Missing Compensation: A Study of Compensatory Mitigation and Fish Passage in Georgia

The widespread degradation and loss of wetlands and streams in the United States (Carpenter et al. 1998; Dudgeon et al. 2006) has prompted increased interest and opportunity for managers to engage in policy creation and permit review with the intention of minimizing and offsetting anthropogenic impacts. The U.S. Army Corps of Engineers (USACE) issues permits for activities subject to regulation under Section 404 of the Clean Water Act (formally known as the Federal Water Pollution Control Act [FWPCA, 1972 as amended]; 33 U.S.C. 1344) that result in the discharge of dredge or fill materials in waters of the United States (e.g., construction of stream–road crossings). Over the past 2 decades, increasing human activities have resulted in a dramatic increase in the number of permits issued for these activities (Hough and Robertson 2009).

Whereas the USACE and U.S. Environmental Protection Agency (USEPA) retain the regulatory authority to regulate the dredge or fill of waters of the United States by administering Section 404 of the Clean Water Act, the USACE is principally responsible for permitting. The U.S. Fish and Wildlife Service (USFWS) and National Marine Fisheries Service engage the USACE through the auspices of the Fish and Wildlife Coordination Act (FWCA 1934, as amended) and U.S. Endangered Species Act (ESA 1973, as amended), but the role of these natural resource agencies is principally an advisory one intended to aid in the evaluation of impacts to fish and wildlife and to identify alternatives before permitting. Although the Clean Water Act is a national policy (FWPCA, 1972), it is administered and applied differently across the 38 USACE districts. Each USACE district employs a more localized interpretation and administration of the Clean Water Act (Doyle et al. 2013). Regional conditions exemplify the heterogeneous application of national policy, where USACE districts develop regional policies that ensure that adverse effects of nationwide permits (NWPs) are minimal, and are specific to the needs of a particular state or subgeography.

Dredge and fill-related impacts to waters can be permitted through individual permits, letters of permission, standard permits, and most commonly, activities that are authorized under NWPs. Approximately 40,000 activities across the United States are reported and approved under NWPs by the USACE annually (USACE 2016). NWPs authorize activities that have minimal individual and cumulative adverse environmental effects to wetlands (including streams and lakes; USACE 2007a). The regulatory goal of the Section 404 program is “no overall net loss” of area, functions, and values of wetlands and waters of the nation's remaining wetlands base (FWPCA 1972; USEPA 1990). To achieve the no-overall-net-loss goal, avoidance, minimization, and mitigation measures are required. The USACE and USEPA oversee compensatory mitigation, which is intended to offset unavoidable adverse impacts after avoidance and minimization measures have been employed. Compensatory mitigation is provided through mitigation banks, in lieu fee programs, or permittee-responsible mitigation, and is required if the extent of impact meets a threshold identified by each USACE district (Hough and Robertson 2009; see USACE 2004 and Table S1 for an example of regional conditions from the USACE Savannah District). Consequently, markets have emerged through which one-third of all aquatic compensation is offset at offsite mitigation banks where restoration, enhancement, and preservation activities take place (Wilkinson and Thompson 2006).

Although much of the recent scientific research focuses on the efficacy and success of compensatory mitigation projects to offset permitted impacts (e.g., Bendor 2009; Lave et al. 2008), relatively less attention has been paid to the congruency among permitted impact amounts, actual impact amounts, and the degree to which impacts receive compensatory mitigation (Mbobi 2005). Our study specifically assesses whether requirements intended to avoid and minimize impacts are being followed, and how the amount of compensatory mitigation for unavoidable impacts compares with the amount of impact that was permitted.

Hundreds of new crossings are constructed annually under NWPs, thereby facilitating evaluation of regulation effectiveness. We evaluated the effectiveness of efforts to minimize and compensate for stream impacts using newly installed stream–road crossings as a case study because crossing construction occurs nationwide and multiple states and USACE districts have requirements that are aimed at minimizing crossing-induced impacts to streams and fish passage (e.g., Washington, Maine, Massachusetts, USACE Savannah District, and USACE Wilmington District). In this study area, impacts to stream–road crossings are tracked and compensated at offsite mitigation banks. Therefore, stream–road crossings offer an opportunity to evaluate multiple stages of the regulatory process with spatial replication, including permitting, implementation, and compensation (see Figure S1 for a more thorough description of permitting process).

Stream–road crossings (hereafter termed crossings) are important in natural resource management because they contribute significantly to the fragmentation of habitat that is important to imperiled species (USFWS 2000). A survey of culverts in an Appalachian watershed, for example, indicated that 97% of culverts were impassable to Brook Trout (Salvelinus fontinalis), resulting in isolation of 33% of Brook Trout habitat (Poplar-Jeffers et al. 2008). Culverts can impede fish movement by: 1) widening stream channels, thereby creating water depths at normal flows that are too shallow for fish to swim (Wargo and Weisman 2006); 2) creating falls at the culvert outlet (outfalls) at heights and velocities that exceed the fish's ability to leap into the culvert (i.e., leap height); and 3) restricting stream flow, thereby artificially increasing water velocities throughout the culvert length and exceeding the maximum sustainable swim speeds of fishes. High water velocities within culverts also can increase bed sediment export from the culvert, thereby eliminating water velocity refugia. Since barrel culverts are designed to pass floods (5–100-y recurrence interval flows), the cross-sectional area at the culvert is usually larger than the bankfull cross-sectional area of the stream, thereby interrupting flow and sediment transport patterns, which can impede fish passage (Wargo and Weisman 2006). Although significant attention is given to consequences of poorly installed crossings on fish passage, effects of crossings on other aquatic organisms (e.g., salamanders) are largely unknown (Ward et al. 2008).

We assessed whether requirements intended to minimize impacts are being followed and whether the amount of permitted impact corresponds to the required amount of compensation using crossings in the Savannah District of the USACE as a case study. In an effort to evaluate the effectiveness of requirements to avoid and minimize impacts to fish passage, this study evaluates whether newly constructed crossings conform to the design specifications in the NWP regional conditions (USACE 2007b), whether requirements result in fish-passable stream–road crossings, and whether the stream length affected by crossing construction corresponds to the stream length proposed for compensation and the stream length for which mitigation credits were purchased. In summary, this analysis is devoted to evaluating whether regionally developed policies to minimize impacts are implemented as intended and the extent to which stream impacts receive compensation.

In June 2007 the USFWS began a concerted effort to evaluate preconstruction notifications (PCNs) from permittees who sought to construct stream–road crossings under a variety of NWPs (Figure S1). Evaluations included stream–road crossings in Georgia that were permitted and constructed following the publication of the 2007 NWP regional conditions (hereafter termed regional conditions; USACE 2007b). Because the focus of the evaluation was on stream–road crossing design, we eliminated a subset of unsuitable PCNs from consideration (e.g., stream buried or transformed into a storm-water detention pond). The resulting set of projects was unevenly dispersed throughout the state, with a majority in the Atlanta metropolitan area. To maximize efficiency of the assessment, we selected for field evaluation only proposed new crossings in the 12 counties with the most PCNs, resulting in a sample size of 45 crossings dispersed across a 9,505-km² area (Figure 1). Construction had been completed at 24 crossings, which represent our sample sites where we conducted additional surveys. Since the regional conditions are specific to Georgia streams, we used the Georgia Department of Natural Resources Field Guide for Determining the Presence of State Waters that Require a Buffer (frequently used by agencies and individuals involved in permitting; GADNR 2006) to determine perennial, intermittent, or ephemeral status. We conducted all crossing surveys in this study between December 2007 and April 2009, more than 72 h after precipitation and within 1 y of construction.

The regional conditions allow new culverts to contain multiple barrels only if the additional barrels are elevated above streambed elevation and designed to pass floodplain flows (regional conditions E.3 and E.6, Table S1). We counted the number of barrels for both pipe and box culverts and measured their dimensions. New culverts in perennial streams must also be embedded to 20% of the culvert height (regional condition E.4, Table S1). To calculate embeddedness, we made all measurements at a point located at 50% of the culvert maximum width. If multiple barrels were present, we measured embeddedness in the barrel with the most inflow. We calculated the embeddedness as a percentage of the total culvert diameter (for pipe culverts) or height (for box culverts) that was filled by sediment (Figure 2).

The regional conditions also require that the nonembedded area of the culvert (hereafter termed culvert area) equal the bankfull cross-sectional area of the stream channel (hereafter termed channel area; regional condition E.3, Table S1). We measured the culvert area for box culverts and the circular reinforced concrete and corrugated metal pipe culverts, and subtracted the embedded area to determine the nonembedded area of the culvert inlet. For instances in which we encountered multiple barrels (box or pipe) near streambed elevation, we summed nonembedded areas of each barrel to determine total culvert area. We used half of the ellipse area to estimate arch-span culvert area. Area under a bridge included the area beneath the bridge frame that would allow flows to pass, including the channel cross-sectional area.

We conducted all crossing assessments after construction. Before construction, multiple channel units (e.g., riffles, runs, pools, and glides) probably existed at the contemporary culvert location. Because channel area and width can vary naturally depending on channel unit type and the presence of flow obstructions (e.g., woody debris and boulders), channel area and width likely varied within the footprint of the newly installed crossing. Therefore, we compared the culvert, arch-span, and bridge areas with multiple channel area measurements at upstream locations that were not affected by culvert construction to assess whether channel dimensions and pattern were altered (regional conditions E.1.). We spaced cross-section measurements in perennial streams at one-half to one times bankfull channel width, starting immediately upstream of the location that was affected by culvert construction, with a maximum of 10 channel area measurements at each site (Figure 3). Three sites had fewer cross-sections because property boundaries or dense vegetation limited access. We did not make cross-section measurements in intermittent or ephemeral streams either because bankfull identification was uncertain or because the geomorphic features necessary for calculating channel area (bed and banks that are distinguishable from the surrounding landscape) were indeterminate. We measured inlet and outlet wetted channel widths at locations where the stream entered and exited the culvert and compared them with wetted channel widths measured at the same locations as the channel cross-sections.

The above analysis provides managers with two methods of comparing stream geomorphic data with dimensions of the road crossings: a statistically rigorous comparison and a basic ranking method. For each site, we made a statistical comparison between culvert area and mean areas from upstream cross-sections using one-sample t-tests. Because the one-sample t-test evaluates whether a mean from a site (the cross-sectional areas) is different from a single value (the culvert area at the same site), we conducted normality tests for each site independently of all other sites using the Kolmogorov–Smirnov test in SAS 8.3, and made transformations when necessary. We also demonstrated the use of a basic ranking method to evaluate compliance. The PERCENTRANK function in Microsoft Excel© evaluates the relative standing of the road crossing area measurement within the channel area data set. Thus, each crossing area measurement is also expressed as a percentage of the channel area values at its respective site that are smaller than the culvert area. PercentRank values between 0 and 100% show whether the culvert area is trending toward being oversized (> 50%), undersized (< 50%), or equal to the mean channel cross-sectional area (= 50%). Managers can interpret areas between 0 and 100% as conforming to the regional conditions depending on how far they deviate from 50%.

We focused on assessing stream conditions for three fish families of management concern that are likely to occur in the study area: Cyprinidae, Percidae, and Cottidae. Common methods that are used to assess fish passability generally rely on estimates of hydraulic conditions within the culvert (USFS 2008; Coffman 2005), the swimmability of fish, and the potential for fish to move upstream into the culvert outlet. We did not attempt a thorough assessment of all factors that could contribute to a crossing's impassability. Instead, we used three simple measurements frequently used as indicators of passability: water depth, distance from culvert outfall to downstream water surface (i.e., leap height), and the extent of sediment accumulation in the culvert. We used a 5-cm depth threshold to define passability because it accommodates body depth for Cyprinidae, Percidae, and Cottidae (approximately 2.5 cm) and additional depth that is necessary to accommodate dorsal and ventral fins, minimize injury, avoid obstacles, and account for variation in fish size. If either the culvert inlet or outlet were shallower than 5 cm, we determined the culvert to be impassable (per Webb 1975; Powers and Orsborn 1985). We also determined leap heights greater than 10.2 cm (4 in.) to be impassable because 10.2 cm is the height that is generally accepted to impede small-bodied fish movement (Percidae and Cottidae; Coffman 2005). We measured fish leap height as the distance from the culvert bottom to the downstream water surface. Culverts with bed sediment throughout 100% of the culvert length are hypothesized to be more fish passable than those without sediment (Coffman 2005, Poplar-Jeffers et al. 2008) because bed sediment provides for water velocity refugia. We measured and expressed linear culvert length that contained bed sediments as a percentage of the total culvert length. We did not make water depth and fish leap height measurements in intermittent or ephemeral streams because water was largely absent.

We measured impacts to waters of the United States (a prerogative of the USACE and USEPA) separately from impacts to habitat (a concern of USFWS). To measure impacts to waters, we measured the stream length with geomorphic impacts due to culvert construction. We measured stream impacts that included the stream length affected by culvert placement, culvert headwalls, and concrete apron; rip-rap placed in the channel and on channel banks; and obvious alteration of channel dimension and pattern, including channel widening or straightening directly resulting from culvert construction. We calculated the total length of stream that contained impacts and compared it with the proposed stream impact length in the PCN.

All PCNs for which compensation was proposed indicated that mitigation credits would be purchased at mitigation banks. Credits are tracked in the USACE's Regional Internet Bank Information Tracking System (USACE 2010). Each PCN indicated the proposed stream impact length and the number of mitigation bank credits that would be purchased, and we were able to search the mitigation bank credit ledgers in the Regional Internet Bank Information Tracking System to determine if and how many credits were purchased on a project-by-project basis. For comparison purposes, we obtained and summed from the PCNs the affected stream lengths proposed for compensation and compared them with stream lengths that corresponded to the mitigation credits reported in the Regional Internet Bank Information Tracking System. Because regional conditions for the Savannah District indicate that compensation is required when total perennial stream impact length is ≥ 30.5 m (100 ft; USACE 2007b), we identified all sites with measured geomorphic impacts ≥ 30.5 m. We calculated the proportion of stream impact lengths that qualified for compensation and compared it with the actual amount of stream length for which mitigation credits were purchased.

Of the 45 crossings selected for field evaluation, we completed construction and surveys at 24 crossings, including 10 reinforced concrete, 7 box, and 3 corrugated metal pipe culverts. We also evaluated three arch-span culverts and one bridge, although measures of embeddedness could not be made for these bottomless structures. Most of the surveyed streams were perennial, and there was general agreement between the USFWS' and applicants' perennial, intermittent, and ephemeral designation. However, applicants identified three streams as intermittent and ephemeral, whereas the USFWS identified them as perennial (Table 1). Of the perennial sites identified by the USFWS, four (20%) contained an identifiable active floodplain (Table S2), although none of these had barrels installed in the floodplain. Of the perennial sites where floodplain presence was either undetermined (n = 3 sites) or absent (n = 12 sites), two had barrels installed in the floodplain. We placed multiple barrels below floodplain elevation and near streambed elevation at seven perennial sites (Table S2). Of the 15 perennial streams with pipe or box culverts, one was embedded (≥ 20% of culvert height) at both the culvert inlet and outlet, and one was embedded only at the culvert outlet (Table S3). All others were not embedded or were embedded < 20% of the culvert height.

We found that road crossings had significant effects on channel morphology. Road crossing areas were generally larger than the mean of the channel area measurements taken upstream (Figure 4). We eliminated one site from the area and width analysis because of stream access difficulty. Road crossing cross-sectional areas were statistically different from channel area measurements at 16 of the 20 perennial sites, and ranked larger than all channel area measurements at 12 sites (Figure 4), meaning that crossings were frequently sized larger than the channel area (Table S4). Crossing inlet and outlet wetted channel widths were statistically different from width measurements taken upstream at all but four box culverts (Figure 5). Crossing inlet wetted channel widths were either wider (3 crossings) or narrower (10 crossings) than all channel width measurements at upstream cross-sections (Table S4). Crossing outlet wetted channel widths were either wider (seven crossings) or narrower (seven crossings) than the range of wetted channel width measurements taken upstream (Figure 6; Table S4). Differences observed between inlet and outlet channel widths and upstream cross-sections are an indication that channel pattern had been altered by culvert construction.

Using our criteria, 18 of our 24 road crossing sites were impassable barriers for fish. Thalweg depths were frequently less than 5 cm deep at the crossing inlet (12 crossings) and outlet (5 crossings), meaning that most crossings were impassable on the basis of depth criteria. Excluding the three arch-spans and the bridge, 86.7% and 66.6% of crossing thalweg depths were less than 5 cm deep at the crossing inlet and outlet, respectively. Bed sediments covered 100% of the crossing length at one bridge, box, and corrugated metal pipe culvert and two arch-span culverts. The remaining 14 culverts either had little or no sediment accumulation, meaning that they were not passable on the basis of sediment extent criteria. Summing across sites, 56% of the total crossing length surveyed in this study did not contain sediment. Leap height measurements indicated that water surface elevation was the same at the outlet and immediately downstream from the crossing at most sites. Crossing outlets were perched at three sites (5.1, 7.6, and 10.2 cm respectively); however, these heights were less than the threshold (greater than 10.2 cm) that is expected to impede fish passage. Of the 24 crossings surveyed, the only ones passable on the basis of all three passability criteria were the bridge, three arch-span culverts, and one box and one corrugated metal pipe culvert.

The measured total stream impact length was 46.0% higher than the amount of impact proposed in PCNs for perennial streams (Table S5), and 23.7% higher than that proposed for intermittent and ephemeral streams (Table 2). Of the total length of perennial stream affected, 30.6% had mitigation credits purchased to compensate for impacts. However, the amount of stream length for which mitigation credits were purchased was 14.8% less than the amount proposed in the PCNs (Table 2). Although most PCNs proposed impacts less than the 30.5-m threshold required for compensation, observed postconstruction impacts at most sites exceeded 30.5 m (Table 3). Summing the measured stream lengths affected only for sites where measured impacts were > 30.5 m, 90.9% (790.5 m) of all measured impacts to perennial streams qualified for compensation. The perennial stream impact lengths for which compensation should have been required but did not receive compensation totaled 524.3 m (the difference between qualifying and actual compensation in Table 2), and is hereafter termed “missing compensation.” Causes of missing compensation included streams that permittees had identified as intermittent or ephemeral in the PCN and therefore did not pursue the appropriate compensation amount for impacts (28.6%), stream impacts proposed for compensation that could not be accounted for in the Regional Internet Bank Information Tracking System (15.1%), and impact lengths that were longer than proposed in the PCN (56.4%).

Several recent scientific investigations have shown mixed ecological and geomorphic success of restoration projects in the southeastern United States (Teels et al. 2004; Sudduth and Meyer 2006) and nationwide (NRC 2001). Fewer studies have examined the percentage of stream impacts (permitted or otherwise) that receive compensation through compensatory mitigation. Our study shows that perennial stream impacts far surpass the amount of stream length reported to be affected in PCNs, leading to significantly less compensatory mitigation than required. Neither intermittent nor ephemeral streams receive the same degree of compensation for impacts as perennial streams (USACE 2004), despite the ecosystem services that they provide (e.g., denitrification) and the importance of headwater streams to downstream hydrogeomorphic (e.g., sediment supply and hydrologic regime) and ecological functions (e.g., trophic subsidies). Although perennial streams have greater legal protection, only 30.6% of the perennial stream impacts in this study received compensation, even though 90.9% of the impacts qualified. When considering these results in tandem with the critical evaluations of stream restoration and compensatory mitigation success, the “no overall net loss goal” of area, functions, and values of wetlands and waters (Clean Water Act 1972; USEPA 1990) is not achieved for crossing-related impacts to Georgia streams.

There might be several reasons why the regional conditions do not translate into fish-passable structures and result in stream impacts that are not accounted for in compensatory mitigation. First, applicants and engineers might disregard the permit in favor of less expensive installation options, knowing that enforcement is unlikely. Similarly, expense might also be the cause of discrepancies between the USFWS' and applicants' perennial, intermittent, and ephemeral stream designations. Because intermittent and ephemeral streams require less compensation (USACE 2004), there is an economic incentive to designate streams as intermittent or ephemeral. Although we have no direct evidence of dishonesty, these hypotheses warrant investigation. Our results show that discrepancies in stream designation account for nearly 28.6% of missing compensation. However, approximately 56.4% of missing compensation can be attributed to impact lengths that were longer than the proposed impact length in the permit application. In addition to the placement of the crossing itself, stream impacts included rip-rap placed in the channel or on channel banks, and obvious channel widening or straightening as a consequence of culvert construction. These types of adverse impacts were rarely accounted for in permit applications. These results highlight the importance of thorough investigations to account for actual stream impact amounts that otherwise would not be included in evaluations of mitigation project efficacy and the no-overall-net-loss goal of the Clean Water Act.

Survey results indicate that most new crossings had multiple design parameters that did not adhere to requirements in the regional conditions. The regional conditions specify that new pipe and box culverts be embedded and use floodplain culverts when a floodplain is present, but not use multiple barrels otherwise. However, culverts constructed by permittees were rarely embedded, did not have floodplain culverts when floodplains were present, and often had multiple barrels in the stream channel. The regional conditions also specify that crossing area must equal the bankfull channel area, and that channel pattern should not be altered by crossing construction. Compared with unaffected upstream reaches, crossing areas were frequently larger than all channel area measurements, making it unlikely that new crossings met the bankfull channel area requirement. Similarly, wetted channel widths in the crossing inlet and outlet usually fell below the range of upstream wetted width measurements, indicating that the channel was narrowed during crossing construction and that channel pattern was altered. Collectively, these results indicate that most projects did not meet the design requirements in the regional conditions.

The components of the regional conditions that specifically deal with crossings were drafted to improve fish passage in all creeks and rivers throughout Georgia. However, most crossings were not fish passable at baseflow on the basis of multiple fish passability criteria that we used in this assessment. Although fish leap heights were less than the threshold expected to impair small-bodied fish movement into crossings from downstream, shallow water depths and the paucity of bed sediments in most new crossings would likely inhibit fish movement for most small- and nearly all medium- and large-bodied fishes. Similar to the results of Coffman et al. (2005), the bridge and arch span culverts were notable exceptions. These bottomless structures contained natural bed sediments and water depths required for passage of small-, medium-, and large-body-sized fishes.

Fish disperse at base flow and higher flows, meaning that passage through natural and artificial structures should be considered for a range of flows. Hydrodynamic models are flexible in that they can incorporate channel dimensions, crossing dimensions, and fish swimmability for a range of flow rates. However, our conclusions are based on passability assessment techniques often used by managers (e.g., Coffman 2005) and probably reflect the types of passability conclusions that managers are likely to make.

Although the survey sites in this study were located in the Atlanta metropolitan area, we expect that the results from this survey are transferable to other U.S. geographies because stream–road crossings are pervasive across the landscape and many states, regions, and federal districts have requirements intended to benefit fish passage through crossings. Results from this study show that stream impacts associated with crossings are largely missing in compensatory mitigation. This is likely because, on a project-by-project basis, impacts routinely end up being both greater than the amounts permitted and larger than the amounts reported to have been received as compensation at mitigation banks. However, the discrepancy between stream impact length and mitigation amounts may be much larger than calculated in this study. We surveyed crossings because they were a comparable subset of the permitted stream impacts in Georgia that could readily be used to evaluate regulation effectiveness and compensatory mitigation. During the time frame of this study, numerous other impacts were authorized but not surveyed, including complete stream burial under parking lots, buildings, and roads; transformation into storm-water detention ponds; and impacts associated with water and sewer infrastructure construction. Additionally, unpermitted impacts to streams likely occur and are neither viewed nor assessed by managers. Collectively, these additional impacts indicate that the actual amount and proportion of stream impacts that are missing in stream compensatory mitigation may exceed our calculations.

The no-overall-net-loss goal is intended to allow wetland impacts to be avoided, minimized, and mitigated. The USEPA and USACE apply this goal at the national scale, and by default, the scale of the nation's Section 404 program. They acknowledge that each individual permit action may not achieve the no-overall-net-loss goal (USEPA 1990). However, we conducted our study at the program scale and included multiple permit actions approved under NWPs in Georgia. We demonstrate that activities that are permitted under NWPs and the associated compensatory mitigation program neither replace lost acreage, nor do they compensate for the functions that wetlands provide, including fish passage. Our analysis demonstrates that the collective impacts from individual permit actions hamper achievement of the no-overall-net-loss goal at the program level.

Much of the emphasis of contemporary literature on compensatory mitigation focuses on the success or failure of mitigation banks and restoration projects to restore lost area, structure, and function (Turner et al. 2001 and sources therein). Although this research continues to make important contributions, we suggest to both scientists and managers that impacts offset at mitigation banks represent only a small fraction of the total stream impact amount that occurs across the landscape. We suggest that additional emphasis be placed on thorough evaluations of permitted and unpermitted stream impacts to develop a more comprehensive evaluation and understanding of stream loss to ensure that the no-overall-net-loss goal of the Clean Water Act is attained.

Table S1. Army Corps of Engineers Savannah District regional conditions for all nationwide permit culverts.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S1 (133 KB PDF).

Table S2. Number of perennial, intermittent, and ephemeral (Int/Ephem) streams surveyed in Georgia by the U.S. Fish and Wildlife Service from 2007 to 2009 that had an active floodplain, floodplain culverts, and multiple barrels near streambed elevation.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S2 (81 KB PDF).

Table S3. Number of box and pipe culverts that were embedded ≥ 20% of the culvert height at the culvert inlet and outlet, and the percentage of the total culvert lengths across all sites that contained sediment. We conducted surveys in Georgia between 2007 and 2009.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S3 (81 KB PDF).

Table S4. Culvert and channel wetted width and cross-sectional areas in perennial streams. We conducted surveys in Georgia between 2007 and 2009.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S4 (45 KB PDF).

Table S5. Stream designation and impact length and culvert design characteristics for surveyed perennial streams. We conducted surveys in Georgia between 2007 and 2009.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S5 (44 KB PDF).

Figure S1. Process through which typical stream–road crossings are approved and compensated. This diagram identifies details that are pertinent to this study and does not attempt to provide a thorough review of the permitting process. Although multiple Army Corps of Engineers districts have similar processes, regional conditions and other details vary across districts.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S6 (344 KB JPG).

Reference S1. [GADNR] GA Department of Natural Resources, Environmental Protection Division. 2006. Field guide for determining the presence of state waters that require a buffer. Atlanta: GADNR.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S7 (2503 KB PDF).

Reference S2. [USACE] U.S. Army Corps of Engineers. 2004. Standard operating procedure, compensatory mitigation, wetlands, open water, & streams. Savannah, Georgia: Savannah District, Corps of Engineers. March 2004.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S8; also available at http://www.sas.usace.army.mil/Portals/61/docs/regulatory/Mitigation_Comp_SOP.pdf (53 KB PDF).

Reference S3. [USACE] U.S. Army Corps of Engineers. 2007a. Reissuance of nationwide permits, final notice. Federal Register. Citation: 72 FR 11091. Document number E7-3960. March 12, 2007.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S9; also available at http://federalregister.gov/a/E7-3960 (631 KB PDF).

Reference S4. [USACE] U.S. Army Corps of Engineers. 2007b. Savannah district 2007 nationwide permit regional conditions.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S10 (115 KB PDF).

Reference S5. [USACE] U.S. Army Corps of Engineers. 2016. Nationwide permit reissuance.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S11; also available at http://www.usace.army.mil/Media/Fact-Sheets/Fact-Sheet-Article-View/Article/1043655/nationwide-permit-reissuance/ (130 KB PDF).

Reference S6. [USEPA] U.S. Environmental Protection Agency and the Department of the Army. 1990. Memorandum of agreement between the Environmental Protection Agency and the Department of the Army concerning the determination of mitigation under the Clean Water Act Section 404(b)(1) Guidelines. Washington, D.C.: USEPA.

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S12; also available at https://www.epa.gov/cwa-404/memorandum-agreement (61 KB PDF).

Reference S7. [USFWS] U.S. Fish and Wildlife Service. 2000. Mobile River basin aquatic ecosystem recovery plan. Atlanta..

Found at DOI: http://dx.doi.org/10.3996/022016-JFWM-017.S13; also available at http://www.ag.auburn.edu/auxiliary/alcfwru/coalition/mrbrecovery.pdf (339 KB PDF).

The authors of this report thank the Savannah District of the Army Corps of Engineers for their continued commitment to work with state and federal agencies to minimize impacts to natural resources. We also thank Eric Prowell for input during early project phases. Special thanks are extended to Caralyn Zehnder, three anonymous reviewers, and the Associate Editor for excellent comments on drafts. Finally, we thank the USFWS for providing the authors time and resources necessary to survey sites.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.