Recovery of acidified waters in the UK

WP 2: Assessment of "Stock at risk"

WP2 formed a large body of research based around two tasks, one focusing on lake status (Task 2.1) and one on stream status and episodes (Task 2.2). The research conducted on stream status and episodes is summarised below, with the results of Task 2.1 being available on a separate page.

Task 2.2 - Regional assessment of stream status and risk of episodes

Determination of study catchments and calculation of Hydrological Response Units (HRUs)

The PEARLS approach determines empirical relationships between stream water concentrations and catchment characteristics in the form of landscape classes, to simulate acidification in unsurveyed river reaches using simple mixing of drainage from each landscape class. Regions selected as having the appropriate data for application of PEARLS were north-west Scotland (relatively unaffected by acid deposition), Galloway and the upper Tywi/Irfon (both with a well known history of acidification) and the Conwy Valley (substantial forestry and acidic headwaters).

Landscape classes were defined on the basis of soils, geology, land cover and land use. The Skokloster classification was found to be too broad for this purpose. New sampling subcatchments were selected in each of the key landscape classes within each region, except in the Twyi/Irfon where 20 existing study sites based on moorland or forest catchments were used. The final number of new study subcatchments was 52 in north-west Scotland, 59 in Galloway and 24 in the Conwy. Sampling campaigns were designed to obtain high- and low-flow water chemistry. During 2001-2002, the Twyi/Irfon subcatchments were sampled twice, Conwy three times and the two Scottish regions four times. Water chemistry was characterized for each landscape class in each region for application of PEARLS. Mixing in the river network is then determined using a topographical network of hydrological response units (HRUs) and river reaches derived from a digital elevation model. PEARLS was then used to determine the probability distribution of ANC falling below selected thresholds for each reach of the river networks, and hence the total river length within or below ANC classes, in the Conwy, the Cree (Galloway) and the Tywi/Irfon.

A key application of the PEARLS approach is then to link it to the dynamic model MAGIC for the prediction of past or future changes in water chemistry across the river network (WP 3, Task 3.2). Invertebrate samples collected at high and low flows were used to explore relationships with measures of episodic water chemistry below.

Characterisation of episodic, circumneutral and acid streams and effects on invertebrate assemblages

The northwest Highlands and Galloway in the southwest of Scotland, Conwy in north Wales and the Llyn Brianne (Twyi/Irfon) catchment in mid Wales are geologically sensitive to acidification but occupy regions with contrasting deposition. Despite the extensive work on acidification in these regions, data on the effects on invertebrates are scarce. In this section, the potential effects of episodic acidification in these regions was examined. We hypothesised from existing data that, if episodic effects are important, invertebrates in streams should reflect high-flow (episodic) chemistry more than base-flow chemistry. To test this hypothesis, benthic macroinvertebrate assemblages, base-flow and episode chemistry were determined in 22-25 streams in each target region. Water chemistry for all streams was determined for low- and high-flow between 2001 and 2002 and invertebrates collected by kick sampling during April 2002.

The streams were grouped by cluster analysis and characterised as acid, episodic or circumneutral. One-way ANOVA (P<0.05) and Tukey's pairwise comparisons (P=0.05) illustrated corresponding variation between the clusters and high-flow chemistry. Variables varying most between episodic, circumneutral and acid groups were pH, ANC, Al, Ca and DOC at high-flow. Variation in invertebrate assemblages between cluster groups were assessed using one-way ANOVA. There was significant (P<0.05) variation between groups in the abundances of Plecoptera and Ephemeroptera in both Wales and Scotland, and in the abundance of Trichoptera in Wales. In canonical correspondence analysis (CCA), pH, ANC, DOC and Al during episodes all explained significant variation in species composition across the Welsh sites, but at base-flow only Al was significant. Al during high-flow was significant in explaining species variation at the Scottish sites, while only DOC was significant during base-flow. Species varying with episode chemistry in both regions were the ephemeropterans Baetis rhodani and Heptagenia lateralis; the plecopterans Isoperla grammatica, Chloroperla torrentium and various species of Nemouridae; and the trichopteran Plectrocnemia conspersa.

These data illustrate how acid-base status varies between sites at high-flow, explaining more variation in invertebrate assemblages than low-flow chemistry. However, effects vary across regions with episodic effects in Wales apparently stronger than in Scotland.

Effects of episodes on invertebrate indicator species

The response of benthic invertebrates to stream recovery from acidification is slow. Acid episodes have been implicated as affecting this response and are being investigated as a factor that might offset recovery. In this section transplantation and intensive quantitative sampling were used to assess episodic effects on an acid-sensitive indicator species in the Llyn Brianne experimental catchment. We hypothesised that strong seasonal variation in invertebrate abundance in the episodic streams would indicate possible episodic effects, while experimental transplantations were used to mimic short-term acid episodes directly. A set of six streams were sampled over a 20 month period for benthic invertebrates, pH, and conductivity, while temperature was also recorded. In addition base-flow and high-flow chemistry were evaluated for the sites.

One-way ANOVA (P<0.05) illustrated clear variation in pH between streams classified as episodic, acidic and circumneutral. Also, differences in ionic concentrations were observed over different flows and between stream types (episodic, acidic and circumneutral). Acid-sensitive mayfly species were found in the circumneutral streams (Ephemerella ignita, Baetis rhodani, Baetis muticus, Baetis vernus, Rithrogena semicolorata, Ecdyonurus spp. and Heptagenia lateralis), but none occurred in the acidic streams. Only B. rhodani occurred in the episodic streams, where densities declined substantially in autumn. Transplantation experiments were carried out with B. rhodani during base-flow (September 2003) and high-flow (April 2004) between the acidic and circumneutral streams. T-tests determined significant (P<0.001) drops in pH, conductivity and temperature during high-flow events, with the exception of conductivity in the acidic streams. One-way ANOVA highlighted significant (P<0.001) variation in pH and conductivity between the episodic, acidic and circumneutral streams. Baetis survival in the control cages during both high- and low-flow was high. T-tests found mortality of B. rhodani in the chronic and episodic exposures increased significantly (P<0.001) during the high-flow experiment. Mortality was also much higher in the chronic exposure compared to the episodic during high-flow. Two-way ANOVA showed that survival varied significantly (P<0.05) between streams at high-flow, with mortality in episodic exposure intermediate between chronic acidity and controls.

These data confirm that episodic exposure to low pH at high-flow can be detrimental to mayfly survival. Density data, by contrast, were equivocal. Seasonal variations in B. rhodani were similar across circumneutral and episodic streams, but only this mayfly species occurred in the latter indicating possible episodic effects on mayfly assemblage composition. Further work is still required on the field effect of acid episodes on species other than B. rhodani.

Development of biological episode response model

Although much evidence indicates that many organisms in acid-sensitive streams are affected by episodic acidification, models that predict the response of stream biota to episodes have proved elusive. This is an important gap that has frustrated attempts to link realistic biological responses to episodic stream chemistry with important hydrochemical models such as PEARLS or MAGIC. Using data from 89 streams in Wales and Scotland, here we evaluate whether high-flow chemistry better predicts invertebrate community composition than base-flow chemistry. We reduced invertebrate composition to principal components and then used stepwise multiple regression to model scores on the first (i.e. major) axis of variation using combinations of acid-base determinands derived from either base-flow or high-flow. In all cases, regressions were evaluated from i) the fit of the modelled relationships to empirical data and ii) their performance in predicting invertebrate PC scores at independent sites (n = 22) reserved from the initial calibration set. Invertebrate principal components reflected well known trends in response to acid-base status, illustrating a shift from acid-tolerant to acid-sensitive species on the first axis.

Regression models using combinations of pH, aluminium, DOC and charge-balance alkalinity always explained 40-60% of the variance in invertebrate score. While determinands measured at high flow on average explained more variance (49-59%) than at low flow (42-57%), differences were modest and in no case statistically significant. All models gave invertebrate scores for the 22 test sites that were highly significantly related to observed scores (r = 0.71-0.88), although there was a moderate tendency in all cases to underestimate acidification effects at the most acid sites. Comparison across determinands illustrated that pH at high flow either alone (r2 = 0.58) or in combination with aluminium (r2 = 0.59) gave the best overall fit to both calibration and test data. Charge-balance ANC performed worst, particularly at low flow (r2 = 0.42).

Overall, these data indicate that, for the purposes of modelling invertebrate assemblages in acid-sensitive streams, combinations of pH and aluminium at high-flow offer the most accurate outcome. However, base-flow chemistry can offers a valuable approximation to biological effects at more extreme flows, although losses in accuracy will be greatest using base-flow ANC as a sole predictor.

Modelling the chemical signature of episodes and risk to biology from catchment character

Evidence is increasing that acid episodes might offset the recovery of acid-sensitive invertebrates in previously acidified streams, but factors influencing the character of episodes and their risks to biota are still poorly quantified. In this section, we assess variations in the chemical signature of episodic acidification between contrasting stream types (n = 89 streams) in Wales and Scotland that were respectively strongly acidified (pH < 5.0), intermediate (pH 5-5.7) and circumneutral (pH > 5.7) at high flow. All the sites were described elsewhere in the report, and respectively came from the Conwy/Llyn Brianne area in Wales, and the Galloway/NW regions of Scotland. Using established methods, we compared variations between these groups in i) the respective percentage contribution by dilution and strong acid anions to episodic acidification at high flow and ii) the percentage contributions of acid anions to total anion concentration during high flow events. We also assessed variations among acid-sensitive invertebrates between these groups.

ANC in some streams increased at high flow over low flow for the events measured (3/48 in Scotland, 17/45 in Wales) - presumably because of increased base cation release. In Wales, contributions to episodic acidification by base cation dilution differed highly significantly (P<0.001) between circumneutral streams (mean = 42%), intermediate (0%) and strongly acidified streams (0%), where strong acid addition was the dominant source of acidification at high flow. By contrast, in Scotland, episodic acidification reflected greater contributions by dilution in all stream types (Circumneutral = 58%; Intermediate = 26% and strongly acidified 44%; N.S. at P < 0.1). As a result, strong acid additions during high flow events were a much greater source of episodic acidification in Wales than in Scotland, even for strongly acidified and intermediate sites (P<0.001). Contributions to anion loading at high flow due to sulphate were significantly greater in Wales (16-24%) than at the Scottish sites (3-5%). Similarly, in Wales (4-7% on average), anion contributions due to nitrate were significantly greater than in Scotland (2-4%). Chloride made up the bulk of the remaining anion loading at high flow, particularly in Scotland.

The apparent consequences of these chemical effects for invertebrates differed between species. Acid tolerant organisms such as Chloroperla torrentium (Plecoptera) and Plectrocnemia conspersa (Trichoptera) were at least as abundant in streams that were intermediate or strongly acidified at high flow as they were at circumneutral sites. By contrast, acid sensitive species such as Baetis rhodani (Ephemeroptera), Heptagenia lateralis (Ephemeroptera; Wales only) and Isoperla grammatica (Plecoptera; Scotland only) were significantly reduced in abundance or absent at strongly acidified and intermediate sites compared with those that remained circumneutral at high flow (P = 0.05-0.001).

Together, these data indicate that acid anions - and still dominantly sulphate over nitrate - drive episodic acidification more in acid-sensitive areas of Wales than in Galloway and NW Scotland. Nevertheless, strong acid additions still contribute significantly to acid episodes at some episodically acidified sites in all regions, which in turn have markedly different invertebrates than those where pH remains over pH > 5.7 at high flow.