The lost history of managed retreat and community relocation in the United States

Managed retreat refers to the abandonment of occupied land and the removal or relocation of population and/or infrastructure out of areas subject to repeated flooding, rising sea level, or other natural hazards (e.g., Hino et al., 2017; Siders et al., 2019; Government Accountability Office [GAO], 2020). Synonymous terms include “planned relocation,” “managed realignment,” “climate migration,” and increasingly, “climigration” (Bronen, 2009; cited in Hamilton et al., 2016). Retreat is increasingly being discussed as an alternative to the current de facto strategy, which focuses on engineering protections and rebuilding in place after successive disasters (Scott and Lennon, 2020). Trends in climate-change impacts suggest impending inundation that may compel, among other adaptations, large-scale retreats from at-risk coastlines and river floodplains. For example, Dottori et al. (2018) predict that flood damages globally may increase 160%–240% and flood fatalities by 70%–83% by 2100, even under a modest +1.5°C climate scenario. In the United States, also by 2100, nearly 500 coastal communities (Dahl et al., 2017) and 4.2 million residents (Hauer et al., 2016) may face disruptive inundation in response to conservative predictions of sea-level rise. Managed retreat away from other hazards such as wildfire has also been discussed (Mach and Siders, 2021). The goal of retreat and relocation is to mitigate the proximal threat in a “managed” fashion—that is, to be planned, strategic, and conscious of synergistic issues such as preserving social capital, equity, serving social justice, and maximizing environmental synergies.

One specific variety of managed retreat—and a focus of this study—involves the wholesale relocation of entire towns, villages, or other social or political units. Community relocation involves additional challenges as well as potential holistic benefits over piecemeal buyouts or just removal from at-risk locations. Wholesale relocation offers a potential mechanism to eliminate flood risk, or other hazards, while maintaining and transplanting intact the social fabric of the community. With some predictions of waves of climate-driven migrants in the medium-term future, many researchers and policy makers are now stressing the need to create actionable frameworks, plans, and tool kits for expanded and perhaps large-scale relocations in the future. Recently, the U.S. GAO (2020) reported on past community relocations in the United States—“GAO identified few communities in the United States that have considered climate migration as a resilience strategy, and [just] two…that moved forward with relocation.” GAO and others seem to be unaware of the much deeper U.S. history of retreat and relocation solutions, including tangible experience, hard-won lessons learned, and living expertise in planning and implementing such projects.

The goal of this article is to review this “lost history” of managed retreat in the United States, including all known cases of complete or significant partial relocation identified to date (Figure 1; Table 1). The focus here is relocation due to inundation and related hazards along rivers and coastlines. The Discussion section of the article places flood-related relocation into a broader typology of population migration and hazard mitigation and introduces examples of relocation due to other drivers, both in the United States and worldwide. The empirical focus of this review emphasizes the importance of real-world, context-specific experience with managed retreat, with practical lessons for future retreat projects.

Although “Managed Retreat” is a new concept in scientific and policy discussions, flooding has been threatening U.S. communities throughout the history of the nation. The political, economic, legal, and social contexts of disaster response have changed dramatically over this time. For example, until the 1920s, disaster response was not considered to be part of the role of the U.S. federal government, and both response and recovery were largely managed by local and state authorities, by private groups such as the Red Cross, and left in significant part to the flood victims themselves. However, through that early history, some flood disasters were severe enough to spur dramatic remedies, including several cases of wholesale community relocations off floodplains.

Niobrara has a twofold distinction: (1) as the earliest well-documented example of U.S. community relocation for flood mitigation and (2) for retreating from the floodplain twice (Carter, 1991). The town was first settled in 1856 at the confluence of the Niobrara and Missouri Rivers, serving as a frontier sawmill and steamboat port along Missouri. The population grew to 475 at the time of the 1880 Census (although Carter, 1991 cites 850 residents), with 3 general stores, 2 drug stores, 2 hardware stores, a harness shop, 2 blacksmiths, 5 hotels, 3 physicians, a school, a church, and 2 newspapers. However, in late March 1881, an ice dam formed on the Missouri River impounding the river’s flow and inundating the floodplain for a distance of approximately 160 km (100 miles) upstream (New York Times, 1881). Niobrara was flooded to a depth of approximately 2 m (approximately 6 ft; Carter, 1991). In the flood’s aftermath, and following intense local debate, residents elected to move to a new site 2.4 km (approximately 1.5 miles) away and on higher ground. By late April 1881, buildings were being jacked up and dragged by horse teams uphill to the new town site (Figure 2); by January 1882, most homes and all of the town’s commercial structures were already in place (Carter, 1991).

The 1881 relocation of Niobrara raised the town above several subsequent floods. But the construction of Gavins Point Dam in 1957, about 50 km downstream of Niobrara, impounded Lewis and Clark Lake. The Niobrara River is one of the major sources of sediment along this stretch of Missouri. Between 1957 and today, the accumulation of approximately 5.1 tonnes/year of sediment has transformed the upstream portions of Lewis and Clark Lake into a sand-choked delta. This accumulation raised the water table beneath Niobrara and gradually inundating the town from below (Vecsey, 1971; Carter, 1991). In 1971, 90% of Niobrara residents voted to accept an Army Corps of Engineers plan to transplant the town to a new site still higher above the river (Sterba, 1974). The Corps began the relocation with the idealistic goal of “building a model rural town” (Vecsey, 1971), but as the process dragged on, acrimony grew. Residents and business owners received buyout offers they believed to be inadequate, and buyouts were mandatory (Carter, 1991). Nonetheless, the new town of Niobrara was rededicated in July 1977, at a cost of $14.5 million and with about two thirds of the population the town had beforehand.

Until the railroads had spread throughout the Midwest, the Ohio River was a major avenue for westward transportation and expansion. Shawneetown was settled on the banks of the Ohio River between 1805 and 1810 and Leavenworth in 1818. When Illinois joined the United States in 1818, over a third of the new state’s population resided in Shawneetown and the surrounding “bottomlands” of Gallatin County (Janes, 1942). All of the Ohio River settlements were subject to frequent flooding. Visiting in 1817, English writer Morris Birkbeck described how “Once a year the inhabitants either make their escape to higher lands, or take refuge in their upper stories until the waters subside, when they recover their position on this desolate sand-bank” (1817; quoted in Beyer, 1938, p. 6). Leavenworth, for example, was flooded 10 times prior to 1937 (Bondy, 1938). Shawneetown built its first levee in 1872 (Janes, 1942) and raised that levee several times afterward only to see it fail again and again during successively larger flood events, most notably in 1898 and 1913.

The winter of 1936–1937 had been warm and wet. The Ohio River and its tributaries first overflowed in western Pennsylvania, and the flood wave grew as it moved downstream (Engineering News Record, 1937). The American Red Cross recorded 1,062,661 people rescued; 698,103 refugees housed, some of them for years; 12,860 structures destroyed and another 60,792 damaged; and an estimated $300,000,000 in damages (approximately $5.3 billion in 2018; American Red Cross, 1938). “Only a Dante could describe in verse, or a Wagner in music, the overwhelming character of the flood at its worst and the amount of human misery it caused” (American Red Cross, 1938, p. 9). Leavenworth was located on a horseshoe bend in the river, and flood flow moved 85 of the town’s 145 homes off their foundations (Lilly Library, 2021), of which 20 homes floated away entirely. The town was “almost a total loss” (Bondy, 1938). In Shawneetown, it had been 24 years since their levee had last failed. As water approached the top of the town’s levee, Shawneetown leaders dynamited a hole so that the town would fill gradually rather than in a catastrophic rush of water (Welky, 2011). This bold move was not a mistake—the 1937 flood crest eventually exceeded the height of Shawneetown’s levee by 2.4 m, putting the town under 7.6 m (25 ft) of water (Work Progress Administration [WPA], 1939).

Visiting Shawneetown in 1817, Morris Birkbeck noted “the pernicious adhesion of the human animal to the spot where it has once fixed itself.” Shaking this adhesion and moving Shawneetown off the floodplain had already been discussed in 1913 (Welky, 2011), but circumstances in 1937 were different. First, the 1927 flood on the Lower Mississippi had increased the U.S. federal role in flood control and response. And the 1937 flood occurred amid the New Deal, with the federal government centrally involved in mitigating the impacts of the Dust Bowl, the Great Depression, and the widespread public suffering of the decade (Klein and Zellmer, 2014). The town relocations of Leavenworth and Shawneetown both saw high levels of outside financial support and involvement; New Deal paternalism extended to flood recovery. A July 8, 1937, article in the New York Times attributed the initial suggestion to move towns “to higher ground” to President Franklin Roosevelt (New York Times, 1937).

Following the 1937 flood, the Red Cross essentially adopted Leavenworth. Construction kicked off already in August 15, 1937. The New-Deal-era WPA built streets, sidewalks, parking, water and sewer, and a new town hall. The Red Cross assigned residential lots “in keeping with former locations plus consideration for grouping of houses of similar type and cost in the same neighborhoods” (Bondy, 1938, p. 8). By the end of 1938, New Leavenworth was complete. The Red Cross estimated that moving the town cost 25% more than rebuilding in place (Table 2).

In Shawneetown, discussions about moving the town began with 2.4–3.0 m of flood water (8–10 ft; Beyer, 1938) still in the streets. At first, with most residents living in tent cities, opposition to the town’s move was sparse (Welky, 2011). A new town site was quickly identified 5.6 km (3.5 miles) west of the old town. WPA hired architect Mary Long Whitmore to design New Shawneetown (Figure 3) and construct a scale model to inspire residents through the long process. Parenthetically, the subsequent extramarital affair between Whitmore and WPA project manager, Lincoln Rogers, so scandalized the locals that the pair was driven from town (Welky, 2011). As Shawneetown’s relocation dragged on, tension grew between local leaders and governmental partners (Janes, 1942; Welky, 2011). Locals criticized WPA designers and engineers as “high-salaried outsiders” (Welky, 2011), whereas WPA leaders perceived the local leadership as amateurish, disorganized, and self-serving (Welky, 2011). Nonetheless, New Shawneetown took shape during 1938 and 1939. WPA physically moved 200–250 of the approximately 425 residential structures in Old Shawneetown to the new site. For residents with homes damaged beyond repair, WPA built new houses. The town’s plan included 900 residential and 77 commercial lots, centered on a grassy Central Mall, which was capped in 1939 by a new Gallatin County Courthouse. As in many later town relocations, transplanting businesses was a challenge, as WPA refused to cover the cost of commercial construction. In the end, Shawneetown leaders used discretionary funding to build a row of commercial structures in the core of the new town.

Three communities relocated during the several decades following the 1937 Ohio River flood: Soldiers Grove, Odanah, and North Bonneville. The Soldiers Grove relocation has been extensively documented. In contrast, Odanah—perhaps the largest relocation in U.S. history—was until recently lost to outside memory.

Soldiers Grove was established in 1850 astride the Kickapoo River. Eight severe floods inundated the town during the 20th century (Tobin, 1992). After 1961, the Army Corps of Engineers began planning for a dam upstream, first estimated to cost $15.4 million, which rose eventually to >$55 million (Tobin and Peacock, 1982). In tandem with the dam, the Corps proposed a $3.5 million levee system to protect downtown Soldiers Grove, triple the value of structures protected (David and Mayer, 1984). Although most floodplain residents in most places press for structural projection regardless of the cost, Soldiers Grove rejected the Corps’ levee proposal because of the disproportionate cost, the residual risk of failure, local levee maintenance costs after construction, and disconnection from the river (Kilian, 1975). As one resident put it, “A levee would turn us from a dying town subject to flooding to a dying town protected from flooding” (Chicago Tribune, August 23, 1984). Instead, Soldiers Grove proposed to take the Corps funding and instead relocate the flood-prone approximately half of the town to land off the floodplain. Town leaders purchased a 0.77 km² (190 acres) site approximately 800 m south of downtown (Becker, 1983). However, in 1977, the Kickapoo dam project was canceled, and with cancellation went the proposed funding for Soldiers Grove’s relocation. In 1978, the Kickapoo river sent another 1.5 m of water through downtown. The 1978 flood was yet another blow to the town, but this time, Soldiers Grove had committed itself to relocation and had detailed plans already on the shelf. Construction began during the fall of 1979 (Becker, 1983), relocating the fire station, 2 other municipal facilities, 24 homes, and virtually the town’s entire business district, about 36 structures (St. Louis Post-Dispatch, 1993; Table 3). An additional 12 residences at the margins of the floodplain were elevated or flood-proofed. About half of the new commercial and municipal buildings incorporated passive solar design (Leruox, 1982)—creating what was billed as the country’s “First Solar Village” (Becker, 1983).

In contrast to the broad interest in and extensive research on Soldiers Grove, the wholesale relocation of Odanah away from the banks of the Bad River has been essentially forgotten. Odanah is one of four unincorporated communities on the Bad River Reservation of the Lake Superior Chippewa (Ojibwe) Tribe. At its peak in the late 1800s, Odanah was home to several thousand inhabitants, subject to repeated and often catastrophic flooding. As of July 2018, only one residential structure remained occupied in “Old Odanah.” Local residents remembered the relocation to New Odanah as beginning in the 1960s, but efforts to find archival records documenting the move have, so far, been largely fruitless. The history here is based on the few documentary sources available, discussions with tribal members, and Geographic Information System analysis of historical and modern imagery and mapping.

Odanah was an established village in 1845 (Milwaukee Journal, 1931; E Leoso, written communication, 2018). In 1854, 500 km² surrounding Odanah were set aside for the Bad River Band of the Lake Superior Chippewa Tribe. Odanah grew, particularly during the booming timber industry of the late 19th century, followed by a period of decline (Jordahl and Booth, 2011). Odanah was flooded in 1873 (U.S. Army Corps of Engineers [USACE], 1955), 1896, 1926, 1941, 1946, 1950, 1960, 2016, and 2018. After 1946, the Corps of Engineers developed a plan to relocate the entire town (then with a population of 1,260; USACE, 1955). This plan was endorsed by the Tribal Council and by the Wisconsin State Planning Board (USACE, 1955). However, funding was not secured, and the Corps’ plan was never implemented.

Nonetheless, Odanah did indeed move to the site recommended under the Corps plan (Figure 4), reportedly during the 1960s and financed primarily by the U.S. Department of Housing and Urban Development. Single-family and multifamily housing units and municipal facilities were constructed in “New Odanah.” Most of these housing units were, and remain, government-owned and are leased to Tribal members. As of 2018, all that remained in Old Odanah was the historical Catholic Church and one occupied residential structure.

North Bonneville was located on the floodplain of the Columbia River, but the town was condemned in order to build a powerplant just below Bonneville Dam. North Bonneville originated as a “shanty and tent town” serving workers building Bonneville Dam in the 1930s (Malcom, 1975). The town reached a population of 643 in 1940, declining thereafter (Reinke, 1991). In 1971, the Corps announced plans to acquire the original town site through eminent domain. The Portland District of the Corps had been involved in previous relocations, for example of Arlington, Oregon (1960 population of 643), upstream on the Columbia River. North Bonneville residents received the announcement of their relocation with enthusiasm, seeing it as an opportunity to build “a brand, spanking, new town” (Mayor Skala; in Malcom, 1975). The Corps originally had planned for a much more limited relocation, but North Bonneville lobbied Congress for a more expansive project scope. Tensions between the town and the Corps multiplied, unfolding as a more than 20-year long battle with 18 lawsuits filed (Kenyon College, 2021).

The North Bonneville relocation was burdened by three challenges. The first was the mismatch between the town’s aspirations and the Corps’ willingness to pay for the spiraling cost. The Corps originally planned $7.6 million to acquire private property in the old town and build infrastructure and municipal facilities in the new town; actual costs totaled $45.8 million (approximately $310 million in 2020 dollars) before the Corps declared its efforts completed (Reinke, 1991). The second limitation was the failure to transplant businesses from the old town to the new. Prior to the move, the town had more than 30 businesses, including “gasoline stations, taverns, restaurants, grocery stores, and motels; clothing, variety, furniture, hardware, and other retail outlets; a beauty parlor, a barber shop, and a shoe repair service; a theater, a real estate office, and a bank” (Reinke, 1991, p. 38). Not a single one of these survived the move (Associated Press, 1980; Reinke, 1991). Third, under pressure from local leaders, New North Bonneville was designed for 1,500 people, but only 158 of the original residents made the move to the new town (Reinke, 1991). This overreach left the town unable to cover the cost of its own infrastructure, and a 1991 court judgment of $365,000 against the town forced it into bankruptcy. Over time, newcomers arrived in North Bonneville, and the population rose to 956 by 2010, but financial exigencies have continued to weigh upon the town.

During the summer of 1993, severe flooding on the Missouri and Mississippi Rivers and many tributaries broke stage records at 73 gages (Myers and White, 1993). Three towns completed wholesale moves off the floodplain after the flooding in 1993: Valmeyer, Rhineland, and Pattonsburg. Relocations after 1993 were possible because (1) the extent of flooding spotlighted public attention on flood recovery and (2) the 1988 Stafford Disaster Relief and Emergency Assistance Act established the Federal Emergency Management Agency’s (FEMA’s) Hazard Mitigation Grant Program (HMGP). FEMA authorized its first HMGP-funded mitigation projects in 1990 and its first floodplain property acquisitions in 1992 (in Allagash, Maine). FEMA buyouts have remained an important funding mechanism for both community relocations and other flood mitigation efforts, combined with other funding sources (e.g., Shabman et al., 1997).

Valmeyer remains the “poster child” for wholesale community relocation as a tool for permanently mitigating flood risk. From its founding in 1909 until 1993, Valmeyer was located on the Mississippi River floodplain about 5 km (approximately 3 miles) from the channel. The town had suffered from recurrent flooding in 1910, 1943, 1944, and 1947 (Watson, 1996). After floods in the 1940s, a levee system was built in 1950 to protect Valmeyer and the surrounding floodplain bottom land. In July 1993, record precipitation across the upper Midwest created a flood wave that rolled down the Missouri and Upper Mississippi Rivers, overwhelming levees and inundating floodplains across the Midwest (McConkey et al., 1993). In Valmeyer, around midnight on August 1, the Fountain Creek levee failed, sending a wave of water across the floodplain and into town. Water reached 4.9 m (16 ft) deep, and over 90% of structures in town were substantially damaged (Knobloch, 2005).

With Valmeyer homes inundated and residents housed in FEMA trailers (locally known as “FEMAville”), the first suggestion of a wholesale relocation of the town is attributed (by Knobloch, 2005) to representatives of FEMA and the local planning commission. On September 8, 1993, more than two thirds of residents voted to move to a neighboring cornfield and surrounding woodland 3.4 km away and approximately 120 m (approximately 400 ft) higher than the original town. Within 8 weeks, citizen committees were planning different aspects of the move. In November 1993, a weekend “charrette” partnered regional architects and planners with Valmeyer locals to explore and plan elements of the new town (Brown, 1993; D Knobloch, written communication, July 23, 2018). The town’s new community development corporation signed a sales agreement for the 200-ha (500-acre) farm parcel on the blufftop. Eventually, an additional $3.25 million from bonds and commercial loans were needed to also purchase mineral rights beneath the new town, which were owned by a neighboring quarry. Financing and planning challenges competed with environmental and regulatory oversight, including archeological surveys, endangered species (e.g., Indiana bat), and geological hazards (in particular, karst sinkholes present in the area).

By April 1995, the first family of New Valmeyer moved in, and by the end of the year, the new town’s infrastructure was largely complete. The new Valmeyer school was completed in April 1996 (Knobloch, 2005). In all, about 60% of the original inhabitants made the move to the new town. Between 1990 and 2000, Valmeyer’s population dropped from 897 to 608, but the population subsequently more than doubled by 2010 to 1,263 people. In contrast, only about 25% of businesses survived the approximately 3-year relocation (Knobloch, 2005). Several industrial employers are now in place in the new town area, and even the quarry and limestone mine—originally forced on Valmeyer for the associated subsurface mineral rights—eventually proved to be a boon. The mine has been opened as Rock City, a facility that leases out more than 550,000 m² of naturally refrigerated storage. The total cost of the relocation (Table 4) was about $100,000 (inflation adjusted to 2018) for each of the approximately 530 residents who made the move. Valmeyer has advantages that some rural communities lack, in particular its proximity to St. Louis (approximately 40 km). But much of the town’s success seems to relate to the strong leadership and community engagement that characterized its move.

Rhineland, Missouri, is located upstream of Valmeyer, along the Missouri River. The town was established by German immigrants in 1853 and incorporated in 1896. The town’s population peaked in 1940 at 204, declining to 157 by 1990. Rhineland is located on the north side of the Missouri River, about 3 km from the river channel. Rhineland and the surrounding floodplain are protected by the 15.6-km-long Tri-County levee system. Rhineland had a history of repeated flooding, including in 1951 and 1986 (Missouri State Emergency Management Agency [MSEMA], 1999). In the 1993 flood, the Tri-County levee failed on July 6, putting 80% of Rhineland under as much as 2 m (6 ft) of water (Boonslick Regional Planning Commission [BRPC], 2015). With the town’s levee breached, 3–4 additional flood crests entered Rhineland during 1993.

Following the 1993 flood, with support from its local planning commission, 98% of Rhineland residents voted to move to a 21-ha (52-acre) site on the bluffs approximately 30 m (100 ft) above the town. This land was purchased for $120,000 (Cummings, 1999), and a lottery was held for the choice of parcels (MSEMA, 1999). Parcels were transferred as a land swap, with residents trading their land on the floodplain for land in New Rhineland. Rhineland physically moved about 50 structures to the new site. The first homes were jacked up in May 1995 and carried by truck or tractor to the blufftop at a cost of $8,000–$10,000 per structure. A few businesses also moved from old to new Rhineland; other commercial structures were moved to the edge of the floodplain in order to maintain access to Highway 94, which runs through the old town (MSEMA, 1999).

The total cost of the Rhineland relocation (Table 5) represents a per capita cost of about $80,000 (2020 dollars). Although all property buyouts in Rhineland were fully voluntary, the town achieved nearly 100% removal from the regulatory floodplain. Most subsequent reports (e.g., MSEDA, 1999) emphasized strong social bonds and consensus in “transplant[ing] as much of the social fabric as they could” (Shinkle, 1995, p. 8).

In 1993, Pattonsburg was located along the Grand River, a tributary of the Missouri River. Pattonsburg was originally established in 1845 at a site higher and farther from the river. Ironically, the town moved in the 1870s to gain access to the railroad to a site that proved highly flood-prone. The town was inundated 33 times in the approximately 100 years leading up to 1993 (MSEMA, 1999). Pattonsburg had more than 1,000 inhabitants into the 1940s, declining to 414 in 1990. As of 1984, the town had at least 25 businesses, including a bank, barber, hardware store, watch repair, pool hall, bowling alley, grocery, and others (Tree Climbers Genealogy Society, 2016).

Early on July 6, 1993, Pattonsburg’s levees failed and water surged into town (Wilkinson, 2005). Longtime inhabitants were familiar with the routine of scraping mud out of homes and businesses, and within days, many were replacing drywall, carpeting, and appliance. However, on July 22, the Grand River peaked a second time. Following this one-two-punch flooding, 92% of residents voted on November 3 to relocate (MSEMA, 1999). Total project costs of $12.5 million included about $6 million for the acquisition of 235 flooded properties, with the remainder to develop a 600-acre site approximately 3 km from the river. Goals for the new town included a “Main Street” feel, a new school, a designated church district, a mix of new and old homes, and green space (O’Malley and Crowl, 1998; Schnell and Haddock, 2004). A memorable aspect of the move was that the vacated downtown was used as the set for the 1998 Civil War–era movie, “Ride with the Devil,” including a climactic scene involving the fiery destruction of the town.

The design of New Pattonsburg has been described as “surreal” (Ziegler, 2013), including “a school that looks like it came out of a science fiction movie” (Schnell and Haddock, 2004). However, Pattonsburg was more successful than most other relocation towns in transitioning businesses to the new town site. The new location is credited with boosting the economic health of the town. The dome-like design for the new school resulted from a U.S. Department of Energy visit in 1994, which promoted energy-saving design elements in the new town (Becker, 2012). In terms of flood mitigation, the Pattonsburg relocation was a success, as multiple flood waves during summer 2017 inundated old Pattonsburg.

Although the focus of this article is on wholesale community relocation related to flooding, a quick review of large-scale buyouts not related to relocation is needed for context. In the aftermath of the 1993 Midwest floods, FEMA acquired and removed 7,700 structures (Pinter, 2005). This number includes properties in Valmeyer, Rhineland, and Pattonsburg, as well as piecemeal buyouts throughout the region (Table 6). Some of the buyouts were concentrated and represent—not relocations of communities—but rather removals. For example, in Cedar City, Missouri, 98 of 115 households accepted buyout offers after 1993 (MSEMA, 1999). As of 2011, just 4 residences remained (Lieb and Salter, 2011). Many residents moved to neighboring Jefferson City, which annexed Cedar City. In contrast, some post-1993 buyouts in Table 6 can be considered partial relocations. Both Winfield, Missouri (MSEMA, 1999), and Grafton, Illinois (Pinter and Rees, 2021), built new neighborhoods to accommodate residents displaced from the floodplain. In contrast, Chelsea, Iowa, voted in 1994 to follow Valmeyer, Rhineland, and Pattonsburg off the floodplain and was allocated $7 million for property acquisitions. But “the reality of the move proved much more contentious than the idea, and just a few homeowners ended up [taking] buyouts” (Saulny, 2008). The town was again heavily damaged during flooding in 2008 and subsequently.

In contrast to wholesale relocations, piecemeal buyouts are often opposed by local leaders because they erode the population and local tax base. In some cases, local residents have sought wholesale relocations but instead saw their communities either significantly reduced or removed entirely. Flooding in 2011 along the Mississippi River inundated part of Olive Branch, Illinois, and all of Pinhook, Missouri. Both communities actively pursued relocation. In Olive Branch, relocation planning included visits to Valmeyer as well as Greensburg, Kansas (rebuilt after tornado damage), and a two-day charrette to plan the proposed new town. Olive Branch received $11.5 in federal and state buyout funding but was unable to find funding for the development of their new site, so that many residents left the community. Pinhook was flooded within days of Olive Branch, after the Army Corps of Engineers opened the Birds Point-New Madrid Floodway, in which Pinhook was located. Pinhook was a historical African American town, with traditions dating back to its establishment in 1940 (Song, 2018). The Pinhook relocation recalls the relocation of another small African American town, Allenville, Arizona, which was moved away from the Gila River in 1978 (Perry and Lindell, 1997). Pinhook actively pursued wholesale relocation after 2011 and, like Olive Branch, was able to secure $829,000 for buyouts and demolition, but nothing for moving. In the end, a few Pinhook residents were able to build 7 houses together in Sikeston, Missouri, 32 km away (Bliss, 2018).

The challenges of the 1993 relocations seem to have quelled the national enthusiasm for wholesale solutions. And the largest obstacles remain unchanged today—in particular, a lack of funding and lack of coordinated multiagency support for relocation and for nonstructural flood control approaches in general (GAO, 2020). However, within the last decade or so, growing climate-driven impacts have raised public awareness and raised the pressure to find managed-retreat solutions.

Gays Mills is located along the Kickapoo River and shares the same history of recurrent flooding as Soldiers Grove. After back-to-back flooding in 2007 and 2008 (FEMA, 2008), residents begin planning a move to higher ground. A FEMA Long Term Community Recovery team convened a 2-day planning meeting and a 1-day design charrette in Gays Mills. This planning identified a 0.12-km² (30-acre) site 1.6 km from downtown, which was purchased in June 2009 (Adams, 2009). A full-time relocation coordinator was hired to manage buyouts and the development of the new site. As of 2010, 26 floodplain residential and commercial structures in Gays Mills had been acquired (Adams, 2010). At the new site, 17 townhomes, 8 single-family homes, and several municipal and commercial structures were built (J Henley, written communication, August 29, 2018), including a grocery store, a new $2.8 million “Mercantile” building occupied by several businesses, and village library and offices, community building, and others (J Henley, written communication, August 29, 2018). The approximately $18 million total cost of the Gays Mills project (Table 7) was funded in roughly equal portions from federal and state grants, loans to the village, and private investment (J Henley, written communication, August 29, 2018). Creative financing solutions were employed, for example, rents from tenants of the Mercantile facility are used by Gays Mills to help repay the project’s loans.

Understudied examples of U.S. managed retreat include the towns around Devils Lake, North Dakota. Devils Lake is a terminal lake occupying a closed depression in the Red River of the North watershed. Lake level depends on local climate and land-cover changes upstream. When Minnewaukan was settled in the 1880s, it built a steamboat landing, but falling water levels forced the landing’s abandonment in the early 1900s (Cummings et al., 2012). Lake levels dropped further during the Dust Bowl era of the early 20th century, eventually putting Minnewaukan as much as 13 km from the lakeshore, before recovery in the 1940s to 1970s.

The surface of Devils Lake began a sharp rise in 1992, rising over 9.6 m by 2012 (Figure 5) and inundating agricultural land and at least 700 structures and several communities around the lake. In 1999, FEMA offered buyouts to the 52 households in Churchs Ferry, and all but two accepted, leading to the near disappearance of the town (Hamilton, 2011). Continued increases in lake level brought water into the streets of Minnewaukan, convincing town leadership to purchase 80 acres of land 1.6 km (approximately 1 mile) to the northwest, and $20 million (Forum News Service, 2015) was secured from the U.S. Department of Education, FEMA, Housing and Urban Development, and the Economic Development Agency (Kadrmas, 2012). The Department of Education provided $6 million for a new school. But just as the relocation of Soldiers Grove, Wisconsin, was “saved” by a second damaging flood, the intended relocation of Minnewaukan was been doomed by stabilization of Devils Lake, driven in part by the completion of two artificial outlets to the lake (Anderson, 2018). Imagery dated July 2017 shows just 7 residential structures at the new town site near the school. Construction of a convenience store and gas station, considered imminent in 2013, was never initiated.

“Unequivocal” success (or even “success”) is a high bar for community relocations to clear, but Silex (population 187 in 2010) may be a recent candidate. The town was established in the 1880s on the floodplain of the Cuivre River. In September 2008, the Cuivre River inundated almost all of Silex (Engel 2008). The idea of a community relocation was introduced to Silex by the BRPC, which had played roles in relocating Rhineland and buyouts in Winfield, Missouri, after the 1993 flood. Interest in the buyout proposal grew from a handful of Silex homeowners at the onset, to roughly one third by mid-2009 (Salter 2009), to approximately three quarters by early 2010 (Schlinkmann, 2010), to—in the end—all but 1 of the 74 residences removed from the floodplain (BRPC, n.d.). By mid-2010, a 10.5-ha (26-acre) site had been purchased for $158,800, located roughly 1 km to the east and 25–30 m higher in elevation (Figure 6). A lottery was held, allowing residents to trade their land in old Silex for parcels at the new site (Schlinkmann, 2010). Of the 50 new residential lots, 41 were occupied by residents from old Silex, 5 houses were moved to the site, and 4 lots were occupied by new residents. An additional 23 households flooded in 2008 left the town.

$3.2 million was secured for Silex’s relocation (C Eggen, BRPC, personal communication, September 15, 2018), including from FEMA, U.S. Housing and Urban Development (Community Development Block Grants), Missouri (Department of Social Security Block Grants), and local bonds. Residents received (1) a new lot in exchange for their old property, (2) assessed preflood market value of their old home, and (3) $25,000 in relocation costs (BRPC, n.d.). One caveat to the success of the Silex relocation was that no nonresidential structures were relocated. Although several lots were set aside, lack of funding and bureaucratic challenges prevented the relocation of any commercial structures. All of Silex’s remaining businesses were severely flooded in 2015.

The relocations of Grundy, Virginia, and Martin, Kentucky, resemble those of Soldiers Grove and Gays Mills, but under the management of the U.S. Army Corps of Engineers. Both towns were severely damaged during the flooding of the Levisa Fork River in 1977. Both towns occupy the valley bottoms in the rugged terrain of the southwestern Appalachian Mountains, a region that has declined economically with a declining coal industry. In Grundy, the State of Virginia creatively combined $96 million in highways funds with a larger pool of federally earmarked funding to acquire and demolish 55 structures in the commercial downtown and for the Corps of Engineers to carve a 5.26-ha (13-acre) site from the mountainside on the west side of the river. Local authorities worked with private developers to draw new businesses to the site, in particular, a new Walmart store. Grundy financed $21.8 million for a two-story parking structure, on which the Walmart sits. Other retailers followed Walmart, and the new commercial site is now essentially fully occupied. Although the Grundy project has been criticized for its high cost and reliance on earmark appropriations (e.g., McLean, 2009), the project successfully reduced flood risk and renewed the local economy.

In Martin, Kentucky (population 634 in 2010), the Army Corps of Engineers is excavating a hilltop for a new municipal and commercial center of the town and using the resulting fill to raise approximately 180 structures 4.9 m higher (Tranum, 2001; Alford, 2009). Initial funding included $33 million (U.S. House of Representatives, 2009), with a few municipal structures built as of early 2018. An additional $80 million were allocated in 2018, with the Federal funds covering 95% of project costs (USACE, 2018). Concern has been expressed over the cost of the Martin relocation project, and it is unclear how this project has met the Corps of Engineers requirement that project benefits exceed total costs (estimated at $159 million; Meadows, 2020). Voluntary and mandatory property buyouts, demolition, and new construction are currently scheduled to be completed by 2023.

Hurricane Sandy in 2012 caused an estimated $62 billion in damages, centered on the coasts of New Jersey and New York. Sandy brought widespread attention and federal and state funding to a variety of disaster recovery and mitigation efforts. New York State established the Governor’s Office of Storm Recovery, providing $650 million under the NY Rising Community Reconstruction Program for communities implementing “innovative rebuilding plans” (Figure 7). Similarly, New Jersey invested in concentrated, neighborhood-scale buyouts through its Blue Acres program.

Oakwood Beach is a neighborhood of small homes built on low-lying ground on the southwestern coast of Staten Island. When Hurricane Sandy hit in 2012, every structure in the neighborhood was flooded by coastal storm surge, as it had been in 2011 during Hurricane Irene. With back-to-back years of flooding, a grassroots coalition pursued a wholesale buyout offer and a plan to return that land to coastal wetlands (Oakwood Beach Buyout Committee, 2015). Initially, Staten Island political leaders resisted buyouts, citing erosion of the local tax base (Koslov, 2016). But the Oakwood Beach homeowners persisted and, in the face of mounting publicity, local and state leaders eventually embraced the initiative. Satellite imagery from 2020 shows most structures in the Oakwood Beach neighborhood removed, with roughly 20 remaining.

Managed retreat and climate-driven relocations have burst into public awareness during recent years (see Introduction). Much recent discussion in the United States has focused on (1) Isle de Jean Charles, Louisiana and (2) the proposed relocation of several coastal villages in Alaska, including the now ongoing move of Newtok. Brief summary reviews are presented here.

Isle de Jean Charles is located in the coastal wetlands of southern Louisiana. Its residents consist primarily of members of the Biloxi-Chitimacha-Choctaw Tribe, with a population that has declined from about 400 down to 85 as of 2018 (GAO, 2020). The threat to Isle de Jean Charles is acute—land loss and frequent inundation, driven by sea-level rise, subsidence, and canal construction and other processes that are leading to the widespread loss of wetlands throughout the Mississippi Delta. Since 1955, the island on which Isle de Charles sits has shrunk >98% from 90 km² to 1.6 km² (Carter et al., 2018). In response, $48.3 million were awarded in 2016 from the U.S. National Disaster Resilience Competition to Isle de Jean Charles, through the State of Louisiana, to relocate the town. Isle de Jean Charles has been the focus of intense press and public interest. Press coverage frequently frames residents as “America’s first climate refugees” and the relocation as the vanguard of an anticipated mass migration away from coastal lowlands (e.g., Milman, 2018). The State of Louisiana, through its Office of Community Development, and local residents assessed several candidate relocation sites, eventually settling on “New Isle,” a 2.1 km² parcel located 64 km inland of the current town. At present, the New Isle land has been purchased and 33 households plan to make the move from Isle de Jean Charles (Baurick, 2020; GAO, 2020), with a planned completion date in 2022. A vocal minority of residents, however, report unhappiness with the new site, the process, and/or the prospect of abandoning the current town location, and do not plan to move.

The other current focus of attention on U.S. managed retreat is the dozen or so Native American villages in Alaska that have considered or are planning to relocate. Of these, Newtok is the farthest along, currently with approximately one third of its residents moved to the town’s new location, and additional construction underway. All of these Alaskan communities are threatened by acute land loss driven by coastal flooding, thawing permafrost, and shoreline erosion (Denali Commission, 2019). Stark photographs of homes undercut and toppling into the ocean have drawn attention for years (Hermann, 2019). Residents of these communities have discussed relocation and have been seeking funding for decades. In Shishmaref, for example, relocation efforts date back to at least 1973, but outside funds were instead directed to shoreline stabilization (Mele and Victor, 2016).

Newtok is a community of approximately 400 residents, most of them Yup’ik tribal members, located along the Ninglick River, along Alaska’s Bering Sea coast. Newtok experienced 200 m (650 ft) of shore erosion 2005–2019. A 2009 GAO study identified Newtok as one of 12 Alaskan villages in imminent danger (GAO, 2009). In 2003, Newtok negotiated a land transfer with the U.S. Fish and Wildlife Service for a new site roughly 14 km away and located on solid bedrock. However, 16 years passed as Newtok struggled to find funding for the relocation. On October 12, 2019, 137 residents from 21 households (approximately 1/3 of the community) moved into the new town, called Mertarvik. As of 2019, $36.3 million had been allocated from various sources (Lester, 2019), primarily from the Congressionally established Denali Commission and state match, managed by the Alaska Native Tribal Health Consortium. Newtok homeowners were paid market value for their homes and were free to move anywhere. New homes in Mertarvik average 110–130 m² (1200–1400 ft²) and represent a significant update from the old town (Kyuk, 2020). Construction of additional homes is ongoing (Kim, 2020), while funding for additional public infrastructure is still needed. Other Alaskan communities look to Newtok as an example, although many of those others struggle even more with the challenges of funding, site selection, and internal consensus. For example, Shishmaref—identified by the Denali Commission as the second most threatened community in Alaska—recently voted to relocate by a narrow margin (89 to 78; Mele and Victor, 2016), and no funding has been secured.

Managed retreat and community relocation, at least in the modern U.S. context, are targeted responses, primarily to coastal and river hazards, within a much broader spectrum of issues and responses to natural hazards (Figure 8). Most clearly, managed retreat can be defined by what it is not—as an alternative to rebuilding in place and engineered solutions such as levees and seawalls. To date, the history of U.S. disaster response has been a history, overwhelmingly, of rebuilding and protecting in place. Pattonsburg, Missouri, flooded 33 times before the back-to-back floods of 1993 convinced most residents to move to high ground. Individual Severe Repetitive Loss structures nationwide have incurred up to 40 taxpayer-funded losses (Pinter et al., 2016). A new nationwide focus on “mitigation,” including passage of the Stafford Act in 1988, added incentives to fix these problems rather than just kicking them down the road to the next flood. As shown in Figure 8, mitigation occupies a broad zone of disaster-response space that includes retreat and removal as well as adaptation (e.g., structure elevations) and protection measures. However, given the inexorable pressure of climate change and sea-level rise, some would argue that managed retreat may represent the next horizon in disaster management.

Figure 8 also highlights the diversity of drivers of community relocations and other types of population displacement. Historical examples range from slow-motion triggers such as drought (e.g., the U.S. Dust Bowl; Parenti, 2011) to temporary displacements (e.g., after Hurricane Katrina in New Orleans; Finch et al., 2010; DeWaard et al., 2016), and other histories. Outside of the present context of managed retreat, past displacements included many examples where the affected populations were moved involuntarily, sometimes brutally so (e.g., Lee, 1966; Black et al., 2011). Justifications for such involuntary resettlements have included dam construction, large infrastructure projects, and urban renewal (e.g., Guggenheim and Cernea, 1993; Scudder, 1993; Mileti and Passerini, 1996; Bronen, 2009; Iuchi, 2014; Obokato et al., 2014). A bedrock gospel of recent U.S. flood mitigation is that projects must be voluntary and motivated primarily by, and for the benefit of, the participating population.

Some of the case studies of migration above involved environmental or climatic drivers, others did not. But climate drivers can be slow and subtle, and anthropologists note that few migrants interviewed during or after resettlement identify climate or climate change as the cause (Black et al., 2011). But even relocations clearly unrelated to flooding or climate can provide useful insights into the social dynamics of population displacement. For example, the largest recent examples of community relocation in the world include Yallourn, Australia (peak population of 5,580; relocated in the 1970s; Wadley and Ballock, 1980), Chuquicamata, Chile (peak population of approximately 24,000; abandoned in 2007), and Kiruna, Sweden (approximately 18,000 residents; relocation ongoing). These were mining towns that moved wholesale in order to access ore beneath the footprint of the town. Each was an order of magnitude larger than the largest floodplain or coastal relocations reviewed in this article. And each of these three examples proceeded, or has been proceeding, expeditiously. The mining relocations added significant financial resources not available to other communities, but a salient characteristic of those moves has been the shared motivation of the population—that their future livelihoods, and future, depended on finding consensus.

The “lost history” of U.S. managed retreat documents, first, the rich diversity of past retreat and relocation case studies. Spanning over 140 years, these examples include changes in social context and policy over time. There are also geographical differences in regional and local culture. And in each community, there has been a wide range of voices and opinions. The summaries here do not do justice to the nuances and drama of each example of moving a town. Nonetheless, some cross-cutting issues can be identified, and some of these may represent lessons for future managed-retreat projects.

Comparing early U.S. relocations like Niobrara—its first move in the 1880s—with recent projects documents both similarities and systematic differences over those approximately 140 years. One key difference is the time needed to implement a relocation. Today, even structure-by-structure mitigation projects involve bureaucratic hurdles requiring many years after a flood to fully navigate—delay times that are widely cited as a national impediment to improving resilience (Pinter et al., 2016). However, the history reviewed here shows that the speed of relocation has not simply increased monotonically with time. Strong local leadership and the support of state and federal partners in Valmeyer and Rhineland put people into homes in the new town sites within approximately 3 years, whereas other relocations and proposed relocations have languished. Perhaps the fastest postflood community relocation on record was outside the United States—Grantham, Australia, had new infrastructure and homes in place within 11 months after fatal flooding in 2011 (Simmonds, 2020). Speed was self-fulfilling in Grantham, with local leaders skillfully using press attention to pressure state and federal into cooperation or, in some cases, benign neglect as local residents seized control of their own relocation.

Another change over time is the increasing quantity and value of infrastructure that needs to be mitigated or moved. Few U.S. residences or other structures today could be simply dragged up a hill by horses, as they were after Niobrara’s 1881 relocation. In a sense, the rising standard of living of U.S. society has made us more brittle and less resilient to natural disasters. In contrast, the “forgotten relocation” of Odanah, Wisconsin, in the 1960s was inexpensive, quick, and widely unknown in part because the community was, and remains, extremely poor. Homes and preexisting infrastructures were simply abandoned. Odanah’s resilience to flooding is still on display—when our research group visited one of the last inhabited structures in the old town, the family told how during the last flood, they had moved all furnishings and possessions to a relative’s house and then returned within days after the last flood water receded (Hersher, 2018).

The cost of relocation projects, when those data are available and adjusted for inflation, varies enormously. The $250,000 cost to relocate Leavenworth, Indiana, after the 1937 flood (= $4.5 million in 2020) is roughly equivalent to $10,000–$15,000 per person relocated. After the 1993 flood, Valmeyer was moved at a cost of approximately $70,000–$80,000 per person and Rhineland at approximately $80,000–$90,000 per person. Valmeyer’s relocation included building a new school and substantial city infrastructure. All U.S. relocations under the leadership of the Army Corps of Engineers have been significantly more expensive, ranging from roughly $130,000 per person for the 1970s relocation of Niobrara, Nebraska, to about $2 million per resident for the 1970s relocation of North Bonneville, Washington. The recent Corps-led projects in Grundy, Virginia, and Martin, Kentucky, have also been enormously expensive, with hundreds of millions of dollars, mostly in the form of federal earmark appropriations, mostly targeted at commercial and municipal structures.

The danger of the high cost of projects like Grundy and Martin is that they may erode public sympathy and political will to undertake managed retreat in the future. However, at the other end of the spectrum, visiting many relocated towns leaves the visitor with the distinct impression that those communities were starved of resources to the point that they never fully recovered their preflood vitality. Odanah is the most extreme example. But even Valmeyer—generally agreed to be the poster child for community relocation in the United States—lacks the couple of blocks of “Main Street” commercial structures that were planned for the center of the new town but never completed. This failure to transition businesses from the old town sites to the new relocated towns is a persistent flaw in all recent U.S. relocations and should be a target of any policy reforms to facilitate the future managed retreat.

Another important issue related to relocation costs is the premium paid to relocate a community compared to simply removing those structures from the floodplain. Where detailed project budgets are available, costs for property acquisitions and demolition of bought-out properties were roughly half to two thirds of total costs, meaning that relocations are approximately 150%–200% more expensive than erasing those communities would have been. But to residents of those towns, the added investment has been richly rewarded in terms of maintaining their cultural fabric, maintaining and even enhancing social capital (e.g., Babcicky and Seebrauer, 2017), and permanently removing the decades-long stigma of repeated flooding from those communities and letting them thrive anew after relocation (Pinter and Rees, 2021).

New relocations are now being planned and implemented without full awareness of past relocations across the United States. With growing public interest in managed retreat, the U.S. GAO (2020) “identified few communities in the United States that have considered climate migration as a resilience strategy, and [just] two…that moved forward with relocation.” A few of past relocation projects are broadly known (e.g., Valmeyer; noted in the GAO report), but some are remembered only locally, if at all. Each project reviewed here represents millions of dollars invested by taxpayers, as well as enormous investment of personal effort, leadership, triumph, and frustration by those involved directly in moving those communities. Thus, a focus on empirical data and experience with managed retreat must continue during and after project completion. Currently, community relocations lack what is common in ecological restoration projects of similar scale—long-term monitoring after completion to assess project outcomes and guide future investments. “Evaluations of retreat outcomes are limited…[and] severely constrained by a lack of data” (Siders et al., 2019, p. 763). The goal of this study was to review the history of past relocation case studies as a resource for scholars of managed retreat, disaster management professionals, as well as local stakeholders contemplating flood risk and the potential for retreat.

Managed retreat has recently burst into widespread academic and policy discussions, and some voices predict waves of climate-driven migrants in the medium-term future. Many researchers and policy makers are now stressing the need for actionable frameworks, plans, and tool kits for expanded and perhaps large-scale relocations in the future. Empirical experience and data have been identified as key needs for these actionable managed-retreat blueprints—“it is incumbent on researchers to synthesize lessons and provide guidance from the accumulating body of knowledge” (Olshansky and Chang, 2009, p. 206) or simply “learn from history” (Marchman et al., 2020; in Scott and Lennon, 2020).

No data were generated that are not included in the tables in this article.

The author declares that no competing interests exist.

How to cite this article: Pinter, N. 2021. The lost history of managed retreat and community relocation in the United States. Elementa: Science of Anthropocene 9(1). DOI: https://doi.org/10.1525/elementa.2021.00036

Domain Editor-in-Chief: Alastair Iles, University of California, Berkeley, CA, USA

Knowledge Domain: Sustainability Transitions