Urban storm water infiltration systems are not reliable sinks for biocides: evidence from column experiments

Soil properties

Stone content

The stone content ranged from (15,pm ,8%) (w/w) at V.18 to (44,pm ,13%) (w/w) at F.3 (Fig. 1a, Table 1). These differences between sites may partly be due to different sources of the raw material used to create the SIS. Further, the stone content increased with depth within the first 15 cm (V.18) and 10 cm (W.10), but remained approximately constant over depth at F.3. Hence, the stone content in the upper layers of the older SIS (W.10 and V.18) was lower than in the lower layers. These depth-related differences at each site may be related to time-dependent developments within the SIS. In the uppermost layers of V.18 and W.10, stone content was comparatively low probably due to input of fine mineral and organic particles by storm water. For the oldest SIS (V.18), this assumption is supported by the field observation of soil material lying on a bricked stone border near the inflow within the SIS.

Bulk density

The bulk density in the upper layers of the different SIS increased in the following order: V.18 < W.10 < F.3 (Fig. 1b, Table 1). At V.18, we observed the strongest change with depth from (1.0,pm ,0.1,hbox {g},hbox {cm}^{-3}) (0–5 cm) to (1.5,pm ,0.1,hbox {g},hbox {cm}^{-3}) (15–20 cm). In contrast, we observed almost no depth-dependent change of bulk density at the youngest site of F.3 ((1.6,pm ,0.2,hbox {g},hbox {cm}^{-3})).

In samples of the older sites of V.18 and W.10, low bulk densities in the uppermost layers compared to deeper layers were probably caused by the activity of macrofauna, an intensive rooting, a higher organic carbon (OC) content and the input of strongly sorted fine material. The older the SIS, the stronger the effect of these factors.

At F.3 the bulk density was relatively high. Here, we supposed an uniform compaction of the soil layer under the topsoil during construction. This assumption was supported by the observation of redox characteristics (iron-red stains next to grey iron-depleted areas) in the soil at approximately 25 cm depth caused by the lack of oxygen due to accumulating water⁴⁵ in compacted soil.

Texture

The mean texture of fine soil at all SIS was very similar: 57–80% (w/w) sand, 16–34% (w/w) silt and 5–9% (w/w) clay, since similar textured materials were used for construction to guarantee solute retention and sufficient hydraulic conductivity³¹. Average clay contents of all SIS were within acceptable ranges of Best Management Practice (BMP) claimed by ATV-DVWK A-138 ((<10%) (w/w) clay)³¹.

The clay fraction at the oldest SIS (V.18) decreased slightly with depth from 12% (w/w) at 0–5 cm to 7% (w/w) at 20–25 cm (Figure S4) while it remained constant with depth at W.10 and F.3. Since a homogeneous texture after SIS construction can be assumed for each SIS, a higher clay content in the uppermost layers suggested fine particle deposition by storm water, as was observed by⁴⁶.

pH

The pH was above 7.0 at nearly all sites and depths (Fig. 1c, Table 1) and was thus within the range of BMP³¹.

The pH increased slightly in the following order: V.18 < W.10 < F.3. These slight pH differences among the SIS could be either due to age (stronger acidification of older soils) or due to different initial pH values of the topsoil materials.

Furthermore, the pH increased with depth at all sites, which was stronger at older sites (V.18 and W.10) than at F.3. These larger depth-gradients at older sites indicate a relationship between pH and age of the sites: older SIS were more acidified in the upper layers than the youngest SIS (F.3) due to the input of acids by precipitation, the production of (hbox {CO}_{2}) by soil respiration and the release of protons and organic acids by roots⁴⁵.

Organic carbon content

The OC content and the slope of depth-gradients within the upper 15 cm increased from (1.15,pm ,0.3%) (w/w) at the youngest SIS to (2.85,pm ,0.9%) (w/w) at the oldest SIS (Fig. 1d, Table 1). This indicates that, in general, OC accumulated during soil development in upper soil layers while the stone content, the bulk density and the pH decreased. Overall, the OC contents of all SIS were within the BMP ranges³¹.

Soil development and spatial variability

All soil parameters described above and especially the development of depth gradients indicate continuous soil development (alteration of important soil properties such as pH, texture, OC content and distribution) in the investigated SIS due to physical and chemical processes and sediment input with storm water. Consequently, biological activity also increased, as shown by the observation of macrofauna at V.18 (earthworms) and W.10 (ants) while no macrofauna was observed at F.3. Overall, age of SIS is reflected in the soil properties after only a few years.

We assumed that the spatial variability of the soils within each SIS was relatively small, as probably only one material was used to build up the upper soil layer. As described above, a low spatial variability within each SIS was confirmed by the low variability of analyzed soil properties of the soil samples ((hbox {n}=4)). Therefore, we assumed in the present study that one soil column with a wide diameter (20 cm) at each SIS can be considered representative of the SIS inflow area. To determine the influence of flow paths on solute transport we thus considered a column experiment without repetition as sufficient to further investigate the differences between the SIS. This is common for such complex column experiments with undisturbed soils (see e.g.^47,48,49) since their performance is very time-consuming.

Percolation experiment

Saturated hydraulic conductivity and water flux

The saturated hydraulic conductivity K(_{s}) ((hbox {cm},hbox {h}^{-1})) differed strongly between the soil columns (Table 2). It was lowest in the youngest SIS F.3 ((0.18,hbox {cm},hbox {h}^{-1})), four times higher at W.10 ((0.66,hbox {cm},hbox {h}^{-1})), and approximately six times higher in the oldest SIS V.18 ((0.98,hbox {cm},hbox {h}^{-1})). Water flow velocities at all sites and time steps were in the optimal range for urban swale systems defined by BMP ((0.006,hbox {cm},hbox {h}^{-1}< hbox {q} < 6,hbox {cm},hbox {h}^{-1}))³¹.

The water flux decreased continuously with time in all soil columns: in W.10 and V.18 approximately 1.5 times and in F.3 approximately 3 times (Table 2, Figure S7 and Fig. S8). This could be due to clogging of pores in the (0.45,upmu hbox {m}) nylon membrane by colloids⁵¹ or dissolved organic matter⁵². In our experiment, the existence of colloids was likely as they could be mobilized by a decreased ionic strength of the infiltrating deionized water⁵³.

Tracer breakthrough

Breakthrough curves (BTCs) of (hbox {Br}^{-}) and (hbox {Cl}^{-}) reached 90–100% at the outlet of all soil columns after percolation of four pore volumes (PV) (Fig. 2). Both ions showed the fastest breakthrough of all tracers suggesting only a weak interaction with the soil matrix; as in other studies, they can be considered as conservative tracers^{54, 55}. Nevertheless, in our experiment, this assumption was challenged by the observation in the youngest SIS (F.3): after the percolation of one PV, only 30% (hbox {Br}^{-}) and (hbox {Cl}^{-}) broke through. Under the assumption that the transport of (hbox {Br}^{-}) and (hbox {Cl}^{-}) is only determined by convection, diffusion and dispersion, 50% of the solutes should have reached the outlet after one PV was exchanged⁵⁶. The 20% deficit indicated that weak adsorption of (hbox {Br}^{-}) and (hbox {Cl}^{-}) occurred in F.3 as was also observed by other researchers, e.g.⁵⁷. In contrast, at W.10 and V.18 substantially more than 50% (hbox {Br}^{-}) reached the outlet after one PV, indicating fast transport of (hbox {Br}^{-}) in macropores. However, since the PV of the individual soil columns are only rough estimates, they are subject to uncertainty. Therefore, the deviation from 50% solute breakthrough at a PV equals one could also be partly explained by the uncertainties in the calculation of the PV.

With the exception of UR at V.18, BTCs of UR and SRB did not reach 100% after four PV. Instead, their breakthrough maxima were (51,pm ,2%) (F.3) and (79,pm ,1%) (W.10) for UR; and (25,pm ,1%) (F.3), (70,pm ,1%) (W.10), and (90,pm ,2%) (V.18) for SRB. Maxima of BTCs at F.3 and W.10 increased in the following order: (hbox {SRB}< hbox {UR} <hbox {Br}^{-} approx hbox {Cl}^{-}). At V.18, this order slightly differed ((hbox {SRB} < hbox {UR} approx hbox {Br}^{-} approx hbox {Cl}^{-})). These results indicated a stronger retention of the fluorescent tracers compared to (hbox {Br}^{-}) and (hbox {Cl}^{-}) most likely due to their higher adsorption affinity⁵⁸. The higher retention of SRB compared to UR was in accordance with literature^{59, 60} and linear sorption coeffients ((hbox {K}_d)-values) for SRB that were about twice as high compared to UR (Table S1 and Figure S11) for all SIS. However, the high discrepancy between the retention of conservative tracers ((hbox {Br}^{-}), (hbox {Cl}^{-})) and UR at F.3 revealed that UR ought to be considered as a non-conservative tracer with respect to its sorption properties. This was described previously for UR application in soils with comparable OC content, texture and pH^{44, 61, 62}.

We observed a faster breakthrough of all substances in W.10 and V.18 than in F.3 as indicated by higher slopes and higher maxima of BTCs (Fig. 2). After the percolation of one PV, 29% (F.3), 73% (W.10), and 93% (V.18) of the (hbox {Br}^{-}) inflow concentrations were reached. Due to the small differences between the chemical soil properties of the SIS and the Kd values of UR and SRB (Table S1), we would have expected similar tracer BTCs. Yet, this was not the case; tracer retention decreased with age from the youngest to the oldest SIS (F.3 (<<) W.10 < V.18). These differences could not be explained by chemical soil properties alone. When they would dominate solute transport, sorption, and thus solute retention should be highest at V.18 where OC content (1.9%) and clay content (8.8%) were highest and pH was lowest (7.1) (Table 1). From this we deduce that solute retention in W.10 and V.18 was controlled less by sorption than by preferential flow in macropores. Existing studies support this finding as they related fast transport of SRB^{42, 43} and UR⁴⁴ in undisturbed soil cores to the existence of preferential flow, too. Furthermore, our results imply the following: when preferential flow in macropores dominates water and solute transport, the differences in the BTCs of substances with different adsorption affinities decrease. This effect led to the largest differences between tracer BTCs in F.3, where matrix flow was dominant, and to the smallest differences between tracer BTCs in V.18, where preferential flow was dominant. In F.3, the matrix flow was high and solutes had a longer contact to a larger number of sorption sites thus intensifying sorption losses. In V.18, the opposite was true: the contact of the solutes to sorption sites was brief as they bypassed the soil matrix with the preferential flow. Additionally, the number of sorption sites may be lower in the preferential flow regions⁴². Furthermore, similar BTCs of conservative and non-conservative tracers with different adsorption affinities indicate strong transport in macropores by preferential flow^63,64,65.

Brilliant blue staining

Roughly 65% of F.3, 38% of W.10 and 8% of V.18 (Figure S10) were stained. Along the cut profile large blue patches were visible for F.3 while only single blue fingers were observable for V.18 (Fig. 3). These macropores, especially in V.18, were most probably caused by earthworms^{45, 66}. In fact, approximately ten individuals were observed on the laboratory columns after saturation from bottom to top. Earthworm holes were visible on the soil surface as well as on horizontal and vertical cuts through the soil column (approximately 2–3 mm, Figure S9). At W.10 several ants were observed that also could cause macropores^{67, 68}. In contrast, no macrofauna was visible at F.3. Overall, brilliant blue patterns supported the assumption of an increasing fraction of preferential flow in the older SIS W.10 and V.18.

Brilliant blue staining also revealed partial water flow on the soil column sidewall⁶⁹ that bypassed a part of the soil column F.3 and entered the soil again further down (Fig. 3). This partial sidewall flow could potentially explain the fast initial (hbox {Br}^{-}), (hbox {Cl}^{-}) and UR breakthrough in F.3 where 5–10% of the tracers was measured in the first sample taken after 30 min (Fig. 2). However, this effect was limited since concentrations remained only initially constant and increased again after approximately 0.5 PV. Interestingly, this effect could not be observed for the stronger adsorbing SRB. Possibly, SRB was initially retained and therefore could not be measured in the first percolates. Similar transport characteristics of UR and SRB in the beginning of the experiment were observed e.g. in the study conducted by⁷⁰.

Biocide breakthrough

At all sites, the biocide breakthrough increased in the following order: OIT < terbutryn < diuron. This order followed the polarity of the substances as indicated by the (hbox {K}_{OW})-values (Table 3) which can be interpreted as a rough estimate for the sorption affinity of non-polar substances to organic matter^{71, 72}. Furthermore, except for diuron at W.10 and V.18, biocides were more strongly retained than the tracers in all SIS. This observation may be explained as follows: the walls of macropores caused by earthworms may be enriched with soil organic matter⁷³. These macropores may provide hydrophobic sorption sites (organic matter) for non-polar substances such as biocides and less for the polar tracers.

Analogous to UR and SRB, the retention of biocides was highest in F.3 (Fig. 2), followed by W.10 and V.18 and was negatively related to the OC content of all SIS (Table 1). Similar to the tracers, the large differences in BTCs of biocides between the SIS cannot be explained by chemical soil properties but rather by soil structure. In F.3, biocides had more intense contact with the soil matrix than in W.10 and V.18 due to a higher fraction of matrix flow. As a result, biocide retention was highest in F.3 despite the lowest OC content. Thus, the influence of biocide properties on their retention was strongly reduced by preferential flow in macropores^{<a data-track="click" data-track-action="reference anchor" data-track-label="link" data-test="citation-ref" aria-label="Reference 74" title="Larsson, M. H. & Jarvis, N. J. Quantifying interactions between compound properties and macropore flow effects on pesticide leaching. Pest Manag. Sci. 56, 133–141.
https://doi.org/10.1002/(SICI)1526-4998(200002)56:23.0.CO;2-N}

(2000).” href=”https://www.nature.com/articles/s41598-021-86387-9#ref-CR74″ id=”ref-link-section-d2224e4144″>74, 75.

Although biocide degradation is unlikely due to the short duration of the experiment (approximately 13–28 h) and due to immediate cooling and freezing of the samples after sampling, these processes could not be completely ruled out. However, if a small amount of biocide degradation occurred, then concentrations would have been reduced by the same amount in all samples. The differences in biocide concentrations between the individual soil columns should not have been affected.

The rapid leaching of organic pollutants through macropores, has also been observed in several studies for pesticides in agricultural soils^{38, 39}. More biocide breakthrough in W.10 and V.18, especially that of diuron in V.18 is in line with findings of⁷⁶, who reported leaching of diuron due to complexation of diuron with dissolved organic matter (DOM). This indicates that DOM molecules have a dual role on pollutant transport in soils: they compete for adsorption of the organic pollutants and enhance their mobility by complexation. But their leaching is enhanced by preferential transport in macropores following intense rainfall⁷⁷ or in strongly structured soils with connection to shallow groundwater^{<a data-track="click" data-track-action="reference anchor" data-track-label="link" data-test="citation-ref" aria-label="Reference 78" title="Brown, C. D., Hollis, J. M., Bettinson, R. J. & Walker, A. Leaching of pesticides and a bromide tracer through lysimeters from five contrasting soils. Pest Manag. Sci. 56, 83–93.
https://doi.org/10.1002/(SICI)1526-4998(200001)56:13.0.CO;2-8}

(2000).” href=”https://www.nature.com/articles/s41598-021-86387-9#ref-CR78″ id=”ref-link-section-d2224e4172″>78.

Important factors of fast solute breakthrough in urban SIS and implications for storm water treatment

Despite similar chemical soil properties between the SIS, W.10 and V.18 showed a much faster tracer and biocide breakthrough compared to F.3. Brilliant blue patterns and saturated hydraulic conductivities indicated prevailing preferential flow conditions in macropores in SIS W.10 and V.18 that lead to fast solute breakthrough. Macropores can be caused by biological activity^{41, 79}, which is supported by the observation of the highest activity of macrofauna (earthworms and ants) in the older SIS W.10 and V.18. Therefore, in our experiment the age of SIS is closely related to macrofauna activity. This observation is in agreement with results from⁸⁰ who found an increasing earthworm population density with increasing age of different urban landscapes. Similar observations were made by⁸¹ in technosols but they concluded that the existence of an initial addition of topsoil is required, which was the case in all SIS of our study.

Beside the age of a soil, the abundance of earthworms is also controlled by bulk density and soil depth⁸². In SIS V.18 bulk density was lowest and soil depth highest, both factors favoring high earthworm abundance. Thus, different relevance of preferential flow in the different SIS may also be due to differences in the construction of the SIS. If a SIS is built with a less thick topsoil layer or a high bulk density, or if other macrofauna-inhibiting factors play a role, the development of the macrofauna may be restricted, making the development of preferential flow paths in aging SIS less likely. Furthermore, climate can influence the population dynamics of earthworms^{83, 84}. Therefore, we conclude that factors supporting higher activity of macrofauna (e.g. earthworms and ants) in SIS may also lead to faster solute breakthrough.

Additional influence on macropore formation can be exerted by the vegetation of the swales, directly by plants with thicker or deeper roots colonizing as development of the swales progresses, or indirectly by influencing the diversity of invertebrates⁸⁵. Due to necessary time for colonization and establishment of species also the age of swales may be linked to the biodiversity. In addition to vegetation and soil structure, the shape of the swales could also have an influence on diversity of invertebrate and thus on macropore formation. Reference⁸⁵ supposed that biodiversity may be higher in round systems than in linear ones. However, we observed rather the opposite, suggesting that age played a greater role for macrofauna diversity and macropore formation than swale shape in our experiments.

Our experiment has shown that SIS can retain biocides by adsorption when the substances have sufficient contact with the soil matrix. Inversely, our study also suggests a decrease in biocide retention capacity in urban SIS due to preferential flow pathways caused by an increasing biological activity and changing soil properties already after 10–18 years of operation. However, biological activity in SIS is also desirable due to several benefits. Higher biodiversity of vegetation and soil fauna may enhance degradation of several organic pollutants and higher infiltration rates, e.g. due to a higher number of macropores, may guaranty fast water infiltration even after long-term operation. Overall, we recommend regular monitoring of the pollutant retention capacity of SIS to detect its reduction in time, which could be done, for example, by tracer experiments.

One approach to address this problem is to treat storm water with special adsorbent materials^{86, 87} before it enters the swales, or integrate additional adsorbent layers in the swales. However, these options are complex and expensive and, furthermore, a large portion of urban runoff often infiltrates diffusely and does not reach the swales at all. Therefore, a more sustainable approach, is to avoid biocide pollution at the source^{88, 89}, which would allow a targeted urban water management by SIS that preserves urban groundwater quality.

Source: Ecology - nature.com