Farmer identities influence wildlife habitat management in the US Corn Belt

1. Farmer wildlife management practices are critical to conserve biodiversity and ecosystem functions across intensively used agricultural landscapes. Policies and initiatives aimed at encouraging these practices have generally focused on economic incentives, with limited effectiveness. 2. Farmer identity theory addresses the emergence of norms, values and perceptions in farm management and can contribute to the development of policies and initiatives that engage more effectively with farmers and farming communities. 3. Here we evaluate linkages between farmer identity and wildlife habitat management practices in the intensively farmed US state of Iowa. 4. We conducted an exploratory factor analysis using data from a survey of over 1,300 Iowa farmers that asked their opinions on what constitutes ‘a good farmer’. We use logistic regression to model relationships between farmer identity factor scores and contextual variables against participation in a set of habitat management practices. 5. Four ‘good farmer’ identity types were identified and labelled as productivist, soil conservationist, wildlife conservationist and civic-minded. Logistic regression

2. Farmer identity theory addresses the emergence of norms, values and perceptions in farm management and can contribute to the development of policies and initiatives that engage more effectively with farmers and farming communities.
3. Here we evaluate linkages between farmer identity and wildlife habitat management practices in the intensively farmed US state of Iowa. 4. We conducted an exploratory factor analysis using data from a survey of over 1,300 Iowa farmers that asked their opinions on what constitutes 'a good farmer'.
We use logistic regression to model relationships between farmer identity factor scores and contextual variables against participation in a set of habitat management practices.
5. Four 'good farmer' identity types were identified and labelled as productivist, soil conservationist, wildlife conservationist and civic-minded. Logistic regression results indicated that these farmer identity types have highly divergent responses to wildlife habitat management practices among Iowa farmers. Recreational factors may supplement identity and are also influential towards habitat production on farms.
6. We conclude that farmer identity theory offers a critical link between social and ecological processes on Iowa farms. The research presented here quantitatively associates farmer identity theory with wildlife habitat management, increasing our understanding of how wildlife habitat practices and individual-level farmer identities interact.

K E Y W O R D S
biodiversity, rural sociology, social-ecological, wildlife conservation, working landscapes

| INTRODUC TI ON
Over the past half century, farming practices in the US Corn Belt have shifted production from hay, small grains and pasture, to exclusively corn (Zea mays) and soybean (Glycine max) production using larger field sizes and reducing overall perennial vegetation cover (Corry, 2016). As a result, wildlife populations such as grassland birds have declined significantly (Evans & Potts, 2015;Shaffer & DeLong, 2019). Efforts to mitigate agricultural impacts towards farmland bird populations are limited by increasing agricultural intensification (Stanton et al., 2018). Solutions have been proposed through reconciling agricultural practices with habitat production through diversification, improved farm design and through formation of coalitions among farmers, citizens and government agencies to increase wildlife-friendly habitat practices (Kremen & Merenlender, 2018;Landis, 2017;Rosenzweig, 2003). Fundamental to this challenge are addressing knowledge gaps in understanding how social and ecological systems interact, and how diverse value systems negotiate landscape management (Ellis et al., 2019;Mastrángelo et al., 2019).
To increase their effectiveness for both stakeholders and wildlife, conservation strategies must innovate to reposition values, agency and well-being as central to farm policy and initiative development and execution (Hicks et al., 2016). One way to address this is to improve understanding of relationships between farmer identity and wildlife conservation to inform conservation strategy development in agricultural landscapes. Here we apply a farmer identitybased approach to understanding farmer decision-making in relation to on-farm wildlife habitat management practices.
Farmer identity represents the cycle of an individual's subjective understanding of good farming practices which are verified through social and ecological feedback (Burton, 2004;Lequin et al., 2019;McGuire et al., 2015;Sulemana & James, 2014). The farmer identity concept represents a relatively new pathway towards understanding how farmer attitudes and behaviours contribute to environmental conservation (Prokopy et al., 2019), and has not been directly assessed in relation to wildlife habitat management.
There is ample evidence that wildlife habitat can exist in the margins of agricultural landscapes and even complement row-crop farming (Quinn et al., 2014), thereby increasing local biodiversity (Fahrig et al., 2011). Agricultural field margins provide a space for fencing, create boundaries for fields and property, act as wind breaks reducing wind erosion of soil and provide habitat for wildlife through forage and cover resources (Marshall & Moonen, 2002;Schulte et al., 2017). Pollination and pest-control services are provided by vegetation situated in field edges and could be incorporated to a greater extent in farm design (Landis, 2017). Within-field farming practices may also serve wildlife habitat functions with crop stubble and residue containing forage material for migrating birds and deer, and provide a space for the establishment of nests for a variety of bird species (Rodgers, 1983). Reduced tillage during spring or fall and increased use of mulching provides forage resources and reduces destruction of nests by farming equipment (Anteau et al., 2011). The tillage of crop fields has been shown to significantly impact nesting birds by either destroying or altering active nest sites and thus hampering brood success (Rodgers, 1983).
Farmers may implement wildlife-friendly practices on their own if they have the resources and intention (van Dijk et al., 2016); however, agri-environmental schemes (AESs) enacted by governments are a catalyst for many efforts to increase and diversify wildlife habitat on farms and address soil and water conservation concerns.
AESs tend to focus on economic reward structures through direct payments and/or technical assistance (Batary et al., 2011). Yearly AES investments total roughly US $2 billion in the United States, and roughly US $3 billion in the European Union (Claassen et al., 2008;Kleijn & Sutherland, 2003). Despite these large investments, agriculture continues to be a major contributor to biodiversity loss, climate change, soil degradation and water pollution (National Research Council, 2010). Furthermore, the benefits of AESs are often temporary as a slight majority of enrolled land in the United States typically transitions back into a row-crop production land when contracts expire and are not renewed, as well as when commodity crop prices exceed the value of these set-aside subsidies (Bigelow et al., 2020).
In addition to these issues of effectiveness and sustained impact for biodiversity conservation on farms, direct payment incentive programs may also create conflict with the production-oriented farmer identity. First, direct payments can introduce counterproductive incentives for farmers to withhold wildlife habitat friendly management actions in the expectation of payment (Engel et al., 2008).
Second, is that while direct payments may position subsidized habitat production embedded into farming landscapes as equal compensation for the opportunity cost of not farming, production-oriented farmers may see these installations not through an economic lens, but rather as a lost opportunity to demonstrate farming skill and know-how on land that has produced row-crops through several family generations (Burton, 2004). Thus, the implementation of wildlife-friendly practices will likely interfere with symbolic interpretations of row-crop farming by production-oriented farmers and will be viewed negatively in this regard. If this is indeed true, farming identity is likely a good predictor of reactions to practices introducing wildlife habitat onto farms and by extension to farm management behaviour.

| Farmer identity theory
Identity theory provides a useful framework for examining the development of attitudes that reflect behaviours and choices in farming, because it addresses the emergence of norms, values and perceptions towards wildlife habitat management on farms (Burke & Stets, 2009;de Snoo et al., 2013). Identity is a crucial source of meaning in individuals' lives, without which would be a life without purpose or structure, resulting in low self-worth and esteem (Burke & Stets, 2009). Thus, identity motivates behaviours that are reinforced by successive social and environmental feedback and reflected appraisals. Identity theory is rooted in symbolic interactionist social theory, which emphasizes how symbolism embedded in language can reveal social structure and the roles individuals occupy in life (Stryker, 1980).

Several complementary lines of identity theory have been devel-
oped and integrated into social psychology research (Carter, 2013).
An identity salience hierarchy has been proposed, where role choices by individuals are based on salience of identity expressed differently across places and social context (Stryker, 1983). 'Principle-level' identities are overall expressions of one's self-reflection, while 'programlevel' identity is associated with a more specific set of characteristics expressed when required by place and social context. A simple example is the coexistence of both professional and family identities, where professional situation choices and behaviours are different from those when holding the role of family member. An identity control theory has also been proposed where once an identity is activated, one's identity meanings are matched with perception of the self by a psychological control system (Burke & Stets, 2009). Other uses of the identity concept exist, such as through the analysis of 'self-identity' as a component of the factors leading to the intentions that predict eventual behaviour (Ajzen, 1991;Fielding et al., 2008). For this study, we focus on the identity salience hierarchy as described by Burke and Stets (2009) because it provides an appropriate framework for examination of multiple identity types which may be expressed independently given different social and ecological feedbacks.
We propose farmer identity here as an extension of the identity salience hierarchy with all farmers expected to desire the principlelevel identity of being 'a good farmer', and then expressing one or more program-level identities dependent upon social-ecological context ( Figure 1). Farmer identity is an attractive concept to socialecological research in agricultural systems, because it provides a theoretical framework to interpret symbol usage both in language and in visible farming practices. Farmers, in a sense, communicate through land use practices to signal group belonging and selfconcept. This is expressed in concepts like 'road farming', which describes farmers focusing more energy on operations visible by peer farmers gazing from their vehicles than on parts of their farm not visible from the road, or using land use practices such as straight crop rows and cleanly tilled fields in order to signal farming success (Burton, 2004;Nassauer et al., 2011). Furthermore, identity is not static or homogenous, rather it is multiple and hierarchical with the potential to be expressed differently given different spatial, temporal and environmental situations (Burton & Wilson, 2006). production-oriented farmers upon regular meetings with peer farmers organized in watershed groups.

| Research framework and objectives -Farmer identity and wildlife habitat
We build on previous research that examines how farmers express identity through farming practices for soil and water quality concerns in Iowa (McGuire et al., 2013, 2015, by adding wildlife habitat as another key component of farm management. Wildlife habitat management practices are defined here as deliberate land management techniques for the benefit of undomesticated animal life. All farmers are expected to hold the principle-level identity as 'a good farmer'. Beneath the good farmer identity, program-level identities are based on which practices the farmer emphasize in management decisions (Burke, 1991;McGuire et al., 2013). Identity verification occurs after farmers receive social and ecological feedback from their farm management decision. For instance, a farmer with clean fencerows and well-fertilized row-crops, both observes the ecological response and receives positive feedback from peer farmers verifying a productivist identity.
Several farmer identity constructs applied in previous research have been proposed and utilized as composites of attitudes and behaviours impacting farmer environmental management (Burton, 2004;Cullen et al., 2020;McGuire et al., 2013;Sulemana & James, 2014). The already mentioned production-oriented identity refers to farmers' wish to be evaluated by success in yield and profit.
Hence, a focus on fertilizer and pesticide application to produce the greenest crop fields and clean fence margins visibly signal these personal characteristics and provide identity verification to the farmer.
A 'productivist' identity was defined as 'the overwhelmingly utilitarian approach to land use based on intensive forms of agricultural production and accompanying attitudes, goals, roles and behaviours that result in a uniform landscape' (Burton, 2004). Several nonproduction-oriented farming identities have also been proposed.
Agricultural producer, agribusiness person and soil conservationist were proposed as an identity hierarchy within British farmers based on the analysis of qualitative interviews and survey analysis (Burton, 1998;Burton & Wilson, 2006). Cullen et al. (2020) suggested forward-looking, and optimistic caretaker identities to join productivist and conservation-oriented identities, which emerged in their factor analysis of a survey of Irish farmers. In previous work in Iowa, four identities were identified through a component analysis of survey responses asking farmers about a range of scenarios of what it means to be a 'good farmer ' (McGuire et al., 2015). The previous four types identified were productivist, soil conservationist, civicminded and naturalist (McGuire et al., 2015). The productivist and soil conservationist constructs were literature based (Burton, 2004), while civic-minded and naturalist were newly introduced and based on the survey analysis.
Our research objective is to assess relationships between farmer identity and behaviours that promote wildlife habitat practices on farms in Iowa. To our knowledge, there is no research on farmer identity and wildlife habitat management practices. We expect the identity verification process to be quantitatively substantiated by assessing farmer responses to wildlife habitat management practices, and thus demonstrate the salience of program-level identities. We follow the program-level identities proposed by McGuire et al. (2015) which found evidence for productivist, conservationist, naturalist and civic-minded, although we propose conservationist and naturalist to be better expressed as soil conservationist and wildlife conservationist. This distinction is made because the term naturalist refers to appreciation and observation of wildlife species while wildlife conservationist refers to active management behaviours to improve habitat for wildlife species. An active management approach is therefore a better representation of the intent of the survey questions. Beyond these program-level identities, several additional factors may supplement identity response in regard to wildlife habitat management. For instance, farmers may also utilize their land for recreational opportunities like hunting and birdwatching, which could impact their self-reflection as farmers and wildlife habitat management decisions (Macaulay, 2015). Hunting is specifically recognized as a rationale for wildlife habitat management on farms and on private lands (Macaulay, 2016). Birdwatching has also been shown to be positively related to conservation behaviours (although not specifically on agricultural land; Cooper et al., 2015;McFarlane & Boxall, 1996). Farmer age and farm size may also influence identity expression, as each of these factors have been shown to predict overall adoption of conservation practices (Prokopy et al., 2019).
Taken together, we are interested in determining if identity constructs along with recreational and demographic features of farms are influencing wildlife habitat management in Iowa. Results will provide improved understanding of farmers' perspectives on use of wildlife-friendly practices on the farmland they manage, and address a research gap in our understanding of how farmer identity leads to adoption of wildlife habitat management by farmers (Prokopy et al., 2019). Improved understanding of farmers' perspectives may ultimately enable the next generation of wildlife conservation initiatives in the working landscapes of the US Midwest by illuminating how different outreach efforts may lead to improved outcomes.

| Research methods
We used survey data from the 2010 Iowa Farm and Rural Life Poll We measured the multi-dimensional 'good farmer' identity through a 31-item question set. Each item was preceded by the statement, 'A good farmer is one who…' and respondents were asked to rate the importance of each statement on a 5-point scale, ranging from (1) 'Not Important at All' to (5)  Control variables of farm demographics were included because they may supplement activation of farmer identity constructs.
Farmer age and total cultivated area are included because differences in these characteristics may also have an effect on the response variable (Prokopy et al., 2008(Prokopy et al., , 2019. Two additional survey items that asked if the farmer had birdwatched or hunted in the previous year were included in the model. These recreational variables may also influence the dependent variable, as they suggest that the farmer actively uses wildlife habitat on their property. All statistical analyses were completed in R using base functions and the psych package (R Core Team, 2019; Revelle, 2019).
Exploratory factor analysis (EFA) with oblimin rotation reduced responses of the 31 good farmer queries into a set of latent constructs.
The EFA is the most suitable dimension reduction technique (as opposed to Principal Components Analysis) because we hypothesize a set of four underlying constructs which will describe program-level identities (Osborne et al., 2014). A scree and parallel plot analysis guided the number of factor solutions in addition to previous research classifying farmer identity into four program-level identities using the same survey (McGuire et al., 2015). Factor scores were estimated using the Thurstone method (regression-based weights) for use as independent variables in a series of logistic regression models. Cronbach alpha reliability coefficients were calculated for each identity construct to evaluate reliability. Logistic regression model diagnostics indicated that no model assumptions had been violated.
Diagnostics included examining for linearity among the predictor and dependent variables, multicollinearity by evaluating the variance inflation factors and influential values within the standardized residuals.

| RE SULTS
Farmer use of the five wildlife habitat management practices ranged from 18% to 74% (Table 1) We found evidence for four program-level identities through the EFA (Table 2) Table 2). The soil conservationist identity construct had high factor loadings for soil and water quality concerns including nutrient run-off, soil erosion, management of soil organic matter and the health of streams. Managing for both profit and environment also loaded onto the soil conservationist identity.
The wildlife conservationist identity construct loaded onto items that would be expected to benefit wildlife, including cover cropping, maintaining wildlife habitat, minimizing pesticides and fungicides and avoiding and minimizing tillage events. The civic-minded identity construct had item loadings that suggested they were active in their community and in farm organizations, were community leaders, helped friends and neighbours with farm tasks and shared knowledge with other farmers.
Results of the binary logistic regression analyses confirmed a divergence in farmer identity types activated in response to the wildlife habitat practices presented (Table 3). The Pseudo R-squared statistics indicate that the models explained between 11 and 16% of the variance in the dependent variables. In the first model, the dependent variable 'Leave brushy and weedy fencerows at least 3 feet wide', was negatively associated with the productivist identity and positively associated with the wildlife conservationist identity, both results were significant (p < 0.01). In terms of magnitude, the odds ratios indicated that a 1 standard deviation (SD) increase wildlife conservationist identity score corresponded to a 79% greater likelihood that the farmer would have reported leaving brushy and weedy fencerows at least 3 feet wide. On the other hand, for the productivist identity construct, a factor score increase of 1 SD was associated with a 23% decrease in likelihood of wildlife-friendly fencerows.
Neither the soil conservationist nor the civic-minded identities were significant in the model.
In model 2, 'Have planted CRP acres to trees for wildlife', the productivist farmer construct was significant and negatively associated with the dependent variable (p < 0.01), while the wildlife conservationist identity was significant and positive (p < 0.001; Table 3).
Similar to model 1, a 1 SD increase in productivist identity factor score was related to a 24% lower likelihood of avoidance of planting CRP land to trees, and a 1 SD increase in wildlife conservationist identity score was related to an 83% greater likelihood of planting CRP land to trees. Again, neither the soil conservationist nor the civic-minded identities were significant in the model.
For model 3, 'Avoid mowing areas such as grass waterways and ditches until late summer to allow birds to nest', the soil conservationist and wildlife conservationist identity constructs were statistically significant predictors (p < 0.01). The odds ratios indicate that a 1 SD increase in soil conservationist identity and wildlife conservationist identity corresponded to a 35% and 42% greater likelihood, respectively, of avoiding mowing. In this model, neither the productivist nor the civic-minded identities were significant.

Results for model 4, 'Have planted prairie grasses and flowers',
showed the productivist identity construct to be negatively associated with planting prairie (p < 0.05), while both the soil and wildlife conservationist constructs were positive predictors (p < 0.05). A 1 SD increase in productivist orientation was related to an 18% lower TA B L E 1 Definitions for farmer identity constructs, control variables and wildlife habitat practices

Definitions based on exploratory factor analysis results References
Productivist A farmer who has the most up-to-date equipment and technology, strives for high profits and yields, clean fields, clear fencerows and tries to get his/her crops planted first Burton (1998) greater likelihood of that behaviour respectively. Again, the civicminded identity was not activated.
In the final model, 'Avoid fall tillage to leave food and cover for wildlife', the productivist identity construct was associated negatively (p < 0.05) and the wildlife conservationist identity was positive (p < 0.01). The odds ratio indicated an 18% decline in likelihood of avoiding fall tillage for each 1 SD increase in productivism factor score and a 79% increase in likelihood for similar increases in the wildlife conservationist factor scores. For the fifth time, the civicminded identity was shown to be inconsequential.
Among the control variables, the birdwatching and hunting variables were significant and strong positive predictors across all models. For birdwatching, the odds ratios indicated that a 1 SD increase translated into increases of likelihood of wildlife-friendly practices TA B L E 2 Exploratory factor analysis results (n = 1,048). The average Cronbach's alpha is reported under grouped items which loaded onto the same factor between 50% and 99%. For hunting, the magnitude of effect size ranged between 37% and 2.18 times more likely to use the practices.

Good farmer items -'A good farmer is one who…'
The size of cultivated area was significant in the 'avoid fall mowing model', but the magnitude of effect was negligible. Finally, farmer age was not a significant predictor in any of the models.

| D ISCUSS I ON
Results provide evidence that farmer identity constructs can be robust predictors of Iowa farmers use of key wildlife habitat management practices. The most consistent predictors were the productivist and wildlife conservationist identity constructs. In four out of five models, the productivist farmer identity was associated with substantially lower likelihood than farmers engaged in wildlife habitat management practices, while the wildlife conservationist identity was strongly related to higher likelihood of wildlife habitat manage- To examine statistically if recreational variables are independent of identity constructs, we conducted a follow-up analysis using a chisquared test to assess differences in overall effect on prediction of weedy suggesting that contributions towards predicting the wildlife habitat management practice was supplementary rather than overlapping.  (Mullendore et al., 2015). This points to an opportunity to engage with farmers by appealing to the wildlife conservationist identity (which may not be the most salient program-level identity) by proposing improvements in recreational opportunities through habitat improvement.
The diverging results between productivist and wildlife conservationist identities have direct policy and advocacy implications.
The negative relationship between productivism and wildlifefriendly behaviours suggests that programs that promote wildlife habitat management practices as something that a productivist- but often only over the short term (Claassen et al., 2008), might achieve longer term benefits by taking identity heterogeneity into account as part of any future strategy.
Our results showed that productivist farmers were less likely to both plant trees to CRP or plant prairie. Resistance to actively increase wildlife habitat on farms may come from a desire to maximize yield, which require maximizing total crop area. It may also stem from a lack of autonomy in the process of deciding where to install wildlife habitat. Bottom-up solutions that put farmers in control of the process might be a pathway to reduce negative reactions to initiatives to increase wildlife habitat. Design processes led by farmers may create potential for innovative solutions responsive to productivist sensibilities that value results rather than prescribed process (Cullen et al., 2018). Abandoning prescriptive-based solutions could be explored using a result-oriented framework to encourage conservation behaviours which might altogether avoid negative responses from production-oriented farmers by engaging with their sense of autonomy (Stock & Forney, 2014).
Relative to prior research on farmer identity, we believe that our approach of using a large set of 'good farmer' questions is more comprehensive than approaches measuring identity through a single or limited number of proxy statements (Sulemana & James, 2014;van Dijk et al., 2016). The concept of a good farmer appears to allow respondents to describe their internal sense-making process with more clarity, especially when provided a large number of well-constructed survey items. The good farmer items used here after demonstrating EFA simple structure were 30 items which addressed a more comprehensive view of identity including attitudes, goals, behaviours and roles. We suggest that future farmer identity research maintain a large number of items for statistical reliability and improve on our questions in order to examine nuances in the identity constructs we proposed here (Osborne et al., 2014).
It is possible that future research could identify a principle identity concept more encompassing, touching on more holistic rationales for farm management. The good farmer concept was developed through in-depth qualitative interviews (Burton, 1998 spatially mediated peer effects were found to influence conservation practice adoption and therefore could be leveraged to increase the chance of initiative success (Kolady et al., 2020).

| CON CLUS IONS
The research presented here extends farmer identity theory into the realm of wildlife management and increases our understanding of how wildlife habitat practices and individual-level farmer identities interact. Farmer identities were tested for activation using five wildlife habitat management practices and with recreational and demographic covariates. Productivist, soil conservationist and wildlife conservationist identity types were readily divided based on their factor scores and statistically significant when regressed against wildlife habitat management practices.

ACK N OWLED G EM ENTS
The Iowa Farm and Rural Life Poll survey was supported in part by the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa, which is supported by USDA-NIFA and State of Iowa funds.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest either personal or financial that has influenced preparation of this manuscript.

AUTH O R S ' CO NTR I B UTI O N S
A.P.D. and J.G.A. conceived the ideas and designed the methodology; the US Department of Agriculture National Agricultural Statistics Service collected the data; A.P.D. and J.G.A. analysed the data; A.P.D. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.