Heavy metal deposition mapping: concentrations and deposition of heavy metals in rural areas of the UK:SID4 interim report covering the period October 2011 �C December 2011. Interim report to the Department of Environment, Food and Rural Affairs by the

Data and Information;Atmospheric Sciences

Malcolm, Heath ; Fowler, David ; Kentisbeer, John ; Hallsworth, Steve ; Lawlor, Alan ; Rowland, Phil ; Guyatt, Hayley ; Thacker, Sarah ; Halford, Alan ; Rogers, Stuart ; Cape, J. Neil ; Sleep, Darren ; Woods, Clive

NERC/Centre for Ecology & Hydrology

学科分类：大气科学

来源: NERC Open Research Archive

【 摘 要 】

The DRIEDUP With Incremental Drought (DRUWID) methodology was used to analyse the existing Environment Agency (EA) National Drought Surveillance Network (NDSN) data. DRUWID models riverine macro-invertebrate community response to flows and physical habitat. Macro-invertebrate communities are represented as an aggregated family-level index (Lotic Index for Flow Evaluation; LIFE) based on historical EA monitoring data. LIFE scores are split into two seasons based on when sampling occurred (spring and autumn). Flow variables are based on gauged daily flows from the National River Flow Archive (NRFA). First, seasonal flow statistics (Q10 and Q95; i.e. high and low flow percentiles, respectively) for standard six-month periods (summer (April-September) and winter (October-March)) were matched with the autumn and spring LIFE scores, respectively (��in-year flow statistics��). Second, LIFE scores were matched with the summer low flow statistics one, two, and three water years before the sampling year (regardless of sampling season). Physical habitat conditions are described using River Habitat Survey (RHS) data (extent of livestock poaching and extent of bank and bed resectioning). The existing NDSN consists of 89 biomonitoring sites. However, the analysis was only undertaken on 61 of the 89 sites as 28 sites had missing flow or physical habitat data. Fifteen of the 61 sites have been classified as Upland sites. Forty six of the 61 sites have been classified as Lowland sites based on existing methodologies. Four models (Spring Upland, Spring Lowland, Autumn Upland, Autumn Lowland) were fitted within a multilevel linear regression modelling framework to provide an integrated analysis of all NDSN sites in each model. Model selection and prediction was achieved via Multi-Model Inference (MMI). Interaction terms (between all flow variables and between low flow and physical habitat variables) were included in all models. Both in-year flow statistics are important for Lowland models but only the in-year low flow is important for Upland models. Spring models are less responsive to low flows than Autumn models. Summer low flows from one and two years before the sampling are important for all models. Spring models have a stronger response than Autumn models. Summer flow three years before sampling has limited influence on Spring Lowland and Autumn Upland models. Both physical habitat variables have some influence with resectioning being the most significant. All interactions terms (i.e. between flow variables and between flow and habitat variables) are present in some models. They have more overall influence on Lowland models than on Upland models, and on Spring models than on Autumn models. The four models highlight that macro-invertebrates in different river types and in different seasons respond differently to flows and hence, potentially to drought. Recovery can take at least two years. Seasonal droughts lead to relatively rapid response once flows return but a 2 to 3 year drought leads to a delayed response. The DRUWID approach is for the LIFE metric only while the drought network does not represent all waterbody types. DRUWID could be run for specific historic, projected or simulated flow regimes. This would allow for the investigation of the impact of climate change, or for the exploration of thresholds beyond which ecosystems are permanently altered. Further work could involve: refining the Upland / Lowland typology, separating out the impacts of natural flow variations from abstraction / compensation and using alternative macro-invertebrate datasets, ecological groups, or other explanatory variables.