Introduction
The economic, environmental and social benefits of farming organically in the USA are increasingly recognized. Over 2 million hectares of US land are currently farmed organically, with organic food sales making up 5% of total US food sales (Willer and Lernoud, Reference Willer and Lernoud2016). Organic farmers report consistent double-digit annual growth in US sales (Willer and Lernoud, Reference Willer and Lernoud2016); a continually shrinking gap exists between conventional and organic yields of US cash crops like corn and soybeans (De Ponti et al., Reference De Ponti, Rijk and Van Ittersum2012), and organic premiums operate well above break-even premiums (Reganold and Wachter, Reference Reganold and Wachter2016). Organic farming methods, along with no-till methods, are also thought to be key in slowing unsustainable soil erosion (Montgomery, Reference Montgomery2007), increasing soil biodiversity (Amundson et al., Reference Amundson, Berhe, Hopmans, Olson, Sztein and Sparks2015) and may generate higher yields than conventional farms as drought conditions increase under climate change (Letter et al., Reference Letter, Seidel and Liebhardt2003). A greater percentage of organic farms in an agriculture-dependent area has also been shown to encourage community economic development and increase social interactions between consumers and farmers (MacRae et al., Reference Macrae, Frick and Martin2007).
Organic farmers also face considerable risks—risks often thought to be greater than those faced by conventional farmers. These risks include input risks, like seed, labor and equipment shortages; external risks, like contamination by genetically modified organisms; economic risks such as a lack of access to, or rapidly changing, markets; and production risks, like crop infertility, pests and diseases (Hanson et al., Reference Hanson, Dismukes, Chambers, Greene and Kremen2004; Constance and Choi, Reference Constance and Choi2010). Perhaps the greatest concern amongst organic farmers—and a major reason why conventional farmers report not transitioning to organic (Bastiaans et al., Reference Bastiaans, Paolini and Baumann2008)—remains weeds (Walz, Reference Walz1999; Mohler and Johnson, Reference Mohler and Johnson2009; Moynihan, Reference Moynihan2011; Zwickle, Reference Zwickle2011; DeDecker et al., Reference Dedecker, Masiunas, Davis and Flint2014; Misiewicz et al., Reference Misiewicz, Shade, Crowder, Delate, Sciligo and Silva2017). Weeds take a physical, psychological and economic toll on organic farmers (Zwickle et al., Reference Zwickle, Wilson and Doohan2014). While conventional farmers can rely on a host of synthetic herbicides to mitigate the risk of weeds, organic farmers must be more creative, relying on crop rotations, mechanical cultivation, manual weeding, beneficial insects and other cultural practices (Hanson et al., Reference Hanson, Dismukes, Chambers, Greene and Kremen2004) to control weeds.
Such cultural practices are often termed ecological weed management, or EWM (Bastiaans et al., Reference Bastiaans, Paolini and Baumann2008). EWM has become a focus of the research community (Zwickle et al., Reference Zwickle, Wilson and Doohan2014; Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016), and studies suggest EWM likely increases weed management effectiveness, reduces the time and labor required to manage weeds and possibly reduces weed seedbanks over time (Jackson, Reference Jackson1997; Liebman et al., Reference Liebman, Mohler and Staver2001; Hatcher and Melander, Reference Hatcher and Melander2003; Gallandt and Molloy, Reference Gallandt and Molloy2008; Anderson, Reference Anderson2010). Despite the many ways in which EWM can be employed, weed scientists and Extension personnel argue that EWM has been poorly adopted by the organic community (Bastiaans et al., Reference Bastiaans, Paolini and Baumann2008), and organic farmers report both a lack of and desire for more research and recommendations regarding such options (Hanson et al., Reference Hanson, Dismukes, Chambers, Greene and Kremen2004; Tautges et al., Reference Tautges, Goldberger and Burke2016).
Previous work by the authors (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014; Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016) and others (DeDecker et al., Reference Dedecker, Masiunas, Davis and Flint2014) shows that organic farmers may resist such recommendations, particularly those coming from Extension. Organic farmers report relying primarily on their own past experience, e.g., what worked in the past, along with trusted sources like neighbors and successful farmers to determine which of the above management techniques to use—as well as how best to deploy them. Extension recommendations, the latest research and science, and National Organic Program standards are rarely mentioned as sources for information (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014). This could very well be a result of Extension offices lacking the appropriate personnel, financial support and knowledge of organic farmers’ needs to make relevant recommendations (Misiewicz et al., Reference Misiewicz, Shade, Crowder, Delate, Sciligo and Silva2017). Indeed many organic farmers prefer local knowledge over expert recommendations (Norman et al., Reference Norman, Bloomquist, Janke, Freyenberger, Jost, Schurle and Kok2000); this despite the USDA's National Food and Agriculture Institute (NIFA, 2017) distributing upwards of 20 million dollars annually for research in organic agriculture.
Familiar sources may not always be reliable however as many organic farmers have demonstrated, at least relative to conventional farmers, an incomplete understanding of weed biology and the mechanisms by which weeds spread (Doohan et al., Reference Doohan, Wilson, Canales and Parker2010). Additionally, many organic farmers have embraced ‘soil balancing’, or the base-cation saturation ratio (BCSR) theory of soil fertility, in managing weeds. Such theories have been empirically contradicted by some (Kopittke and Menzies, Reference Kopittke and Menzies2007) and termed by others to be merely ‘pseudoscience’ (De Decker et al., Reference Dedecker, Masiunas, Davis and Flint2014; Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014). Despite evidence to the contrary, soil balancing is often heavily encouraged by consultants who not only offer the necessary amendments, but wield far more influence in local organic communities than do Extension and University personnel—at least in central Ohio. Regardless, organic farmers have been shown to rely on a ‘limited suite’ of mechanical weed control methods, ignoring more diverse ‘information-intensive’ and ‘integrated’ weed management strategies, strategies that may prove more successful in reducing weed pressures (Walz, Reference Walz1999; De Decker et al., Reference Dedecker, Masiunas, Davis and Flint2014: 529).
Due to the complicated tradeoffs that organic farmers face regarding risks and benefits (Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016), the long-standing disconnect between organic farmers and the research community, and what may be a misunderstanding of weed biology spread by some consultants and organic farmers, encouraging organic farmers to seek out the best available EWM information is crucial. At the same time determining the factors that lead some organic farmers to seek out such information and others to resist it is equally important. In the past, researchers (Lyson, Reference Lyson2012) and Extension personnel (Rogers, Reference Rogers2010) have relied heavily on the diffusion of innovation theories to describe and model how conventional farmers adopt different agricultural practices. Such theories have been less successful however in modeling how ‘soft’ technologies like knowledge—especially knowledge about EWM—diffuse through organic communities (Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016). As such, we test here an alternate model, an adaptation of Griffin et al.’s (Reference Griffin, Neuwirth, Dunwoody and Giese1999) risk-information seeking and processing (RISP) model, to identify and examine the individual and social characteristics associated with organic farmers’ EWM information-seeking (IS) behavior.
Literature review
RISP is an adaptation of Chaiken et al.’s (Reference Chaiken, Liberman, Eagly, Uleman and J1989) Heuristic–Systematic model of processing and Azjen's (Reference Ajzen1991) Theory of Planned Behavior. Systematic processing is defined as comprehensive and analytical, a mode in which individuals seek to integrate all useful information into their judgments; heuristic processing, on the other hand, is less cognitively demanding and relies on simple decision rules and less information (Chaiken et al., Reference Chaiken, Liberman, Eagly, Uleman and J1989). The RISP model has been used to predict the extent to which individuals will seek out information about and then analyze a risk, e.g., organic farmers who rely more on systematic processing may be more likely to seek out information about EWM. RISP has been shown to successfully predict risk-information behavior across a variety of contexts, from health (Yang et al., Reference Yang, Aloe and Feeley2014a) to river flooding (Griffin et al., Reference Griffin, Yang, Ter Huurne, Boerner, Ortiz and Dunwoody2008), the environment (Kahlor et al., Reference Kahlor, Dunwoody, Griffin and Neuwirth2006) and climate change (Kahlor, Reference Kahlor2007; Yang et al., Reference Yang, Rickard, Harrison and Seo2014b).
RISP incorporates up to seven principal factors to predict behavior. These include: (i) individual characteristics, including socio-economic and demographic factors, as well as an individual's relevant hazard experience, (ii) perceived hazard characteristics, such as the perceived risks and benefits of a hazard, (iii) individuals’ affective response to risk, like the worry or dread associated with a risk, (iv) informational subjective norms, such as social pressure to know more about the risk, (v) information insufficiency, or the level of knowledge one believes is necessary to engage a risk, that individual's current knowledge and the gap between, (vi) an individual's perceived capacity to gather information and (vii) beliefs about the usefulness of information, as well as an individual's trust in risk management. While the original RISP included all seven factors, as little as two [e.g., current knowledge and informational subjective norms (Yang et al., Reference Yang, Aloe and Feeley2014a)] have been shown to explain a significant portion of the variance in IS behavior.
In the current study, we use a modified RISP model to examine five factors that previous research (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014) suggests may predict organic farmers’ EWM IS behavior (see Fig. 1): (1) a farmer's personal (hazard) experience with the weed, or the amount of time a farmer has been managing a weed; (2) the hazard characteristics, or both the risks and benefits farmers perceive to be associated with the weed; (3) individuals’ affective response, i.e., farmer's worry, regarding the weed; (4) informational subjective norms, or the extent to which neighbors and friends urge the individual to know about the weed; and (5) farmers’ perceived knowledge gap, or the gap between what a farmer believes they should know (sufficiency threshold) and what they currently know (current knowledge) about managing the weed, also termed information insufficiency.
Fig. 1. EWM RISP model.
In this work, we did not expressly examine farmers’ information processing behavior (e.g., whether they demonstrated heuristic or systematic processing of information); however, active, goal-directed seeking of information has been shown to lead to systematic processing (Kahlor et al., Reference Kahlor, Dunwoody, Griffin and Neuwirth2006).
Research questions and hypotheses
Based on previous RISP studies (Griffin et al., Reference Griffin, Dunwoody and Neuwirth1999; Kahlor, Reference Kahlor2007), our own work examining the mental models of organic farmers (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014; Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016), and our work with organic farmers in central Ohio, we hypothesized a number of relationships to exist with regard to our five factors. With regard to farmers’ personal experience managing a weed, we expected that the more experience a farmer had managing a particular weed, the less risk they would perceive it to have. This is not to say that the risks of a weed would necessarily decline in reality, or that a farmer would become better able to mitigate those risks over time, only that the risk perceptions associated with it may decline based on the greater number of years the farmer demonstrated success managing or overcoming that risk. Put differently, we assume that a farmer must be successful in managing risk or else they would no longer be farming. As such:
H(1): The perceived risks of a weed will be negatively associated with the farmers’ greater personal experience managing that weed.
The relationship between farmers’ experience and the perceived benefits of a weed is not so clear. While some farmers may come to believe that a weed does help stem soil erosion or increase soil organic matter (SOM) over time, others may perceive the opposite. As such, we are only able to ask the following research question:
RQ(1): How does a farmer's greater personal experience managing a weed affect their perceived benefits of that weed?
With regard to an individual's affective state regarding a risk, Griffin et al. (2004) argue that one's worry is immediately influenced by their perceived characteristics of the hazard. Such characteristics, good or bad, pleasing or painful, generate ‘risk as feelings’, and these feelings precede and inform individuals’ judgment and decision-making (Slovic et al., Reference Slovic, Finucane, Peters and Macgregor2004). Indeed, one's dread regarding a hazard may be the key determiner of their risk perceptions and their avoidance or acceptance of a hazard (Slovic, Reference Slovic1987). High dread is often associated with risks that are new, difficult to see, and out of one's control, characteristics often associated with the weed seedbank.
As such, we would expect farmers to report greater worry if they perceive the risks of a weed to be high, for instance, if they expect the weed is likely to disrupt their harvest operations, decrease their yield or dramatically increase the time and labor they need to invest in managing the weed. Whereas, we would expect the farmers’ worry to be negatively associated with the benefits they perceive the weed to have, such as its perceived ability to reduce soil erosion or increase SOM. We thus hypothesize:
H(2a): Farmers’ affective state (worry) will be positively associated with the perceived risk of the weed.
H(2b): Farmers’ affective state (worry) will be negatively associated with the perceived benefits of the weed.
Just as affect precedes and informs one's judgment, the RISP model proposes that individuals’ affective state may also influence the degree to which they believe they have sufficient knowledge or information to manage a risk, i.e., whether or not they meet a knowledge sufficiency threshold (Griffin et al., Reference Griffin, Dunwoody and Neuwirth1999). Griffin et al. (Reference Griffin, Dunwoody and Neuwirth1999) argue that the more a person is worried about a particular risk, the greater the confidence they will want to have in their knowledge of that risk. Additionally, the more a person is worried about a risk, the more likely it may be that they believe they lack sufficient knowledge necessary to manage that risk. As such we hypothesize that:
H(3a): Farmers’ information insufficiency will be positively associated with the farmers’ affective state (worry).
At the same time, we might expect that the more information a farmer believes they already have may lessen the degree to which they believe there is still information to acquire. As such:
H(3b): Farmers’ information insufficiency will be negatively associated with the farmers’ current knowledge.
Organic farmers have already been shown to rely extensively on networks of friends and neighbors for information about EWM (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014). Accordingly, we would expect social norms, particularly expectations from other successful farmers that an individual seeks out information about how to manage weeds, to affect an individual's own perceived knowledge gap. Thus, we hypothesize that:
H(3c): Farmers’ information insufficiency will be positively associated with informational subjective norms.
The ultimate goal of this work is to elucidate those factors that lead organic farmers to not only recognize they lack information about EWM but also to seek such information out. The RISP model predicts that both the greater an individual's information insufficiency regarding a risk and the informational subjective norms encouraging IS, the more likely they are to seek out information about managing that risk. As such, our final two hypotheses are:
H(4a): Farmers’ IS behavior will be positively associated with information subjective norms.
H(4b): Farmers’ IS behavior will be positively associated with information insufficiency.
One additional and important note about this study is that it purposely focused organic farmers’ attention on their single most problematic weed. This is especially useful as previous interviews, surveys and focus groups (Zwickle, Reference Zwickle2011) showed that most organic farmers identify only one weed as representing the bulk of their weed pressure and concerns. Additionally, we expect that improved knowledge and management of the most problematic weeds would also lead to the mitigation of farmers’ less worrisome weed species.
Methods
Sample
To test our five-factor framework, we collected data via an online survey delivered to a random sample of 3000 organic farmers, identified from the US National Organic Certification List. While relying solely on an online survey may introduce some self-selection bias into our sample, we believe it to be minimal as a 2017 (USDA, 2017a) report shows that 72% of farmers either own or lease a computer, nearly 40% of certified organic operators listed an email address in the Organic Integrity Database (USDA, 2017b), and the USDA National Agricultural Statistics Service now ‘highly recommends’ farmers respond to the online version of the 2017 Census of Agriculture (King and White, Reference King and White2017).
A total of 574 respondents’ surveys were returned for a response rate of 19.1%. Twenty respondents did not actively manage weeds on their farm and thus were removed from the analysis. Of the remaining 554 respondents, 361 reported their gender, with 284 reporting as male (78.7%) and 77 as female (21.3%). Additionally, of those respondents who reported their age, 150 out of 326 (46.0%) reported an age between 35 and 66 yr. One hundred and thirty-four (37.1%) participants reported having a bachelor's degree; 92 (25.5%) had a graduate or professional degree; and 134 had at least a high school diploma (37.1%). Of those who reported their farm's average gross sales (n = 356), 69.4% (n = 247) reported sales >$50,000; 60 (16.9%) reported between $50,000 and $100,000, 134 participants (37.6%) reported sales between $100,000 and $500,000, and 53 (14.9%) reported sales >$500,000.
Measures and factors
All of the survey questions used to test our RISP model were drawn from previous tests of the original RISP model (Griffin et al., Reference Griffin, Dunwoody and Neuwirth1999; Kahlor, Reference Kahlor2007), but were refocused for examining farmers’ weed perceptions and weed management behavior (see Table 1). As in the previous work, we used both unidimensional and multidimensional factors. Farmers’ Personal Experience, Current Knowledge and Worry were measured using a single survey question, while farmers’ Perceived Risks and Perceived Benefits of the weed, Informational subjective norms, Information insufficiency and IS behavior were measured using multiple survey questions. We used principal components analysis to group multiple questions into factors that would account for the most variance. All of the factors were represented as a single component except for Perceived Risks, which made up two components: Harvest and Yield Risk, and Labor and Time Risk. Questions pertaining to these four specific types of risks were informed by our previous work (Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016). Our measure of personal experience drew from the National Transition to Organic Survey (2017) and how the USDA classifies individuals as ‘beginning farmers or ranchers’, i.e., those with <10 yr experience farming or ranching. Our participants could select either <1 yr, up to 10 yr, or more than 10 yr experience managing their most problematic weed.
Table 1. RISP survey questions, descriptive statistics and principal components analysis (PCA) results
Response option was a five-point Likert scale from (1) very unlikely/strongly disagree to (5) very likely/strongly agree, unless otherwise noted.
In addition to standard demographic questions, farmers were also asked to identify from a list of nine commonly reported sources of information (Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016) the three sources they used most often, as well as assess five statements about their beliefs regarding weed management and two statements about their weed pressure and weed seedbank.
Analysis
The Statistical Package for the Social Sciences (SPSS) and specifically hierarchical multiple regression were used to analyze the data. To examine the research question and hypotheses, we regressed farmers’ IS behavior on the following blocks of variables in an order informed by Griffin et al. (Reference Griffin, Yang, Ter Huurne, Boerner, Ortiz and Dunwoody2008) and the RISP model: (1) personal experience; (2) hazard characteristics; (3) worry and informational subjective norms; (4) current knowledge and (5) information sufficiency threshold. To avoid reliability issues and ceiling effects (Cohen et al., Reference Cohen, Cohen, West and Aiken2013), current knowledge was entered in the block previous to sufficiency threshold so that the latter variable represented the difference between the two. When insufficiency threshold was the dependent variable, we entered current knowledge in the first block so the remaining variables predicted the variance in the threshold not accounted for by current knowledge (Griffin et al., Reference Griffin, Yang, Ter Huurne, Boerner, Ortiz and Dunwoody2008).
To examine H(2), we regressed farmers’ affective state on the blocks (1) personal experience and (2) hazard characteristics, and to examine RQ1 and H1, we conducted analysis of variance (ANOVA).
Paired sample t-tests were used to distinguish which risks and benefits farmers perceived to be greatest, and ANOVA was used to distinguish between the characteristics of farmers based on perceived knowledge gaps and IS behavior.
Results
Research questions and hypotheses
Only 26.1% (n = 130) of our respondents identified having <1 yr managing their most problematic weed, while 27.9% (n = 139) had between 1 and 9 yr, and 46.0% (n = 230) reported having at least 10 yr experience. The amount of personal experience a farmer reported managing a particular weed showed no significant negative (or positive) association with either their perceived harvest and yield risk (F = 0.05, df = 2, p > 0.05) or the risk of that weed to their labor and time (F = 0.36, df = 2, p > 0.05). Thus, H(1) was rejected. Regarding RQ1, farmers’ personal experience with a weed was also not significantly associated with the perceived benefits of that weed (F = 0.18, df = 2, p > 0.05).
Based on paired-sample t-tests, respondents reported the greatest risk of their most problematic weed to be an increase in the time and labor they spent weeding (t = 9.36, df = 462, p < 0.01), followed by the likelihood that weed may reduce their crop yield and interfere with harvest operations (t = 6.65, df = 476, p < 0.01), respectively. No significant difference (t = 1.16, df = 454, p > 0.05) was found between farmers’ perceptions of the benefits of that weed, i.e., between the likelihood of it stemming erosion or improving SOM.
Respondents reported moderate worry regarding their most problematic weed with 95% of farmers reporting worry between 3.20 and 8.20 on a ten-point scale (with 0 = no worry at all and 10 = a lot of worries). Amongst those problematic weeds mentioned by more than 15 farmers, the greatest worry was associated with bindweed (6.73), followed by foxtail (6.47) and giant ragweed (6.20) (see Table 2). Redroot pigweed (5.55) and grasses (5.63) inspired the least amount of worry. Regarding those hazard characteristics that may have led to such worry, multiple hierarchical regression demonstrated that farmers’ worry was positively associated with a weed's perceived risk to harvest and yield (β = 0.53, t = 6.96, p < 0.01), but not with its risk of increasing the farmers’ time and labor spent weeding (β = 0.10, t = 1.39, p = 0.17) (see Table 3). Thus, H(2a) was partially supported. H(2b) was supported as farmers’ worry demonstrated a significant negative association with a weed's perceived benefits (β = −0.15, t = −2.03, p = 0.05). Grasses, which inspired relatively low worry compared with other weeds, were thought to be the most beneficial weed in both stemming erosion and improving SOM.
Table 2. Farmer characteristics for each weed
Highest characteristic is in bold; lowest characteristic is in italics. Each Risk and Benefit is the product of its perceived probability and consequences (e.g., Harvest × Harvest%).
a Score is an average of the three informational subjective norms questions from Table 1.
Table 3. Multiple hierarchical regression of affective state: worry [standardized regression coefficients (β)]
With regard to information insufficiency, 37 (7.7%) farmers reported having just enough knowledge to manage their most problematic weed. One hundred and twenty-two (25.5%) of the farmers reported having more than enough information and 320 (66.8%) reported needing more information. Multiple hierarchical regression showed farmers’ current knowledge to have no negative association with information insufficiency (β = −0.06, t = −0.71, p = 0.48), thus H(3b) was rejected (see Table 4). Put differently, the amount of information farmers reported having about a particular weed had no significant association with the amount of additional information they reported needing to manage that weed.
Table 4. Multiple hierarchical regression of information insufficiency and information-seeking behavior [standardized regression coefficients (β)]
With regard to those weeds mentioned most often, farmers reported the least current knowledge about giant ragweed (52.67) and the greatest current knowledge about grasses (63.60) (see Table 2). The largest perceived gap between current knowledge and the knowledge necessary to manage a weed effectively also belonged to giant ragweed (23.11); the smallest gap belonged to grasses (4.73).
Multiple hierarchical regression showed that information insufficiency was positively associated with the farmers’ affective state (β = 0.23, t = 2.43, p = 0.02) and informational subjective norms (β = 0.20, t = 2.41, p = 0.02). Thus, both H(3a) and H(3c) were supported; farmers’ worry about a weed and strong social norms encouraging them to seek out information about that weed were both positively associated with needing additional information. Farmers reporting giant ragweed (3.39) as their most problematic weed reported the strongest norms; the weakest norms were associated with grasses (3.19) and pigweed (3.19).
As RISP predicted, H(4a) and (4b) were both supported with significant positive associations found between farmers reporting seeking information about weed management and both informational subjective norms (β = 0.34, t = 4.17, p < 0.01) and information insufficiency (β = 0.31, t = 2.55, p = 0.01). Put differently farmers reported seeking out more information about their most problematic weed when they reported stronger social norms encouraging that behavior and when they reported a lack of sufficient knowledge.
Overall the proposed models explaining information insufficiency and IS behavior explained approximately 20% of the variance in each model (R 2information insufficiency = 0.20 and R 2IS behavior = 0.21) (Table 4). Gender and education, the latter found in previous research to affect organic farmers’ EWM behavior (Tautges et al., Reference Tautges, Goldberger and Burke2016), were both non-significant (p > 0.05) here and actually reduced the amount of variance explained in both models.
Sources of information
Regarding sources of information, organic farmers used most (see Table 5) often their own trial and error (58.3% of mentions; 3.24 on a four-point Likert scale where 1: useless and 4: very useful) and friends and neighbors who farm (44.4% of mentions; 2.62). The sources mentioned least by farmers were ACRES publications (7.9%) and websites they liked (4.2%), while local Extension agents were identified as the least useful source of information (1.56). Farmers reported using 19.0% (SD = 20.8) of their total time spent farming researching how to manage weeds (including reading and thinking about weeds, and attending workshops and field days); this proportion was unrelated (p > 0.05) to the weed identified, farmers’ worry, perceived risks and benefits or social norms.
Table 5. Sources of information mentioned most by farmers
a A total of 1205 mentions (72.5%) out of 1662 possible mentions (n = 554 × 3 sources).
b Likert scale, 1 (useless) to 4 (very useful).
Beliefs about weed management and ‘soil balancing’
Figure 2 displays the frequency of responses for five statements about farmers’ beliefs regarding weed management. Over 23% of farmers (91 out of 391) reported using weeds to ‘read’ what nutrients needed to be added to the soil, and 38.5% (148 out of 384) agreed that weeds indicated what nutrients were missing in the soil and upon application of those nutrients weeds would stop growing. In total, 78.4% of farmers (301 out of 384) agreed that understanding weed biology made a difference in managing weeds effectively; only 10.2% (39 out of 384) disagreed.
Fig. 2. Organic farmers’ beliefs about weed management and control.
Figure 3 displays the frequency of responses for farmers’ perceived changes in weed pressure and the amount of weed seeds in the soil. Positively, 81.8% of farmers (302 out of 369) reported their weed pressure was either declining or staying the same, while 74.3% of farmers (274 out of 369) reported the number of weed seeds in their soil was either declining or staying the same.
Fig. 3. Organic farmers’ perceptions of weed pressure and weed seeds in soil.
Discussion
The principal goal of this study was to test a five-factor framework, informed by Griffin et al.’s (Reference Griffin, Dunwoody and Neuwirth1999) RISP model, to identify those factors that may lead organic farmers to think they require additional information about EWM and to seek out such information to better manage their most problematic weed. The results support previous work testing the RISP model (e.g., Kahlor, Reference Kahlor2007) and many of the relationships hypothesized to exist between such factors based on our own previous work (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014; Zwickle et al., Reference Zwickle, Wilson, Bessette, Herms and Doohan2016). The perceived risks of farmers’ most problematic weed were positively associated with their worry, while the weed's perceived benefits were negatively associated with farmers’ worry. Both farmers’ worry and informational subjective norms were positively associated with perceived information insufficiency; and information insufficiency and informational subjective norms were positively associated with the farmers’ EWM-IS behavior.
The negative relationship hypothesized to exist between the amount of personal experience a farmer had managing a weed and their perceived risks of that weed was not supported. Nor did we find a positive relationship between a farmer's current knowledge and their perceived knowledge gap or IS behavior—indeed most farmers reported needing a similar amount of knowledge regardless of how much knowledge they reported already having.
Specific risk perceptions played different roles in inducing worry and promoting IS behavior, roles consistent with our previous research (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014). Farmers who perceived a higher risk of weeds to their harvest and yield expressed more worry, but did not report a lack of knowledge or the need to gather more information as a solution. Conversely, those farmers who perceived weeds as increasing the time and labor they spent in the fields were more likely to report needing additional information to manage them. We expect this difference may be the result of farmers seeing their time and labor as manageable, a tractable problem that can be (at least somewhat) controlled by applying different tactics or strategies. Not only does previous work suggest that organic farmers are management-centered rather than ecology-centered (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014), but one's ability to control a hazard is a key factor in reducing its dread—and thus its worry, as well as increasing its acceptance as a risk (Slovic, Reference Slovic1987). This is in opposition to the risk of weeds to a farmer's harvest and yield, which may be seen as more holistic, ecological problems, and less influenced by farmer's management strategies; organic farmers have previously reported that terms like yield—and profit—are not appropriate for their decisions (Mccann et al., Reference Mccann, Sullivan, Erickson and De Young1997). Such a view may explain greater worry regarding these risks that are seemingly out of farmers’ control (Slovic et al., Reference Slovic, Finucane, Peters and Macgregor2004). Additionally, the risk of weeds and weed seeds damaging farmers’ products as well as their reputation and relationships with buyers is certainly considerable, and yet also out of the farmers’ control at the point of delivery or purchase.
Another explanation for this difference in risk perceptions may be the time lag between when a management strategy is employed and when the consequences of weeds manifest. Farmers often begin making weed management decisions immediately previous to or at the outset of the planting season. At this point, and immediately following, high weed pressure can be managed via greater time and labor spent weeding. Thus, seeking—and finding—information about different EWM tactics can lead to direct, measurable benefits and a relatively clear, though not necessarily positive, cause–effect relationship. Farmers’ harvest operations on the other hand and the resulting yields are both far more uncertain and occur at the end of the season, often long after farmers’ weed management decisions have been made. Seeking out information at this point is, at least for the current season, futile. This separation of outcomes from decisions may reduce the perceived value of IS with regard to harvest and yield, at the same time increasing farmers’ worry regarding both. Indeed, people are familiar with and have shown themselves to be adept at managing close cause–effect relationships; at the same time, we struggle with the decisions that are characterized by large time lags and low proximity (Gardner and Stern, Reference Gardner and Stern1996).
Farmers who perceived the benefits of weeds, on the other hand, were less worried and more likely to assume that they have the necessary knowledge to manage them effectively. This was especially apparent with regard to grasses. Farmers who reported grasses as their most problematic weed reported little worry, high knowledge and low social pressure with regard to acquiring knowledge about how to manage them, and perceived the greatest benefits with regard to that weed increasing SOM and stemming erosion. One exception to this pattern regarding grasses was foxtail, which inspired high worry and high social pressure, and was perceived to have low benefits. These findings suggest that researchers and Extension may not want to prioritize organic research on grasses with the exception of foxtail, instead focusing their research and communication efforts on ragweed, bindweed—and foxtail, far more worrisome and less (obviously) beneficial weeds.
That farmers perceived any benefit stemming from their most problematic weed may speak to philosophical differences between conventional and organic farmers (Sullivan et al., Reference Sullivan, Mccann, De Young and Erickson1996; Stofferahn, Reference Stofferahn2009). Though recent studies comparing conventional and organic farmers’ perceptions of weeds in the USA are sparse, organic farmers often report motivations for transitioning from conventional methods that suggest a more positive view of weeds and have already been shown to perceive the benefits of weeds to soil health (Jabbour et al., Reference Jabbour, Zwickle, Gallandt, Mcphee, Wilson and Doohan2014). They may be unaware of additional weed benefits, however. For instance, Henckel et al. (Reference Henckel, Börger, Meiss, Gaba and Bretagnolle2015) found that organic farms may generate a positive ‘metacommunity effect’ in providing habitats for less frequent weed species and enhancing weed diversity across landscapes. While these weeds may decrease yields, they also: (i) support critical ecological services like pollination (Henckel et al., Reference Henckel, Börger, Meiss, Gaba and Bretagnolle2015), (ii) improve biodiversity, not just of weeds, but also of insects, microbes and birds (Marshall et al., Reference Marshall, Brown, Boatman, Lutman, Squire and Ward2003; Gabriel et al., Reference Gabriel, Sait, Kunin and Benton2013) and perhaps most importantly, (iii) provide habitats for bees (Bretagnolle and Gaba, Reference Bretagnolle and Gaba2015). Supporting pollination and increasing bee habitat and health both readily come up in our own conversations with local organic farmers.
Ensuring that the information coming from University researchers and Extension makes these benefits more salient may be an effective way to overcome the worry of conventional farmers considering transitioning to organic and to reduce the worry of those already having transitioned, and reducing worry remains a critical goal of the agricultural community (Rosmann, Reference Rosmann2005; Browning et al., Reference Browning, Westneat and Mcknight2008). Previous research has shown that most farmers battle high levels of stress, worry and depression (Browning et al., Reference Browning, Westneat and Mcknight2008; Stallones et al., Reference Stallones, Doenges, Dik and Valley2013; Tiesman et al., Reference Tiesman, Konda, Hartley, Menéndez, Ridenour and Hendricks2015), levels linked to farmers’ long, demanding work hours, weather-contingent planning and harvest schedules, isolated working conditions and reduced access to affordable, high-quality healthcare services (Browning et al., Reference Browning, Westneat and Mcknight2008). Additional research is necessary, but framing weeds as a benefit instead of a risk, or simply helping farmers identify the thresholds at which weeds may begin to threaten their yield, may help to reduce farmers’ worry, particularly in organic communities.
Our results demonstrate that organic farmers also rely extensively on their families, friends and other successful farmers to make EWM decisions; in fact, it is their top source of information outside their own experience. Additionally, social networks that encourage farmers to seek out information about weed management can lead farmers to both believe they lack the necessary information to manage weeds effectively and seek out more and better information. Indeed, these norms were the strongest positive indicators of farmers’ IS behavior. Conversely, Extension, USDA personnel and University educators are rarely identified or included in these networks, and Extension was reported to be the least useful source of information here. While such exclusion may have once been the result of bias, criticism and scientific opposition to organics (Constance and Choi, Reference Constance and Choi2010), its persistence today is more likely the result of insufficient personnel, their dwindling financial support and a ‘lack of basic understanding of organic agricultural practices’, organic farmers’ needs and existing research (Misiewicz et al., Reference Misiewicz, Shade, Crowder, Delate, Sciligo and Silva2017: 14). Misiewicz et al. (Reference Misiewicz, Shade, Crowder, Delate, Sciligo and Silva2017) add that even if the pertinent research is being completed, it may not be accessible to farmers, or because organic farming relies on processes that are intensely local, it may not translate from one area or crop to another.
This lack of understanding, support, personnel and relevant research does little to build trust between organic farmers and Extension personnel, and trust is key. It is commonly accepted that trust is critical in not only improving the public's understanding of science (Wynne, Reference Wynne1980) and effectively communicating about risk (Kasperson et al., Reference Kasperson, Golding and Tuler1992) but also in reducing (and increasing) the perceived risks (and benefits) of a hazard (Siegrist, Reference Siegrist2000). Trust is a multidimensional theoretical concept—a composite of competence, care, fairness and openness (Poortinga and Pidgeon, Reference Poortinga and Pidgeon2003), but often involves the sharing of similar values and similar understandings of a situation, something called ‘salient value similarity’ (SVS) (Earle and Cvetkovich, Reference Earle and Cvetkovich1995; Poortinga and Pidgeon, Reference Poortinga and Pidgeon2003). High SVS has been shown to increase trust and correspondingly increase (and reduce) the perceived benefits (and risks) of a hazard (Siegrist et al., Reference Siegrist, Cvetkovich and Roth2000: 355). Extension personnel and research scientists may be inadvertently contributing to greater mistrust by misunderstanding the organic community, communicating disparate values and focusing on short-term research priorities as opposed to the long-term research projects organic farmers desire (Misiewicz et al., Reference Misiewicz, Shade, Crowder, Delate, Sciligo and Silva2017). This lack of trust between the organic community and the traditional Extension and research community may then be leading to greater (and less) perceived risks (and benefits) of weeds, as well as discouraging organic farmers from seeking out information about EWM strategies that could be effective on their farm.
One particular area where different understandings exist between the two communities regards the ability of weeds to act as indicators of soil nutrients, or nutrients that are lacking. In this study, a considerable portion—nearly 40%—of farmers reported using weeds either to ‘read’ what nutrients the soil may be missing or to guide their nutrient management decisions. Furthermore, it was these farmers who reported relying on soil balancing who demonstrated the largest information insufficiency (see Table 6). This is especially important to recognize as the farmers most likely to need information are also those most likely to be turned off by research scientists and Extension personnel who ignore BCSR and soil balancing or worse yet admonish those farmers who practice them. The research community may be missing a key opportunity to build trust by constructively engaging those farmers and conducting local, accessible research that works to either support—or refute—such beliefs.
Table 6. Characteristics of organic farmers most needing and most wanting EWM information
Characteristics of respondents (n = 146) reporting knowledge gap higher than the median (30)–not including those farmers who reported a negative gap, and those farmers reporting an IS behavior score higher than the median (3.5) (n = 221). Characteristics are significantly different from those reporting gaps and IS scores below the median at *p < 0.05, **p ≤ 0.01 or ***p ≤ 0.001, unless otherwise noted.
Recommendations: organic specialists and active decision support
This study's results suggest that in order to increase organic farmers’ IS behavior—and eventually their knowledge (both real and perceived) about EWM—we must increase the number of USDA, Extension and University personnel who share similar values, goals and common understandings with organic farmers. One means of doing so would be to create positions in the form of ‘organic specialists’, individuals who have:
(i) extensive personal experience farming organically and knowledge about organic farmers’ day-to-day operations,
(ii) existing ties in established networks of organic farmers,
(iii) specific knowledge of farmers’ most problematic weeds (for instance, bindweed, foxtail and giant ragweed),
(iv) exposure to and an appreciation for soil balancing and BCSR beliefs—as well as the ability to communicate with farmers about how their beliefs may or may not align with on-farm observations and
(v) access to, understanding of and the ability to communicate with farmers about government and University-funded research programs.
By engaging with the organic farming community, these specialists would build trust between themselves and the organic community, while also likely generating more relevant research priorities and results, and greater dissemination and understanding of research findings.
This study also provides some initial direction as to whom specifically these specialists might target in the organic community. Table 6 lists characteristics of those individuals who reported most needing information (largest knowledge gap) and those most seeking it out (high IS behavior). The former reported less experience managing their most problematic weed, but spend a comparatively larger amount on weed management equipment and labor, while feeling little control over weeds and weed seeds and perceiving increasing weed pressure and growing weed seedbanks. These characteristics describe a farmer navigating a possibly tenuous situation. Those farmers most seeking out information—it is important to remember that farmers seeking out information do not always report needing information—operate relatively larger farms, are newer to farming, sell more products wholesale relative to farmers’ markets, and spend more money on equipment. These characteristics describe farmers who may have access to more financial resources and easier distribution networks. While certainly more research is needed, it may be that as organic farmers’ weed pressures—and accordingly their worries—increase they move from a position of seeking out information to believing such information is either unavailable or inconsequential. The most effective risk communication efforts then may be those that engage farmers before they cross that threshold.
The above study also suggests that active decision-support frameworks may aid organic farmers in comparing different EWM strategies. Such frameworks, particularly those that rely on ‘value-focused thinking‘(Keeney, Reference Keeney1992), work to specifically identify farmers’ values and allow for the evaluation of different strategies in terms of their performance with regard to farmers’ most important values. These frameworks have demonstrated success in providing a means for balancing complicated tradeoffs (Bessette et al., Reference Bessette, Arvai and Campbell-Arvai2014), and could allow farmers to engage the labor, harvest and yield risks discussed above as well as the benefits of different EWM practices and strategies. Clarity around the short and long-term performance of different EWM strategies may also be a key in assisting farmers to overcome or at least stay motivated in the face of short-term weed pressure and the worry that accompanies it. We hypothesize that such clarity may increase the adoption of EWM strategies that require greater up-front costs but perform better over the long run.
Acknowledgments
This work was supported by Organic Transitions Award No. 2015-51106-24193 from the USDA National Institute of Food and Agriculture. We also acknowledge and appreciate the help of Guang Han, Hugh Walpole, and two anonymous reviewers.
