Acidic episodes retard the biological recovery of ... - Wiley Online Library

Global Change Biology (2007) 13, 2439–2452, doi: 10.1111/j.1365-2486.2007.01437.x

Acidic episodes retard the biological recovery of upland British streams from chronic acidification R E N A T A A . K O W A L I K *, D . M . C O O P E R w , C . D . E VA N S z and S . J . O R M E R O D * *Catchment Research Group, Cardiff School of Biosciences, Cardiff University, Cardiff CF10 3US, UK,

wCentre for Ecology and Hydrology, Crowmarsh Gifford, Wallingford, UK, zCentre for Ecology and Hydrology, Bangor, UK

Abstract We tested two predictions required to support the hypothesis that anthropogenic acidic episodes might explain the poor biological response of upland British streams otherwise recovering from acidification: (i) that invertebrate assemblages should differ between episodic and well-buffered streams and (ii) these effects should differentiate between sites with episodes caused by anthropogenic acidification as opposed to base-cation dilution or sea-salt deposition. Chronic and episodically acidic streams were widespread, and episodes reflected acid titration more than dilution. Nonmarine sulphate (16–18% vs. 5–9%), and nitrate (4–6% vs. 1–2%) contributed more to anion loading during episodes in Wales than Scotland, and Welsh streams also had a larger proportion of total stream sulphate from nonmarine sources (64–66% vs. 35–46%). Sea-salts were rarely a major cause of episodic ANC or pH reduction during the events sampled. By contrast, streams with episodes driven by strong anthropogenic acids had lower pH (5.0 0.6) and more dissolved aluminium (288 271 lg L1) during events than where episodes were caused by dilution (pH 5.4 0.6; 116 110 lg Al L1) or where streams remained circumneutral (pH 6.7 1.0; 50 45 lg Al L1). Both biological predictions were supported: invertebrate assemblages differed among sites with different episode chemistry while several acidsensitive species were absent only where episodes reflected anthropogenic acidification. We conclude that strong acid anions – dominantly nonmarine sulphate – still cause significant episodic acidification in acid-sensitive areas of Britain and may be a sufficient explanation for slow biological recovery in many locations. Keywords: acid deposition, biodiversity, episodic acidification, invertebrates, reversibility

Received 12 July 2005; revised version received 8 December 2006 and accepted 22 May 2007

Introduction The acidification of sensitive freshwaters by acidic precipitation is among the most intensively researched of all global change effects. At its peak during the 1970s and 1980s, large areas of Europe and North America were affected causing widespread changes in freshwater hydrochemistry, species’ extirpation, altered ecological function and attendant economic damage (Muniz, 1990). Although chemical data now show unequivocally that reversal is underway (Evans et al., 2001; Fowler et al., 2001; Harriman et al., 2001; Fo¨lster & Wilander, 2002; Davies et al., 2005), evidence of any accompanying biological recovery is still patchy (Tipping et al., 2002; Monteith et al., 2005). This misCorrespondence: Prof. S. J. Ormerod, e-mail: [email protected]

r 2007 The Authors Journal compilation r 2007 Blackwell Publishing Ltd

match is particularly marked in recovering streams and rivers – where acid-sensitive species often occur only sporadically or represent a fraction of taxa that were previously lost (Soulsby et al., 1997; Bradley & Ormerod, 2002a). One of the major needs in acidification research is to explain the apparently nonconvergent trends between chemical and biological recovery, and candidate hypotheses are now emerging (Yan et al., 2003; Ledger & Hildrew, 2005). One increasingly favoured is that acidsensitive species are prevented from recolonizing recovering streams because of the continued effects of acidic episodes – transient periods of reduced pH generated during rainstorms or snow melt over hours to weeks (Davies et al., 1992; Bradley & Ormerod, 2002a). By markedly increasing acidity and metal concentrations, episodes were considered key drivers of the initial biological effects of acidification, particularly in 2439

2440 R . A . K O W A L I K et al. running waters. Episodes might still affect sensitive organisms despite increasing mean base-flow pH if conditions became toxic at high flow – for example, because strong acid anions led to the release of toxic metals such as aluminium. Antecedent deposition conditions, residual catchment pools of mobilisable anions and depleted base-cations in soils or runoff could all engender such effects (Stoddard et al., 1999; Alewell et al., 2001; Campbell & Eagar, 2002; Lawrence, 2002). The deposition of sea-salts – sodium and chloride from natural marine sources – can also contribute to episodic acidification by displacing H 1 ions or aluminium from acidified soils (Harriman & Wells, 1985). Moreover, soils remain sensitive to such effects due to the persistence of base-cation depletion (Evans, 2005). Any biological effects of sea-salt episodes would be particularly interesting because they would implicate natural deposition processes in prolonging acidification effects despite a decline in anthropogenic sulphur deposition. A major recent advance in acidification research has been the development of methods to quantify contributions to acidic episodes from strong anthropogenic acidity as opposed to natural organic acidity or basecation dilution (Kahl et al., 1992; Wigington et al., 1992, 1996; Laudon & Bishop, 1999; Bishop et al., 2000; Lepori et al., 2003a). These methods have improved the understanding of chemical recovery on both sides of the Atlantic (Laudon & Bishop, 2002; Laudon et al., 2002). They might also aid the interpretation of episodic effects on organisms (Lepori et al., 2003b), particularly in assessing whether anthropogenic contributions to acidic runoff are still sufficient to damage biota. Data purporting to show continued episodic effects on organisms are few, mostly descriptive, and provide no information about the causes, driving anions or wider ecological effects (Raddum et al., 2001; Bradley & Ormerod, 2002a; Kowalik & Ormerod, 2006). In this paper, we quantify the contributions to episodic acidification from base-cation dilution, sea-salts, and strong acid-addition in a large sample of streams across acid-sensitive areas of Scotland and Wales. We also appraise consequences for aquatic invertebrates. The work was designed to test the hypothesis that anthropogenic acidic episodes can explain the absence of acid-sensitive species from recovering streams and we made two predictions about results that would provide support. Firstly, continued episodic effects on assemblages should be accompanied by detectable differences in invertebrate assemblage composition between episodic and well-buffered streams (prediction 1). Second, any such effects should differentiate sites where episodes are driven by anthropogenic acidification as opposed to base-cation dilution or sea-salt addition (prediction 2).

Our focus on aquatic invertebrates reflects their importance in previous acidification research, their species richness and range of acid sensitivity (Wade et al., 1989), their life cycles and potential exposure to episodes in winter/spring (Lepori et al., 2003a; Kowalik & Ormerod, 2006), their capacity for dispersal into recovering sites (Petersen et al., 2004) and their recognized value as indicator organisms (Rutt et al., 1990).

Methods

Sites and study areas Our emphasis was on extensive comparisons across multiple sites of contrasting sensitivity to acidification and regions of contrasting acidic deposition. The data were collected from second- and third-order streams in the northwest Scottish Highlands (n 5 24), Galloway in south-west Scotland (n 5 25), the Conwy catchment, north Wales (n 5 19), and locations in central Wales centred on the Llyn Brianne experimental catchments (n 5 21; Fig. 1, Appendix A). Sites covered a range of landscapes, including those expected to be acidified, and others that were less sensitive to acidification due to geology or land use. Exact site selection reflected a combination of logistics, vehicle access to low order streams, and the availability of previous research information on acidification. All the target regions have been affected previously by acidification except the relatively more pristine Scottish Highlands. However, S and N deposition, and rainfall acidity, have declined because emissions

Scotland NW Highlands

Galloway

Wales Conwy Llyn Brianne

Fig. 1 The survey regions in Scotland (NW Highlands and Galloway) and Wales (Conwy and Llyn Brianne).

r 2007 The Authors Journal compilation r 2007 Blackwell Publishing Ltd, Global Change Biology, 13, 2439–2452

A C I D E P I S O D E S R E TA R D B I O L O G I C A L R E C O V E R Y peaked in the late 1970s (Fowler et al., 2001). From estimates at adjacent representative locations (Conwy 5 41 sites, mid-Wales 5 26, Galloway 5 55, and the NW Highlands 5 98) in mostly standing waters (http:// critloads.ceh.ac.uk; Hall et al., 2004), critical loads for ANC (o20 mEq L1) are now exceeded in very few catchments in the NW Highlands (6%), and less than half of those in either Wales (30%) or Galloway (42%). Recovery conditions across large parts of the study area are thus finely balanced, so that any episodic effects could be important.

Field sampling Water samples were collected from each site during storm events (Galloway, October 2001; northwest Highlands, Conwy and mid-Wales, January/February 2002) and at base-flow (northwest Highlands and Galloway, April 2002; Conwy and mid-Wales August 2002). This approach has been effective at characterizing event chemistry in biological studies (Lepori et al., 2003a) and also avoids the difficulties of accurately determining mean or extreme chemistry at acid-sensitive sites from random sampling (Brewin et al., 1996). Samples were analysed for Na1 , K 1 , Ca2 1 , Mg2 1 , NH41 , Cl, 2 n1 , and dissolved organic carbon (DOC). NO 3 , SO4 , Al Alkalinity was determined by Gran titration, anions by ion chromatography (Dionex, Camberley, UK), and base cations and metals by Atomic Absorption Spectrophotometry (AAS) (Perkin Elmer, Waltham, MA, USA) after filtration on-site at 0.45 mm into nitric acid fixative. DOC was determined by continuous flow colorimetry with UV digestion (Skalar autoanalyser System) and Al colorimetrically using pyrocatechol violet (Dougan & Wilson, 1974). pH was recorded on-site using a handheld combined pH/EC/TDS meter with an electrode specially designed for low-ionic strengths (Hanna Instruments HI 991300, Leighton Buzzard, UK). Charge-balance acid neutralizing capacity (ANC) was calculated as the sum of base-cations minus the sum of acid anions (Na1 1 K 1 1 Ca2 1 1 Mg2 1 NO 3 Cl ). We determined nonmarine sulphate (i.e. SO2 4 excess SO4 5 xSO4) by subtracting marine SO4 (mSO4) from total SO4 based on the ratio of SO4 to Cl in seawater (when computed in equivalents 5 0.104). Invertebrates were collected during April 2002 from all sites using semiquantitative kick samples (standard net with 1 mm mesh size) of 2 min duration in riffles and 1 min in margins. This strategy samples most major habitats and is a well-calibrated method sufficient to detect differences between sites (Weatherley & Ormerod, 1987; Bradley & Ormerod, 2002b). Samples were preserved on-site by adding 100% industrial methylated spirit (IMS) to the sample volume. In the laboratory,

2441

samples were hand sorted, preserved in 70% IMS and major groups were identified to species or genus for most insects (Plecoptera, Ephemeroptera, Trichoptera, and Coleoptera) or to family in cases where taxonomy was difficult or larvae were insufficiently well developed (e.g. Diptera).

Chemical data analysis Initially, we categorized streams into those with baseflow pHo5.7 ( 5 acidic), pH 5.7–6.99 (episodic), and pH7 (circumneutral). Upland British streams with base-flow pH7 usually have minimum pHo5.7, recognized as a threshold at which many acid-sensitive taxa become scarce (Sutcliffe & Carrick, 1973; Weatherley & Ormerod, 1991). For each of these stream categories, we calculated mean values for major acid– base determinands at high and low flow. We also wished to assess variations among regions or stream groups in the calculated loss of alkalinity due to base-cation dilution, and in the percentage contribution of major acid anions to any titration effects. For this we used Kruskall–Wallis tests. These analyses were carried out using data derived from the ionic concentrations in mEq L1 of alkalinity (Alk), the cations NH41 , Na1 , K 1 , 2 Ca2 1 , Mg2 1 , Aln 1 , and the anions Cl, NO 3 , SO4 2 (including both marine and excess SO4 ), and organic acids OA. For calculations, aluminium was given a charge of 2 1 (Sullivan et al., 1989) while OA were estimated using a value of 5 mEq mg1 of DOC, the approximate amount expected to act as strong organic acids (Munson & Gherini, 1993; Bishop, 1996). We determined the overall percentage loss of alkalinity attributable to dilution for the measured events at each site relative to base-flow after Kahl et al. (1992): Alk dilution ¼

ðððSBClow SBChigh Þ=SBClow ÞAlklow Þ ; ðAlklow Alkhigh Þ ð1Þ

[Eqn (1); BC, base cations]. Based on their common origin in weathering, this procedure assumes that total base cations should dilute in proportion to alkalinity (mostly bicarbonate) during storm events. While small discrepancies are possible due to exchange between base cations and protons in surface soils during events (Bishop et al., 2000), values for dilution substantially o100% indicate the addition of strong acids from deposition or catchment processes, and hence titration effects. The ratio between alkalinity and the sum of BC (alkalinity/SBC), set with a lower limit of zero, was used as a measure of the relative importance of titration (Kahl et al., 1992). The relative contributions of acid anions (AA: NO 3, 2 mSO2 4 , xSO4 and OA) to any titration effects during


2442 R . A . K O W A L I K et al. storm-flow were estimated from their proportionate contribution to anion loading at each site (anion/ SAA). Because the sum of anions should equal the sum of cations (with minor errors due to missing contributions from minor ionic species or mismeasurement), these relative contributions are usually calculated as anion/SBC because cations can be determined analytically with more certainty (Kahl et al., 1992; Lepori et al., 2003a). Once the chemistry of episodic acidification was known at each site, we plotted the proportionate contribution to anion loadings from nonmarine sulphate (xSO4/SAA) against the percentage of alkalinity loss at high flow due to dilution (from Eqn (1)). This identified groups of sites where episodic acidification was driven mostly by elevated nonmarine sulphate with minor base-cation dilution (Group 1 sites) or mostly by dilution (Group 3 sites). Intermediate sites (Group 2) had episodic acidification driven by small amounts of basecation dilution with elevated concentrations of anions other than sulphate (mostly organic acids). Sites in Group 4 were circumneutral and were not subject to major episodic acidification. We compared acid–base determinands between these groups using oneway analysis of variance (ANOVA) after appropriate transformations to homogenize variances. These same groups were also used in comparisons of invertebrate assemblages across sites. As much of the work was carried out in relatively maritime location of western Britain, we expected Cl to contribute substantially to anion loading at each site and to be balanced mostly by Na1 from marine sources. However, additional loss of ANC might occur where Cl moved more rapidly through catchment soils than Na1 , thus mobilizing acid cations (including H 1 and Aln 1 ) through ion exchange (Harriman & Wells, 1985). The reverse effect can occur when stored Na1 is released into runoff to balance anions, thereby contributing to ANC. We appraised these possible ‘sea-salt’ effects on episodic acidification from the relative changes in ionic concentrations of Cl and Na1 between low and high flow at each site. Locations where Cl increased more substantially than Na1 indicated potential acidifying effects. We determined the proportion of ANC loss that could be explained by Cl in molar equivalents for each site, and on average across sites and among acid, episodic, and circumneutral sites within regions. We also compared contributions to ANC loss among Groups 1 to 4 identified as above.

both marginal and riffle samples. Assemblages were first ordinated in CANOCO using detrended correspondence anlysis (DCA), among the most established of all ordination methods (ter Braak & Smilauer, 1998). DCA produces an objective ordination of samples along orthogonal axes that reflect similarity in taxonomic composition without being artificially constrained to fit any environmental variable. Differences between sites in scores on each axis, thus, reflect differences solely in assemblage composition. To test predictions 1 and 2 (see ‘Introduction’), we assessed variations in DCA scores between streams characterized by contrasting episode chemistry (Groups 1–4) using one-way ANOVA with Tukey–Kramer range tests. We evaluated trends on the first two DCA axes with major acid–base determinands, including alkalinity loss due to dilution, contributions to ANC loss attributable to Cl, and xSO4/SAA, using Pearson’s correlation coefficients. Variations in abundance were appraised for a range of common species of known sensitivity to acidity (Weatherley & Ormerod, 1987; Wade et al., 1989; Rutt et al., 1990). These were:

Invertebrate data analysis

Causes and chemistry of episodes

We analysed data from the whole macroinvertebrate assemblages and from individual taxa summed from

Streams categorized initially as acidic or episodic comprised over 50% of the Scottish sample and almost 90%

(i) acid-tolerant species from the Plecoptera [Chloroperla torrentium (Pictet), Leuctra hippopus (Kempny), L. inermis (Kempny), L. nigra (Olivier), Nemurella picteti (Klapa´lek) and Amphinemura sulcicollis (Stephens)] and the Trichoptera [Plectrocnemia conspersa (Curtis) and Rhyacophila dorsalis (Curtis)]; (ii) acid-sensitive taxa from the Ephemeroptera [Baetis muticus (L.), B. rhodani (Pictet), Ecdyonurus spp. Eaton, Rhithrogena semicolorata (Curtis) and Heptagenia lateralis (Curtis), the Plecoptera (Isoperla grammatica (Poda)), and four other acid-sensitive species (the trichopteran Hydropsyche siltalai Do¨hler, the coleopterans Hydraena gracilis Germar and Elmis aenea (Mu¨ller), and the crustacean Gammarus pulex (L.)]. We assessed variations among stream groups (i.e. Groups 1–4) with contrasting episode chemistry using ANOVA as above. Abundance data were first log10(x 1 1) transformed to homogenize variances. All invertebrate analyses were made on data pooled across regions because parallel investigations discounted the possibility that regional and acid–base effects on invertebrates were confounded (R. A. Kowalik & S. J. Ormerod, unpublished data).

Results


A C I D E P I S O D E S R E TA R D B I O L O G I C A L R E C O V E R Y

2443

Table 1 Mean (SD) pH, alkalinity (mEq L1), Aln 1 4 (mg L1) and DOC (mg L1) at base-flow and storm-flow of streams in Scotland and Wales grouped by base-flow pH (acidic o5.7; episodic 45.7–o7.0 and circumneutral 7.0) Wales

Scotland

Determinand

Stream-type

Base-flow

Storm-flow

Base-flow

Storm-flow

pH

Acidic Episodic Circumneutral Acidic Episodic Circumneutral Acidic Episodic Circumneutral Acidic Episodic Circumneutral

5.3 6.3 7.4 27 109 902 50 9 7 7.4 5.3 3.6

4.7 5.1 6.7 22 4 308 247 122 50 4.2 3.3 5.0

5.4 6.7 7.5 11 153 908 193 57 28 7.4 5.9 4.9

4.5 5.2 6.7 25 3 428 293 109 50 9.1 7.0 6.1

Alkalinity

Aln 1

DOC

(0.3) (0.2) (0.3) (27) (89) (444) (51) (3) (3) (6.5) (3.8) (2.4)

(0.3) (0.4) (0.2) (15) (39) (250) (195) (114) (37) (2.9) (2.4) (3.7)

(0.3) (0.2) (0.4) (16) (101) (850) (171) (52) (26) (5.9) (4.2) (4.7)

(0.4) (0.6) (1.1) (24) (36) (634) (340) (102) (47) (5.6) (4.4) (5.4)

Sample sizes in each group are given in Table 3.

of the Welsh sample (but note that site selection was not random). In general, mean alkalinity and mean pH were similar within these groups across regions during both base-flow and storm-flow, when highly acidic conditions were widespread (Table 1; Fig. 2). A small number of Scottish circumneutral streams with baseflow pH47 also had pHo5.7 during storm-flow (n 5 5). The percentage loss of alkalinity during episodes attributable to base-cation dilution was considerably less on average in Wales (median 7%) than Scotland (median 43%; Kruskall–Wallis test, H1, 88 5 17.21, Po0.001). Base-cation dilution increased significantly in the order acidic streamsoepisodicocircumneutral in both Wales (medians: acidic 5 2%; episodic 5 8%, and circumneutral 5 70%; Kruskall–Wallis test, H2, 38 5 11.79, Po0.001) and Scotland (medians: acidic 5 0.6%; episodic 5 31%, and circumneutral 5 63%; Kruskall–Wallis test, H2, 47 5 17.81, Po0.001). Thus, episodic acidification in most episodic and acidic streams in all regions was dominated by titration. Most of the anion load was due to Cl during stormflow in both regions (Wales 72% vs. Scotland 78%), but was overwhelmingly balanced by Na1 . On average, only 11% and 18% of ANC loss could be ascribed to Cl, respectively, in Scotland and Wales, but in neither case were there significant variations among acidic, episodic, and circumneutral sites (Ho1.2, NS; Table 2). Local effects from sea-salts were apparent, but Cl accounted for 450% of ANC loss at only two of 25 Scottish and six of 31 Welsh locations where pH fell below 5.7 at high flow. Otherwise, acid anions differed considerably between regions with sulphate (H2, 87 5 64.8, Po0.001), excess sulphate (H2, 87 5 Po0.001) and nitrate (H2, 87 5 12.6,

Po0.001) all contributing more to events in Wales than Scotland irrespective of stream types (Table 3; Fig. 3). Probably as a consequence, Aln 1 concentrations in Wales increased between base-flow and storm-flow by 5–10 compared with a 2 increase in Scotland (Table 1). The reverse trend was true for DOC, which increased during storm-flow in Scotland to concentrations around twice those in Wales (F2, 88 5 10.96, Po0.001). Organic acids were at least as important as excess sulphate in the acid anion load of acidic and episodic Scottish streams during storm-flow and varied significantly from circumneutral streams (Table 3; H2, 47 5 7.8, Po0.05). Total sulphate varied significantly in contribution to anion loading among acid, episodic, and circumneutral streams at high flow in both Wales (H2, 38 5 12.8, Po0.01) and Scotland (H2, 47 5 7.2, Po0.05). Contributions from excess sulphate did not vary formally among these stream types, but there was other evidence of a role in episodic acidification: whereas the contribution of excess sulphate to anion loading fell due to dilution between base-flow and storm-flow in circumneutral streams on average by 10% (Scotland) to 15% (Wales), contributions to anion load at storm-flow in episodic and acidic streams were maintained or increased (Fig. 4). The plot of xSO4/SAA against the percentage of alkalinity lost through dilution effectively separated sites at which episodes were driven by nonmarine sulphate (xSO24 1 5 Group 1) from those where episodes reflected other anions (e.g. mostly organic acids 5 Group 2) and those where episodic acidification reflected mostly dilution (Group 3; Fig. 5). Acid–base chemistry varied progressively and significantly among


2444 R . A . K O W A L I K et al.

(a)

Wales e

a

800

800 ANC (µEq / L)

(d) 1000

ANC (µEq / L)

(c) 1000

600 400 200

200 0 −200

800

800 Al (µg/L)

(f) 1000

Al (µg/L)

(e) 1000

600 400

0

0

(g)

(h)

30

30

25

25 DOC (mg /L)

200

DOC (mg /L)

200

20 15 10 5

c

400

−200

400

Scotland e

600

0

600

a

8.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0

pH

8.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0

pH

(b)

c

20 15 10 5

0

> 7.0

5.7