Site selection and study design

The studied streams were located in and around the city of Austin, Travis County, in central Texas, USA. Austin is in the transitional area between the Central Texas Plateau and Blackland Prairie ecoregions (Omernik 1987). It has a semiarid climate with an average annual precipitation of 810 mm, spread relatively evenly throughout the year. However, the climate is characterized by high-intensity, short-duration precipitation events and a high average rate of pan evaporation (1956 mm/y).

An upstream–downstream study design was chosen (Burton 2001) with matched pairs of upstream (reference) and downstream (treatment) sites within the same stream reach (Fig. 1). This design provides an extra degree of certainty that the variation present in the results is caused by the treatment and not by inter- or intrastream differences. The treatments were concentrations of PAHs in stream sediments immediately above and below parking lots that had been sealed with coal-tar sealant. The distance from the parking-lot discharge point to the downstream sampling site, type of discharge (pipe, grassy swale, rip-rap channel), age of sealant, and traffic use in the parking lots all were potentially important variables in explaining PAH concentration in stream sediments, but the number of sites and replication necessary to evaluate these variables were beyond the scope of this initial field study.

Seal-coated parking lots immediately adjacent to ≥2^nd-order streams in urban Austin were identified based on observable wear patterns with high-resolution (pixel = 0.15 m) aerial photography in a geographic information system (GIS). Areas of coal-tar-sealed parking lots that receive consistent vehicle wear appear lighter in color than other parts of the parking lot, and light areas show where the sealant has worn off the pavement. The city of Austin has been screening PAH levels in stream sediments with enzyme-linked immunosorbent assay (ELISA) methods since 1997 and has identified a number of streams with PAH levels of concern. Seven of these streams had adjacent coal-tar-sealed parking lots that drained directly to the stream and appropriate sampling habitat upstream and downstream of that drainage and were selected for our study.

The selected streams had catchments ranging from 5 to 310 km² and development levels from 4 to 48% impervious cover. Most of the development in the catchments was urban in nature, and no agricultural land uses were noted at the time of our study. In May 2004, pool sediment chemistry, habitat quality, and riffle and pool benthic macroinvertebrate communities were surveyed in sites upstream and downstream of all 7 parking-lot discharge sites.

Data collection

Habitat was assessed at each site using the US Environmental Protection Agency (EPA) Rapid Bioassessment Protocol (RBP) Habitat Quality Index (HQI; Barbour et al. 1999). This protocol is a visually based method that allows general comparison of habitat quality among stream reaches. Stream physicochemical variables were measured with a multiprobe (Hydrolab Quanta, Hach Company, Loveland, Colorado), and discharge was measured using a hand-held Marsh–McBirney Flomate flow meter (Hach/Marsh–McBirney, Frederick, Maryland). Benthic macroinvertebrates were collected in both riffle and pool habitats. Riffle benthic communities were assessed using a modification of the EPA RBP (Barbour et al. 1999), in which a Surber sampler (0.093 m², mesh = 500 μm) was used instead of a kick-net at 3 discrete riffle locations. Detritus and associated biota collected in the Surber from the bottom, middle, and top of the riffle were placed in 1-L jars and preserved in the field with 90% ethanol. Sediment-dwelling benthic communities were assessed by collecting 3 discrete ½-L sediment composite grabs from the top 10 cm in representative areas in the study pools using a Teflon scoop. Sediment and associated biota were placed in 1-L jars and preserved in the field with 90% ethanol.

Sediment chemistry was evaluated in each pool by collecting a 1-L grab composite sample from depositional areas around the pool using a Teflon scoop. Samples were stored at 4°C, submitted to the testing laboratory within 5 d of collection, and analyzed by the laboratory within 15 d of collection. Sediments were analyzed for 16 parent PAHs using gas chromatographic/mass spectrophotometric (GC/MS) techniques according to EPA method 8270C (USEPA 1996). In addition to PAHs, several other variables were measured, including pesticides and polychlorinated biphenyls (PCBs) (EPA 8081A and 8082, respectively), total organic C (EPA 5310B, in which organic C was oxidized into CO₂ and H₂O and measured with an infrared analyzer), trace metals (EPA 6020), Hg (EPA 7471A), and total petroleum hydrocarbons (TNRCC 1005). All analyses were done by DHL Analytical in Round Rock, Texas, using standard laboratory methods (APHA 1995, USEPA 1996, TNRCC 2001).

Data analysis

Benthic macroinvertebrates were sorted from associated detritus in the laboratory according to EPA RBP laboratory methods (Barbour et al. 1999), with the exception that all organisms were sorted and no subsampling method was used. Insects were identified to the lowest practical taxonomic level, usually genus, by city of Austin staff using standard taxonomic keys (Berner and Pescador 1988, Pennak 1989, Epler 1996, Merritt and Cummins 1996, Westfall and May 1996, Wiggins 1996, Needham et al. 2000, Thorp and Covich 2001, Stewart and Stark 2002). Chironomidae and noninsect taxa were identified to genus by M. Winnell (Freshwater Benthic Services, Petoskey, Michigan).

Nine community attributes (metrics) were used to evaluate differences between upstream and downstream sites: density, taxon richness, Diptera richness, % dominant taxa, % Chironomidae density, number of noninsect taxa, the Hilsenhoff Biotic Index (HBI), % tolerant taxon density, and number of intolerant taxa. The metrics were selected a priori based on traditional categories of benthic macroinvertebrate community structure (taxonomic richness, composition, and tolerance; Resh and Jackson 1993, Merritt and Cummins 1996, Barbour et al. 1999, Karr and Chu 1999) and experience with these metrics in Austin-area streams. Abundances from the riffle and pool samples were converted to density measurements. Densities in riffles were expressed as ind./m²; densities in pools were expressed as ind./L.

Within-habitat variability was not evaluated for this paper. Therefore, the 3 benthic-community replicate samples within each habitat in each stream were combined into single composite samples to allow a more robust statistical comparison among streams. The Wilcoxon matched-pairs test (Statistica, version 7, Statsoft, Tulsa, Oklahoma) was used to distinguish significant differences between upstream and downstream data sets for community metrics, taxon responses, and other environmental variables. Our data set was relatively small, so significance was interpreted using α = 0.1 (Helsel and Hirsch 1995); p values from the analyses are reported to facilitate interpretation, and values <0.20 were considered to evaluate trends. Means and standard deviations were calculated and are presented for purposes of interpretation, but the Wilcoxon matched-pairs test is a nonparametric test of distribution, not means. The Spearman rank-order correlation coefficient (Statistica) was used to evaluate correlations between metrics.

SQGs were used to assess potential toxicity of PAHs in stream sediments and to compare levels of contamination among study sites. The PEC is the value above which adverse effects on sediment-dwelling organisms are likely to be observed; this value was developed from a consensus-based review of the literature (MacDonald et al. 2000). MacDonald et al. (2000) calculated TPAH as the sum of the concentrations of 13 parent (unsubstituted) PAH compounds, where ½ the value of the detection limit was used as the concentration for any nondetected compounds, and the PEC for TPAH was set at 22.8 mg/kg. In our study, TPAH was calculated as the sum of the concentrations of the 16 EPA priority pollutant PAHs (Table 1; USEPA 2002) plus 2-methyl naphthalene; nondetected values were not included in the sum. This method gives results that are comparable to, but slightly greater than, TPAH calculated by the method of MacDonald et al. (2000).

Equilibrium partitioning sediment benchmark toxicity units (ESTBU; DiToro and McGrath 2000, USEPA 2003) also were used to assess levels of contamination. This approach uses equilibrium partitioning theory to generate toxicity units (ESBTUs) for mixtures of PAHs in sediment and is applicable to a wide range of aquatic organisms. ESBTUs >1 suggest toxicity, whereas values <1 are acceptable for protection of benthic organisms. The EPA ESBTU method recommends summing the concentrations of 34 PAHs for this approach. However, our laboratory analysis was limited to the 16 EPA priority PAH pollutants. Therefore, a recommended adjustment factor was applied to compensate for this discrepancy (USEPA 2003), and the increase in ESTBUs from upstream to downstream was used to predict community changes at the downstream sites using linear regression with ln(x)-transformed data.

GIS was used to evaluate land use and impervious cover at the catchment and local scales for each of the study sites. Local scale refers to a buffer width of 300 m on each side of the stream centerline (600 m total). For the downstream sites, the length of the buffer was the stream distance from the downstream to the upstream site, which ranged from 175 to 1100 m. For the upstream sites, the maximum distance between sites (1100 m) was used as the standard buffer length. City of Austin landuse categories and methods for measuring impervious area were used to compare landuse variables among study sites.

Sediment chemistry

TPAH concentrations were markedly higher at downstream than upstream sites in 5 of the 7 study streams (Table 1, Fig. 2). Downstream TPAH values were above the PEC of 22.8 mg/kg at 2 streams (Barton and Walnut); downstream TPAH values ranged from <1 to 10 mg/kg in the remaining streams (Fig. 2). Differences in PAH concentration and ESBTUs between upstream and downstream sites were minimal in 2 streams (Buttermilk and Manor). The remaining 5 streams were chosen for further evaluation.

Individual PAH compounds showed the same pattern as TPAH, with minimal variation. Concentrations of 6 of the 9 PAH constituents for which PECs have been proposed (including benzo[a]pyrene [B{a}P], possibly the most carcinogenic of the PAH parent compounds) were above the PEC at the downstream sites in Walnut and Barton (Table 1). Fluoranthene was generally the dominant PAH in all samples, followed by chrysene, pyrene, and benzo(b)fluoranthene, all of which are relatively high-molecular-weight PAHs and are EPA priority pollutants (USEPA 2004). Each of these PAHs has distinct chemical and toxicological characteristics, but increases in TPAH were generally proportional to increases in PAH parent compounds. Therefore, TPAH values are used to represent PAH contamination throughout our study.

Other environmental variables at study sites

Water-quality variables, such as pH, specific conductance, and dissolved O₂ were relatively consistent within streams, and there were only minimal differences between upstream and downstream sites (Table 2). On average, conductivity was 17 μS/cm higher and temperature was 1.7°C lower at downstream sites than at upstream sites (p = 0.04, p = 0.08, respectively; Table 2). Sediment grain size was not measured in the pool sediment samples. However, from observations documented on field sheets, grain size was relatively uniform between upstream and downstream sites; some streams had predominantly fine sediments and others had predominantly gravel-size sediments. The EPA HQI ranged from 57 to 89% of reference values and was supportive of optimal or suboptimal biological-condition categories at all study sites. HQI values were, on average, 8% lower at downstream sites than at upstream sites (Table 2).

Concentrations of trace metals, pesticides, and polychlorinated biphenyls (PCBs) generally were below values of concern (Table 2). Arsenic was elevated at the upstream (17 mg/kg) and downstream (28 mg/kg) sites at Wells Branch (above the threshold effect concentration [TEC] of 9.79 mg/kg, below the PEC of 33 mg/kg). Cu was between the TEC and the PEC at the upstream site at Barton (38 mg/kg). Pb was slightly higher, on average, at downstream sites (p = 0.04) but did not approach the TEC of 35.8 mg/kg at any site. Organochlorine pesticide concentrations were uniformly low at all sites (only 4 detected values, all below the TEC, distributed among 3 downstream sites and 1 upstream site).

Parks/open space was the predominant local-scale land use at both upstream and downstream sites (34% and 32%, respectively). Single-family residential landuse was more important at upstream than at downstream sites (29% vs 14%), whereas multifamily residential land use was more important at downstream than upstream sites (28% vs 9%). The percentages of the remaining landuse classes were either relatively similar between upstream and downstream sites or were small. Percent impervious area was slightly higher at the downstream sites than at the upstream sites (36% vs 28%).

Benthic macroinvertebrate community response

Four biological measures (2 with p < 0.10) indicated that the benthic macroinvertebrate community in pools was degraded at the downstream sites relative to the upstream sites (Table 3). Three of the measures (taxon richness, Diptera richness, and number of noninsect taxa) were strongly correlated (r > 0.90). Density also indicated significant degradation at the downstream sites when Wells Branch was excluded from the analysis (p = 0.07). This site had the lowest downstream TPAH concentration (4.00 mg/kg), and all of the biological measures indicated little or no biological change in its downstream biological community relative to its upstream community.

Three biological measures (all with p < 0.10) indicated that the benthic macroinvertebrate community in riffles was degraded at the downstream sites relative to the upstream sites (Table 3). Average densities were 2× higher at upstream than at downstream sites in both habitats (pools: ∼200 vs ∼100 ind./L, riffles: ∼4000 vs ∼2000 ind./m²). A small, but significant decrease in % dominance was noted at the downstream sites relative to the upstream sites (p = 0.08). However, the decrease in % dominance was not considered degradation because this metric generally responds positively to environmental stress. Some of the biological metrics from the riffle community were correlated but not to the extent observed in the pool community (r < 0.90).

Benthic macroinvertebrate organismal response

Across the 5 study streams, 74 taxa were collected from riffles and 78 were collected from pools. Ten taxa in the pool community and 13 taxa in the riffle community showed an apparent treatment effect (p ≤ 0.20; Table 4). In pools, individual taxon responses were noted primarily among the Chironomidae. Average Chironomidae densities were 270 ± 209 ind./L at upstream sites and 137 ± 171 ind./L at downstream sites. Densities of 3 genera (Paratendipes, Cladotanytarsus, and Ablabesmyia) were significantly lower at downstream than at upstream sites (p < 0.10), and densities of 4 other genera were apparently, but not significantly, lower at downstream sites (p ≤ 0.20). Densities of Stenelmis beetles were significantly lower at downstream sites than at upstream sites, whereas the amphipod Hyallela and the snail Physa were apparently, but not significantly, lower at downstream sites (p ≤ 0.20). These organisms were well represented in the 5 study streams and were present in at least 3, but usually 4 or 5 streams.

Individual taxon responses were more diverse in riffles than in pools. Thirteen taxa showed an apparent treatment effect (p ≤ 0.20; Table 4), and densities of 6 of the 13 taxa were significantly lower in downstream sites than in upstream sites (p < 0.10). Across both habitat types, taxa with density decreases at downstream sites represented 6 orders of insects, including some typically sensitive groups (Zygoptera, Trichoptera, Ephemeroptera, Coleoptera), 6 Diptera (5 Chironomidae, 1 Empididae), and 1 snail (Helisoma anceps). The only taxon that appeared to have a positive response to the treatment effect was Oligochaeta, which had higher density in pool sediments at the downstream sites in 4 of the 5 study streams (Fig. 3). However, this response was not significant (p = 0.42). Wells Branch was the only study stream in which Oligochaeta density was higher at the upstream site than at the downstream site (21 and 6 ind./L, respectively). This stream had the lowest downstream PAH concentration.

PAH toxicity units correspond to community changes

ESBTUs were calculated and compared to community changes from upstream to downstream sites to evaluate predicted toxicity of PAHs in pool sediments. Increases in ESBTU explained 76% of the taxon losses in pool communities at the 5 study streams (R² = 0.76, p = 0.05; Fig. 4A). Little Walnut was the only site at which taxon richness increased (1 taxon) from upstream to downstream. Increases in ESBTU also explained 72% of the reduction in density between upstream and downstream pools (R² = 0.72, p = 0.07; Fig. 4B). Wells Branch was the only study site at which density increased (+90 ind./L) from upstream to downstream.

Sediment chemistry

PAH hot spots have been documented mostly below point-source discharges, where contaminated sediment from industrial wastes historically has been dumped into large receiving waters (Merrill and Wade 1985, Catallo and Gambrell 1987). However, some of the highest concentrations of PAHs found in the Austin area (1000–3500 mg/kg TPAH) cannot be explained by industrial wastes or by traditional urban runoff sources, such as tire wear, automotive fluid leakage, and atmospheric deposition (Steuer et al. 1997, Dickhut et al. 2000, Mahler et al. 2005, Neff et al. 2005). PAH concentrations in suspended sediment in runoff from coal-tar-sealed lots (mean = 3500 mg/kg TPAH) are 65× higher than concentrations in suspended sediment in runoff from unsealed lots (mean = 54 mg/kg TPAH), and total yields from sealed lots (average of coal-tar- and asphalt-sealed lots) are 50× higher than from unsealed lots (Mahler et al. 2005). PAHs from unsealed lots represent background automotive PAH sources, such as exhaust (atmospheric deposition), fluid leakage (used motor oil), and tire and brake wear (Mahler et al. 2005). In our study, TPAH was 4 to 216× higher at downstream sites than upstream sites and cannot be explained by the diffuse discharge from common urban runoff sources, such as roads and atmospheric deposition, in the small drainage area between the paired sites in each study stream. We propose that the primary source of the higher PAHs at the downstream sites was the sealed parking lots that drain to these streams.

Some investigators have successfully identified sources of PAHs in the environment using ratios of PAH constituents, or fingerprinting methods (Khalili et al. 1995, Dickhut et al. 2000, Yunker et al. 2002), whereas others have found these methods inconclusive or insufficient (Marvin et al. 2000, Bucheli et al. 2004, Zhang et al. 2005). We attempted to identify the sources of PAH in the sediments of our study streams using ratio methods, but we were unsuccessful and found no significant clustering of field data with known source data. A comprehensive analysis of a large suite of parent PAH compounds and alkyl homologues is often necessary for fingerprinting methods because certain forms of PAHs are more resistant to weathering than others (Burgess et al. 2003, Thorsen et al. 2004). Alternative, relatively complex sourcing methods that use multiple lines of evidence have been proposed to separate specific sources of pyrogenic PAHs in the urban environment (Brenner et al. 2002, Larsen and Baker 2003, Zhang et al. 2005). Our inability to associate PAH contamination in our study streams with coal-tar sealant might have been because we analyzed only the 16 EPA priority PAHs in field sediments or because extensive weathering and mixing with other materials occurs as the coal-tar sealant abrades and moves from parking lots to stream systems. Nevertheless, the location of these parking lots relative to the streams, the large increases in PAH concentrations below the lots, and absence of alternative high-PAH source material in the local catchment suggest that parking-lot sealants are the most probable source of the PAHs in the downstream sites.

The observed increases in PAH concentration immediately below coal-tar-sealed parking-lot inflows should be useful for understanding downstream trends in PAH contamination because similar inflows could be present in a wide range of development conditions. Tributary hot spots are likely to be more common and to have much higher concentrations of PAHs than presently recognized PAH problem areas in large receiving waters throughout the USA because of the close proximity of high-concentration parking-lot sources to streams with small catchments that provide minimal dilution by clean sediments from erosion.

Benthic macroinvertebrate response

A significant treatment effect was found for ∼½ of the biological metrics, including taxon richness (Table 3), and densities of 9 taxa (Table 4) in both pool and riffle habitats. Taxon richness was probably the most robust of the metrics used in our study. It often is used as a master variable for measuring community health because it has broad ecological implications (Resh and Jackson 1993, Rosenberg et al. 1997). Losses of sensitive or keystone species can have significant effects on ecosystem function and can lead to a wide range of indirect effects, such as trophic cascades, which can result in degradation of more-resistant species at lower trophic levels (Casper 1994, Fleeger et al. 2003).

Density generally is not used as a metric in bioassessments because of the high natural variability and patchiness of benthic macroinvertebrate communities (Allan 1995, Barbour et al. 1999, Karr and Chu 1999). However, density was used as a measure of community viability in our study because of the documented toxic potential of PAHs (Catallo and Gambrell 1987, Beasley and Kneale 2002, Den Besten et al. 2003, USEPA 2003). The lower macroinvertebrate densities in pools and riffles downstream of parking lots could indicate toxic PAH effects, such as mortality or reduced fecundity, at these sites. The significant downstream decrease in density in 2 types of habitat is striking when one considers the wide variation in TPAH concentrations at the study sites. Sediment TPAH was above the PEC at 2 sites (Barton and Walnut) and well below the PEC at 3 sites, but upstream-to-downstream effects were observed at 4 of the 5 sites. The relationship between PAH toxicity units and change in density in the pool habitat also was striking (Fig. 4B) because the increase in toxicity units corresponded with large decreases in density. Streams with <1.0 ESBTU had minimal density changes (losses of <50 ind./L of sediment), whereas the 2 streams with >1.0 ESBTU had large density losses from upstream to downstream (>250 ind./L). Our data set is very small, but our results suggest that the PAH hot spots in at least 2 of the study streams are having toxic effects on the benthic community in these sediments.

One potential response to perturbation is a relative increase in density or richness of more tolerant species as more sensitive species are extirpated. Several oligochaete species, particularly Lumbriculus variegatus, have the ability to biotransform and degrade PAHs, therefore they are potentially less toxic to the organism (Leppanen and Kukkonen 2000). In our study, the densities of oligochaetes increased at downstream sites in 4 of 5 streams. We also expected densities of Chironomidae to increase at downstream sites because many taxa in this family have high tolerances for poor environmental conditions (Merritt and Cummins 1996, Barbour et al. 1999). However, overall chironomid density and densities of at least 6 individual taxa decreased at downstream sites. Den Besten et al. (2003) also documented decreases in Chironomidae abundances in the Holland Deip as PAH concentration increased. Chironomidae bioaccumulate PAHs, and their exposure is heavily influenced by sediment quality and their feeding habits (Bott and Standley 2000). Chironomus tentans biotransforms fluoranthene, the most dominant PAH in our stream sediments, to a more toxic metabolite rather than a less toxic form (Schuler et al. 2004). The results of our study suggest that Chironomidae may be an important indicator of PAH effects in field sediment studies, and their tolerances should be investigated more closely.

The TPAH concentrations in our study streams were typical of the Austin area, where values range from low (<1 mg/kg) to relatively high (32 mg/kg). TPAH concentrations at sites below parking lots were higher than at sites above them in all study streams, but biological effects were not observed at the site with lowest TPAH concentration (Wells Branch, 4 mg/kg). The downstream Wells Branch site had the same parking-lot runoff effects, including hydrological pulses and any non-PAH chemical constituents associated with this land use, as the other study streams, strengthening the hypothesis that PAH toxicity was the variable most strongly influencing the biological degradation observed. This hypothesis is further supported by the clear relationship between biological degradation and levels of PAH toxicity at the downstream sites (Fig. 4A, B).

Other environmental variables

The study streams differed substantially with respect to environmental variables other than PAH level (e.g., catchment area and % impervious cover). The negative effect of PAH-contaminated parking-lot runoff was obvious despite the inherent variability of the streams. However, our conclusions are supported by correlations between biological degradation and PAHs in sediment rather than by manipulation of PAH levels; thus, potential confounding effects of other within-stream variables must be considered. Physicochemical variables (pH, dissolved O₂, temperature, and conductivity) varied very little among sites, but conductivity was slightly (17 μS/cm) higher and temperature was slightly (1.7°C) lower at the downstream sites than the upstream sites. The differences in both of these variables were small when compared to natural temporal variability or instrument precision. HQI was slightly lower at the downstream sites than the upstream sites, but the difference was very small considering expected precision of this qualitative method (Hannaford and Resh 1995). Moreover, the magnitude of biological degradation between upstream and downstream sites was not correlated with HQI, and all sites were within the suboptimal or optimal categories, suggesting they were capable of supporting an acceptable biological condition.

Measures of sediment quality other than PAHs did not indicate any values of concern or important upstream-to-downstream patterns. Pb concentrations was slightly higher at downstream than upstream sites, but all values were well below TEC. Increases in metal concentrations at downstream sites might have been related to runoff from parking lots (Maltby et al. 1995) but did not correspond to the increases in PAHs or to the biological degradation at the downstream study sites. Percent impervious area in riparian buffers was slightly higher at downstream than at upstream sites, and the fact that the study sites were selected on the basis of the presence of a parking lot adjacent to a stream biased the downstream sites toward higher impervious area. However, neither parking-lot area nor % impervious area explained biological degradation at the downstream sites. Considering the complex mix of urban stressors on the benthic communities in these streams, it is likely there were other variables that contributed to the degradation observed at our downstream sites. Nevertheless, ESBTUs explained the observed biological degradation better than any other measured variable. As is typical of the sediment chemistry of most urban streams in Austin and in the nation (Lopes and Furlong 2000), PAHs were the most common pollutant in the sediments in our study and were found at levels closest to acute toxicity criteria.

The authors of a number of pertinent studies have concluded that sediment-bound PAHs, particularly those from pyrogenic sources, might not be as bioavailable as would be predicted on the basis of concentrations derived with traditional laboratory extraction methods and national guidance associated with those concentrations (Paine et al. 1996, Volkering and Breure 2003, Thorsen et al. 2004, Jonker et al. 2005). However, growth and development of the model amphibian Xenopus laevis were negatively affected when larvae were exposed to coal-tar sealant at nominal concentrations similar to those in our study streams (Bryer et al. 2006). The significant negative response to the medium exposure (30 mg/kg TPAH) agrees with national guidance (PEC = 22.8 mg/kg TPAH), but the results from the low exposure (3 mg/kg TPAH) suggest that amphibians may experience chronic effects of exposure to PAH concentrations well below guidance values (Bryer et al. 2006). In our study, the site with the lowest toxicity (Wells Branch, 0.79 ESBTU) showed little to no apparent biological effect of PAHs, and this result supports the EPA's ESBTU as a measure of potential toxicity (Table 1). However, community changes were apparent at 4 of the 5 study streams despite the fact that only 2 of the streams had values above the PEC (Barton and Walnut). This result suggests that PEC guidance might not be sufficient to predict toxicity. Our experimental design precludes statistical verification of individual stream effects. However, our results and those of Bryer et al. (2006) suggest that sediment-bound PAHs, particularly those from coal-tar sources, are causing effects at, and possibly below, guidance values.

We have shown that PAH concentrations in stream sediment downstream of coal-tar-sealed parking lots can be significantly above background concentrations and that those concentrations could be a primary factor in the biological degradation observed in the 5 study streams. Our results, when combined with those of other recent work on coal-tar sealants (Mahler et al. 2005, Bryer et al. 2006), provide a compelling link among PAH contamination in urban receiving water bodies, biological degradation, and parking-lot sealants.

Recommendations

Our study focused on the range of PAH values commonly detected in streams in the Austin area. We evaluated a newly identified source of PAHs, coal-tar-sealed parking lots (Mahler et al. 2005), and coal-tar sealants are used extensively in the USA. Verification studies are needed to evaluate the relationship between PAH hot spots and parking lots in urban streams in other parts of the country and to document the geographical scope and extent of the effects observed in our study. Van Metre and Mahler (2005) documented strong increases nationally in PAHs in urban lake cores over the past 20 y. It is logical to assume that recent suburban growth, with its parking-lot-intensive development, is contributing significantly to these trends. The city of Austin's practice of sampling sediment in small urban drainages and sequential upstream tracing of contaminant sources led to the documentation of remarkably high concentrations of PAHs and eventually resulted in the identification of coal-tar sealant as an important source of PAHs in the environment (City of Austin 2005). We recommend monitoring sediment and biological communities in tributaries to water bodies in which sediment PAH concentrations are increasing to identify PAH hot spots, compare hot spot concentrations with receiving-water concentrations, and evaluate potential biological degradation.

We attempted to explain the magnitude of PAH contamination at the downstream study sites with spatial data. Neither total area of sealed parking lot nor its proximity to sampling locations were significantly correlated with PAH concentrations in the sediments at the downstream sites. A complex mix of age of sealant applied, amount of traffic on lots, local rainfall patterns, flow paths, local stream hydrology, and sediment deposition patterns probably contributes to high variability in the movement of PAHs from parking lots to stream sediments. All of these factors are potential avenues of subsequent research. Our study focused on the biological responses to these PAH hot spots rather than on the specific physical and chemical processes that created them. Understanding the processes that create hot spots will require far more sites and replication of sites under different local conditions. Research on the fate of PAHs from sealant runoff may be particularly useful when designing specific best-management practices (BMPs) that could address this problem through remediation, structural and regulatory pollution controls, and catchment protection programs. The city of Austin has chosen to ban the use of coal-tar-based pavement sealants. This regulatory BMP is designed to stem a 40-y trend of increasing PAHs in its local receiving waters and to address high PAH concentrations identified in tributaries.