! SOCIAL ATTRACTORS IN AGRO -ECOLOGICAL SYSTEMS: AN ENHANCED PERSPECTIVE ON THE RESILIENCE OF NITROGEN FERTILIZER POLLUTION By Matthew Kaleb Houser A THESIS Submitted to Michigan State University in partial fulfillment of the requireme nts for the degree of Sociology -Master of Arts 2015 !ABSTRACT SOCIAL ATTRACTORS IN AGRO -ECOLOGICAL SYSTEMS: AN ENHANCED PERSPECTIVE ON THE RESILIENCE OF NITROGEN FERTILIZER POLLUTION By Matthew Kaleb Houser The most significant contributor to exce ss environmental nitrogen (N) in the US is agricultural fertilizer application . Using a mixed methods approach, this study identifies social processes that drive Michigan corn farmersÕ application of nitrogen in excess of crop demand. We use HattÕs (2013) recently developed s ocial attractors framework to conceptualize excess N application as a resilient practice actively reinforced by ongoing structural and organizational influences. The social attractors framework significantly improves resilience theoryÕs conceptualization of social systems and may facilitate social scientistsÕ capacity to engage in coupled systems research. Despite its analytical potential, the social attractors framework has yet to be applied empirically. This work contributes to the ag ricultural pollution mitigation literature , as well as explores the usefulness of social attractors for social -ecological systems (SES) research. Our findings indicate that the greater number of acres planted, more reliance on personal experience in nitrog en application decisions, and being unaware that nitrous oxide is a greenhouse gas all increase the likelihood of apply ing nitrogen in excess of crop demand. The social attractor framework proved useful in identifying these influential processes, theorizin g their relationship to broad social values, organizations and social structures and conceptually framing N rate as an actively resilient practice in a n nonlinear SES feedback system . !"""!ACKNOWLEDGMENTS The faculty and students of the Sociology Department have all been gracious guides along the path to this degree. Of these individuals, my committee members stand out for their support. Dr. Stephen Gasteyer, who was kind or maybe fooli sh enough to accept me into the program as an under -qualified applicant ha s given pleasant and patient insight since then. Dr. Sandra Marquart -PyattÕs willingness to answer my seemingly endless questions with unmatched enthusiasm and humor has inspired both my intellectual pursuits and spirits. Dr. Diana Stuart, the Chair of thi s thesis, is another individual who was either kind or foolish. She offer ed a young graduate student much needed funding for what will be the majority of my time at Michigan State. Without her generosity, the wonderful experiences of my past two years woul d not have been realized . Since this crucial point, her feedback and encouragement was p ivotal to the successful complet ion of this paper. The endlessness of my parentsÕ love and support is incomprehensible and my gratitude cannot begin to be communicated . What more could a child ask for? The desire to eventually close the geographical distance between us spurs my work onward. My fianc”e Lauren Õs compa nionship was indispensible. As I have become more engrossed in acade mia, her patience for my changing char acter amaze s me. The need to be in your presence every evening compelled me to work determinedly during the day s. My love for my parents and Lauren continually reminded me that any academic setbacks faced were but minor trials in an already successful life . !!"#!TABLE OF CONTENTS LIST OF TABLESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉV LIST OF FIGURESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.Vi KEY TO ABBREVIATIONSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.....Vii INTRODUCTIONÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.1 RESILIENCE THEORYÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...4 Social Attractors ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...........................6 SYNTHETIC NITROGEN'S ENVIRONMENTAL CONSEQUENCESÉÉ......10 The Resilience of Nitrogen Pollution ÉÉÉÉÉÉÉÉÉÉ..................12 MODELSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.14 Samples ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ....15 SOCIAL ATTRACTORSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...17 Individualism/Personal Experience...................... ..........................................17 ProfitsÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉ...ÉÉÉ...18 Reported Kn owledge of ConsequencesÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉ..20 Information SourceÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .22 Acres of Corn Grow nÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉ.. 22 QUANTITATIVE METHODSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...25 Dependent VariablesÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉ...25 N-Rate............................................. ....................................................25 Awareness of N 2O as a greenhouse gas (GHG)ÉÉÉÉÉÉÉÉ. 25 Indep endent VariablesÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉ26 Personal ExperienceÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉ.......26 Profit ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...........26 Infor mation Sourc esÉÉÉÉÉÉÉÉÉÉÉ... ÉÉÉÉÉÉÉ.26 Reported KnowledgeÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉ..26 Acres of Corn ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. ..27 THE SOCIAL ATTRACTORS FRAMEWORKÉÉÉÉÉÉÉÉ..................29 RESULTSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ31 Discussion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ................33 CONCLUSIONÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..................38 BIBLIOGRAPHYÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...É....40 !!#!$%&'!()!'*+$,& !! '-./0 !12!304567-/!,8904"07:0 ;;;;;;;;;;;;;;;;;;;;;;;;;;;1< !!'-./0!=2!346>"?5 @;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;@@=A !!'-./0 !B2!C05:4"9?"#0!D05E/?5 !;;;;;;;;;;;;;;;;;;;;;;;;;;;@@=F!!'-./0!G2!)"7-/!H6I0/@ ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;@@B=! !!#"!LIST OF FIGURES !)"JE40!12!D05"/"07:0;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;@@ G!!)"JE40!=2!&6:"-/!*??4-:?64 5!D05"/"07:0 ;;;;;;;;;;;;;;;;;;;;;;;@1B !!)"JE40!B2!KL96?M05"N0I!&,&!)00I.-:O! 346:055;;;;;;;;;;;;;;;;;;@BA !!)"JE40!G2!)"7-/!H6I0/ ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;BB !#""!P,Q!'(!*++D,R*'%(S& !!!+H3T+05?!H-7-J0U07?!34-:?":05 !!VKVTV4007M6E50!V-5 !!S=(TS"?46E5!(8"I0! !!STS"?46J07 !!S3&TS67W96"7?!&6E4:0 !!&,&T&6:"-/W,:6/6J":-/!&L5?0U5 !! !1!INTRODUCTION ! In 1973, resilience theory was debuted by C.S. Holling in his seminal paper Resilience and Stability of Ecological Systems. Resilience theoryÕs co ncept of ecological systems that do not achieve peak equilibrium of function but rather fluctuate between different ecological attractors or stable domains of ecological processes radically altered ecological thought (Curtin and Parker 2014). Resilience th eory has since been used by multiple discipline s to understand ecological as well as social -ecological systems (SES ) (Folke 2006) 1. However, applications of resilience theory to social systems (e.g. Adjer 2000; Gunderson and Holling 2001) have been critiq ued for failing to capture the influence of individual/collective agency, social structural power and offering an overly functionalist , or mechanical, depiction of society (Cote and Nightingale 2012; Davidson 2010; Hatt 2013 ; Hornborg 2009 ). In response to these critiques , Hatt (2013) developed the concept of social attractors. Functioning similarly to ecological attractors, social attractors are intended to enhance resilience theoryÕs ability to concept ualize the influence of power and agency within social systems. We apply HattÕs (2013) framework to Michigan corn farmersÕ use of nitrogen (N) fertilizer . Agricultural N fertilizer pollution, a form of nonpoint source (NPS) pollution , significantly degrades environmental quality and poses health risks to hum ans. It is estimated that nearly all freshwater and coastal zones in the US are degraded by N pollution (Baron et al. 2012) and that 20% of drinking water in agricultural regions contains N levels beyond the safe drinking limit (Agrawal et al. 1999). Thoug h numerous nonpoint source (NPS) pollution mitigation strategies and policies have been attempted, !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1 Social -Ecological Systems is also often referred to as Coupled Human and Natural !!=!their success has proven difficult to evaluate (Knowler and Bradshaw 2007; Sergerson and Walker 2002). The inability to reduce the prevalence of agricultural nutrient runoff leads Morton and Brown (2010) to argue that more attention needs to be paid to Òthe persistent and difficult problem of nonpoint source pollutionÓ (3). This persistence may be better understood through further investigation of social fact ors that motivate farmers to continue inefficient nutrient management practices. In this paper the determinants of applying nitrogen in excess of crop demand, which is often considered the leading cause of nitrogen leaching and oxidization 2 (Broadbent and Rauschkolb 1977; Grant et al. 2006; Halvorson et al. 2008; Hoben et al. 2011; McSwiney and Robertson 2005; Miller et al. 2009; Robertson et al. 2013), is inspected using HattÕs (2013) social attractors framework. The social -ecological features of agricultu ral N application and pollution is re cognized through the social attractors framework, as are the social forces that actively compel, constrain or justify farmers Õ decisions. This exploration of the social determinants of a farmerÕs N application decision s contributes to agro -food studies and may serve to guide policies and programs aimed at reducing nitrogen fertilizer pollution. Further, as HattÕs (2013) social attractors approach to understanding resilience in SESs has n ot yet been empirically applied, our research will begin to explore the usefulness of this concept for SES research. It is our aim to suggest the framework Õs potential to (1) enrich the understanding of social drivers of ecol ogically impactful behavior (2) provide a means for more social scientists to engage in SES or coupled systems research and (3) reveal how individual -level decisions and !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!2 See Snyder (2009) for a full review !!B!behavior are embedded and thus partially determined by contextual factors at multiple levels. !G!RESILIENCE THEORY C.S. Holling Õs (1973) resilience theory altered the traditional notion of an ecological system that could attain a peak equilibrium, or single ÒclimaxÓ stable state of biophysical processes. Instead, it Òillustrated the existence of multiple stability domains or multiple basins of attract ion in natural systems and how they relate to ecological processes, random events (e.g. disturbances) and heterogeneity of temporal and spatial scalesÓ (Folke 2006 , 252). In resilience theoryÕs conception of ecosystems, resilience refers to the capacity of a system to undergo disturbances, yet still maintain its current set of functions before shifting to a new stable state, with new processes and functions (Holling 1973 ). Resilience in ecosystems is often illustrated using a ball in a basin (see Figure 1 ), which represents a system in a stable state, placed between empty adjacent basins representing potential new stable states. In each basin, various social or ecological variables known as attractors can cause the ball , which signif ies the ecosystem , to shi ft into a neighboring steady state. The width or steepness of each basin represents an ecosystemÕs resilience . !!!!!!!!!!!!!! !"#$%&'( )'*&+","&-.& 'Adapted from Peterson (2000, 326) !!X! In 1993, an interdisciplinary research group at the Beijer Institute developed a framework to understand the linkages between social and ecological approaches titled Social and Ecological Systems , or SES (Curtin and Parker 2014) , using it to examine a number of topics (see Folke 2006 for a brief review) . The SES framework emphasizes the mutual exchanges between humans and the ec osystems in which they are embedded (Gunderson and Holling 2002; Holling 1973). This concept has led to a fuller recognition within the scientific community of interactions between social and natural systems (Cote and Nightingale 2012). Despite this pot ential , past applications of resilience to the social component s of a SES (e.g. Adger 2000 ; Adger et al. 2005; Gunderson and Holling 2001 ) have recently been critiqued (Cote and Nightingale 2012; Davidson 201 0; Hatt 2013; Hornborg 2009). These critiques focus on the failure to incorporate a nuanced understanding of the multiple social dynamics within a SES that may drive ecological ly impactful behaviors or shape the desirability of a given ecological stable state. Influenced heavily by a functionalist persp ective, it has been argued that traditional SES applications reify conceptions of society and thus obscure constraining or motivating factors such as power, social norms, ethical standpoints, etc., which drive individualsÕ, communitiesÕ or populationsÕ act ions (Cote and Nightingale 2012; Hatt 2013; Hornborg 2009). In general , these scholars call for the inclusion of the motivating and /or constraining factors that influence a social behavior and to acknowledge an individual Õs or a sub -populationÕs agency in consent , dissent, or adaptation to a given action or event. !!Y!Social Attractors In response to these critiques, social scientists have started to further develop resilience theoryÕs conception of society (e.g. Cote and Nightingale 2012; Hatt 2013 ). Within this effort, Hatt (2013) developed the concept of social attractors . Social attractors function analogously to ecological attractors, but instead are social devices or processes , such as policy , laws or rhetoric, that Òpersuade, attack, dissuade, justif y, or neutralize actions or projectsÓ (34). They are defined as Òdiscursive (i.e. rhetorical) formulations that serve as reference points in social processes associated with the construction, mobilization, establishment, contestation, and resistance of pow erÓ (Ha tt 2013, 34). Hatt envisions social attractors operating in a critical realist conceptualization of society . In contrast to a functionalist social system model, critical realism depicts social relations as stratified along three levels: social str ucture (e.g. capitalism ), organization (e.g. the state) and social interaction s ( e.g. individuals /civil society ) (Archer et al. 1998; Bhaskar 1998; Hatt 2013; Joseph 2000 , 2002). This model posits that the layers of society are in a dynamic relationship of mutually shaping feedbacks. O rganization s and social interactions are embedded within and shaped by social structure s. Social structures are in turn reproduced or transformed through organizational and individualsÕ actions . Social structureÕs influence on individuals is not direct, but rather mediated through organizations. Organizations, such as the state or institutional policies, shape social activities through laws or normative practices to reproduce social structure. Bill Freudenberg has emphasized th e use of middle range theories in environmental sociological research, which are theoretical frameworks that are testable through the !!F!integration of empirical analysis (Merton 1949). To achieve this Òmiddle range,Ó this analysis will focus on the individua l level, using theory to link observed social attractors at this level to the organizational/structural levels in which it is embedded. While the influence of social structure and organizations may constrain or compel behaviors, the means through which t hey exert influence can also enable the contestation of their influence. As Hatt (2013 , 36)) notes, Òthe processes of organizing consent may also create opportunities for constructing [counter] movements and resistanceÓ (citing Carroll 1990). According to Gramsci (1971), diverse techniques exist to organize consent for, or potentially contest, social activities and policies . It is from this notion of Gram sciÕs that the social attractor s concept was inspired . Hatt (2013) depicts them as the social devices, r hetoric, policy, etc., that motivate , coerce or justify individualÕs actions. To illustrate the concept of social attractors in a SES , Hatt (2013) uses a n anecdotal example of humans living in close proximity to a lake. In this example, nature , property and conservation are identified as three social attractors, along with two traditional ecological attractors. Hatt (2013) argues that through these three discursive constructions , social actors organized either consent or dissent for given practices that affect the lake Õs biophysical stable state . For example, through reference to conservation, individuals justified or opposed the use of powerboats on the lake , which potentially increases pollution . Conservation is a value that has Òthe status of being a pol itically correct orientation that cannot be denied without some sense of stigmaÓ ( Hatt 2013, 37). While HattÕs example focuses primarily on social attractors that are discursive, the potential for more formally articulated and forceful attractors, such as policy or !!?2.%& '!1G!MODELS The SES nature of N pollution demands joint consideration of both social and ecological factors. Based on the need for further understanding of the latter, this study gives priority to conceptualizing social determinants through the social attractors framework. Social attractors are conceptualized as operating in non -linear feedback loops . This study identifies what social attractors are related to N application on corn . To determine th is w e use qualitative and quantitative data from interviews and surveys of Michigan corn farmers in 2011 along with an extensive review of relevant literature . Survey and interview data are used concurrently in our results section. A mixed methods approach is called for based on the limited amount of relevant prior research. Qualitative analysis, along with a review of relevant literature, will reveal what determinants are important. Following this, we use survey data to construct quantitative variables to represent the determined social attractors and test their effects on the likelihood of over -apply N using logistic regression . This approach tests the applicability of the social attractors determined in qualitative analysis to the wider sample through qua ntitative methods. Adhering to HattÕs (2013) nonlinear feedb ack loop conceptualization, w ater quality and atmospheric conditions are understood to be affected by N application rate s which in turn affect social attractors , although the exact effects are no t measured in this paper . This intention of this study is to reveal important determinants of behavior related to inefficient nitrogen use. Future work can expand on these findings to include observed effects of ecological factors. !!1X!Samples Qualitative d ata were gathered from four focus groups and 40 in-person interviews with Michigan corn farmers. Focus groups were conducted in four Michigan counties: Branch, Calhoun, Kalamazoo and St. Joseph . These four counties represented 14% of MichiganÕs total corn production in 2011 (Kalamazoo (2.9%), St. Joseph (4%), Calhoun (3.5%), Branch (3.7%)) (USDA 2011). Each focus group consisted of participants identified by local M ichigan State University Extension (MSUE) agents, recruited at local farm meetings, or invite d by another participant. Participation ranged from five to eight farmers in each group. The same list of questions guided each discussion . Interviews were conducted between January and May 2011. MSUE facilitated initial contact s, and additional participan ts were recruited through snowball sampling. In total, 11 farmers in Calhoun County, 9 in Kalamazoo County, 12 in St. Joseph County, and 8 in Branch County were interviewed. Of those interviewed , 23 were commercial corn farmers, 11 were strictly seed corn farmers (produce the seeds for commercial corn) , and 11 grew both seed and commercial corn. Interviews were conducted on -farm, using an interview guide, and recorded whenever possible. Interview and focus group questions focused on factors influencing fert ilizer application, willingness to reduce N fertilizer, and interest in a potential offsets program. Data for quantitative analys es were collected through a mail survey conducted during the spring of 2011. With the assistance of the National Agricultural Statistical Service , 1,000 surveys were mailed to corn farmers in the four southwest Michigan !!1Y!counties mentioned above. Based on acreage, farms were divided into four strata. Different sampling rates were applied to each stratum with the intention that the final sample would adequately represent the different strata. Of the 1,000 mailed surveys, 274 were returned (27.4%). No significant differences exist between this studyÕs respondentsÕ farm size, irrigated acres, age, education, and farm income and that o f respondents to a 2008 statewide survey of corn and soy farmers with a 56.4 percent response rate (Jolejoy 2009). Based on response rates related to varieties of corn grown, this analysis applies to un-irrigated commercial corn farmers and thus may not ge neralize to other varieties (e.g. seed corn farmers). The social drivers of nitrogen application rate s that emerged as common themes are used to construct relevant social attractors. Potential social attractors determined via past literature compliment th ese. While not all are discursive, th ose discussed below represent social attractor s based on (1) their conception of the potential to simultaneously motivate appropriate or excessive application rates and (2) their theoretical link to social structures or organization. In the section following th e construction of social attractors, their quantitative equivalents are outlined along with statistical m odels . !1F!SOCIAL ATTRACTORS Individualism/Personal Experience Interviews indicate that farmersÕ references t o past N application experience s are used to justify rejecting external N rate recommendations (see Table 1). Personal experience is conceptually an expression of individualism, which is the widespread American value legitimizing individual choice and pers onal freedom (Inglehart 1997; Sampson 2001). In centralizing personal autonomy, it justifies the marginalization of other social influences (Oyserman et al. 2002). This was reflected in f armersÕ comments. For instance, one commented, ÒI wouldnÕt say [ferti lizer dealers, buyers, other farmers or extension] really influence [our application rate]. I mean they might a little bit, but I guess I know what our ground is capable of and they might think they know. Ó Survey results show that personal experience is a prominent influence on the application rate for the majority of farmers. Compared to other social sources of information ,6 personal experience was most often cited as somewhat or very impo rtant (85%).7 The prominence of farmersÕ reliance on personal expe rience over external information indicates its relationship to individualism. Through its link to individualism, personal experience is conceptualized a s a discursive social attractor that is deployed to justif y excluding the influence of external informat ion sources on farm management decisions. As it is discursive, the concept of individualism that is embedded within reference to personal experience does not exist at any level of society exclusively. Its deployment by farmersÕ may reflect influence from sources at the organizational level on N decisions and/or a means to justify a self -determined practice. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!6 The three following personal experience were (1) fertilizer dealers (53%), (2) universities (48%), (3) other farmers (33%). 7 Percentages do not equal 100 %, as rankings were not mutually exclusive. !!1!-99/":-?"67!"5!J0704-//L!56!:/650!?6!?M0!.6??6U!-7I!%!JE055!E5E-//L!%!I67]?!>00/! ?M-?!%!:6E/I!:E?!-!/6?!^6>!7"?46J07_!`"?M6E?!5"J7">":-7?/L!40IE:"7J!UL!L"0/I@a !!*5!6>!4"JM?!76`;%]U!9E??"7J!?M0!/0-5?!-U6E7?!?M-?!%!?M"7O !%!:-7!-7I!5?"//!J0??"7J!?M0!U-8!L"0/I!>64! 0:676U":5;>46U!-7!0:676U":!5?-7I96"7?!"?b5!76?!>0-5"./0!^?6!40IE:0!>04?"/"N04!-99/":-?"67_@a !!\%!`6E/I!/6#0!^?6!40IE:0!>04?"/"N04!-99/":-?"67_!.0:-E50!"?!5-#05!U670L c-5!/67J!-5!`0!-407]?!:65?"7J!6E450/#05! L"0/I@!\ !!ÒIf I put fertilizer on and I lose it then itÕs not economical . . . so IÕm going to put it on the most economically friendly way. I donÕt want to throw my money away.Ó !!!=1!similar to awareness of N 2O consequences ,8 finding it not significantly predictive of conservation practice adoption. This may be a result of limitations (economic or otherwise) on farmersÕ ability to adopt conservation practices. Knowledge related to N 2O may driv e behaviors. Its possession is likely determined by information networks in which a farmer is embedded . Information networks, as with other types of organizations, are mediators of social structures influence on individuals. Their influence reflects the structure that supports them, and thus the information they provide may motiva te social actions that recreate this structure . What organizations supply this information to farmers is unclear. Though unable to theorize about a particular organizations interest in offering this knowledge, our investigation does indicate what influence this knowledge has at the individual level. For reported awareness of N 2O as a GHG to act as a significant deterrent , farmers will likely have to believe in anthropogenic climate change. Recent studies of farmers show this might not be the case , however . Surveys of Midwestern corn farmers show that 54% of Iowa (Arbuckle et al. 2013 ) farmers and 66% of Indiana farmers (Gramig et al. 2013) believe in climate change. However, out of these 54% in Iowa, only 35% perceive human actions to be a cause, and this i s even lower in Indiana where a mere 8% of those 66% see this as the current situation . If we can generalize to other farming communities in the Midwest, i t is likely that the majority of respond ents in our study see climate change to be unassociated with human activities . Thus, awareness that N2O is a GHG may not affect application, as it does not capture farmersÕ perception s of anthropogenic climate change. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!8 Baumgart -Getz, Prokopy and Floress (2012) calls this variable ÔcauseÕ. It is also included in research conducted by Esseks and Kraft (1988). !!==!Information Source We also identified primary sources of information related to N application as a social attractor. Similar to the influence of information networks mentioned above, information from specific organization s or institution s reflect s the biases or interests of these organizations and thus function to re create specific social structures. Fertilizer dealers likely have an economic incentive to recommend high rates, especially compared to sources of information such as university extension educators or literature created in different social arenas . As illustrated in Stuart et al. (2012), man y farmers recognize that information provided by fertilizer dealers is biased . Despite this recognition , many farmers in our sample consider fertilizer dealers the most important source for information (27.2%). This was followed by university recommendatio ns with approximately 22% rating it as their most important source. The source farmers trust will likely direct their final application decisions. Other s have show n the importance of farmersÕ social connections in influencing N decisions (Lubell and Fulto n 2008; Prokopy et al. 2008). More generally, in their meta -analysis of conservation ado ption literature Baumgart -Getz et al. (2012) find information about best management practices to be the best predictors of practice adoption. We identify primary source s of information about fertilizer application as a social attractor given that they are social processes that shape farmerÕs actions and likely reflect the interests of organization structures from which they emerge . Acres of Corn Grown Though not discuss ed thoroughly in interviews or focus groups, national changes in farm structure represent a potentially powerful social attractor. The increasing capital !!=B!intensity (i.e. more machines, inputs, etc.) in agriculture reflects the structural force of capitalis m ( Lewontin 2000) and has compelled or enabled farmers to manage more land. This is reflected in the growing size of US farms. As of 2007, the midpoint 9 acreage harvested since 1982 has risen 114%, from 500 to 1,071 acres. This is particularly true for Mid western farms, including Michigan, which have seen the greatest percentage increase in acreage. Additionally, cornÕs midpoint acreage increased more than any other field crop , jumping 300% (from 200 to 600 acres) between 1987 and 2007 (MacDonald et al. 201 3). This change in American farm structure functions at the organizational level, as it possibly mediates the influence of structure on the individual. Capitalism (a social structure) dr ives changes to the national farm structure (organizational) and we e xamine its affects on operators Õ actions (individual). Larger farms could enable or prevent a reduction in N rates. Some studies have considered acres farmed an indicator of economic capital which is thought to measure the capacity to adopt ÔriskyÕ practic es such as conservation practices (Baumgetz et al. 2012, Daberkow and Mcbridge 2003). Alternatively, more acres could also result in increased risk related to N application as more crops planted equates to greater potential profit losses in poor years. N o ver -application is a form of risk management to avoid the potential consequence of unprofitable yields (Sherriff 2005 , Stuart et al. 2012 ), and thus may be a result of changing farm structures. Farm size functions as a social attractor in that it is driven by !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!9 Midpoint measurements are a form of medians. As MacDonald et al. (2013) explain, Òmidpoint acreage is the median of the distribution of acreage by farm size, as opposed to the more commonly reported median of the distribution of farms by farm sizeÓ (6). This measure is used to reduce the influence that a recent spike in the number of very small farms have on average acreage. !!=G!social structural influence on farming that may compel or motivate individualÕs appropriate or excessive N application rate s. The social attractors noted above are considered because of their conceptual links to broader social structures /organizations . They function as indirect influences from a social structure, such as acres of cornÕs link to capitalism or information sources connection to institutions or organizations. Others (e.g. economics and personal experience) also reflect the rhetorical attrac tors that Hatt (2013) focuses on . Further, all potentially motivate either appropriate or excessive N rates and thus incorporate the nuances of individual agency. If we were applying HattÕs (2013) framework precisely , a theoretical framing of these social attractors Õ nonlinear feedback relationships to N application would follow. Instead, a quantitative analysis is used to assess the noted social attractorsÕ influence on the likelihood of over -applying N. Though t his application flattens the conceptual capacity of attractors to be simultaneously motivators of both excessive and conservative application rates, it provides certain benefits. We are able to verify statistically if a connection exists between influence and actions, as well as make assertions as to the general influence a social attractor has on N application (i.e. does it tend to drive excessive or appropriate rates). Our model remains true to HattÕs (2013) nonlinear feedback loop framework for social attractors, and thus embeds its analysis of s ocial actions within an SES framework. Further, our model pulls from the process approach to social resilience indicated by HattÕs (2013) work , as we attempt to reveal the social processes that reinforce the persistence of N over -application. !=X!QUANTITATIVE METHODS To investigate the statistical effects of the five identified social attractors on N application rate, logistic regression is used to predict the effect of variables on the probability that farmers applied N above the generalized threshold rate me ntioned above (> 120 lbs./acres = 1). In this section, t he variables used in this model are outlined. See Table 3 for descriptive results Dependent Variables N Rate : The primary dependent variable used in these models is pounds per acre of N fertilizer app lied to commercial un -irrigated corn. Respondents were asked to write in their fertilizer application rate in lbs./acre for the most recent year they grew corn. Fertilizer application rate was recoded into a binary variable (> 120 lbs./acres = 1) to reflec t the generalized threshold rate at which fertilizer application exceeds crop N demands and N 2O emissions begin to increase substantially (Bouwman et al. 2002; Hoben et al. 2011). Commercial un -irrigated corn fertilizer application rates ranged from 1.5 lb s./acre to 400. 10 The mean, at 136.3 lbs./acre is approximately 14 lbs. above the 2011 Michigan average of 122 lbs./acre (USDA 2012). Of all the farmers in our sample growing un -irrigated commercial corn (152) 11, 61.8 percent applied above the estimated thre shold rate (120 lbs./acre). Importantly, Stuart et al. (2012) show that commercial corn farmers apply significantly less N than seed corn farmers. Awareness of N 2O as a greenhouse gas (GHG) . Awareness of N 2O as a GHG measures farmersÕ knowledge of N 2O to be a greenhouse gas (Yes/No dichotomous response). It is !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!10 Five responses of 0 were dropped. Farmers that do not apply nitrogen are an exception and likely represent a minor subgroup of corn farmers (e.g. organic). 11 Sample sizes were reduced based on missing cases. !!=Y!used in a linear path model, where an independent awareness variable (discussed below) is used to predict it and then both are used to predict N -rate as the primary dependent variable. Independent V ariables Personal Experience: Personal experience measures the farmerÕs self -reported importance of previous farming experience in determining current N rate, with higher scores indicating more importance. Profit: The motivation of profit is measured usin g farmersÕ self reported importance of a balance between costs of returns in determining their N rate. Higher scores indicate more importance. Information Sources : Three variables are included in the quantitative model. The effect of fertilizer dealers as the farmerÕs most important source is measured against those that use university recommendations . Other sources (50.4%) is included for the purpose of direct comparison between university recommendations and fertilizer dealers as sources of information. 12 Respondents choose only one most importance source from a list. These are therefore dummy variables. Reported Knowledge: Reported knowledge of consequences is assessed using two variables in the quantitative analysis: Reported awareness of N 2O and report ed awareness of N 2O as a greenhouse gas (GHG). Awareness of N 2O measures farmersÕ awareness of N2OÕs link to nitrogen fertilizer application. Awareness of N 2O as a GHG measures !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!12 Other sources include: Other farmers, magazines, company fieldman, private con sultant and a write in category of Ôother. Õ !!=F!farmersÕ knowledge of N 2O to be a greenhouse gas. Both are binary variables (i. e. awareness is measured against being unaware) as the survey questions was a Yes/No response. The effect of a farmerÕs knowledge of nitrogenÕs link to N 2O and N 2O as a greenhouse gas are expected to effect application rate in a linear path, where awarenes s of N 2O precedes and thus predicts awareness of N 2O as a GHG, which are then both used to predict odds of applying in excess. Acres of Corn: Acres of Corn is constructed from farmersÕ reported acres of corn grown. Higher values indicate more acres grown. Results will suggest how nationwide changes in farm structure impact farmersÕ N application rate in our sample. All descriptive information (e.g. range, standard deviation, mean) for variables is available in Table X. Table 3 Descriptive Results Variables Mean Standard Deviation Range Dependent: N-Rate Avg. Rate: 49.93 1.5-400 lbs./acre* 136 lbs./acre Awareness of N 2O as a greenhouse gas (GHG) 45% .50 0-1 Independent: Personal Experience 3.31 .85 1-4 Profit 2.93 1.14 1-4 Information Sources: % Used as Main Source: 1. Fertilizer Dealers 27% .45 0-1 2. University Recommendations 20% .40 0-1 3. Other 52% .50 0-1 Reported Knowledge: % Aware: 1. Awareness of N 2O 37% .48 0-1 !!=>":"07?!:6E7?5!igBA[@!! ! !=Z!THE SOCIAL ATTRACTORS FRAMEWORK Social attractors operate in a nonlinear feedback loop where they drive actions that have ecological impacts that in turn affect other social attractors. However, with only one wa ve of data, the causal connections proposed in this nonlinear feedback loop cannot be assessed. This may be accomplished in future analysis. In Figure 3, we depict the social attractor variables driving N application rate s to be below or above 120 lbs ./acre. As shown, rates above 120 lbs ./acre encourage further changes in aquatic and atmospheric states towards undesirable consequences such as climate change and hypoxia. Application below 120 lbs ./acre mitigates the agricultural drivers of these ecosystem changes. These ecological states are considered to impact social attracto rs in a non-linear feedback loop , although the exact effects are not examined in this study. The social and ecological attractors that may drive application rate s but are not measured in our study are depicted at the far left of the figure in gray. Our model, as with all models , is an abstraction from the complexity of reality. We indicate other potential social and ecological attractors not tested in this study to illustrate the large r context and to indicate other variables for future work . !!BA! !B1!RESULTS Results of the logistic regression model predicting nitrogen application are shown in Table 3. The social attractors of acres of corn, both reported awareness variables (N2O and N 2O as a GHG) and personal experience all significantly predicted the likelihood of applying above the N threshold rate. Results indicate that larger farms are more likely to over -apply nitrogen. Further, the odds N will be over -applied increase s as farmers rely more on past experience (and rely less on outside sources of information) . As expected, awareness of fertilizerÕs relationship to N 2O predicted farmersÕ aware ness of N 2OÕs relationship to climate change. 13 When predicting the odds of applying in exces s, bo th aware ness of N 2O and N2O as a GHG were significant. Farmers who are knowledgeable about N as a GHG are less prone to apply in excess. In contrast , those that reported they were aware that fertilizer is a source of N2O were actually more likely to o ver -apply . All results are shown in Table 1. Depicted in Figure 4, we use the social attractor framework to comprehend how a farmerÕs N application s are embedded in a nuanced SES. First, N applications are situated within a layered society that asserts do wnward pressure to actively reinforce resilience. FarmersÕ actions are further embedded within a broader ecosystem, the state of which is influenced through their N application rates, among other practices. In turn, in a perpetual non -linear feedback loop ecosystem feedbacks created by environmental N (or lack thereof), such as water or climatic states, operate reflexively to influence farmerÕs N application practices or the processes reinforcing them (indicated by left facing curved arrows). For instance, states associated with climate change, such as sporadic weather !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!13 Results can be found in N 2O model. !!B=!patterns like heavy rainfalls , may increase N rates through causing more leaching. Farmers excess N rates (>120 lbs./acre) are actively reinforced by the increasing farm sizes associated with capitalist production, the widespread social value of individualism, embodied here as personal experience, and their awareness of N 2O. The continuation of this practice feeds excess agricultural N to the environment, thus spurring on the creation of undes irable aquatic and atmospheric states. Awareness of N 2O as a GHG was shown to disrupt excess N application. 927,&'@ 2!!"-2,'A1;&, !!H6I0/2 !i1[ !)"7-/ !&6:"-/!*??4-:?645!940I":?"7J2 !*`-40!6>!S=(! -5!-!VKVT1 !S!*99/":-?"67! D-?0!g!1=A! /.5f-:40!T!1 !304567-/!,89 04"07:0 !!1@<=1dd !!!iA@X=A[ !346>"?5 !!A@Z== !!!iA@1YZ[ !*`-40!6>!S =(!1F@XFddd !B@YZZdd !!iF@Y!S =(!-5!VKV !!A@BZ!k647 !!1@