Large wind turbines are often blamed for mortality and declines in bats, whilst developers claim limited or no impacts. A recent report looks at some of the evidence, and just how much we still do not know.
Whether or not bats are affected by wind turbines is much argued at Planning Inquiries. For many, the data produced are suspect. Central to the arguments is normally the chapter on wind turbines in the 2012 Bat Conservation Trust’s non-prescriptive survey guidelines.1 Endorsed by the Statutory Nature Conservation Organisations (SNCOs), the guidelines suggest a standardised approach, and the use of professional judgement (supported by reasoned proof) when varying from the standards. This is rarely substantiated. Essentially, the existence of the guidance has been used to undertake surveys in a non-standard manner.
Because the wind farm chapter needed a little more rigour and validation, Fiona Mathews2 at Exeter University was commissioned in 2010 by several Government departments, SNCOs and RenewableUK to look at the factors affecting bats at turbines and, hopefully, to put a little more science behind the arguments on possible turbine impacts.
There are currently c5492 large turbines (turbines > 100MW working capacity) at 1069 onshore operating wind projects in the UK.3. It is important to understand what we know about their potential effects on bats and other species, and how planners should use this when reviewing planning applications.
Turbine history
The first commercial wind farm opened in the UK in 1991. Turbine building programmes in the USA and Germany produced the first reports of noticeable impacts on birds and bats at wind farms.2.
Understanding potential impacts on protected and other species is required by the EU EIA Directive 85/337/EEC4 for all proposals for wind farms ≥ 5 turbines, or > 5MW capacity. In addition, the Habitats Directive 92/43/EEC 5, when translated into UK law, means that deliberate killing or injuring of bats, or deliberate disturbance that would significantly affect their local distribution and abundance (the exact wording varies between England and Wales, and Scotland) is illegal.
A European bat study6 showed that bats were being found dead around turbines in Europe. 20097 saw the first documented UK windfarm mortalities..
In response, the SNCOs produced a series of interim guidance for potential turbine sites.8,9,10 This categorised risks to individual species. The exact basis for this was unclear, due to limited data assessments and evidence.10, 2 If bat population estimates are unknown, then impacts on population viability are equally uncertain. Essentially, it is hard to prove, or disprove a possible impact
BCT followed the SNCO guidance with its own document.1 The use of non-prescriptive guidance means that the quality of data provided in Environmental Statements and EIA assessments is highly variable. It had been hoped that the updated 2016 guidance11 would incorporate the new study2 : providing tighter, more reasoned, methods for surveying sites with potential wind farm developments. Mathews2 was published after BCT’s update.
There was a lot riding on Mathews: clarity on methods, assessments of mortality at windfarms, and understanding the key factors affecting casualties and bat activity. The Press Release12 stated that the findings would be used to revise and update existing advice and guidelines on site assessments.
How far is Mathews was a significant step forward? Can it help move from indicative to prescriptive guidance?
The Mathews et al Report2 (MR)
Background
Some 6 years in the making, the study had three initial aims:
Determine whether bats are killed by wind turbines in the Great Britain.
Establish the species and sex distribution of any bat casualties at wind turbines.
Assess the relationships between casualty rates and bat activity (acoustic data), habitat and local weather, to improve future mitigation strategies.
Pre-Mathews, most studies were either based on incidental data collection 13,14,6 or other non-standard searches or studies 15 . It was hard to isolate the key factors potentially affecting mortality at turbines. If deaths associated with turbines had a local, or wider, impact was unclear. Establishing this would be a major step forward in practice and in policy making.
MR chose a stratified random sample of areas with wind farms and habitat types. This required sites to have at least 6 working turbines >50m tall, though smaller ones ended up being included in 9 cases. MR noted that their 2010 selection was non-random.
MR selected the late summer/ early autumn period, when bats are relatively active, for their single month’s sampling, rather than the whole bat activity year. They recognised that their results were limited to that period. This affected their generalisability in prescriptive guidance.
The sample used 46 (2620 turbines) of the 204 UK operational windfarms. By publication in 2016, this had swelled to 1069 operational windfarms. As MR’s sample was unrepresentative at the outset, it unlikely that this has changed: results represent their sample, and may have limited generality beyond that.
Typically, field surveys at a prospective wind farm site occur in one year only, often in a short part of the survey year 1. The extent of year-to-year variation, and the generalisability of the data collected, is often moot. To address this, MR looked at four repeat sites, surveying one in three years (2011-2013), and in three in two (2012-13).
To tease out the potential effects of environmental variables on the numbers of bats occurring at or around a site, MR quantified the surrounding habitats (≤1500 & 2500m from the centre and edge of the site).
Sampling
MR surveyed 3 turbines at each wind farm for corpses. Using dogs and handlers, 100m radii were searched for bat carcasses in the early morning every 2-3 days. Trials showed that dogs were more than three times as efficient at finding carcasses, and took 25% of the time less-efficient human searchers did. With scavengers removing corpses, time matters.
Blind trials-putting out corpses to be found, showed only 75% of corpses were discovered. So, not finding corpses is not the same as no bats killed. This affected the statistical analyses.
In addition they used Song Meter SM2 bat recorders at the base and, where possible, on the turbine nose (nacelle) of the same 3 turbines at each wind farm: to detect and compare the scale and identity of bat species recorded at ground level and at height.
Few wind farm studies admit that SM2 have detection limits of <30m2. As most turbines are >60m tall, ground-based recorders are not catching what is going on in the potential Collision zone of the whirling rotor blades.
SM2s were also placed on the ground in ‘control areas’: areas of similar habitats < 1939m away. In 2013, weekly walked surveys were carried out on survey sites, to compare methods usually used1 with the SM2 results.
Background results
Were bats killed near turbines? The answer was yes: 120 corpses were found, including the first European record of a Natterer’s bat Myotis natteri at a turbine fatality. Almost all corpses were within 30m of the turbine tower. With c 75% search efficiency, it was likely that more bats were killed.
What was the identity and sex ratio of bats killed at turbines? Of the 120 corpses, 51 were soprano pipistrelles Pipistrellus pygmaeus, 46 were common pipistrelles Pipistrellus, 11 were noctules Nyctalus noctula, with single Nathusius’s pipistrelle Pipistrellus nathusii , brown long-eared bat Plecotus auritus and Natterer’s bat. 8 pipistrelles were unidentified 46 were common pipistrelles, 11 were noctules, with single nathusius’s pipistrelle, brown long-eared bat and natterer’s bat. 8 pipistrelles were unidentified, as was the remaining corpse. There was no sex bias, nor age bias: young bats were no more likely to collide than older bats.
What were the relationships between casualty rates and bat activity, habitats and weather? The answers were not clear-cut. Most were hedged around with qualifiers and caveats, precluding tightening guidance.12
Bats were highly unpredictable: there was huge variation in bat activity within and between sites from day to day or visit to visit, including by a factor of 6 over 4 nights at one site. There is no such thing as a reliable quick visit to a site; this applied to static recorders (ground and at height) and to transect surveys. Year-to-year data were also highly variable on the same site. As many bats surveys regularly stick data from 2 different years together to make a single April-October ‘season’ this is a basic fault in data used for planning.
There was large spatial variation in activity levels within sites, and no clear link between numbers of bats at control sites and turbine sites, nor was there a link between activity at control sites and turbine mortalities. Put simply, proxies are of little use in predicting impacts on site.
Just as there were differences between sites, there were differences between recorded activity levels on the ground and at height (recorded on the nacelle). Activity levels on the ground were greater than at height, species also differed between the two. They were effectively sampling different elements of the bat populations on the ground and at height.
Detailed results
MR undertook a range of complex analyses. These came with a health warning: their sample was statistically limited, especially when looking at predictors of the presence or absence of bat deaths at turbines, activity levels and mortality rates, so that results should be “considered exploratory”.
They added one additional rider (important for developing guidelines): the project was not designed to assess pre-construction survey activity, but might offer some insights. Having collected no pre-construction data, this should be obvious, but the comparisons with control and other sample sites has clouded the aspirations of the sponsors, and needs thinking about at all stages of the project results.
With caveats in place, MR looked at habitat, weather (temperature and wind speed), height and activity levels as predictors of any mortality, or rates of mortality at operational wind farms. As many of these variables were interconnected, it was hard to factor out the individual contributions in predicting the occurrence or rate of mortality.
Mortalities
Almost all the turbines were large, but there was an increased 18% risk of mortality with each 1m increase in blade length. No other individual factors (nacelle height, year of study, time in study period, age of site, number of carcass searches) contributed significantly to mortality risk. Mortality rates showed the same basic trend: big-bladed turbines had higher casualty rates.
The relationship between the number of casualties per site and the number of turbines was significant: more occurred at bigger sites, so that risk increases with site size.
Weather
All recording took place in basically suitable weather conditions (>10C at dusk, low wind speed, minimal rain (<2.5mm/hr)) teasing out any general pattern for bats in general was difficult. Deaths occurred, or failed to occur, on suitable nights.
Wind speed was biased by height: a 4m/sec wind at ground height was typically 8m/sec at 60m. Relying solely on ground data would mis-represent what was going at rotor height.
All 120 deaths occurred at a range of low wind speeds (as recorded on the ground), every one was at <6m/ sec. There were also many nights with low winds, but no deaths.
There was no general link with temperature, as deaths occurred over a range of temperatures >10C. Most suitable nights had no deaths.
Activity levels: at the turbine
Activity levels varied significantly from site to site, and from day to day. Activity levels at ground level were significantly more than those at height for each species. There was no link between total casualty rates (i.e. collisions at rotor height) and bat activity levels recorded at ground level. Relying only on ground data would result in under recording both Nathusius’s pipistrelle and noctules, and bias estimates of site risk and species composition.
There was a significant link between the presence of bat casualties and activity levels at the nacelle.
Activity levels: control sites
Activity levels at control sites were not linked to turbine pass rates or the number of casualties. They were poor proxies.
Bats and habitats
There was no link between the presence or absence of casualties and habitat type at the wind farm, nor with habitat cover.
There was a negative link between the number of bat casualties and the % of broadleaved and mixed woodland cover within 1500m. This was strongest for soprano pipistrelles. It was suggested that availability of woodland offered foraging opportunities, keeping bats away from turbine areas. Noctules differed, responding positively to the total amount of coniferous woodland.
Key issues
MR answered some questions, but not all. As the 46 sites were from an unrepresentative sample, and by 2016 the number of wind farms and turbines has more than doubled, skewed more and more towards Scotland3 to what extent it represents the current reality is unclear. This also means that the statistical warnings that applied to the initial study will be even more critical.
What lessons can be learnt from a single month’s sampling in the whole bat year? Were these strong enough to produce prescriptive guidance as had been hoped?
Bats die in and around turbines. MR estimated casualty rates between 0- 5.25 bats per turbine per month for their short study period. With inadequate data it is hard to put a figure forward for the whole active bat year per turbine.
MR sampled only a fraction of the turbines in any one site. They recognised spatial patchiness in the data; areas with high mortality within a large site may well have been left unsurveyed.
Death was selective: soprano pipistrelles and noctules were found more frequently than their estimated relative national abundance. These proportions were higher than indicated by ground-based recording, and close to the proportions recorded by the nacelle-based recorders. Put simply, MR recognised that ground-based recorders were missing most of the bats flying at rotor height. Does this matter? Yes. As they put it: “This therefore raises concerns about assessments of the relative abundance of some medium and high flying specialists if monitoring is conducted only at ground level.”
BCT1 suggests that undertaking at-height recording (in the potential rotor zone) needs to be justified, so is rarely done. MR showed that many ground-based data sets for potential windfarms may well be unsafe.
If at-height recording is normally ignored, is this the only problem for routine data collection? BCT 1, 11 set out minimum numbers of visits to sites based on the non-substantiated risk categories. MR looked at the number of days of recording on ideal weather nights- rarely the case- in Britain, concluding that a minimum of 12 ideal nights of ground recording and 19 ideal nights of at-height recording were needed to estimate 80% of maximum activity rates. That is not in either BCT guidance.
If finding 19 ideal nights is a bit daunting, or time is short, then it is not uncommon to slot a lot of recorders onto a site. Does this blitz approach work? MR concluded not: “further bias may be introduced as a result of bat surveys being too short in duration”.
They summarised their concerns: “at least a third of surveys conducted to current specifications are not detecting the full composition of species at a site”.
It seems that routine data sets are non-robust, especially if missing at-height data, and ignoring limits in detection distances for most machines.
The very high variability in the data, and the limited understanding/ background for assessing impacts/risk, means that claims of no impact at potential developments cannot be substantiated. This is critical when most surveys focus solely on ground-based data collection. This neither samples the rotor zone, nor does it significantly predict rotor level collision rates.
Recommendations
What sort of recommendations did MR put forward? MR’s main management recommendation was standardized post-construction surveys, followed by curtailment based on wind speed. This has one or two issues. MR stated said not to extrapolate in the absence of data for the rest of the April- October bat activity period. Given this, perhaps there should perhaps be at least two years of detailed surveys- as repeat year sampling showed that data from just one year was highly divergent from other years. Knowing what is ‘typical’ is still very unclear; one year may well not be. In similar circumstances bird guidance requires at least 2 years of data, or more.16 MR suggested feathering blades at low speeds might reduce mortalities- but more data are needed to substantiate this recommendation.
MR listed 10 research priorities needing work. Most of these are fundamental: offering the possibility of closing the gap from surmise to robust statistical predictability- something MR recognised was not possible with most of their data.
If 12 the SNCOs want to focus on case-by-case assessments, then they need a fall-back of a body of reliable methods and data to indicate no impact. MR has shown that valid standardised methods and data are basically missing from current wind farm surveys. A lot more rigorous research, and validation of methods and data, is needed. Until then, the protocols used by the SNCOs should come with a health warning, they do not provide the hard material claimed by would-be developers. Instead, the material uses as proof of no possible impacts is unsafe. Ground-based data don’t cut it. In terms of the hopes of the SNCOs, prescriptive methods still seem a little way off.
References
- Hundt. L. (2012) Bat Surveys: Good Practice Guidelines, 2nd edition,. London, BCT.
- Mathews, F., Richardson, S., Lintott, P. & Hosken, D. (2016) Understanding the Risk to European Protected Species (bats) at Onshore Wind Turbine Sites to inform Risk Management. University of Exeter.
- RenewableUk website. http://www.renewableuk.com
- Environmental Impact Assessment (EIA) Directive 85/337/EEC. Brussels, Belgium.
- Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora. Brussels, Belgium.
- Rodrigues, L., L. Bach, M. Dubourg-Savage, J. Goodwin, & Harbusch, C. (2008) Guidelines for consideration of bats in wind farm projects. UNEP/EUROBATS Secretariat, Bonn, Germany.
- Jones, G., R. Cooper-Bohannon, K. Barlow, & Parsons, K. (2009) Scoping and method development report: determining the potential ecological impact of wind turbines on bat populations in Britain.
- Natural England. (2009) Bats and single large wind turbines. Natural England Technical Information Note TIN059
- Natural England (2012) Bats and onshore wind turbines. Natural England Technical Information Note TIN051
- Natural England (2014) Bats and onshore wind turbines. Natural England Technical Information Note TIN051
- Collins J. (ed) (2016) Bat Surveys for Professional Ecologists – Good Practice Guidelines, 3rd edn. London, BCT.
- BCT (2016) http://www.bats.org.uk/news.php/342/bats_and_wind_turbines_research_project
- Rydell, J., L. Bach, M. Dubourg-Savage, M. Green, L. Rodrigues, & Hendenström, A. (2010) Bat mortality at wind turbines in northwestern Europe. Acta Chiroptirologica. 12: 261-274.
- Rydell, j., Engstrom, H., Hedenstrom, A., Larson, J.K., Pettersson, J. & Green, M. (2012) The effect of wind power on birds and bats- a synthesis. Stockholm, SEPA.
- Bach, L., P. Bach, K. Eckschmitt, K. Frey, & Gerhardt, U. (2013) Bat fatalities at different wind facilities in northwest Germany. CWE2013, Stockholm, Sweden.
- Scottish Natural Heritage. (2014) Recommended bird survey methods to inform impact assessment of onshore wind farms. Perth, SNH.
Tim Reed
Head of Tim Read Ecological Consultants Ltd.