This article evaluates the role that different levels of control for combined sewer overflows have in addressing the recreational water quality objectives of the Toronto Inner Harbour of the Toronto and Region Remedial Action Plan. Three models are used to establish the predictive methodology: the Infoworks model for the combined sewer service area, the Hydrologic Simulation-F model for the remainder of the watershed, and the MIKE 3 computer code to evaluate Lake Ontario response to control. Each model was calibrated with E. coli densities observed respectively in sewer discharges, instream, and in the Inner Harbour. Two indices are used to evaluate the response of water quality in the Inner Harbour – fraction of the surface area achieving Blue Flag status, and portion of the swimming season (June to August) above recreation objectives. Analyses of control options led to the recommendation that virtual elimination of combined sewer overflows (one overflow per season control strategy) should be pursued, rather than the lower level of control of 90% volumetric control, which is the minimum provincial environmental requirement. Implementation of priority projects for improving water quality along the Toronto waterfront, including the Don River and Central Waterfront project, are integral to delisting Toronto as a Great Lakes Area of Concern.
A major source of degraded water quality in the Toronto and Region Area of Concern (AOC) is polluted stormwater runoff and combined sewer overflows (CSOs), which contain a mixture of stormwater and untreated sewage, and which are discharged from outfalls into Toronto’s waterways after heavy rains or snowmelts. Historically, this has been of particular concern in the Don River and Toronto’s Inner Harbour (Howell et al., 2018; RAP, 1989a).
The Wet Weather Flow Master Plan (WWFMP) (see MMM, 2003) was developed based on the principles of managing wet weather flows (WWF) on a watershed basis, recognizing rainwater as a valuable resource, and with a hierarchy of management practices and controls, starting with “at source”, followed by “conveyance” and finally “end-of-pipe” controls (D’Andrea et al., 2004).
The WWFMP (2003) included a 25-year implementation plan valued at $1 billion (CAD), which identified projects and initiatives for implementation in five year periods, with the first priority being the improvement of water quality along Toronto’s waterfront and at the City’s beaches. This priority supports actions identified within the Toronto and Region Remedial Action Plan (RAP), under the Canada-Ontario Agreement respecting the Great Lakes Basin Ecosystem and the Great Lakes Water Quality Agreement, and is considered integral to improving water quality conditions and delisting Toronto and Region as a Great Lakes AOC. Other key elements of the WWFMP’s 25-year implementation included watercourse restoration, remediation of basement flooding, municipal operations, greening new development, and public education.
An Environmental Assessment (EA) Study (MMM, 2012), the Don River and Central Waterfront Project (DR and CW project) was completed in 2012, to address recommendations from the WWFMP (2003) related to management of CSOs. The DR and CW project integrates wet weather flow (WWF) management systems to capture and treat stormwater discharges and CSOs from all the combined sewer outfalls to the Lower Don River, Taylor-Massey Creek (a tributary to the Don River) and Toronto’s Inner Harbour (D’Andrea et al, 2004; Appendix, Figure S1 and Figure S2 [available in the online supplementary files]).
The original RAP problem definition for beach postings as outlined in the Stage 1 RAP report entitled, Environmental Conditions and Problem Definition (RAP, 1989a) was: frequent beach postings as a result of stormwater and CSO contamination. Specific remedial actions were identified (RAP, 1994) which addressed water quality concerns related to beaches and to CSOs within the AOC. The DR and CW project addresses the effect of WWF on E. coli densities in the lower Don River and the Inner Harbour from discharges originating in the combined sewer service area (Appendix Figure S1). In addition to addressing impairments in receiving water quality within the Toronto and Region AOC, the DR and CW project will improve conditions for multiple additional impairments including aesthetics, eutrophication and toxic contaminants. The emphasis in this paper is on recreational water quality as reflected by E. coli densities, and as a surrogate for both whole body contact (e.g. swimming within the open water), as well as incidental contact (e.g. sailing, kayaking, and dragon boating, and wind surfing).
This article presents the modeling methodologies used in the EA study and evaluates the role that different levels of control of CSOs have in addressing the recreational water quality objectives in Toronto’s Inner Harbour.
Three levels of CSO control are presented in this paper for the System Concept Plan (Appendix, Figure S3):
Base Case - Existing Conditions for 2031 Population Growth
90% volume capture applied to the Base Case
One CSO on average, per season applied to the Base Case.
The Ontario Ministry of the Environment, Conservation and Parks (MECP) 90% volume capture procedure F5-5 requirement (MECP, 2016), involves interception and treatment of 90% of wet weather flows generated in the combined sewer system, and is equivalent to control requirements of the US EPA and of European jurisdictions. In terms of RAP targets, the control level of one CSO per season, is interpreted as virtual elimination of CSO’s to the Lower Don River, and Inner Harbour. Based on statistical considerations and local variability from combined sewer catchment to catchment, the criteria of one CSO per season reflects a statistical average across all outfalls for a typical (average) year - 1991 was used in this assessment.
The Base Case also used assumptions about the effects of implementation of source and conveyance controls on discharge volume and quality from the sewer systems, including: (1) Mandatory Roof Downspout Bylaw; (2) Increase tree canopy cover across the City; (3) Green Roof implementation; and (4) Increase opportunities for exfiltration systems in the conveyance system, where soils are suitable. These source and conveyance controls along with the 2031 population estimates were applied to all three CSO control scenarios. 2031 population data is considered, as it influences the characteristics and dry weather flow rate of sewage, which is diluted by rainwater and causes CSOs.
The improvement in receiving-water quality due to control efforts was assessed using changes in indices related to E. coli densities. Two indices were used in the assessment: (i) portion of the swimming season in the Inner Harbour above recreational water quality objectives (100 E. coli 100 ml−1) and (ii) fraction of the surface area of the Inner Harbour achieving Blue Flag status.
Blue Flag is an internationally recognized eco-label awarded to beaches that meet or achieve criteria/standards within four areas: water quality, environmental education, environmental management, and safety and services. One imperative criteria for a beach to be awarded Blue Flag status is that a beach’s water quality have an E. coli density equal to or less than 100 E. coli 100 ml−1, 80% of the time. Eight of the City’s 11 beaches are routinely awarded Blue Flag status. As such, this study adopted the criteria as the basis for one of the Inner Harbour water quality indices (defined as the “Blue Flag” index).
The Provincial Water Quality Objective for E. coli density for swimming at designated beaches is 100 E. coli 100 ml−1, calculated as a geometric mean of 5 samples collected within a one month period (MECP, 1994). For recreational water quality at designated beaches, Toronto Public Health compares the 100 E. coli 100 ml−1 objective value to a rolling geometric mean of 10 samples collected at 5 locations at a specific beach on two successive days. For purposes of this study, these two indices are based on the seasonal E. coli density at each station relative to the 100 E. coli 100 ml−1 value. The season is defined as from June 1 to Sept 1 of a specific year (92 days).
Modelling framework used in this assessment
This study used a similar suite of models to those used in the WWFMP study (D’Andrea et al., 2004). The U.S. EPA’s Hydrologic Simulation Model: HSP-F (Bicknell et al., 2001) was used to continuously simulate hydrologic processes and pollutant generation and transport processes for the separated sewer catchments and rural areas within the watershed (Appendix, Figure S2) upstream of the combined sewer area (MMM, 2012). The Infoworks Model (Wallingford, 2006; MMM, 2012) was used to create a detailed model of the combined sewer system and run in continuous mode for water quantity and quality evaluation in the Don River and Inner Harbour watersheds. Discharges calculated by the Infoworks model were used as a time series input to the HSPF model for the Don River or to the MIKE3 model.
The Danish Hydraulic Institute (DHI) MIKE–3D model (DHI, 2002) was used to predict lake circulation patterns and water quality impacts along the waterfront and within the Inner Harbour (MMM, 2012). The whole lake version of the model was run and internally linked to nested representations resulting in a finer scale model for the Toronto waterfront. The whole lake model resolution of 2430 m uses a 3:1 nesting with smaller grids based on 810 m, 270 m and 90 m grids (Appendix, Figure S4). The waterfront model utilized the time series of flow and associated water quality concentrations from the HSP-F watershed model for the Don River and from the combined sewer area Infoworks model as inputs.
Model calibration studies were undertaken by comparing model calculations for flow and water quality in storm water discharges, the Don River prior to its discharge, and the Inner Harbour, building on the calibration procedures documented in D’Andrea et al. (2004). The HSPF and the Infoworks models were used to estimate the flow characteristics frequency duration curve and total volume of annual runoff in the Don River (MMM, 2012), obtaining similar results to the WWFMP study (D’Andrea et al., 2004). The Event Mean Concentration (EMC) approach was used (D’Andrea et al., 2004) for stormwater runoff from catchments with separated watersheds, while the pollutant concentrations in combined sewer overflows were calculated based on the mixing of stormwater entering the combined sewer system with raw wastewater flowing in the sewer system. The EMC values are applied to events when they occur, while dry weather values are used for runoff from stormsewers during the inter-event period.
The E. coli calibration (Appendix, Figure S5) obtained in the WWFMP study (MMM, 2003) for discharges from the City’s rivers and creeks to Lake Ontario formed the basis for the calibration of the watershed models (HSPF, Infoworks) in this study. E. coli levels in the Don River watershed were elevated above those of the other watersheds (Etobicoke Creek, Mimico Creek, Humber River, Highland Creek, Rouge River) discharging through the City to Lake Ontario. This is caused by the large combined sewer service area which discharges to the Don River. Model calibration studies for E. coli in the Inner Harbour are described below.
The seasonal flow volumes and E. coli densities discharged to the Inner Harbour calculated by these watershed models for two years (2007 and 2008) are summarized in Appendix Table S1. The E. coli density in the river blend dry weather and wet weather flows from all upstream areas, including separated stormwater systems and CSO discharges from the combined sewer service area (Appendix Figure S1a). E. coli density in discharges to the slips are larger than the river because the slips receive direct overflows whereas CSO discharges to the river are effectively diluted by upstream flows.
Receiving water (Toronto Inner Harbour) focused studies
Current velocities (speed, direction) were measured at three locations (Eastern Gap, Western Gap and Gibraltar Point) using Acoustic Doppler Current Profilers (ADCP), to provide data for calibrating the hydrodynamic portion of the MIKE3 model. Appendix Figure S7 provides field measurements for the Western Gap station, for the two-month monitoring period. The ADCP measured current velocities at different depths in the water column of the lake from near the lake bottom where it was moored, using the acoustic Doppler principle. Temperature was collected at the ADCP deployed depth.
A sensitivity analysis of the number of vertical layers and other model parameters is provided in MMM (2012). Lake Ontario is over 250 m deep, with the thermocline extending down to about 60 m at the time of fall over-turn; many trials (MMM, 2012) have shown that adequate representation is required of the upper 60 m while the area below the thermocline can be simulated as a single layer of water. A summary of the model calibration parameters is presented in Appendix (Table S2). At a whole-lake scale, the seasonal variation in Lake Ontario water levels and the inflow of the Niagara River are also included.
The MIKE3 model uses a flow and concentration time series of substances as input data from each discharge location, to calculate loadings entering the Inner Harbour and the adjoining Coastal Zone of Lake Ontario. The HSPF and Infoworks models were used to compute loadings for each of the 53 discharge locations along the Lake Ontario waterfront between the Humber River and Bluffers Park inclusive for three meteorological years. Field data for 2007 and 2008 were used for model calibration studies. 1991 was used as the reference year for comparison of the CSO control options; 1991 represents a typical meteorological year (D’Andrea et al., 2004), and was also used to design the CSO interception system. For discharges to the waterfront beyond the Humber River or Bluffer’s Park, the loadings from the WWFMP (MMM, 2003) study were used as the representative time series.
The Toronto Inner Harbour was sampled weekly for E. coli during 2007 and 2008 at 17 locations (Appendix, Figure S6), to create a data set for calibrating the Inner Harbour E. coli model. Hydrologically, 2007 was relatively a dry year while 2008 was relatively a wet year. The sampling period was for the “swimming season” (1 June to Labour Day); the index “percent of time the location is above the PWQO for E. coli (100 E. coli 100 ml−1)” was used to assess the field data.
Calibration of Lake Ontario hydrodynamic model
The comparison of model simulations with field observations (i.e. model calibration studies), uses qualitative summaries of the degree of fit observed using graphical methods (a graphical comparison of model calculations with field observations) followed by statistical summaries. For the measured ADCP and predicted current speed and direction time-series in the Western Gap, the model speed predictions show good response to events, albeit the amplitude does not always reach the same high levels of speed as the observations (Appendix Figures S7a). The model predictions for directions (Appendix Figures S7b) show good correlation with the ADCP data, albiet there are events and periods when the ADCP data is opposite to the predictions. As only two months of data (October – November 2007) are available, this study limited itself to undertaking model calibration studies to characterize circulation from the Lake Ontario Coastal Zone to the Inner Harbour.
In terms of statistical techniques, Schwab (1983) and Beletsky et al. (2006) have pioneered the use of the Fourier Norm or Fnorm, a test of variance between observations and model predictions, focusing on currents at a lake wide scale, whereas this study focuses on the coastal (i.e. nearshore, shallow) zone. The Fnorm test is more severe than the simpler RMSE test of single vector components, because both components are tested synoptically. The following Fnorm values were found in this study: Western Gap is 0.86; Gibraltar Point is 0.96 and Eastern Gap is 1.00 for the Fall 2007 monitoring period. A Fnorm value of zero would be a perfect match between the model and the observed data, whereas a value of 1 means the error between vectors is the same as the vector magnitude, and a value greater than 1 means poor agreement with the model. This range of values for Fnorm have been observed in other hydrodynamic studies of the Toronto waterfront, and are attributed to the variability in Lake Ontario currents, winds, temperature in the nearshore zone, and the complexity of the shoreline and the shallow depth of water. It is concluded that the model’s strength is in looking at general trends.
For the open lake, there is good agreement between temperature observations and model simulations (Appendix, Figure S8), at a current meter deployment and water treatment plant intakes located in 15–20 m depth of water. The model captures the timing of upwelling events, as well as the magnitude of changes in lake temperature.
Influences on hydrological residence time of the Inner Harbor
On a season long basis, the model predicts that the net circulation through the Inner Harbour is from the Humber Bay through the Western Gap into the Inner Harbour, and out the Eastern Gap. The exchange flows are driven both by nearshore currents and especially (Hlevca et al., 2018) by upwelling events. The main direction of the Don River plume is initially to discharge west toward the middle of the Inner Harbour, but to mix with Inner Harbour water and to be redirected toward the Eastern Gap. Lake – bay flows are approximately a factor of 10 to 20 larger than river discharges from the Don River (Table 1). Lake – bay flows into the Inner Harbor dilute the pollutant loading from the Don River by about an order of magnitude. The summer season hydrological residence time of the Inner Harbour (Table 1) was approximately a week in 1991 and 2007.
Calibration of the E. coli Inner Harbour model
Seasonal data summary of 2007 and 2008 observations for the period 1 June to 31 August, as well as the model predictions for the 17 Inner Harbour locations, are provided in the Appendix (Table S3).
Examination of the Inner Harbour E. coli data for 2007 and 2008 together with the modelling studies lead to the observation that there are three groups of stations (Appendix, Table S3). The station locations near the Western Gap and Offshore form one group, the locations along the north shore and the Don River form another and the final group is around the Eastern Gap. For 2007 the Western Gap and Offshore area were assessed as having an average of approximately 30% of the swimming season above the PWQO guideline, while the Don River Group average about 55% of the time and the Eastern Gap group averaged 20%. During the wetter 2008 year, the grouping is not as distinct, although the Don River group is higher overall.
The model decay rate (removal rate constant) for E. coli was adjusted during the calibration phase to match the observed data. On a station by station basis, the assessment of the model’s predictions was problematic because the model over predicted the E. coli Index at some locations and under predicted posting times at other locations (Appendix, Table S3). Similar over and under trends in agreement between observations and model simulations were observed using a sensitivity analysis of the E. coli removal rate constant.
The E. coli calibration is analyzed as scatter plots in Figure 1 for 2007 (a relatively dry summer) and in Appendix Figure S9 for 2008 (a relatively wet summer). The largest values of the E. coli Index are located in the areas influenced by the Don River plume while the lower values of the E. coli Index are located near the Western Gap, which is especially influenced by Lake Ontario inflows through the Western Gap. These plots incorporate the cumulative effects of model calibration for the hydrodynamic model as well as in-lake biochemical processes. The results show a good agreement, indicating that a quite good calibration has been achieved, using the criterion of the degree of fit observed graphically.
Over a swimming season (defined as 1 June to 1 September, 92 days), the observed index is 38% while the model forecast index is 34% for 2007 (a relatively dry year). The observed index is 55% while the model forecast index is 62% for 2008, a relatively wet year. Since the model forecasts agree with field observations for these two years (Appendix Table S3), this further strengthens the conclusion that a good model calibration has been achieved.
Effects of combined sewer overflow control on inner harbour water quality
The spatial changes of the “Blue Flag Index” in the Inner Harbour to the two CSO control levels are provided in the Appendix (Figure S10). The Base Case (Appendix Figure S10a) indicates that there is a limited extent of the Inner Harbour (ca 5% of its surface area) which meets Blue Flag criteria (that the water area is swimmable 80% or more of the swimming season). The MECP Procedure F 5-5 intercepts sufficient volume of CSO and stormwater discharges that the Blue Flag criteria increases to ca 50% of the Inner Harbour’s surface area (Appendix Figure S10b).
The second control, one overflow per season or less, intercepts an additional volume of CSO and stormwater discharges increasing the portion of the Inner Harbour which meets Blue Flag criteria to ca 70% of the Inner Harbour’s surface area (Appendix, Figure S10c).
Since the lake-harbour exchange flows (Table 1) are approximately an order of magnitude larger than inflows from the Don River plus discharges to the Inner Harbour (Appendix Table S1), the exchange flows plus in situ biochemical removal processes of E. coli reduce the blended influent E. coli densities by greater than an order of magnitude. These conditions are represented spatially by the model forecasts of Appendix Figure S10a, which indicate that only a small portion of the Inner Harbour meets recreational water quality criteria during the reference year. The one overflow per season control strategy provides a significant improvement in Inner Harbour water quality (Appendix, Figure S10c).
Implications for decision making and the Remedial Action Plan
Comparison of control options
The minimum level of CSO control that is required by environmental agencies is represented by the second control option (90% Volume Capture – see Methods section). The third control option addresses the level of control expressed in one of the RAP’s Specific Actions (Stormwater Action 12: Reduce and virtually eliminate CSOs to receiving waters, Toronto RAP, 1989b). The environmental benefit of these two control levels in the Inner Harbour was presented spatially in the previous section; this section examines an additional basis – an economic basis, as a tool for decision making, especially as it relates to the RAP Specific Action.
An assessment of the improvement in Inner Harbour water quality as a function of storage volume and its associated costs was undertaken (MMM, 2012). Figure 2 presents the index “portion of the swimming season above recreational objectives” as a function of WWF Storage costs. The storage costs are proportional to the volume of storage required for each of the two CSO control options (F5.5 and 1 overflow per season). On Figure 2, the first dot (Y-axis) represents the null option (zero CSO control), the middle dot (X-axis value of ca 0.7) represents the F5.5 control level, and the third dot (X-axis value greater than 0.8) represents the 1 overflow control level. Appendix Figure S11 provides the Blue Flag index (Y axis) as a function of WWF storage costs (X Axis).
Both curves (Figure 2, Appendix Figure S11) indicate that an increase in volume of storage (and hence the cost of storage) provides a proportional benefit for improving receiving water quality. But even with essentially almost complete capture of CSO’s (dot on each figure, X-axis value greater than 0.8), there is still a substantial portion of the Inner Harbour area which does not meet Blue Flag criteria, due to E. coli levels in stormwater discharges in the Upper Don River watershed (both within the City of Toronto and municipalities outside the City of Toronto).
The DR and CW EA Study (MMM, 2012) recommended that a CSO control level of one CSO or less per season be the level of CSO control for discharges to the lower Don River and Inner Harbour. This recommendation received very favourable feedback from the public and stakeholders during the EA study’s intensive public consultation process. As it is related to delisting the Toronto and Region AOC, this control level is equivalent to “virtual elimination of CSO discharges.” The storage requirements for one CSO control strategy was increased by about 30% to provide a safety factor which addresses factors such as environmental variability, including climate change. The main benefitting receiving waters of this project will the Lower Don River and the Inner Harbour and all the recreational activities that presently occur within the Inner Harbour or could potentially occur in the future when the DR and CW project is fully built and commissioned.
As there are presently no designated beaches within the Inner Harbour, recreational activities that will benefit from this project include users of the open waters (canoeing, sailing, etc.), Blue Flag beaches located on Lake Ontario shoreline of Toronto Islands which are influenced by the E. coli density in waters from the Inner Harbour which flow toward them, and potentially future designated beaches within the Inner Harbour. Intercepting CSOs will reduce flottables and improve the aesthetics and public health of the water within the slips of the north shore of the Inner Harbour. The three beaches which are currently not Blue Flag Beaches, are mainly impacted by E. coli discharges from other rivers (Marie Curtis Beach impacted by Etobicoke Creek; Sunnyside Beach by the Humber River; Rouge Beach by the Rouge River), and are at a large enough distance from the Inner Harbour that they will not significantly benefit from the DR and CW project.
Implementing the Don River and Central Waterfront Project
The DR and CW project is the City’s most significant WWFMP end-of-pipe project to improve water quality and is an implementation priority within Toronto Water’s 10 year wet weather flow capital program. The EA Study for the DR and CW project was completed in 2012
The project consists of three integrated tunnels (22 km in total) connected to 12 underground vertical storage shafts, 27 connections to outfalls, seven off-line storage tanks, an Integrated Pumping Station at the Ashbridges Bay Wastewater Treatment Plant, and a new wet weather flow high-rate treatment facility to be built on a future landform project south of Ashbridges Bay. The DR and CW project will also help service future growth and provide redundancy for the Coxwell Sanitary Trunk Sewer. A schematic of the DR and CW project components is provided in the Appendix (Figure S12).
This $1.5 billion project (City of Toronto, 2017) is being implemented in stages over 25 years and once it has been fully implemented, it will virtually eliminate the release of CSO discharges into the Don River, Taylor Massey Creek, and Toronto’s Inner Harbour, as well as reduce polluted stormwater discharges. Interception of these discharges will also reduce the associated loadings of nutrients, suspended solids and associated heavy metals. The ultimate impact of this project on improved water quality in these waterbodies will be significant and will also contribute to improved aquatic recreational uses and fish habitat.
Progress in advancing the implementation of the DR and CW project includes completion of preliminary design for the system of tunnels, shafts, and off-line storage tanks in 2015. In the City of Toronto capital budget, construction of the first phase, the Coxwell Bypass (Appendix Figure S12), is scheduled to start in 2018 and take ca seven years.
This article addresses the specific remedial action identified for the Toronto and Region AOC – to reduce and virtually eliminate CSOs to receiving waters, and is based on the completion of an EA study for the DR and CW project in 2012. This article uses modelling tools to forecast the recreational water quality improvement expected in Toronto’s Inner Harbour to different levels of CSO control. Two indices related to E. coli levels in Toronto’s Inner Harbour are used as the basis for assessing recreational water quality improvements.
Calibration studies for the hydrodynamic and E. coli model for the Inner Harbour and coastal zone of Lake Ontario achieved a satisfactory agreement between model forecasts and field data. Several factors including the complex shoreline of the Toronto Inner Harbour’s environs mean that it is often beyond the abilities of the hydrodynamic model to resolve the currents at the resolution used in this set of simulations. It is concluded that the model’s strength is in looking at general trends.
The calibration of the water quality model was assessed in two ways. Firstly, a comparison of observed spatial variations within the Inner Harbour showed a good agreement with model forecasts. In addition, the model’s summer season forecasts bracket field observations for a relatively dry year (2007) and a relatively wet year (2008). It is concluded that a good E. coli model calibration has been achieved.
On a season long basis, the model predicts that the net circulation through the Inner Harbour is from Humber Bay through the Western Gap into the Inner Harbour, and out the Eastern Gap. Lake-bay flows are approximately a factor of 10 to 20 larger that discharges from the Don River (Table 1), and the summer season hydrological residence time of the Inner Harbour was approximately a week in 1991 and 2007. These exchange flows plus insitu E. coli removal processes, reduce the Inner Harbour E. coli levels by greater than an order of magnitude relative to the influent E. coli levels.
An analysis of alternative levels of CSO control lead to the conclusion that one overflow per season control strategy should be used to implement the DR and CW project. This selected control strategy is based on several factors including the areal extent of improvement forecast for the Inner Harbour. This control strategy plus lake-harbour exchange flows and internal biochemical removal processes are the basis for the forecast improvements.
The authors acknowledge the assistance of the many colleagues and co-workers involved in this study, including Pat Chessie, Allen Li, James Yacoumidis, Lou Di Gironimo, Brian Buchanan, Tracy Ehl, Frank Quarisa, Ellen Leeste, Mahesh Patel, Dr Howard Shapiro, and the RAP team, including Shari Dahmer, Laud Matos, and Nadine Benoit. The significant assistance of the editorial staff of the journal is also gratefully acknowledged.
Supplemental data for this article can be accessed on the publisher’s website.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/uaem.