Identifying and Quantifying Landowner Perspectives on Integrated ﬂood Risk Management A

Identifying and quantifying landowner perspectives on integrated ﬂood risk management A. Milman1 , B.P. Warner2, D.A. Chapman3 and A.G. Short Gianotti4

1 Department of Environmental Conservation, University of Massachusetts, Amherst, MA, USA 2 Department of Geography and Environmental Sciences, University of New Mexico, Albuquerque, NM, USA 3 Department of Psychology and Brain Sciences, University of Massachusetts, Amherst, MA, USA 4 Department of Earth and Environment, Boston University, Boston, MA, USA

Correspondence Abstract Prof. Anita Milman, Department of fl Environmental Conservation, University of Understanding landowner perspectives on ood mitigation is an essential step fl Massachusetts, Holdsworth Hall, Amherst, towards minimising con ict and ensuring public support and compliance. To MA 01003, USA illuminate landowner perspectives on flood mitigation and the drivers of those Email: [email protected] perspectives, we surveyed landowners in the Deerfield River Watershed, Massa- chusetts (USA). Confirmatory factor analysis shows landowners differentiate DOI: 10.1111/jfr3.12291 between the physical and the policy pathways through which flood mitigation strategies provide protection. Multiple regression analyses indicate landowner Key words support or opposition for each of those pathways is related both to the risks a Governance and institutions; integrated landowner faces and to their broad worldviews. Lastly, cluster analysis indicates flood risk management; land management; public engagement. variation in patterns landowner of support or opposition to the physical and policy pathways through which flood mitigation provides protection. Findings point to how issues beyond flood impacts, including understandings of riverine processes, assessments of responsibility, and interpretations of the relationship between government and private property drive landowner perspectives on flood mitigation policies.

Introduction 2016). Implementation of IFRM involves five key principles: (1) managing the water cycle as a whole, (2) integrat- Floods are the most damaging natural hazard. In 2013 ing land and water management, (3) maximising the use of alone, hydrological disasters caused $53 billion USD in floodplains and adopting a mix of flood mitigation strate- damages and were responsible for 46.5% of natural gies, (4) ensuring a participatory approach, and (5) adopting disaster-related deaths (Guha-Sapir et al., 2013). In integrated hazard management (Grabs et al., 2007; World response to the rising costs of flooding (Pielke, 1999; Meteorological Organization, 2011). Loucks et al., 2008; Kreibich et al., 2015), across much of Inherent in these principles are a number of proposi- Europe and the United States (Tunstall et al., 2004; Erdle- tions. The first is the acceptance of some level of flood nbruch et al., 2009; Hartmann, 2011; Heintz et al., 2012; impacts. In recognising the inevitability of flood impacts van Herk et al., 2015; Morris et al., 2014; Rouillard et al., and seeking to minimise damages when flooding occurs, 2015; Buchecker et al., 2016) the paradigm for addressing IFRM implies not all properties and peoples will be fully floods has shifted from controlling or blocking flows (flood protected. The second is the adoption of a basin-wide defence) to ‘living with floods’ and decreasing overall vul- approach to managing floods (Holstead et al., 2015; Rouil- nerability to floods through basin-wide water and land lard et al., 2015; Hartmann and Spit, 2016), where riparian management strategies. lands with large floodplains and/or lands located in key Integrated flood risk management (IFRM) encompasses areas are viewed as tools for absorbing flows and energy in a broad range of strategies that are rooted in the recogni- order to reduce downstream risks. The third is a reliance tion that absolute protection from floods is impossible and on controlling and coordinating land-use as well as river that aim to reduce potential human and socio-economic management strategies. Non-structural flood mitigation losses while preserving ecosystems and biodiversity (Grabs measures, such as restoration of floodplains, removal of et al., 2007; Tempels and Hartmann, 2014; Challies et al., embankments, riparian buffers, and restrictions on

J Flood Risk Management (2017) © 2017 The Chartered Institution of Water and Environmental Management (CIWEM) and John Wiley & Sons Ltd 2 Milman et al. encroachments are land based. Lastly, IFRM relies on the perspectives. Our findings point to how issues beyond past private provision of public goods (Tompkins and Eakin, flood impacts, including understandings of riverine pro- 2012; Geaves and Penning-Rowsell, 2016). In implementing cesses, assessments of responsibility, and interpretations of non-structural flood mitigation measures on private prop- the relationship between government and private property erty, landowners are providing public flood mitigation ben- drive landowner assessments of flood mitigation policies. efits sometimes at the expense of their ability to use that Thus increasing acceptance and effectiveness of IFRM property. These propositions point to the critical role of requires better understanding not only of risk but also of private property, and the individuals who own those prop- broader questions of worldviews and governance. erties, in IFRM. Implementation of IFRM has very real consequences for landowners; and landowners, through their attitudes and Landowner perspectives on flood land management practices, influence the effectiveness of mitigation IFRM strategies. IFRM involves changes in how water and debris flow through a basin and thus how landowners expe- Prior research indicates that public preferences regarding rience floods on their property. Yet flows through other flood mitigation strategies are mixed. While different mem- portions of the basin also depend on how individual land- bers and segments of the public view both structural and owners respond to flood risks on their own properties. In non-structural measures positively (Kenyon et al., 2008; using their land and in protecting their land from flood Rouillard et al., 2015; Buchecker et al., 2016), traditional risks, landowners may engage in practices that are synergis- structural measures, including clearing of watercourses and tic with an IFRM approach and provide public flood miti- channelling of rivers, are often preferred (Rasid and Haider, gation benefits or may also engage in practices that serve to 2002; Rasid et al., 2008; Lara et al., 2010; Harries and protect their property yet increase flood risks elsewhere in Penning-Rowsell, 2011). Insights into preferences for flood the basin (Milman and Warner, 2016). Beyond direct phys- mitigation strategies can be gleaned from a variety of theo- ical impacts, landowners can influence IFRM through their retical and methodological approaches including: multi- political voice. IFRM has been contested on a variety of attribute utility, protection-motivation, and governance occasions (Zaugg, 2002; Næss et al., 2005; Scolobig et al., values. Each of these bodies of literature has a unique focus 2008; Harries and Penning-Rowsell, 2011; Buchecker et al., and approach. While these approaches address overlapping 2016; Challies et al., 2016) and landowner dissatisfaction issues, they remain largely unconnected. can lead to pressure to change public flood mitigation poli- From a utility-maximisation perspective, preferences for cies (Emery and Hannah, 2014). flood mitigation strategies can be seen to reflect a prioritisa- Understanding landowner perspectives on different tion of the multiple benefits and costs associated with each aspects of flood mitigation is thus an essential step towards strategy (Keeney and Raiffa, 1993) and multi-attribute util- addressing the distributional consequences of various flood ity theory allows the assessment of how landowners make management activities, minimising political conflict, and decisions about these tradeoffs. Flood mitigation measures ensuring greater public support and compliance (Becker vary in the reduction in flooding achieved as well as et al., 2014). Yet little is known about the degree to which impacts on the environment, aesthetics, economic circum- landowners support IFRM and the reasons for their sup- stances and social practices (Kenyon, 2007), thus individual port or opposition. To illuminate the multiple dimensions preferences will vary with context (Scolobig et al., 2008). of landowner perspectives on IFRM and the drivers of Landowners, tenants, and estate managers that are con- those perspectives, we interviewed and surveyed land- cerned about appearances value the aesthetics of controlled owners in the Deerfield River Watershed, Massachusetts. and tidy rivers over the messy or changing features associ- The first stage of our analysis reveals that landowners’ ated with non-structural flood mitigation measures views on specific flood mitigation strategies stem from their (Rouillard et al., 2015). Individuals concerned about envi- perception of the physical mechanisms through which the ronmental impacts prefer non-structural and ecological strategies provide protection and their ideas about the river management measures (Zaugg, 2002; Lara et al., 2010; appropriate role of government in flood mitigation and Buchecker et al., 2016). Social actors with strong economic assistance. The next stage of our analysis shows that land- or urban-development interests (Scolobig et al., 2008) tend owner support or opposition for flood mitigation strategies to be less supportive of restrictions on land use or on the is related both to the flood risks a landowner faces and their use of productive agricultural land for non-structural miti- broad worldviews regarding government and the role of gation (Holstead et al., 2015; Rouillard et al., 2015). humans in nature. The third stage of our analysis indicates While utility maximisation focuses on interests, a robust that landowners are heterogeneous and, even among those literature on human behaviour points to the influence of who support or oppose IFRM, there is a diversity of psychological variables in determining preferences.

Perceptions of flood mitigation strategies are closely tied to of policies (i.e. regulations or incentives) used and his or understandings of risk and of adaptation (Rogers and her attitudes towards governance processes (Fischer, 2009; Prentice-Dunn, 1997; Grothmann and Patt, 2005; Lindell Glenk and Fischer, 2010). For example, while one study and Perry, 2012). Empirical studies on protection- shows landowners in New Brunswick (Canada) and Maine motivation (i.e. the reasons people undertake protective (USA) agree rules and regulations are needed in order to actions against potential hazards) have demonstrated that promote the public good and that both incentives and regu- perceived risk and concerns about future flooding influence lations are useful (Quartuch and Beckley, 2014), other stud- an individual’s motivation to undertake protective action. ies have shown regulatory approaches to be contentious However, an individual’s expectations regarding his or her (Schaaf and Broussard, 2006; Cocklin et al., 2007; Larson, own capacity to undertake protective action as well as the 2009). Moreover, there is a lack of consensus on how gov- efficacy of those actions in reducing risk (i.e. his or her ernment can best protect the public good while respecting adaptation appraisal) are more consistent predictors of private property rights (Raedeke et al., 2001; Milman and individual action (Blennow and Persson, 2009; Bubeck Warner, 2016). et al., 2012; Becker et al., 2014). As related to perceptions Drawing upon both the theory of cognitive hierarchies of public flood mitigation, the corollary is that an indivi- and cultural theory, researchers have demonstrated individual’s support for particular flood mitigation strategies will dual’s attitudes towards policies are related to the underly- depend on whether they believe the strategy is likely to be ing worldviews people hold (Glenk and Fischer, 2010; effective in reducing the risk and if the government or Offermans and Valkering, 2016). In other words, support other implementing organisation has the capacity to suc- or opposition towards environmental policies is derived cessful implement the strategy. For example, research has from an individual’s values and understanding regarding shown that landowners who doubt the effectiveness of non- how the world does and/or should function, including both structural flood mitigation measures express low support views on the relationship between humans and the environ- for those measures (Howgate and Kenyon, 2009; Rouillard ment and views on the manner in which humans behave et al., 2015; Buchecker et al., 2016), and that landowners and interact with one another (Fischer, 2009). Individuals with high trust in governmental authorities perceive the who hold a collectivist worldview see responsibility as effectiveness of non-structural measures to be higher shared and have an awareness of interconnections between (Buchecker et al., 2016). their actions and their neighbours. These individuals are Research on protection-motivation has also found that more supportive of regulation (Janota and Broussard, 2008; appraisals of adaptation are moderated by perceived Quartuch and Beckley, 2014). In contrast, those who hold responsibility for protection (Paton, 2003; Lindell and an individualist worldview see individuals as responsible Perry, 2012). Expectations regarding the roles and responsi- for their own decisions and success and are less supportive bilities of government and individuals influence perspec- of environmental regulation (Fischer, 2009). In terms of tives on the appropriateness of flood mitigation strategies flood mitigation, soft-engineering, or natural approaches (O’Brien et al., 2009; Adger et al., 2012). While multiple seem to be more preferred by individuals who place a studies find broad agreement among the public that gov- higher value on the welfare of others than those who are ernment has a large responsibility to reduce flood risks focused on self-enhancement (Glenk and Fischer, 2010) (Næss et al., 2005; Terpstra and Gutteling, 2008; Butler and and people’s preferences for natural over structural flood Pidgeon, 2011; Adger et al., 2012; Box et al., 2013), other management stem from a feeling of obligation or responsi- studies show that government policies are increasingly bility to help neighbouring communities at risk of flooding shifting responsibility for managing flood risks to indivi- (Howgate and Kenyon, 2009). Willingness to implement duals (Meijerink and Dicke, 2008; Butler and Pidgeon, natural flood mitigation is also related to concerns about 2011; Bichard and Kazmierczak, 2012). Moreover, opinions expectations of shared responsibility across the watershed are mixed regarding the nature of the role of government. (Holstead et al., 2015). Structural measures that clearly provide public goods are In summary, the different theoretical approaches that consistently viewed as falling under the purview of state or have been used to investigate landowner views on flood federal government (Butler and Pidgeon, 2011; Bichard and mitigation have not been connected. Our research inte- Kazmierczak, 2012; Mees et al., 2013; Barraqué, 2014) while grates across theoretical approaches to investigate how views regarding the role of government in land use pla- landowner flood mitigation objectives, appraisals of flood nning and the regulation of development are more conflict- mitigation strategies, and governance values combine to ing (Geaves and Penning-Rowsell, 2016). determine their perspectives on flood mitigation, including Moreover, sociological research illustrates that an indivi- whether or not they are likely to support IFRM. In doing dual’s endorsement of public policies, including flood miti- so, we explicitly recognise heterogeneity among landowners gation measures, reflects his or her beliefs about the types and use the individual rather than the community as the

J Flood Risk Management (2017) © 2017 The Chartered Institution of Water and Environmental Management (CIWEM) and John Wiley & Sons Ltd 4 Milman et al. unit of analysis. We also recognise that flood mitigation Our analysis of landowner perceptions of flood manage- strategies do not necessarily occur in isolation and land- ment approaches is based on a mail survey of riparian landowners may support implementation of more than one owners in the 10 of the 17 townships in the Deerfield River strategy. Thus our analysis provides a more comprehensive Watershed that suffered the most damage from Tropical look at the multi-faceted nature of landowner perspectives Storm Irene (Milman et al., under review). Semi-structured than has previously been provided. interviews that were conducted used to guide the survey design. Thirty-eight private riparian landowners were identified from tax assessor data using a random stratified sam- pling strategy. Interviewees were asked about prior and Study area and approach expected flood impacts, flood mitigation strategies To investigate landowner perspectives on IFRM and the employed by the property owner and by the town, and fl drivers of those perspectives, we conducted a survey of responsibilities for ood protection (Milman and Warner, riparian landowners in the Massachusetts portion of the 2016). These data ensured the survey encapsulated the fl Deerfield River Watershed (Figure 1). The study area, range of perspectives on ood management held by land- which includes over 777 km2 of primarily rural land, is owners in the study area. The mail survey was developed ’ home to approximately 30 000 residents whose livelihoods and implemented following Dillman s Tailored Design are based mainly on agriculture, timber, tourism, and man- Method (Dillman et al., 2014). Private non-incorporated fi ufacturing. The river main stem and its tributaries are con- riparian landowners were identi ed by overlaying tax asses- fi fined by steep valleys in the uplands areas yet have a flatter sor data from Mass GIS (Massachusetts Executive Of ce for fl and more open downstream topography. The study area Administration and Finance, 2015) with FEMA oodplain has a long history of floods that have altered the physical, maps. Survey recipients were randomly selected from the political, and economic landscapes. Most recently, during population of landowners with properties within the FEMA ‘ ’ Tropical Storm Irene, landslides, riverbank failures, river- Flood Zone A or within 30.5-m riparian buffer zones. fl bed incision, and sedimentation caused substantial damages The mail survey catalogued the impacts of past oods ’ to roads, bridges, culverts, parks, croplands, houses, busi- and rural landowners experiences with and opinions fl nesses, and hospitals. Townships in the Deerfield River regarding ood mitigation strategies on their land and Watershed experienced more damage during and after the across the watershed as a whole. The survey also collected storm than any other township in Massachusetts (Gartner parcel and demographic data. The survey response rate was et al., 2015; Magilligan et al., 2015; Northeast States Emer- 42.3%, with 463 of the 1093 landowners in the sample pop- gency Consortium, 2015) ulation returning valid questionnaires. More than 55% of

Riparian Parcels Identified Town in Study Area Deerfield River Watershed

0 5 10 Kilometers

Figure 1 Deerfield River Watershed, Massachusetts. Shading indicates towns included in the study: Ashfield, Buckland, Charlemont, Colrain, Conway, Deerfield, Hawley, Rowe, and Shelburne.

© 2017 The Chartered Institution of Water and Environmental Management (CIWEM) and John Wiley & Sons Ltd J Flood Risk Management (2017) Identifying and quantifying landowner perspectives 5 respondents reported flood-induced damage; 27% said that, Modification), and (4) restrictions on human development compared to other concerns about their property, floods that affects natural processes (Limiting Floodplain Develop- are either a big or the most pressing concern; and of the ment). River Process Modification includes actions within 45% who have not previously been impacted by floods, over the riparian area that seek to directly alter flows or geomor- 50% expect to experience some kind of flood damage in the phic processes, such as dredging, channel modification, or future. Milman et al. (in review) provide a detailed sum- channel armouring. Limiting Floodplain Development refers mary of the survey data. to preventing encroachments in the floodplains that may In order to gain a comprehensive understanding of land- serve to block flows or impede natural riverine processes. owner perspectives on strategies for flood management, the To statistically test our hypothesis that landowners see survey data were analysed in multiple stages. A confirma- flood mitigation strategies as comprised of four factors tory factor analysis (CFA) was performed to examine how (i.e. latent variables) and whether or not those factors are landowners distinguish between flood mitigation activities reliably measured by the survey items in the data set and whether the items in the survey reliably measure those (i.e. observed variables), CFA was performed using LISREL distinctions. Using the results from the CFA, multiple 8.7 (Jöreskog and Sörbom, 2004; for an explanation on regression analyses were utilised to examine the contextual CFA see Brown and Moore, 2012); for an example of the and attitudinal variables that are predictive of landowner use of CFA in a similar context, see Graham et al., 2014). preferences for flood mitigation strategies. Finally, cluster All four specified factors were allowed to correlate and were analyses were performed to explore how preferences for scaled by setting their variances to one. Table 1 displays the management strategies combine and co-occur in the opi- list of survey items used to measure the four factors and nions and attitudes of landowners. the reliability estimates for each of the four factors. For additional details on the CFA model see Appendix S1, Sup- porting Information. CFA: landowner interpretation of flood The model is a good fit to the observed data, mitigation strategies χ2(72) = 186.09, P < 0.001, Root Mean Square Error of Survey results indicate that landowners hold heterogeneous Approximation (RMSEA) = 0.06, Confirmatory Fit Index views of flood mitigation strategies. Individual landowners (CFI) = 0.96, confirming that landowners indeed view the four view some flood response strategies as appropriate and hypothesised factors as distinct and that the survey items can others as inappropriate. Further, there are predictable pat- be used to reliably measure them. The CFA model also identi- terns in how a landowner views multiple flood mitigation fies relationships between the four factors. Landowners who strategies. For example, a landowner who thinks channel believe the government should assist landowners in addressing armouring is an appropriate response to flooding is likely flood risks (Flood Assistance) are likely to also support flood to respond favourably to straightening the river channel. mitigation strategies involving human intervention in riverine This is because the landowner interprets these strategies as processes (River Process Modification); while landowners who having similar characteristics. Drawing on the aforemen- believe that government regulations of landowner actions (Reg- tioned literature on utility maximisation, protection- ulation) should be a part of flood mitigation are likely to sup- motivation and government values, as well as our interview port restrictions on human development that affects natural data analysis (Milman and Warner, 2016), we hypothesised processes (Limiting Floodplain Development). There is no cor- that landowner support or opposition to flood mitigation relation between the support for government flood assistance strategies is based on their support or opposition to the (Flood Assistance) and government regulation (Regulation); this underlying policy and physical pathways through which paradox has been shown to exist in other high-profile, flood- those strategies reduce flood risks. prone areas and in flood mitigation policy (Burby, 2006). These pathways can be described as comprised of four Lastly, landowners who support regulation (Regulation)and factors (Table 1). Two factors are related to the responsibil- restrictions on human development that affects natural pro- ity and domain of government: (1) government assistance cesses (Limiting Floodplain Development) are less likely to to help landowners address flood risks (Flood Assistance), support human intervention in riverine processes (River Proc- (2) government regulation of landowner actions (Regula- ess Modification). tion). Flood Assistance includes any form of financial, tech- nical, or other support provided to landowners while Regression analyses: drivers of landowner Regulation refers to legal rules or directives that dictate or support for each of the four factors of flood constrain the actions landowners undertake on their prop- mitigation erty. The other two factors are related to the efficacy of different physical mechanisms for flood risk mitigation, Having identified the four factors used by landowners when (3) human intervention in riverine processes (River Process interpreting flood mitigation strategies, analyses were

Table 1 Survey items measuring each of the four factors of flood mitigation Factor Composite reliability Survey items* Flood Assistance 0.83 1. It is primarily the responsibility of individual landowners † to protect their land from floods. 2. The government should provide assistance to landowners located in areas that experience frequent flooding. 3. The government should provide assistance to landowners in advance of a flood event to help them protect against future impacts. 4. The government should provide assistance to landowners as they rebuild or restore their land after a flood event. Regulation 0.69 1. Governmental regulations usually produce good results for landowners. 2. Government is needed to settle disputes among property owners. 3. The government should be able to regulate how landowners use their land in order to address flooding. River Process Modification 0.76 1. The government should armour stream banks using engineering strategies, such as rip-rap or concrete. 2. The government should straighten or move the river. 3. The government should clean debris and remove logjams. 4. The government should dredge rivers regularly. 5. The government should let rivers run their course; restrict additional human intervention in the river bed † and surrounding areas. Limit Floodplain Development 0.78 1. Restrict use of land designated as floodplain. 2. Restrict all new developments on land designated as floodplain.

*Table A1.1 of Appendix S1 contains summary statistics for each survey item used in the confirmatory factor analysis. † The survey questions for these items were phrased such that agreement with the survey question represents opposition to the latent variable; hence, the factor loadings have negative signs. performed to predict landowners’ support (or opposition) nuanced and preferences are situation-specific (Glenk and to each of them based on a combination of contextual and Fischer, 2010; Flint et al., 2013), rather than using broad- attitudinal variables. Contextual variables were included to based philosophical measures, we measure landowner determine how a landowner’s socio-economic circum- worldviews based on their answers to survey questions spe- stances and perception of risk influences attitudes towards cific to the locational and social context of our study. The flood mitigation activities. Contextual variables included variables included in the regression analyses along with an demographic information on education and whether the explanation for their inclusion are shown in Table 2. The land was purchased for affordability (proxy for income), survey items used to measure these variables and summary land use, and flood risk. statistics for those survey items are included in Table A2.1 Attitudinal variables were included to determine how a of Appendix S2 . landowner’s underlying views on environmental govern- To analyse the relationship between each of the four fac- ance influence support for flood mitigation activities. We tors of flood management and contextual and attitudinal draw on prior research that indicates preferences for con- variables, a score representing each survey respondent’s crete policy options are informed both by worldviews support for each of the four factors was computed. These regarding interactions between humans and by worldviews scores range from complete opposition (score of −2) to regarding the relationship between humans and nature complete support (score of +2). Regression analyses to pre- (Fischer, 2009). From this, attitudinal variables were dict landowner support (or opposition) to each of the four selected to represent whether a landowner holds a collectiv- factors were then performed using SAS 9.4. Table 3 displays ist versus individualist worldview and whether a landowner the full models including standardised and unstandardised holds an environmentalist versus an anthropocentric regression coefficients. Due to missing data points for some worldview (Table 2). As an individual’s worldviews are landowners, sample sizes differ across the four analyses.

Table 2 Variables regressed on each of the four factors of flood mitigation Category Justification for inclusion Variable name Survey question Demographic (Schaaf and Broussard, 2006; • Education • What is the highest degree or level Janota and Broussard, 2008; of education you have completed? Fischer, 2009; Hornsey • Purchased land for • How important is affordability for et al., 2016) affordability why you currently own land in Western Massachusetts? • Primary residence • Do you use this land as a primary residence? Land use (Janota and Broussard, 2008; • Agriculture • How important are agriculture, Kenyon et al., 2008; Scolobig timber, or game as reasons for why et al., 2008; Holstead et al., you currently own land in Western 2015; Rouillard et al., 2015; Massachusetts? Milman and Warner, 2016) • Recreation • How important is recreation as a reason for why you currently own land in Western Massachusetts? • Environmental • How important is environmental protection protection as a reason for why you currently own land in Western Massachusetts? • Investment • How important are investment reasons for why you currently own land in Western Massachusetts? Flood risk (Grothmann and Reusswig, • Length of ownership • How long have you, personally, 2006; Kenyon et al., 2008; owned this land in Western Bubeck et al., 2013; Becker Massachusetts? et al., 2014; Buchecker et al., • Prepared for floods • Overall, how prepared are you to 2016; Milman and recover from future flood events? Warner, 2016) • Past erosion • Looking back at past flood events, to what degree did you experience erosion damage on your land? • Future damage • How likely do you think it is that there will be extensive flood damage to your land during the next 5 years? Collectivist worldview (Janota and Broussard, 2008; • Watershed approach • Indicate extent of agreement: Flood Terpstra and Gutteling, 2008; mitigation measures that benefit Howgate and Kenyon, 2009; the larger community should be Butler and Pidgeon, 2011; chosen even if they make certain Box et al., 2013; Hornsey et al., individuals worse off. 2016; Milman and Warner, • Responsible for • Indicate extent of agreement: I am 2016; Offermans and neighbours responsible for ensuring actions I Valkering, 2016) take on my land do not reduce the capacity of my neighbours to respond to floods Environmentalist versus (Zaugg, 2002; Janota and • Prioritise protection of • How should Massachusetts prioritise anthropocentric worldview Broussard, 2008; Lara et al., environment or its goals for river management? 2010; Tam and McDaniels, protection of human (Environmental versus human 2013; Buchecker et al., 2016; systems protection) Milman and Warner, 2016; Offermans and Valkering, 2016)

Reduced models that only include the variables identified who possess a bachelor’s degree or higher, are less support- as significant predictors in the full models are included in ive of flood assistance while landowners who said afforda- the Table A2.2 of Appendix S2. bility was important to their decision to purchase their land Landowner support for Flood Assistance is predicted pri- and those who use their land in the Deerfield River Water- marily by the contextual variables evaluated. Landowners, shed for their primary residence are more supportive of

Table 3 Regression models predicting landowner support for each of the four factors of ﬂood mitigation River Process Limiting Future Flood Assistance Regulation Modiﬁcation Development Full model (n = 282) Full model (n = 282) Full model (n = 282) Full model (n = 282) Adjusted R2 = 0.14 Adjusted R2 = 0.32 Adjusted R2 = 0.30 Adjusted R2 = 0.20 b SEBb SEBb SEBb SEB

Demographic Education −0.26* 0.13 −0.13 0.26* 0.11 0.13 −0.28** 0.11 −0.14 0.08 0.12 0.04 * Purchased Land for Affordability 0.10 0.04 0.13 0.03 0.04 0.04 0.04 0.04 0.05 0.02 0.04 0.03 Land use ** Primary residence 0.49 0.16 0.17 0.09 0.13 0.04 0.24 0.14 0.08 0.19 0.15 0.07 Agricultural land use −0.03 0.04 −0.04 −0.04 0.03 −0.06 −0.02 0.04 −0.04 −0.04 0.04 −0.07 Recreational land use 0.02 0.05 0.03 0.05 0.04 0.07 −0.03 0.04 −0.04 −0.01 0.04 −0.02 Environmental protection 0.06 0.06 0.07 0.05 0.05 0.06 −0.03 0.05 −0.03 0.12* 0.05 0.15 * Investment 0.01 0.04 0.02 −0.03 0.04 −0.05 0.08 0.04 0.12 −0.06 0.04 −0.09 Flood risk Length of ownership <0.01 <0.01 −0.01 <0.01 <0.01 0.04 0.01 <0.01 0.09 <0.01 <0.01 0.06 * * Prepared for ﬂoods −0.14 0.06 −0.13 −0.06 0.05 −0.06 −0.11 0.06 −0.10 −0.06 0.06 −0.05 ** ** Past erosion 0.05 0.10 0.03 0.26 0.09 −0.18 0.13 0.09 0.08 −0.26 0.09 −0.18 ** ** Future damage 0.18 0.06 0.21 0.04 0.05 0.06 0.13 0.05 0.15 0.09 0.05 0.11 Collectivist worldview ** ** *** Watershed approach −0.02 0.05 −0.02 0.32 0.05 0.37 −0.12 0.05 −0.14 0.22 0.05 0.25 * Responsible for neighbours 0.06 0.07 0.05 0.11 0.06 0.1 −0.07 0.06 −0.06 0.15 0.06 0.13 Environmentalist versus anthropogenic 0.03 0.03 0.07 −0.09*** 0.02 −0.21 0.14** 0.02 0.31 0.09*** 0.03 −0.21 worldview

* ** *** P ≤ 0.05; P ≤ 0.01; P ≤ 0.001.

flood assistance. Landowner support for flood assistance process modification. Flood risk also matters: landowner also increases with expectations for future flood damage support for river process modification increases with expec- and decreases with confidence in their preparedness for tations for future flood damage and with greater length of flood events. Attitudinal variables were not significant pre- ownership (marginally significant). Landowner support for dictors of landowner perspectives on flood assistance. river process modification decreases with their sense of pre- These effects were significant in both the full regression paredness for future flood events. In terms of attitudinal model, as well as the reduced model, with the exception of variables, landowners with a more anthropocentric world- education, which was only marginally significant in the view are more supportive of river process modification reduced model. while landowners with a more environmentalist worldview Landowner support for Regulation is predicted by both and landowners with a more collectivist worldview are less the contextual and attitudinal variables evaluated. In terms supportive of river process modification. of contextual variables, holding a bachelor’s degree or Landowner support for Limiting Floodplain Development higher is associated with greater support for regulation is also predicted most strongly by the attitudinal variables while having experienced past erosion damages is associ- evaluated, though contextual variables also have an influ- ated with lower support for regulation. In terms of attitudi- ence. In terms of contextual variables, owning the land for nal variables, landowners with a more anthropocentric the purpose of protecting the environment leads to greater worldview are less supportive of regulation than land- support for limiting development while past erosion dam- owners with a more environmental worldview, while land- age leads to less support for limiting development. In terms owners with a more collectivist worldview are more of attitudinal variables, landowners with a more anthropo- supportive of regulation than landowners with a more indi- centric worldview are less supportive of limiting floodplain vidualist worldview. development while landowners with a more environmen- Landowner support for River Process Modification is also talist worldview and landowners with a more collectivist predicted by both the contextual and attitudinal variables worldview are more supportive of limiting floodplain evaluated. Landowners who possess a bachelor’s degree or development. higher are less supportive of river process modification In comparing across the models, several findings stand while landowners who purchased their land as an invest- out. The first is that education has a significant influence ment or who use their land for their primary residence on landowner perspectives on three of the four factors of (marginally significant) are more supportive of river flood mitigation. Landowners who have attained higher

Strongly 2.0 Agree

1. 5 Flood Assistance

Regulation 1. 0

River Process Modification 0.5 Limiting Floodplain Development Neutral 0.0

-0.5

-1.0 Support Flood Mitigation for

-1.5

Strongly -2.0 Disagree Non- Flood Moderated Flood Development- Environment- interventionist Defense Defense centric centric (n = 49) (n = 70) (n = 138) IFRM IFRM (n = 109) (n = 92)

Figure 2 Landowner perspectives on flood mitigation strategies. Landowners were grouped into five categories based on patterns in their degree of support or opposition to each of the four factors of flood mitigation. Longer bars indicate greater support or opposition. The number of landowners in each category is listed along with the category name at the bottom of the chart. Table 4 describes these categories qualitatively. levels of education are more supportive of Regulation and matters. Generally landowners who are more concerned less supportive of Flood Assistance and River Process. The about future impacts are more supportive of River Process second is that, with the exception of primary residence and Modification and Flood Assistance, yet landowners who Flood Assistance, landowner perspectives do not seem to have experienced past erosion damages are more support- vary by land use. Thirdly, concern about flood impacts ive of Regulations and Limiting Floodplain Development.

Table 4 Landowner perspectives on flood mitigation strategies Cluster name (number of survey respondents) Perspectives on flood mitigation Non-interventionist (n = 49) These landowners have a laissez faire attitude towards flood mitigation. They see little need for action to address flood risks and are opposed to government intervention. Flood defence (n = 70) These landowners are likely to favour flood-defence over integrated flood risk management. They are strongly opposed to regulation and to limiting floodplain development, likely due to concerns about private property rights. However, they view government as responsible for protecting private property and for assisting landowners and support the use of engineering and other interventions in riverine processes to reduce flood risks. Moderated flood These landowners are likely to favour flood-defence over integrated flood risk management, defence (n = 138) though are moderate in their opinions and recognise that encroachment in floodplains can increase flood risk across the watershed. Development-centric These landowners are likely to support integrated flood risk management, though recognise the IFRM (n = 109) challenge of balancing needs across the watershed. They view management of land use practices at the watershed-scale as the ultimate solution but are sympathetic to risks posed to existing development. Thus while they support government assistance for landowners, regulation, and limiting floodplain development, they are only moderately opposed to human intervention in riverine processes. Environment-centric These landowners are likely to support integrated flood risk management and favour letting nature IFRM (n = 92) run its course. They view management of land use practices across the watershed-scale as the only solution and are willing to allow damage to existing development in order to reduce risks over the long term.

Lastly, attitudinal variables related to collectivism versus landowners are more strongly opposed to Flood Assistance individualism and environmental versus anthropocentric and to River Process Modiﬁcation, suggesting those land- worldviews are particularly important for predicting sup- owners believe more strongly that humans should move port for Regulations and Limiting Floodplain Development. away from rivers and let nature run its course.

Cluster analysis: co-occurrence of support for Discussion the four factors of flood mitigation The four factors used by landowners to interpret flood mit- Landowner support is essential for the success of IFRM, yet igation are not mutually exclusive; rather they can be con- we do not have a robust understanding of why and in what currently implemented. Moreover, landowners may ways individual landowners support or oppose IFRM. Our support or oppose more than one. To understand how the study of landowner perspectives on flood risk management four factors combine in the minds of landowners, we con- in the Deerfield River Watershed shows that flood risk ducted a cluster analysis. This analysis grouped landowners management cannot be understood by relying on utility based on similarities in their support and opposition to the maximisation, protection-motivation, or governance the- four factors. Groups were identified using a k-means clus- ories in isolation. Rather, we must consider how the key tering approach and PROC FASTCLUS in SAS 9.4. The aspects of cognition each of those theories highlights (eco- analysis was run multiple times with the data randomly nomics, psychology, and worldviews) combine to influence sorted during each run to produce alternative solutions. an individual’s support or opposition to flood risk manage- There was a high degree of similarity across the alternative ment strategies. solutions in how landowners were clustered. The selected For example, our analysis shows that landowners con- solution converged in 10 iterations. sider the physical pathways through which flood mitigation Results from the cluster analysis indicate the four factors strategies provide protection as distinct from the policy combine in the minds of landowners in five differing ways, pathways used by government to implement those strate- or, phrased differently, landowner perspectives can be cate- gies. This finding suggests adaptation appraisal, a key con- gorised as falling into one of five groups. Figure 2 depicts the sideration of protection-motivation theory, inherently degree of support or opposition to each of the four factors of encompasses consideration of government values as well as flood mitigation for each group and Table 4 provides a quali- the flood mitigation strategy itself. If landowners perceive tative description of the perspectives of landowners in each government regulation as ineffective, they will be less likely category. While all combinations of support and opposition to support flood mitigation strategies that hinge upon regu- to the four factors of flood mitigation exist, support or oppo- lations, regardless of the physical pathway (Holstead et al., sition to Regulation and Limiting Floodplain Development 2015). Similarly, if landowners assumes flood management and support or opposition to Flood Assistance and River entails actions that affect river behaviour rather than actions Process Modification tend to covary, while support or opposi- that influence the human-environmental relations that pro- tion to River Process Modification tends to have a reciprocal duce risk, and they will be less supportive of policies such as relationship with Limiting Floodplain Development. While Limiting Floodplain Development (Cook et al., 2016). By these correlations were also identified via the CFA analysis, more explicitly considering how landowners evaluate both the clustering develops a typology of landowner perspectives. the physical and the policy pathways, we can develop more The patterns of support and opposition to each of the nuanced understandings of exactly what aspects of flood four factors within each category also point to how that mitigation strategies landowners view as effective. group of landowners is likely to respond to IFRM. The Our research also builds on research that have shown non-interventionist landowners are staunchly opposed to that landowner expectations of government shapes support all four of the factors. The Flood Defense landowners sup- or opposition to flood mitigation policies (Chamlee-Wright port River Process Modification and are opposed to Regula- and Storr, 2010) by demonstrating that landowners see tion and Limiting Floodplain Development and thus likely Flood Assistance as unique to Regulation, and their support favour a flood defence rather than an IFRM approach to for each of the two factors is uncorrelated. This distinction flood mitigation while the Moderate-Flood Defense land- points to the fact that landowners do not simply support or owners support River Process Modification, are opposed to oppose government involvement in flood protection; Regulation, yet moderately support Limiting Floodplain rather, they hold more nuanced views of social contracts in Development and thus would prefer a strategy that incorpo- relation to floods. It also reflects that individuals can hold rates elements of both flood defence and IFRM. The contradictory views, and often expect government to miti- remaining two categories of landowners both are likely sup- gate their risks while limiting the actions government can portive of IFRM, though the Environment-Centric IFRM take to provide those services.

Additionally, our analyses indicate that support or opposi- flood mitigation. As such, additional research is needed to tion for Flood Assistance, Regulation, River Process Modifica- identify cognitive processes or considerations that explain tion,andLimiting Floodplain Development is predicted not or can predict why landowners fall into each cluster. only by the contextual variables representing the landowner and the particular risks that landowner faces, but also by attitudinal variables representing that landowner’s worldviews. Conclusion In terms of contextual variables, our analysis supports findings from prior research that shows support for government Our research in the Deerfield River Watershed reveals that policies is related to education (Schaaf and Broussard, 2006) principles embedded in IFRM may themselves impede its and yet challenges the tendency to stereotype landowners implementation. Successful IFRM requires coordinating based on land use (Arneth et al., 2014). Our findings also actions across the watershed and thus is predicated on the support research that shows landowners who have previously support of private property owners. Yet the impacts of experienced flood damages or who expect future flood IFRM are not equal across landowners, and landowners impacts support different types of flood mitigations strategies hold a plurality of perspectives on IFRM. In seeking to compared to landowners who expect lesser flood impacts minimise costs of damages across the watershed, an IFRM (Kenyon et al., 2008; Adger et al., 2012; Buchecker et al., approach to flood mitigation presumes there is a widely 2016), yet our results also point to important nuances. While understood, transparent, and accepted prioritisation of par- utility maximisation and protection-motivation theories ticular strategies for flood management. Landowners who would lead us to expect erosion damage to predict support own floodplain land have the potential to implement natu- for Limiting Floodplain Development because protecting ral flood mitigation measures on their property are asked floodplains can reduce erosive energy in rivers, our results to assume risks or to take on changes in their land-use reveal the opposite for many landowners in the Deerfield practices in order provide flood protection for others. In Watershed. This finding may reflect concern among land- addition to the common IFRM critiques of uncertainty in owners that Limiting Floodplain Development will prevent our understanding of flood risk (Faulkner et al., 2007), fail- them from engaging in individual practices on their land that ures in estimating the values and impacts of floods to dif- could protect them from future erosion. In this case, land- ferent stakeholders (Hall et al., 2003), or ineffective owner utility and flood risk mitigation must be understood compensation for those asked to assume risks (Thieken within the context of their attitudes towards governance. et al., 2007), our research shows that there is a fundamental In terms of attitudinal variables, our study finds collectivist disagreement among landowners about the role for govern- versus individualist and environmental versus anthropocen- ment in flood mitigation, which can impact the develop- tric worldviews have greater predictive power than contextual ment and implementation of IFRM policies. variables among our study population. However, such attitu- We thereby argue that flood mitigation policies need to dinal variables have received scant attention in the flood risk incorporate a multifaceted understanding of landowner management literature. As studies have found contextual perspectives. More research is needed to advance the goal variables alone often cannot explain what leads some indivi- of reducing flood risks. Improved understandings of differ- duals to support certain flood mitigation strategies while ences in landowner worldviews and understandings of river others do not (Bracken et al., 2016), it may be that these atti- processes and how they influence support for flood mitiga- tudinal variables are indeed the largest driver of landowner tion strategies and management actions on their own prop- opinions. Thus there is a strong need for further developing erty is needed to identify potential conflict and to develop theoretical understandings of the drivers of attitudes, better policies that minimise it. This research should investigate a mechanisms for predicting attitudes, and outreach and policy variety of flood contexts to understand the role of geogra- that explicitly recognises and addresses attitudes. phy, culture, and history on landowner perspectives and Lastly, our analysis highlights variation in how support should also explore how outreach or dialog can influence or opposition for differing physical and policy pathways for landowner views. In addition, equity considerations need to flood mitigation combine. In the study area, two clusters of be addressed, particularly when productive floodplain land landowners both support Limiting Floodplain Development uses (typically rural-based) are marginalised to provide and are opposed to River Process Modification, and thus security to prioritised land uses (typically urban and envi- ostensibly would support IFRM. Yet those groups of land- ronmental protection). We must improve our understand- owners vary in their support for Assistance and thus their ing of if and how mechanisms can be developed to better support for IFRM may well depend on how policies for compensate landowners who reduce the risks across the promoting IFRM are designed. To our knowledge, there are watershed by experiencing increased flood risk to their no studies that have sought to determine the relationship property. We also need to understand how governance between support for physical and for policy pathways for values influence acceptance of compensation mechanisms.

These difficult research agendas are interdisciplinary and Bracken L.J., Oughton E.A., Donaldson A. et al. Flood risk man- based in the social sciences but cannot be accomplished agement, an approach to managing cross-border hazards. Nat without an understanding of flood risk that resides in disci- Hazards 2016, 82, (2), 217–240. plines of fluvial geomorphology, hydrology, and meteorol- Brown T.A. & Moore M.T. Confirmatory factor analysis. In: R. ogy. We are hopeful that such collaborations may minimise H. Hoyle, ed. Handbook of structural equation modeling. the creation of winners and losers among landowners, and New York, NY: Guilford Press, 2012, 361–379. will reduce flood risks and protect river ecosystems for Bubeck P., Botzen W.J.W. & Aerts J.C.J.H. A review of risk per- future generations. ceptions and other factors that influence flood mitigation behavior. Risk Anal 2012, 32, (9), 1481–1495. Bubeck P., Botzen W.J.W., Kreibich H. & Aerts J.C.J.H. Detailed Acknowledgements insights into the influence of flood-coping appraisals on mitigation behaviour. Glob Environ Change 2013, 23, (5), – We thank the many interviewees and survey respondents 1327 1338. who participated in this research. We also thank Michael Buchecker M., Ogasa D.M. & Maidl E. How well do the wider fl Roberts for his assistance with data entry and management. public accept integrated ood risk management? An empirical 55 This material is based on work supported by the National study in two Swiss Alpine valleys. Environ Sci Policy 2016, , – Institute of Food and Agriculture, U.S. Department of Agri- (2), 309 317. culture, the Massachusetts Agricultural Experiment Station, Burby R.J. Hurricane Katrina and the paradoxes of government disaster policy: bringing about wise governmental decisions and the Department of Environmental Conservation, under for hazardous areas. Ann Am Acad Pol Soc Sci 2006, 604, project number MAS00023. The contents of this project are 171–191. solely the responsibility of the authors and do not necessar- Butler C. & Pidgeon N. From ‘flood defence’ to ‘flood risk man- ily reflect the official views of the funding agencies. agement’: exploring governance, responsibility, and blame. Environ Plan C Govern Policy 2011, 29, (3), 533–547. Supporting Information Challies E., Newig J., Thaler T., Kochskamper E., & Levin-Keitel M. Participatory and collaborative governance for sustainable flood risk management: an emerging research agenda. Envi- Appendix S1: Methods and Results of Confirmatory Factor ron Sci Policy 2016, 55, (2), 275–280. Analysis Chamlee-Wright E. & Storr V.H. Expectations of government’s Appendix S2: Methods and Results of Regression Models response to disaster. Public Choice 2010, 144,(1–2), 253–274. Cocklin C., Mautner N. & Dibden J. Public policy, private land- References holders: perspectives on policy mechanisms for sustainable Adger W.N., Quinn T., Lorenzoni I., Murphy C., & Sweeney J. land management. J Environ Manage 2007, 85, (4), 986–998. Changing social contracts in climate-change adaptation. Nat Cook B., Forrester J., Bracken L., Spray C. & Oughton E. Com- Clim Change 2012, 3, (4), 330–333. peting paradigms of flood management in the Scottish/Eng- Arneth A., Brown C. & Rounsevell M. Global models of human lish borderlands. Disaster Prevention Manage 2016, 25, (3), decision-making for land-based mitigation and adaptation 314–328. assessment. Nat Clim Change 2014, 4, 550–557. Dillman D.A., Smyth J.D. & Christian L.M. Internet, phone, mail, Barraqué B. The common property issue in flood control and mixed-mode surveys: the tailored design method. Hobo- through land use in France. J Flood Risk Manage 2014. doi: ken, NJ: John Wiley & Sons, 2014. 10.1111/jfr1113.12092. Emery S.B. & Hannah D.M. Managing and researching floods: Becker G., Aerts J.C.J.H. & Huitema D. Influence of flood risk sustainability, policy responses and the place of rural commu- perception and other factors on risk-reducing behaviour: a nities. Hydrol Process 2014, 28, (18), 4984–4988. survey of municipalities along the Rhine. J Flood Risk Manage Erdlenbruch K., Thoyer S., Grelot F., Kast R., & Enjolras, G. 2014, 7, (1), 16–30. Risk-sharing policies in the context of the French Flood Pre- Bichard E. & Kazmierczak A. Are homeowners willing to adapt vention Action Programmes. J Environ Manage 2009, 91, (2), to and mitigate the effects of climate change? Clim Change 363–369. 2012, 112,(3–4), 633–654. Faulkner H., Parker D., Green C. & Beven K. Developing a Blennow K. & Persson J. Climate change: motivation for taking translational discourse to communicate uncertainty in flood measure to adapt. Glob Environ Change 2009, 19, (1), risk between science and the practitioner. Ambio 2007, 36, 100–104. (8), 692–704. Box P., Thomalla F. & van den Honert R. Flood risk in Fischer A. On the role of ideas of human nature in shaping atti- Australia: whose responsibility is it, anyway? Water 2013, 5, tudes towards environmental governance. Hum Ecol 2009, 38, (4), 1580–1597. (1), 123–135.

Flint C.G., Kunze I., Muhar A., Yoshida Y. & Penker M. Explor- change. Nat Clim Change 2016, 6, 622–626. doi: 10.1038/ ing empirical typologies of human–nature relationships and nclimate2943. linkages to the ecosystem services concept. Landsc Urban Howgate O.R. & Kenyon W. Community cooperation with natu- Plan 2013, 120, 208–217. ral flood management: a case study in the Scottish Borders. Gartner J.D., Dade W.B., Renshaw C.E., Magilligan F.J., & Burass Area 2009, 41, (3), 329–340. E.M. Gradients in stream power influence lateral and down- Janota J.J. & Broussard S.R. Examining private forest policy pre- stream sediment flux in floods. Geology 2015, 43, (11), ferences. Forest Policy Econ 2008, 10, (3), 89–97. 983–986. Jöreskog K.G. & Sörbom D. LISREL 8.7 for Windows [Computer Geaves L.H. & Penning-Rowsell E.C. Flood risk management as software]. Lincolnwood, IL: Scientific Software Interna- a public or a private good, and the implications for stake- tional, 2004. holder engagement. Environ Sci Policy 2016, 55, (2), 281–291. Keeney R.L. & Raiffa H. Decisions with multiple objectives: pre- Glenk K. & Fischer A. Insurance, prevention or just wait and see? ferences and value trade-offs. Cambridge, UK: Cambridge Public preferences for water management strategies in the con- University Press, 1993. text of climate change. Ecol Econ 2010, 69, (11), 2279–2291. Kenyon W. Evaluating flood risk management options in Scot- Grabs W., Tyagi A. & Hyodo M. Integrated flood management. land: A participant-led multi-criteria approach. Ecol Econ Water Sci Technol 2007, 56, (4), 97–103. 2007, 64, (1), 70–81. Graham S., Barnett J., Fincher R., Hurlimann A., & Mortreaux Kenyon W., Hill G. & Shannon P. Scoping the role of agriculture C. Local values for fairer adaptation to sea-level rise: a typol- in sustainable flood management. Land Use Policy 2008, 25, ogy of residents and their lived values in Lakes Entrance, (3), 351–360. Australia. Glob Environ Change 2014, 29,41–52. Kreibich H., Bubeck P., Van Vliet M. & De Moel H. A review of Grothmann T. & Patt A. Adaptive capacity and human cogni- damage-reducing measures to manage fluvial flood risks in a tion: the process of individual adaptation to climate change. changing climate. Mitig Adapt Strateg Glob Change 2015, 20, Glob Environ Change 2005, 15, (3), 199–213. (6), 967–989. Grothmann T. & Reusswig F. People at risk of flooding: why Lara A., Saurí D., Ribas A., & Pavon D. Social perceptions of some residents take precautionary action while others do not. floods and flood management in a Mediterranean area (Costa Nat Hazards 2006, 38,(1–2), 101–120. Brava, Spain). Nat Hazards Earth Syst Sci 2010, 10, (10), Guha-Sapir D., Hoyois P. & Below R. Annual disaster statistical 2081–2091. review 2013: the numbers and trends. Brussels, Belguim: Cen- Larson K. Social acceptability of water resource management: tre for Research on Disasters, 2013. a conceptual approach and empirical findings from Port- Hall J.W., Meadowcroft I.C., Sayers P.B. & Bramley M.E. Inte- land, Oregon. J Am Water Resour Assoc 2009, 45, (4), grated flood risk management in England and Wales. Nat 879–893. Hazards Rev 2003, 4, (3), 126–135. Lindell M.K. & Perry R.W. The protective action decision model: Harries T. & Penning-Rowsell E. Victim pressure, institutional theoretical modifications and additional evidence. Risk Anal inertia and climate change adaptation: the case of flood risk. 2012, 32, (4), 616–632. Glob Environ Change 2011, 21, (1), 188–197. Loucks D.P., Stedinger J.R., Davis D.W. et al. Private and public Hartmann T. Contesting land policies for space for rivers – responses to flood risks. Int J Water Resour Dev 2008, 24, (4), rational, viable, and clumsy floodplain management. J Flood 541–553. Risk Manage 2011, 4, (3), 165–175. Magilligan F.J., Buraas E. & Renshaw C. The efficacy of stream Hartmann T. & Spit T. Legitimizing differentiated flood protec- power and flow duration on geomorphic responses to cata- tion levels – consequences of the European flood risk man- strophic flooding. Geomorphology 2015, 228, 175–188. agement plan. Environ Sci Policy 2016, 55, (2), 361–367. Massachusetts Executive Office for Administration and Finance. Heintz M.D., Hagemeier-Klose M. & Wagner K. Towards a risk MassGIS Datalayers. 2015. Available at: http://www.mass.gov/ governance culture in flood policy—findings from the imple- anf/research-and-tech/it-serv-and-support/application-serv/offi mentation of the “Floods Directive” in Germany. Water 2012, ce-of-geographic-information-massgis/datalayers/layerlist.html 4, (1), 135–156. Mees H.P., Driessen P.J. & Runhaar H.C. Legitimate adaptive van Herk S., Rijke J., Zevenbergen C. & Ashley R. Understanding flood risk governance beyond the dikes: the cases of Ham- the transition to integrated flood risk management in the burg, Helsinki and Rotterdam. Reg Environ Change 2013, 14, Netherlands. Environ Innov Soc Transitions 2015, 15,84–100. (2), 671–682. Holstead K., Kenyon W., Rouillard J., Hopkins J., Galan-Diaz, C. Meijerink S. & Dicke W. Shifts in the public – private divide in Natural flood management from the farmer’s perspective: cri- flood management. Int J Water Resour Dev 2008, 24, (4), teria that affect uptake. J Flood Risk Manage 2015. doi: 499–512. 10.1111/jfr3.12129. Milman A. & Warner B.P. The interfaces of public and private Hornsey M.J., Harris E.A., Bain P.G. & Fielding K.S. Meta- adaptation: lessons from flooding in the Deerfield River analyses of the determinants and outcomes of belief in climate Watershed. Glob Environ Change 2016, 36,46–55.

Milman A., Short Gianotti A.G., Warner B.P. Flooding and pri- Rouillard J.J., Ball T., Heal K.V. & Reeves A.D. Policy implemen- vate land: perspectives and responses among rural landowners tation of catchment-scale flood risk management: learning in Western Massachusetts. Manuscript submitted for from Scotland and England. Environ Sci Policy 2015, 50, review. 2016. 155–165. Morris J., Beedell J. & Hess T.M. Mobilising flood risk manage- Schaaf K.A. & Broussard S.R. Private forest policy tools: a ment services from rural land: principles and practice. J Flood national survey exploring the American public’s perceptions Risk Manage 2014, 9, (1), 50–68. and support. Forest Policy Econ 2006, 9, (4), 316–334. Næss L.O., Bang G., Eriksen S., Vevatne, J. Institutional adapta- Scolobig A., Broto V.C. & Zabala A. Integrating multiple per- tion to climate change: flood responses at the municipal level spectives in social multicriteria evaluation of flood-mitigation in Norway. Glob Environ Change 2005, 15, (2), 125–138. alternatives: the case of Malborghetto-Valbruna. Environ Plan Northeast States Emergency Consortium. Flood. 2015. Available C Govern Policy 2008, 26, (6), 1143–1161. at: http://nesec.org/flood/ Tam J. & McDaniels T.L. Understanding individual risk percep- O’Brien K., Hayward B. & Berkes F. Rethinking social contracts: tions and preferences for climate change adaptations in bio- building resilience in a changing climate. Ecol Soc 2009, 14, logical conservation. Environ Sci Policy 2013, 27, (0), 114–123. (2), 12. Tempels B. & Hartmann T. A co-evolving frontier between land Offermans A. & Valkering P. Socially robust river management: and water: dilemmas of flexibility versus robustness in flood role of perspective dependent acceptability thresholds. risk management. Water Int 2014, 39, (6), 872–883. J Water Resour Plan Manage 2016, 142, (2), 04015062. Terpstra T. & Gutteling J.M. Households’ perceived responsibil- Paton D. Disaster preparedness: a social-cognitive perspective. ities in flood risk management in the Netherlands. Int J Water Disaster Prevent Manage 2003, 12, (3), 210–216. Resour Dev 2008, 24, (4), 555–565. Pielke R. Jr. Nine fallacies of floods. Clim Change 1999, 42, (2), Thieken A.H., Kreibich H., Muller M., & Merz B. Coping with 413–438. floods: preparedness, response and recovery of flood-affected Quartuch M.R. & Beckley T.M. Carrots and sticks: New Bruns- residents in Germany in 2002. Hydrologic Sciences Journal wick and Maine forest landowner perceptions toward incen- 2007, 52, (5), 1016–1037. tives and regulations. Environ Manage 2014, 53, (1), 202–218. Tompkins E.L. & Eakin H. Managing private and public adapta- Raedeke H.A., Rikoon J.S. & Nilon C.H. Ecosystem management tion to climate change. Glob Environ Change 2012, 22, and landowner concern about regulations: a case study in the (1), 3–11. Missouri Ozarks. Soc Nat Resour 2001, 14, (9), 741–759. Tunstall, S., Johnson, C., & Penning-Rowsell, E.. Flood hazard Rasid H. & Haider W. Floodplain residents’ preferences for non- management in England and Wales: from land drainage to structural flood alleviation measures in the Red River Basin, flood risk management. Paper presented at the World Con- Manitoba, Canada. Water Int 2002, 27, 132–151. gress on Natural Disaster Mitigation, New Delhi, India, 2004. Rasid H., Duffy K. & Steuck J. Floodplain management in La World Meteorological Organization Integrated flood manage- Crosse, Wisconsin: newspaper discourse vs. floodplain resi- ment. Geneva, Switzerland: World Meteorological Organiza- dents’ preferences. Focus Geogr 2008, 51, (1), 7–16. tion, 2011. Downloaded March 1, 2014 https://www.wmo.int/ Rogers R.W. & Prentice-Dunn S. Protection motivation theory. pages/prog/dra/documents/IFM_Factsheet_EN.pdf In: D.S. Gochman, ed. Handbook of health behavior research Zaugg M. More space for running waters: negotiating institu- 1: Personal and social determinants, Vol. xxviii. New York, tional change in the Swiss flood protection system. GeoJour- NY: Plenum Press, 1997, 113–132. nal 2002, 58, (4), 275–284.