Skip to contents

FAQs

1. What ecological components can the hetoolkit be used to assess?

In principle the toolkit can be used to help analyse data on any ecological receptor of interest. Depending on the receptor, manual importing and processing of your ecological data may be required before you progress onto using the toolkit’s various functions. This is one of the advantages of the toolkit being in the R environment: personalised code and manual workarounds can be interwoven with the toolkit’s functions to best suit your analysis.

2. What ecological data can be downloaded by the hetoolkit?

The toolkit currently only provides functions to import data from macroinvertebrate samples, including indices (import_inv()), taxon-specific abundances (import_inv_taxa()) and contextual environmental information (import_env()). However, macrophyte, diatom and fish data can be manually downloaded from the EA’s Ecology & Fish Data Explorer and then imported into R for subsequent processing and analysis. There is a bulk downloads page (https://environment.data.gov.uk/ecology/explorer/downloads/) which makes it possible to import data ‘en masse’ into R and then filter to your sites or area of interest. All of this, including the downloading of data, can be done in R (see https://www.rdocumentation.org/packages/utils/versions/3.6.2/topics/download.file).

3. What hydrological datasets can be downloaded and which are the most appropriate to use?

Observed (gauged) river flow data can be downloaded from the National River Flow Archive and the EA’s Hydrology Data Explorer using the import_flow set of functions. These data can be appropriate to use if flow variation at a gauging station is representative of that at a paired ecology site, and if the focus of the analysis is on responses to historic flows, rather than flow alteration. Otherwise modelled flow data should be used, which will require model outputs to be manally imported into R.

4. Which is best to use – surface water or groundwater hydrological models/datasets, or both?

In parts of the country covered by well-calibrated, EA-approved groundwater models, these are the best source of modelled flow data. Elsewhere tools such as QUBE or rainfall-runoff/lumped catchment models may be used. If you are uncertain about the best available source of flow time series data, you should consult a local EA hydrologist.

5. Can the hetoolkit be used to download water quality data?

Yes, the import_wq() function allows you to download data for determinands of interest from the EA’s Water Quality Archive.

6. Can I import my own datasets?

Yes, you can import your own data and then use the hetoolkit for processing and analysis. You may first need to format your data to ensure it is compatible.

7. Can I just use the first stages of model development to download and prepare datasets?

Yes, you can use any or all of the functions in the toolkit as you see fit. You may choose to use the toolkit for relatively simple actions such as running RICT predictions, importing and plotting flow data or creating HEV plots. The toolkit is not exclusively for the development of HE models.

8. Is the use of the hetoolkit a ‘must’ for water company assessments?

It is not obligatory to use the toolkit for all water company HE assessments, although some applications (e.g. local flow constraint proposals) demand a modelling approach with a comparable or greater level of statistical robustness to that offered by the workflow of the toolkit. Ultimately the toolkit makes it easier to collate, process and analyse HE data, and it is therefore in the interest of water companies and their consultants to use it.

9. What range of ecological modelling tools or assessment approaches can be used?

The toolkit primarily facilitates the building of mixed regression models using ‘pooled’ monitoring data. These can take the form of generalised linear models or generalised additive models. The latter, which may be abbreviated as GAMs, GAMMs or HGAMs, are more versatile as, among other things, they can capture both linear and nonlinear ecological responses. The case study in this website’s article library ‘Macroinvertebrates in upland rivers’ illustrates how to undertake generalised additive modelling.

other modelling approaches may be complementary or in specific cases more suitable. These include hydraulic-habitat modelling and multivariate analyses of community composition. There is a case study on the latter (‘Diatoms’) in this website’s article library. Even for approaches not using mixed regression models the toolkit may still be useful for data collation and processing.

10. Is the hetoolkit designed only for the assessment of invertebrate impacts?

No, the toolkit is designed for use with any ecological receptor of interest. The toolkit’s functionality is more conducive to the analysis of invertebrate data because invertebrates are the best flow bioindicators: they form diverse communities comprising taxa with distinct and well-established flow preferences. Most HE analyses and models therefore tend to use invertebrate data. However, other aspects of the ecology may need to be considered depending on the study.

11. Are there any case studies for macrophytes, diatoms or fish assessments using the hetoolkit?

There are toolkit case studies on macrophyte and diatom responses to flow in this website’s article library. We are planning work to explore how fish data can be best represented in HE models.

12. Can the same flow time series data be used for analyses of different ecological receptors?

Yes, the same flow data can be used to build multiple models based on different ecological receptors, although the selection of optimal flow windows may differ (see below). Depending on the scope of the study it can make sense to build a large HE dataset that includes all relevant flow and environmental data plus multiple ecological receptors, each of which is then used for a different model.

13. What period of antecedent flow should flow statistics represent?

The optimal antecedent flow period over which to calculate statistics will vary from model to model, and should be informed by statistical approaches and/or expert judgement. Statistical approaches include sensitivity analysis and moving window analysis (for which there is a case study in this website’s article library), while expert judgement can draw on the literature and knowledge of species life cycles and phenology. The toolkit provides the flexibility to calculate flow statistics for a wide range of time periods, at various intervals and lags.

14. How many sites should an HE model incorporate?

There is no hard-and-fast rule, but all other things being equal, the more sites the better. However, ensuring that sites provide good coverage of the environmental gradients of interest (e.g. flow alteration) is more important than the absolute size of the dataset. For instance, a dataset of 25 sites exposed to varying levels of abstraction pressure is likely to be preferable to a dataset of 50 sites that are all minimally impacted or all heavily impacted. Even with a large number of sites there is no guarantee of producing a useful model that can predict ecological responses to flow or flow alteration with a high degree of confidence.

15. Is there a minimum number of samples per site required?

Again there is no hard-and-fast rule, and again all other things being equal the more samples per site the better. Typically for an invertebrate-based model a minimum of six samples straddling at least three years is preferable, but this may depend on data availability and the advantage of a pooled modelling approach is that the relationship derived at any one site can ‘borrow strength’ from the wider dataset.

16. What geographical scale can/should an HE model cover – site, waterbody, whole catchment, regional or national?

A site-specific model should only be developed if long-term (15-20+ years of continuous annual monitoring) flow and ecology data are available. Realistically, models built at a waterbody scale are unlikely to be able to draw on sufficient monitoring data or capture an adequate range of flow pressures. Whole-catchment models are likely to be preferable in this regard, although model extents do not necessarily need to be defined by watersheds. Regional models, particularly those specific to a particular pre-defined river type, offer the greatest promise, capturing a wide range of conditions (or calibration range) whilst also still allowing local knowledge to play an important part in model development. At a national scale flow-ecology relationships may be masked by very broad environmental gradients, and HE models are largely untested at this scale.

17. What environmental variables should be included in a model?

Any environmental variable that could, based on the sites in the model, confound the derived relationship between flow and ecology should be included. Commonly these include channel modification, which can be represented by RHS data, and water quality, which can be represented by historic monitoring data or outputs from water quality models. Other relevant environmental data include land use variables (e.g. % arable land cover in catchment), substrate information (e.g. % fine sediment recorded during macroinvertebrate sample collection) and suspected presence of invasive species. Where these data are not used site screening will need to be more stringent, to exclude sites where confounding pressures are prevalent.

18. For invertebrates must studies use LIFE or can other indices, community or species data be used?

It is not obligatory to use LIFE as your ecological receptor, and stronger relationships with flow and flow alteration have been derived using WHPT ASPT in some models. WHPT ASPT also has the advantage of providing a direct measure of changes in WFD classification. However, community-averaged indices, such as LIFE and WHPT ASPT, are relatively insensitive to shifts in abundance, and their responses may mask or lag behind species-level impacts. An index-based HE model may therefore not be sufficiently precautionary, at least in isolation, for certain applications, such as identifying a threshold level of abstraction for the purposes of proposing a local flow constraint. Analyses of community compositional responses can therefore be valuable, and the toolkit’s import_inv_taxa can be used for this purpose. If you require help processing invertebrate abundance data, including harmonising records at different taxonomic levels, you can contact .

19. Can the hetoolkit be used to build a hydroecological model for a temporary river?

Yes: although the case studies on this website provide examples from perennial rivers, many of the toolkit’s functions can still be used to process and analyse data for studies on temporary rivers. However, it is important to remember that temporary rivers‒ those that naturally cease to flow at one or more points along their length for at least part of the year‒ exhibit characteristically different flow-ecology relationships. On perennial watercourses flow magnitude exerts a fundamental influence on river ecology, with low and high flow events particularly significant. On temporary rivers flow duration replaces magnitude as the critical aspect of the flow regime, so flow metrics (e.g. Q95) and biomonitoring tools (e.g. LIFE scores) developed on perennial rivers will not be transferable to these systems.

These differences are important to bear in mind when undertaking a temporary river HE assessment, and will likely necessitate a bespoke modelling approach. The definition of reference conditions and good ecological status and the development of biomonitoring indices (with the exception of the Plant Flow Index, PFI) for temporary rivers are currently ongoing, and are unlikely to inform the Environment Agency’s regulatory position for several years to come. In the interim, studies should make use of experimental approaches and expert interpretation, informed by the now-substantial body of scientific literature on temporary rivers. Please contact for further advice.

20. Can the functions in the hetoolkit be expanded beyond what is currently on Github?

We anticipate that the toolkit will continue to evolve as its uptake increases. If you have any particular suggestions for functions you would like to see added please contact .

21. What model/R code evidence will the EA require to review HE models?

This will depend on the area for application: where the toolkit is used as part of a standard WINEP Water Resources Investigation a clear and concise overview of the key stages of data compilation and processing and model specification will typically suffice. For higher-risk applications such as local flow constraint work, comprehensive and suitably annotated R code should be shared, along with all relevant output including model diagnostics.

22. Who do I contact if there are issues with running any parts of the hetoolkit?

Please contact the EA’s National Hydroecology Team () and/or APEM (). If you find a bug you can also log an issue (and, if possible, a reproducible example) at https://github.com/APEM-LTD/hetoolkit/issues

23. Who do I contact for general advice and support for the hetoolkit?

Please contact the EA’s National Hydroecology Team ()

24. Can we send in our own case studies to add to the articles library – if so what format and where to send?

You are welcome to share your toolkit case studies, whether they detail the full development of an HE model or simply use of certain aspects of the toolkit’s functionality. If possible please send them in R Markdown format to .