Summary – Day 1 Martyn Kelly

Summary (day 1) Data Analyses Feasibility checks Validation steps Freshwater Marine Analyses R v Excel Regression methods Categorical methods Multivariate analyses Other methods Feasibility checks Validation steps Ranges of boundaries Freshwater Marine Best Practice Handbook Next steps

Data Ideally: lots of data spanning long gradient Need advice on what to do when this is not possible Better advice on excluding outliers (or not …) Is it better working with raw uncapped metrics? Merging types? May be impossible to obtain missing parts of gradient Work with neighbours combination of BQEs? Combining sub-elements (mac & phytobenthos) Which BQE? 1OAO suggests most sensitive

Analyses R v Excel – both needed Multivariate analyses may be needed Coastal – flushing/turbidity as factors in some cases? Multiple pressures Regression Type 1 v Type 2 Quantile – pros and cons? Categorical methods When? Minimisation of mismatch ... Attractive …

Feasibility checks Needs to be emphasised Recognise limitations of toolkit Preliminary analyses to confirm that relationships exist (boxplots etc) When is toolkit not useful? Causal relationships? Confirm limiting nutrient / exclude co-linearity Size of dataset Short gradients Extrapolation may be okay if high r2. Circularity – when P is used to set boundaries “Alternative benchmarking”

“Road map” Clearer directions for early stages When is toolkit useful? When is it not? When to use what methods? Categorical Mismatch Quantile Criteria to guide users to appropriate choice of method Choice of BQE more important than choice of statistical method Experiment with as many as possible Some parts of scripts need more explanation.

Validation steps Needs to be emphasised Underpin statistics with ecological “reality check” Need suggestions for approaches Ideally independent BQE with causal link to eutrophication Also check against literature Comparison against broad type ranges

Ranges of nutrient boundaries How useful are these? Issues with Broad Typology Issues with wide ranges in rivers But useful in many cases … Political dimensions from presenting wide ranges … Ranges need to be applicable over large areas but take account of how thresholds are applied Make IC datasets available to MS to fill out data ranges Heliana will continue to work with MS to develop relationships for TRAC

Alternative methods Need suggestions … How much detail? Can you write up your ideas, with links to detailed descriptions / examples?

Structure of handbook Much material will be moved to annexes Emphasise links with other projects, especially MARS Include words about possible effects of climate change Include “FAQs” “I’ve only got 12 samples, what should I do?” interim “help desk”

Next steps (2017) JRC facilitate collaboration between MS on transboundary water bodies Look at mismatches between biology and nutrients Co-ordinate with MSD experts Do we need tables of nutrient ranges; if not, how do we present this summary information?