Motorway runoff transports mineral and organic materials to aquatic ecosystems which may provoke changes in the latter's chemical, physical and biological characteristics. We measured the water quality of a road runoff discharge and its effect on the periphyton communities in a small upland stream. The study was carried out during a summer low flow period (June to September 1998). We did not record notable, or lasting, effects of road runoff on physical and chemical parameters. Similarly, road runoff discharges did not change the mass (biomass, chlorophyll a) or the functioning (net primary production, respiration) of periphyton. The development of this community was most sensitive to rainfall and river flow conditions. The absence of impact possibly results from the fact that our monitoring was not carried out at times of more extreme road runoff. In fact, traffic and pollution, brought to the stream, were quite low and discharge dilutions were generally high during the study period. Thus this study should be continued to determine effects when discharges have potentially a high impact on the receiving water quality.
Road runoff, generated during rainfall events, may be discharged into freshwater ecosystems and contain minerals and organic matter (Hvitved-Jacobson and Yousef, 1991). The importance of these contributions depends on three principal factors (Thomson et al., 1997): 1) climatic (dry weather period, length and intensity of precipitation), 2) human (traffic density, oil component, road maintenance, accidents), and 3) technical (nature of road drainage catchment and road pavement, storm drainage and sewerage system). These discharges constitute an intermittent pollution source, varying not only in time but also in space (Chocat et al., 1993). Studies concerning assessment of runoff water quality have been greatly developed since 1980 (Cathelain et al., 1981; Balades et al., 1985; Legret et al., 1997; Sansalone and Buchberger, 1997; Thomson et al., 1997; Wu et al., 1998; Drapper et al., 2000; Pagotto et al., 2000; Shinya et al., 2000). The results gathered on different motorway sites in France and abroad, have led to the definition of major pollutants and the characterization of the principal factors controlling the levels of the pollutants.
Few studies have been carried out concerning the impact of motorway runoff on aquatic biota (Boisson, 1998). This may be due partly to the intermittent character of the pollution, its space-time variability and from the multitude of physical, chemical and biological factors which condition receiving water response (Seager and Maltby, 1989; McCahon and Pascoe, 1990). The few studies that have been carried out confirm the complexity of the problem and the difficulty of offering precise answers. The results obtained are variable and differ depending on the motorway sites, receiving water and physical, chemical or biological parameters measured.
In this study we have monitored the quality of motorway runoff over a whole summer season at regular intervals, and the impact of runoff on the biological quality of the receiving water. We used epilithic periphyton as bioindicators. Localised at the sediment and bulk water interface, these communities can have a significant impact on biogeochemical cycles (the mineralising of organic matter, enzyme hydrolysis activities) and ecosystem dynamics (Mulholland et al., 1994, 1995; Paerl and Pinckney, 1997; Allan, 2001). Any modification in environmental conditions may have important repercussions on the community and consequently on the aquatic ecosystem as a whole. Pollutants may affect both structural (e.g., biomass, community composition) and functional characteristics (photosynthetic and heterotrophic activity) (Masseret et al., 1998; Sabater et al., 2002).
The following criteria were used for site selection: 1) a wide surfaced road for large volumes of runoff, 2) dense traffic (> 10,000 average annual daily traffic units) offering problems of traffic flow to obtain maximum car pollution, 3) maximum area of hard surfaces and the absence of sewerage systems (grassy ditches, infiltration tanks, oil separators) so that the hydraulic impact would be optimised and the transfer of pollution to the receiving water would be maximal, 4) a receiving stream of good quality to maximise the impact, and of low flow to minimise the dilution of discharge, and 5) a site near enough to the laboratory to carry out chemical and biological analyses soon after sample collection.
The site chosen was situated on a small upland mountain stream (Chabanty, order 2) in the Monts du Forez (the Puy-de-Dôme area). The stream drains a 580 ha catchment, dominated by granite and is very slightly urbanised (4 localities). The catchment is largely covered with meadow. At the site, the stream receives rainwater drained from the surface of a 3,660 m2 section of motorway (the A72, Clermont-Ferrand to Saint-Etienne). The collection system is largely made up of cement gutters, street inlets and embankments.
Two sampling sites were selected, one above and one below the motorway runoff (Figure 1). They were selected for comparable environmental characteristics, such as light intensity, current velocity and grain size distribution. The first was 60 m upstream from the motorway discharge and constitutes the reference station, the second was 40 m downstream from the discharge where the flow was sufficiently turbulent to allow good mixing of runoff. The substratum was mainly composed of sands and cobbles. Riparian forest occurred on the left bank (alder and fir), and since the stream was less than 2 m wide, it was largely shaded. Runs were the dominant physical structure, and average current velocity was less than 15 cm s−1 during low flows (discharge < 10 l s−1) and between 27 and 45 cm s−1 in periods of flood (discharge: 40 l s−1).
Material and methods
The study site was carried out during a summer low flow period (17 June to 9 September). Stream discharges were continuously monitored downstream of the road runoff. There the stream is channelled over a length of 5 m allowing the installation of a flow meter (ISCO 3230). The stream water was sampled upstream from the discharge where 4 instantaneous samples were taken during dry weather (5, 19 and 25 August and 1 September 1998). A 24 h sequential sample was taken during wet weather (1 August 1998). During rainfall events, runoff flowing into the stream was measured (ISCO 3230) and the water was sampled every 2 m3 of flow. On site rainfall was recorded with a recording rain gauge (Precis Mécanique). All data were recorded with a data logger (CR2M).
The physical and chemical characterisation of stream water and its discharge was carried out in an earlier study (January to April 1998), only a few parameters were taken into account during that study. The pH, conductivity, total suspended solids (TSS), biochemical oxygen demand (BOD) and chemical oxygen demand (COD) were measured in raw water; NH4+, NO2−, NO3−, PO43−, SO42− and Cl− concentrations were measured in water filtered through a 0.45 μm filter. Analyses were carried out using standardised methods (AFNOR and ISO).
Periphyton communities were sampled using artificial substrata. This technique permitted the standardisation of periphyton development conditions and reduced the natural variability of the substratum, so that the effects of the motorway discharge could be isolated. Glass slides, sanded on one side, (7.8 × 2.4 × 0.4 cm) were used as artificial substrata. These were assembled in 3 series of 6 slides fixed on blocks. Three sets were installed at each sampling site in similar conditions of light and current velocity. The slides were submerged on 22 May 1998, and sampling began on 17 June 1998. Sampling of the initially deployed set was carried out each week then until 9 September 1998. At each sampling date 6 slides were removed per site (i.e., 2 slides per block), then quickly transferred to the laboratory for analysis.
The periphyton were characterised using three structural and two functional variables. The ash free dry mass (AFDM) was taken as an indicating parameter of the total community biomass (autotrophic and heterotrophic organisms) and chlorophyll a with pheopigments as representative parameters of algal biomass. Periphyton were collected by brushing the slide surface in a known volume of water from the stream. Aliquots of the solution were filtered (Whatman GF/C glass fibre filters) for measurement of chlorophyll a and biomass community. For analyses of chlorophyll a, filters were immersed for 12 h in 90% acetone and the absorbance of the extracts was read in a spectrophotometer (Lorenzen, 1967). Community biomass (AFDW) was measured as the difference in weight between filters dried at 60°C for 24 h and combusted at 500°C for 2 h.
Net primary production (NPP) and respiration (R) were measured as functional variables to represent community autotrophic and heterotrophic activity. The light and dark bottle oxygen method was used as described by Wetzel and Likens (1991). Every slide sample was placed in airtight Pyrex flasks (BOD flasks) filled with stream water. Initial dissolved oxygen (DO) levels in each flask were measured (WTW oxymeter Oxi 538, StirrOx G probe). The flasks were then sealed airtight and placed in a thermostat bath adjusted to stream water temperature. Light was provided by two florescent Mazda Fluor Presiflux tubes that provided approximately 60 μE·m−2 s−1 of light. After two hours of incubation, DO content was measured again. The NPP was estimated through the difference with initial oxygen content. The flasks were resealed and wrapped in aluminium foil for artificial darkness then incubated for a 2 h incubation in the bath after which a final measurement of DO was taken. Oxygen consumed by respiration was determined by the difference with the second measurement.
The number of acquired results (13 sampling dates, 2 sampling sites, 5 parameters, and 3 replicates) allowed us to carry out statistical analyses on the set of data. A two-way analysis of variance (repeated measures ANOVA) was carried out on all the results so as to verify the variability of gaps observed between the sampling sites and measurement sampling dates. Differences were considered significant if p = 0.05 (Fisher test). Difference with 0.05 < p < 0.10 were considered marginally significant.
The average annual daily traffic at the site was quite low (< 13,000 vehicles per day). However it was higher during the study period due to the rise in motorway use during the holiday period (16,000 to 20,000 vehicles per day).
Precipitation was relatively low during the study period (103 mm), representing about one tenth of the annual rainfall (1,213 mm an−1). Thirty-one rainfall events were recorded (Figure 2), but only eight of these resulted in sufficient water runoff to be considered. The main characteristics of these rainfall events are summarised in Table 1. Two events are particularly interesting as they followed quite long dry periods (more than 10 days), the most striking event being the storm of 27 July (measured rainfall: 36.6 mm; dry period: 23 days).
Chabanty flow remained below 40 l s−1 during the study. It decreased progressively during the study period to reach very low values in August, that is, less than or equal to 1 l s−1. Flow was rapidly and easily influenced by rainfall (Figure 2) such as during the storm event of 27 July 1998 when the average daily flow went from 3 l s−1 to 35 l s−1. During the 8 main rainfall events, when intensity was highest, runoff discharges of between 3.6 m3 and 131 m3 occurred for periods of an hour. Thus, during the rainfall events these discharges represented 3 to 92% of stream flow (Table 2).
Physical and chemical characteristics of Chabanty water
Chabanty water was relatively cool, well oxygenated and weakly mineralised (Table 3). Chemical oxygen demand and BOD, as well as concentrations of NH4+, were slightly elevated as a result of the presence of cows in the meadow upstream from the study site. The animals drank from the stream and released mineral and organic particles. Rainfall during the sampling could result in significant increases in TSS (251 mg l−1), COD (96 mg l−1) and NH4-N (597 μ g l−1). The physical and chemical characteristics of the water during these days were comparable to those observed over a longer period from January to April 1998 (authors' unpublished data). Heavy metal concentrations (Pb, Zn, Cd, Cr, Cu, Fe and Mn) were low and under the quality threshold for drinking water (WHO recommendations) (authors' unpublished data).
Physical and chemical characteristics of runoff
The first rainfall event on 25 June 1998 and another on 27 July were not sampled due to sampler malfunction. Runoff water was alkaline with low mineral and organic matter content (Table 4). This result confirms those obtained from early sampling from January to April (18 analysed rainfall events). The runoff heavy metal concentrations (Pb, Zn, Cd, Cr, Cu, Fe and Mn) were also low (authors' unpublished data).
Periphyton community response
Periphyton biomass (AFDM) did not differ between the two sites over the duration of the study (Anova, p = 0.1559). The AFDM varied slightly during the study period (Figure 3); only on 18 August 1998 were the AFDM values statistically different between the two sites (p = 0.0003). Biomass developed little between 17 June and 7 July (< 3.5 g m−2). During that period, flow fell from 40 l s−1 to 21 l s−1, but rose on occasions following three rainfall events (Figure 2). The latter were of low amplitude (< 12 mm) and the increase in river flow was not great. At these times the runoff was strongly diluted in the stream (3 to 16%; Table 2). Biomass then increased progressively until 18 August (6.5 g m2), while flow diminished to reach its lowest value (< 1 l s−1). The growth of periphyton slowed after the heavy rainfalls of 27 July and 1 August (which resulted in major increases in stream flow, 36 l s−1 and 32 l s−1, respectively). The stream regained its low flow 12 days before the rainfall events of 22 to 23 August. The amplitude of these events was moderate (< 10 mm), but their effects seemed more pronounced particularly at the downstream sampling site where the biomass fell to 3 g m−2.
Periphyton chlorophyll a was low (< 9 mg m−2) and less downstream of the motorway discharge (p = 0.0259). In contrast to AFDM, chlorophyll a progressively diminished during the study period (Figure 3), but only values obtained on 30 June were statistically different between sites (p = 0.0014). This difference was, however, quite small. The decrease was slowed down by rainfall events, and most of them even resulted in a slight increase. It should be noted that on 22 July the artificial substrata of the upstream sampling site were covered by only a few mm of water, which may have resulted in a significant decrease in chlorophyll a.
The AFDM/chlorophyll a ratio was generally higher than 400 (Figure 4) and increased progressively during the study period to reach high values (i.e., > 2,000). This shows the dominance of heterotrophic organisms within the periphyton and/or the accumulation of organic matter (exsudats, exopolymers, and exogenic organic material).
Overall, pheopigments were comparable between the two sampling sites (p = 0.3400). However, they were significantly higher between 11 August and 1 September compared with values obtained earlier at both sites (p = < 0.001) (Figure 3). Thus the chlorophyll a/pheopigments ratio progressively lowered during the study period, which shows the ageing of autotrophic components within the periphyton (Figure 4).
Net primary production and R were relatively low (NPP: < 50 mg O2 h−1 m−2; R: < 9 mg O2 h−1 m−2; Figure 5). The values were significantly lower at the sampling site situated downstream from the motorway discharge (NPP: p = 0.0037; R: p = 0.0183). The NPP development is compared to that of AFDM in Figure 3; a notable fall in values was observed on 18 August when AFDM underwent a significant increase. This observation supports the hypothesis that an accumulation of organic matter within the periphyton led to suffocation of the community and a reduction in photosynthetic activity. Respiration did not show significant trends in development over time. Values fluctuated during the study period (Figure 5). However, as with chlorophyll a and NPP, most of the rainfall events resulted in a slight increase in respiration.
Our study site offered numerous advantages including containing low flows relative to the motorway discharge, a good physical and chemical quality, a large road catchment, a waterproof rainwater collection system and the absence of sewerage systems. Also, the period was interesting, that is, long dry spells, storms and low water. Thus we believed that the motorway discharges should provoke notable mechanical and chemical stress on the periphyton. However, chemical stress was relatively limited as motorway traffic was low to moderate (< 20,000 annual average daily traffic); accumulated pollution on the road surface between two rainfall events was not high. The principal chemical parameter values analysed in runoff (conductivity, TSS and COD) were among the lowest that have been observed on motorways (Legret et al., 1997; Thomson et al., 1997; Wu et al., 1998; Drapper et al., 2000). Polluting element content in runoff equally depends on the frequency and the intensity of rainfall, as well as the length of preceding dry spells (Thomson et al., 1997). With the exception of the 27 July storm event (36.6 mm) maximum precipitation intensity was relatively low (< 4 mm h−1), low in total amount (< 10 mm) and/or quite short in duration (> 3 h). Because of this, most precipitation did not lead to washing, nor intense transportation of elements from the road surface. At the beginning of the study, stream flow was higher (> 20 l s−1), so road runoff was highly diluted (3–16%) and contributed only slightly to the rise in stream flow.
While the content of pollutants is the main factor conditioning periphyton response, it could also depend on other physical, chemical and biological factors that act either antagonistically and/or synergistically. As well, the duration of perturbations is a factor to consider. If high discharges are short in length, polluting substances may not have the time to penetrate the biofilm (Lock et al., 1981a). The microorganisms situated at the base of the mat, protected from diffusion by the organisms higher in the matrix, might be little exposed to pollutants, (Lock et al., 1981b). After rainfall, an increase in nutrients could stimulate periphyton growth (Peterson et al., 1994; Biggs, 1995; Mosisch and Bunn, 1997; Dent and Grimm, 1999; Biggs and Smith, 2002). However, the length of time exposed to such conditions might be insufficient for cells to absorb nutrients for future cell growth (Peterson et al., 1994). Similarly, the increase in current velocity may have a positive effect by eliminating dead cells, grazers and toxic substances from the periphyton (Mosisch and Bunn, 1997). Chemical stress may have a number of effects on periphyton: it may bring about a change in algal taxonomy which can be seen through a reduction in sensitive species and an increase in tolerant species (Amblard et al., 1990; Napolitano et al., 1994; Genter, 1995; Ivorra et al., 2000; Gold et al., 2003) and may reduce grazing, which leads to a stimulation in the growth of periphyton (Maltby et al., 1995; Genter and Amyot, 1994; Forrow and Maltby, 2000).
Thus, it may prove difficult to quantify motorway runoff perturbation effects on aquatic biota. For example, at Chabanty, we cannot state that road runoff had an impact on the streambed periphyton biomass or functioning. In fact, we have seen that, downstream from the discharge, chlorophyll a, NPP and R were clearly lower while the total biomass and the pheopigments were comparable to those at the upstream sampling site. Furthermore, the periphyton at both sampling sites went through the same growth pattern during the study period, and discharges from the eight main rainfall events did not have a specific effect on the downstream station. This last point leads us to suggest that the development of periphyton in our stream is strongly influenced by river catchment natural pluviometry and hydrology. At the downstream sampling site, the low values for chlorophyll a, net primary production and respiration are probably linked to the presence of slightly different environmental characteristics than those at the upstream sampling site (e.g., flow).
The influence of river flow variations is more easily observed on the total biomass of periphytic communities (AFDM). At the beginning of the study, stream flow (20–40 l s−1) generated high enough current velocity (< 50 cm s−1) to limit the development of periphyton and the accumulation of biomass (Horner and Welch, 1981; Horner et al., 1990; Biggs, 1995). Flow conditions at this time were favourable to algae development, and thus chlorophyll a was more abundant. The higher flows probably resulted in the elimination of senescent cells, favouring trophic exchange with the water column and light penetration in the mat (Ghosh and Gaur, 1991; Biggs, 1996; Biggs and Stokseth, 1996). During the summer dry spell (July–August) current velocity dropped (< 10 cm s−1) and facilitated the growth of periphyton (AFDM 4–8 g m−2). On the other hand chlorophyll a was lower during this period and pheopigments increased appreciably. Similarly, the AFDM/chlorophyll a ratio was high (Biggs and Close, 1989) and increased during the period of study. From these results, we suggest that during the late summer low flows periphytic biomass might be influenced by inorganic particle deposits on the periphyton (Graham, 1990; Fayolle et al., 1998). In covering the biofilms, they might contribute to the senescence of algae cells, thus explaining the fall in chlorophyll a and the rise in degradation products (i.e., pheopigments at both sites). These particle deposits are certainly released as a result of current velocity during flow increases (Biggs and Close, 1989; Biggs, 1996; Biggs et al., 1999), which would explain the fall in AFDM after the rainfall event of 22–23 August. The effect is less visible for chlorophyll a, because the autotrophic component is generally more solidly fixed to their support (Blenkinsopp and Lock, 1994).
This hypothesis is supported by the results of NPP and of R. The measurements show slightly significant fluctuations over time and a relatively stable ratio over the period of study (NNP/R, 2–6). This shows a relationship between equally stable autotrophic and heterotrophic organisms, the latter not being responsible for the rise in recorded biomass. Finally, the progressive rise in the NPP/chlorophyll a ratio may be explained by a change in the taxonomic composition of communities and their adaptation to new environmental conditions (low current velocity, biofilm plugging, temperature increases).
The results of this study confirm the complexity of the problem and the difficulty of predicting accurately the effects of road runoff on freshwater ecosystems. Similar studies should be carried out on sites where a relatively high discharge of motorway runoff occurs to a stream with very good quality water and low flow. This would then allow a better analysis of impact on stream ecosystems.
Accumulation phenomena and the persistence of certain substances found in discharge in the environment (suspended solids, organic and mineral micro pollutants) were not taken into account in this work. These substances may accumulate in different compartments of the ecosystem (Forrow and Maltby, 2000), be released during flood, dispersed, then released further downstream at various distances. The understanding of transfer mechanisms, dispersion, retention and transformation of pollutants in different compartments of the ecosystem is incomplete. It would be informative to use other biological indicators to identify risks taken by receiving water downstream from road runoff.
This study was carried out within the framework of a research project on Environment and Urban Engineering piloted by the Laboratoire Central des Ponts et Chaussées (French Ministry of Public Works and Transports). We thank Alain Fournier, Corrine Pierrat; Sebastien Roux and Marc Danjean for their technical assistance, and Brendan Keenan for the translation and proof reading.