the dene widens out into a large area of meadow just before it reaches the sea.   The stream’s path to the sea is, however, blocked by mine waste that was dumped from the coal mines that used to line the Durham coast.  This forces the stream to turn ninety degrees south for a few hundred metres before finding a way through and, gradually, trickling and percolating through the beach. The mines have all gone now and the sea is gradually eroding this compacted mass of waste.  Before the waste arrived, apparently, there was an area of saltmarsh at the mouth of the burn.   Now, there is a freshwater marsh, dominated by reeds (Phragmites australis).  When the mine waste finally goes, maybe the saltmarsh will return.  Meanwhile, Castle Eden Burn has no grand finale: it ends on a whimper not a bang.

We climbed a narrow, steep pathway up through gorse and brambles onto the clifftops overlooking these final stages of Castle Eden Burn to get a view that was, in light of all that had passed through my mind earlier, oddly symbolic.  The stream flowed almost due east until it encountered the bar, and the gentle arc which it then describes looks just like a question mark.   How ironic, I thought, for a stream that raises more questions than answers to sign off in that way ….


Denemouth, at the end of Castle Eden Dene, just above the point where the stream joins the North Sea.