- Top of page
The influences of medium-range climatic cycles on marine and continental ecosystems are increasingly well described (Dayton et al. 1992; Harrison 2000; Holmgren et al. 2001). By contrast with the El Niño Southern Oscillation (ENSO), however, there is less evidence for ecological patterns that reflect its Atlantic analogue – the North Atlantic Oscillation (NAO). In terrestrial ecosystems, the NAO has been linked to the phenology of birds or plants (Forchhammer, Post & Stenseth 1998; Post & Stenseth 1999; Przybylo, Sheldon & Merila 2000), the survival and demography of large mammals (Milner, Elston & Albon 1999) and the effects of northern snow cover on predator–prey interactions (Post et al. 1999a, 1999b). In marine systems, planktonic communities reflect year-to-year patterns in temperature, nutrient upwelling or fluctuating salinity driven by the NAO (Fromentin & Planque 1996; Lindahl et al. 1998; Nehring 1998; Planque & Taylor 1998; Belgrano, Lindahl & Hernroth 1999; Hagberg & Tunberg 2000; Hanninen, Vuorinen & Hjelt 2000). In lakes, the NAO affects plankton through temperature, wind-induced mixing, ice cover and altered food web dynamics (George & Harris 1985; Straile & Geller 1998; George & Hewitt 1999; George 2000; Straile 2000). Parallel and synchronous changes across large areas show that such effects are probably widespread (Straile & Adrian 2000). So far, however, and in contrast to lakes, few studies have assessed whether the NAO affects rivers and river organisms (Elliott, Hurley & Maberly 2000; Monteith, Evans & Reynolds 2000).
As the alternation of differences in atmospheric sea-level pressure (SLP) between the Azores and Stykkisholmur (Iceland), the NAO drives complex weather patterns over a cycle that is now roughly decadal and of increased intensity since the mid-1970s (Hurrell 1995; Hurrell & van Loon 1997). Hurrell’s (1995) NAO index describes these pressure fluctuations though time, negative values being accompanied by cold, dry and calm winters in NW Europe, whereas positive values are accompanied by milder winters, strong westerly winds and rainfall up to 30% above annual average (Hurrell 1995). With variations in precipitation and temperature so marked in the NAO, knock-on effects on river ecosystems might be expected. First, variations in river temperature follow closely those of air temperature (Mohseni & Stefan 1999). Secondly, for some regions rainfall variations in the NAO influence annual river discharge (Cullen & de Menocal 2000; Hanninen, Vuorinen & Hjelt 2000). Variations in discharge, in turn, can drive ecologically important changes in rivers such as floods, droughts and the dilution or mobilization of important ions (Reynolds, Emmett & Woods 1992; House & Warwick 1998; Jarvie, Whitton & Neal 1998; Bishop, Laudon & Kohler 2000). In the case of the ENSO, effects such as these influence river organisms (Puckridge, Walker & Costelloe 2000; Mol et al. 2000).
In this paper, we assess year-to-year changes among aquatic macroinvertebrates over 14 years in eight independent streams in western Britain in relation to the NAO. Following previous studies at the same sites and elsewhere, we focus on persistence as a measure of stability or environmental constancy (Townsend, Hildrew & Schofield 1987; Weatherley & Ormerod 1990a; Holomuzki & Biggs 2000). We defined persistence as a characteristic of the whole community revealed by the relative constancy in the rank abundance pattern and composition of assemblages through time (Holling 1973; Connell & Sousa 1983). Elsewhere, we have assessed long-term trends at the same sites in invertebrate abundance and richness (Bradley & Ormerod 2001).
All the streams were in the experimental catchment area around Llyn Brianne reservoir, mid-Wales (52°8′N 3°45′W), described previously by Stoner, Gee & Wade (1984) and Weatherley & Ormerod (1987). The streams are all second or third order, with catchments at altitudes of 215–410 m and of 15–264 ha, consisting either of upland sheep-pasture or plantations of sitka spruce (Picea sitchensis Carriere) and lodgepole pine (Pinus contorta Douglas; Rutt, Weatherley & Ormerod 1989; Table 1). The underlying Ordovician and Silurian shales, mudstones and grits are base-poor with low buffering capacity. Overlying brown podzolic soils, stagnopodzols and peats provide limited buffering so that runoff is soft (mean total hardness 3·9–18·8 mg CaCO3 L−1) and in some cases acidified (mean pH 4·6–6·9; Rutt et al. 1989). The streams form groups classifiable as acid conifer (LI1, LI2, LI8), acid moorland (LI5, CI1, CI4) and circumneutral moorland (LI6, LI7), between which faunal communities differ substantially (Weatherley & Ormerod 1987; Table 1). Typically, the circircumneutral moorland streams are species-rich and characterized by ephemeropterans, trichopterans and plecopterans, whereas acid forest streams are species-poor and dominated by acid-tolerant plecopterans. Acid moorland streams are intermediate between these extremes. The catchments of three further study streams (L14, C12 and C15) were limed artificially over a decade ago, but these are not considered further here (Weatherley & Ormerod 1990a; Bradley & Ormerod 2001).
Table 1. Catchment characteristics and chemistry for the study streams at Llyn Brianne. Chemical determinands are long-term annual means from 1985 to 1998 (adapted from Rutt, Weatherley & Ormerod 1989)
|Site code||Catchment land use||Catchment area (ha)||Mean pH||Total hardness (mg CaCO3 L−1)||Filterable aluminium (mg L−1)|
|LI1||Acid stream in conifer forest (c. 40 years old)||264||4·9|| 7·0||0·40|
|LI2||Acid stream in conifer forest (c. 40 years old)||194||4·9|| 6·9||0·43|
|LI8||Acid stream in conifer forest (c. 25 years old)||112||5·4|| 7·9||0·23|
|LI5||Acid stream in moorland|| 66||6·0|| 8·5||0·04|
|CI1||Acid stream in moorland|| 15||5·2|| 3·9||0·10|
|CI4||Acid stream in moorland|| 71||5·5|| 5·4||0·12|
|LI6||Circumneutral stream in moorland|| 82||6·9||15·7||0·05|
|LI7||Circumneutral stream in moorland|| 72||6·9||18·8||0·04|
The climate at Llyn Brianne is temperate, with mean stream temperatures rarely outside the range of 0–20 °C, and mean annual rainfall c. 1900 mm (Weatherley & Ormerod 1990a, 1990b). Stream substrata vary between gravel (> 2–16 mm) and bedrock, with bryophytes the only submerged macroflora; in marginal areas, Juncus spp. and Sphagnum spp. form the most abundant vegetation (Rutt et al. 1989; Weatherley & Ormerod 1990a).
- Top of page
Several studies have shown that environmental factors act across catchments to influence the persistence of river invertebrates (Townsend et al. 1987; McElravy, Lamberti & Resh 1989). Weatherley & Ormerod (1990a) observed such supra-catchment effects at the sites in this study from 6 years’ of data, but at the time no explanation was available. This longer run of 14 years’ data shows not only that synchronous fluctuations in persistence have continued, but also that they are consistent across contrasting catchments, taxa and stream habitats in the Lynn Brianne experimental area. On the basis of correlative evidence, the North Atlantic Oscillation provides a region-wide effect of the type required to explain the observed pattern: low year-to-year persistence in rank abundance and composition in all catchments was associated with positive phases of the NAO winter index while high persistence was associated with negative phases. Although this study is not yet long enough to link stream invertebrate communities unequivocally to the NAO through several cycles (cf George & Harris 1985; Straile & Geller 1998; George & Hewitt 1999; George 2000; Straile & Adrian 2000; Straile 2000), these data are the first to show that a link might exist. So far, evidence that the ENSO affects river organisms is from fish rather than invertebrates, and implicates droughts rather than floods (Puckridge, Walker & Costelloe 2000; Mol et al. 2000). At these Welsh sites, persistence among invertebrates was lower in wet phases of the NAO than in dry phases.
Fluctuations in persistence at Llyn Brianne involved two different ecological effects. Spearman’s rank coefficients showed that relative abundances across species were more constant during cooler, drier phases of the NAO, but dissimilar during warmer and wetter phases. In turn, Jaccard’s coefficients showed that high and low persistence in rank abundance were accompanied, respectively, by high and low constancy in species composition. These effects might well be linked – for example if shifts in relative abundance were sufficiently large to reduce some species to undectable levels. We can discount sampling error as an explanation for these trends, which in other long-term studies can explain apparently significant community variation between census years (Arnott et al. 1999). First, sampling error would be randomly distributed across samples, years and streams, rather than being responsible for highly synchronous variations across streams and habitats. Secondly, sampling error among aquatic invertebrates is usually greatest among rare taxa and complex habitats, where sampling is most difficult (Parsons & Norris 1996; Cao, Larsen & Thorne 2001). In our case, these effects were apparent because communities from structurally complex marginal habitats and rare species had the lowest overall persistence (Weatherley & Ormerod 1990a). Nevertheless, year-to-year variation in persistence among rare taxa and margins echoed exactly persistence in common taxa and riffles. We conclude that changes in persistence through time at Llyn Brianne have been real.
Few studies have assessed persistence among river invertebrates, particularly over timescales similar to this study. Although none have linked patterns with climatic cycles, those available provide valuable insight into the relationship between persistence and environmental variation. In general, persistence is greatest where environmental conditions are relatively constant (Robinson, Minshall & Royer 2000) and where taxa are adapted to the prevailing environmental regime (Miller & Golladay 1996). For example, in the United Kingdom, Townsend, Hildrew & Schofield (1987) showed that persistence was greatest where streams had low discharge, constantly low summer temperatures and pH regimes that were acid and stable. By contrast, persistence is least where environmental conditions fluctuate or are characterized by pulse, press or ramped disturbances (Meffe & Minkley 1987; Lake 2000). Those disturbances known to influence persistence include changes in catchment character – for example the replacement of seminatural forest by agricultural development (Brewin, Buckton & Ormerod 2000) – or particular catchment-scale events such as pesticide use (Hutchens, Chung & Wallace 1998) and fires (Richards & Minshall 1992). Changes in flow conditions also affect persistence, for example where regimes shift from long-term stable to short-term unstable due to freezing conditions or floods (Matthaei, Uhlinger & Frutiger 1997; Bradt et al. 1999). Flow effects like this occur at a variety of scales from the whole stream down to the individual patch (Death 1996; Matthaei & Townsend 2000). On all of this evidence, variations in persistence at Llyn Brianne between different climatic phases of the NAO would be consistent with a link between persistence and environmental variability.
With the NAO influencing western European rainfall, and discharge variation in turn affecting the persistence of invertebrates, it might be expected that the effects of flow would be central to our results. No discharge data were available directly from the study sites, although average discharge in rivers just 40 km away was indeed larger in positive NAO years by up to 18%. However, extreme flows were no larger, and discharge was no more variable in positive NAO years than in negative NAO years. Thus, on our evidence low persistence cannot have reflected the effects of pronounced floods (Lake 2000). More importantly, there were no direct correlations between discharge pattern and indices of persistence in the Llyn Brianne streams. Therefore, direct discharge effects cannot alone be sufficient to explain varying invertebrate persistence. We cannot discount subtle effects since the influences on flow on aquatic invertebrates are many and varied (Hart & Finelli 1999; Lancaster 1999; Holomuzki & Biggs 2000; Doisy & Rabeni 2001). Those species which increased or decreased at Llyn Brianne under different phases of the NAO provide clear examples: local density and net-spining activity in the caddis Hydropsyche siltalai reflects current velocity (Statzner & Bretschko 1998); Paraleoptophlebia submarginata is a marginal specialist characteristically occurring in areas of low hydraulic stress (Mobes-Hansen & Waringer 1998); Elmis aenea belongs to a group with well-characterized preferences on current velocity (Dietrich & Waringer 1999), and Nemurella picteti shows clear microdistributional responses to changing flow conditions (Lancaster & Hildrew 1993). We suggest that further assessment of discharge and flow pattern might well explain how invertebrate communities respond to the NAO.
Variations in discharge not only have direct effects on stream organisms, but also indirect effects through changes in stream chemistry. As Townsend et al. (1987) noted, persistence is greater in stable chemical environments. At soft-water sites such as those in this study, acid episodes are potentially important responses to flow that result from increased base-cation dilution, and from increased titration effects due to anions mobilized from catchment soils (Bishop et al. 2000). There is clear experimental evidence that acid episodes in some Llyn Brianne streams affect stream invertebrates (Ormerod et al. 1987) and might still offset biological recovery from acidification (Bradley & Ormerod 2001). However, several lines of evidence show that fluctuations in invertebrate persistence due to the NAO were not related to episodicity. First, some of the most pronounced variations in persistence occurred in circumneutral moorland streams – where buffering is greatest, and pH never falls below 6–6·5. Acid episodes therefore do not offer the regionwide effect necessary to explain our data. Secondly, no measure of acid-base status across years in any stream type was linked to the NAO. Thirdly, measures of persistence were not correlated with either pH, aluminium or calcium concentration. Other aspects of stream chemistry vary with discharge in ways that might be important ecologically, for example because interactions between droughts, floods and temperature affect the mobilization of nitrogen and phosphorus from catchment soils (Reynolds, Emmett & Woods 1992; House & Warwick 1998; Jarvie, Whitton & Neal 1998). Across the 22 lakes and streams in United Kingdom Acid Waters Monitoring Network, negative phases of the NAO through 11 years have been associated with increased nitrate concentrations in surface waters (Monteith et al. 2000). In turn, low temperature at the surfaces of catchment soils appear to result in increased nitrate loss into runoff, perhaps because plant or microbial retention is lower (Monteith et al. 2000; B. Reynolds, unpublished data). At present we have no data on whether such effects on nutrient release are widespread, nor on what their influence on stream productivity or invertebrate stability might be.
In addition to rainfall, NAO cycles are also reflected in temperature variations due to air movements from different sources, and because cloud cover or atmospheric dust affect radiation budgets (Moulin et al. 1997; Forchhammer et al. 1998). Such temperature variations affect upland British rivers enough to affect the emergence times of salmonid fish (Elliott et al. 2000). Thermal regimes, in turn, affect invertebrate persistence (Townsend, Hildrew & Schofield 1987). At Llyn Brianne, variations in stream temperatures between months, years and catchments reflect air temperature and insolation, with variations sufficient to affect the emergence periods of some stream insects (Weatherley & Ormerod 1990b). Effects are subtle, however, and there is so far no evidence that they affect communities. Among the species responding to the NAO at Llyn Brianne, some such as Chloroperla tripunctata have clear thermal tolerance, but in this instance increased abundance during warm phases would be contrary to expectation in this cold-water species (Elliott 1988). Nevertheless, increased winter temperatures affect the timing of insect lifecycles and oviposition success of adults (Chen & Folt 1996), so that warm phases of the NAO could translate into population performance and hence low measures of persistence between years. As with examination of flow responses to the NAO, we suggest that temperature effects should figure in the search for processes linking variations in river invertebrate communities to the NAO.
These results provide further evidence that ecosystems in northern and western Europe are affected by fluctuations in the NAO and its associated climatic effects. In addition to lakes and stream fishes, our data show that the NAO probably affects stream invertebrates. In keeping with the well-known challenges of testing hypotheses about large-scale and long-term effects on whole ecosystems, our data are of necessity correlative and require further investigation of the processes involved (Manel, Buckton & Ormerod 2000; Ormerod & Watkinson 2000). Nevertheless, apparent effects of the NAO occurred in replicate streams of contrasting chemistry and catchment land-use and might therefore be widespread.
Pronounced variations in invertebrate persistence that follow the NAO have both fundamental and applied importance. In fundamental terms, these data not only confirm previous ideas that persistence in invertebrate communities reflects environmental variability, but also show that persistence can vary through time within the same river system: constancy is not a fixed property of a given location. In applied terms, our data confirm Weatherley & Ormerod’s (1990a) view that variations in aquatic communities risk confounding or obscuring the effects of other long-term trends such as recovery from acidification, eutrophication or the effects of climatic change (Chen & Folt 1996; Lancaster et al. 1996; Lawlor et al. 1998; Monteith, Evans & Reynolds 2000). We recommend that researchers designing long-term monitoring programmes consider such effects carefully.