Journal of Environmental Psychology 60 (2018) 55–62
Contents lists available at ScienceDirect
Journal of Environmental Psychology
journal homepage: www.elsevier.com/locate/jep
Herbicide resistant weeds as place disruption: Their impact on farmers'
attachment, interpretations, and weed management strategies
Katherine Dentzman
Washington State University – Department of Crop and Soil Sciences, 509 E. Circle Drive East Lansing, MI, 48824, USA
A R T I C LE I N FO
Handling Editor: Leila Scannell
1. Introduction
Farmers are theorized to have especially high levels of place at-
tachment—that is, they feel a strong affective connection to their farm
(Beckley, 2003, pp. 105–126). Several studies have supported this, al-
though the impact of attachment on farmers' behaviors and manage-
ment activities varies (Gosling & Williams, 2010; Lincoln & Ardoin,
2016; Marshall, Park, Adger, Brown, & Howden, 2012). One way to
clarify these inconsistencies is to analyze how place attachment leads
farmers to hold varying interpretations of their farm, for instance as a
place that needs to be protected or a place that can be easily replaced
and therefore requires little protection. These interpretations may in
turn impact farmers' management choices. This article specifically ad-
dresses how place attachment relates to the interpretation and man-
agement of farm-specific place disruptions. Farmers frequently en-
counter place disruptions—events that threaten their farms'
distinctiveness, self-efficacy, and continuity (Anton & Lawrence, 2016).
Theoretically, such place disruptions trigger strong feelings of place
attachment, which in turn impact how the disruption is interpreted and
acted upon. An especially worrisome potential place disruption af-
fecting farmers across the United States is the spread of herbicide re-
sistant weeds.
In conventional agriculture systems, herbicides along with herbi-
cide-tolerant corn and soybean varieties have dominated U.S. agri-
culture as the primary method of weed control (Livingston et al., 2015).
This has resulted in herbicide resistant (HR) weeds as a product of
natural selection (Livingston et al., 2015). These weeds present a ser-
ious problem for farmers, resulting in crop yield loss, reduced com-
modity prices, lower land values, and increased weed management
costs (Norsworthy et al., 2012). It seems possible that HR weeds are a
serious form of disruption on farms; however research shows that
farmers are reticent to address the problem with recommended best
practices.
While academic extension educators recommend the use of
E-mail address: katie.dentzman@wsu.edu.
https://doi.org/10.1016/j.jenvp.2018.10.006
Received 27 March 2018; Received in revised form 9 October 2018; Accepted 16 October 2018
Available online 24 October 2018
0272-4944/ © 2018 Elsevier Ltd. All rights reserved.
T
integrated weed management (IWM) practices to lessen the develop-
ment and impact of resistant weeds, such practices have seen relatively
little adoption (Webster & Sosnoskie, 2010). Rather, the majority of
U.S. farmers continue to rely on current and hoped-for future herbicides
(Bonny, 2016; Livingston et al., 2015). Reasons for this include the
increased complexity and time involved in applying IWM, which re-
quires the combination of various weed management techniques such
as tilling, crop rotation, and use of pre-application chemicals to delay
herbicide resistance (Barman et al., 2014). However, there may be
additional reasons for adopting or foregoing IWM. Several studies have
found that farmers tend not to adopt IWM unless they have experienced
herbicide resistance on their own farm, suggesting a connection between
place disruption and attachment (Bonny, 2016; Llewellyn, Lindner,
Pannell, & Powles, 2004). This observation is particularly relevant
given that herbicide resistance problems are not uniform across the U.S.
Geographically, severe infestations of herbicide resistant weeds
occur more frequently in the Southern U.S. (Heap, 2016; Livingston
et al., 2015). This pattern is due in part to the climactic character of the
South, which in tandem with socio-economic expectations encourages
mono-cropping and allows weed populations to thrive (Ervin &
Jussaume, 2014). Taking into account this geographical distribution of
herbicide resistant weeds I work from the hypothesis that increased
presence of herbicide resistant weeds can be viewed as a form of place
disruption that triggers farmers' attachment to their farms and influ-
ences their weed management behaviors. Such a relationship would
help explain why farmers do not adopt IWM until they have herbicide
resistance issues on their own farm. However it is also possible that the
effect operates in the opposite direction, in particular because the ac-
tions of the farmers themselves impact the presence of HR weeds on
their farm. Therefore attachment may precede HR weed infestation.
However it is unlikely that preexisting levels of place attachment would
impact farmers' potential HR weed infestation; glyphosate is used as the
primary weed control method by nearly all corn and soybean farmers in
the U.S. and the high mobility of HR weed seed can result in the
K. Dentzman
infestation of even very well managed land (Livingston et al., 2015).
In this article, I draw on a range of place attachment dimensions to
help understand the processes of attachment and interpretation that
farmers are going through when they experience place disruption in the
form of herbicide resistant weeds. In clarifying this relationship, I ad-
dress two primary research questions; 1) Do herbicide resistant weeds
constitute a place disruption that may impact different dimensions of
farmers' place attachment? and 2) Can different dimensions of farmers'
place attachment affect their interpretation and management of her-
bicide resistant weeds? As part of my fist research question, I also in-
vestigate the impact of perceived threat of HR weeds versus the actual
geographical threat of HR weeds. More broadly, using exploratory
factor analysis and structural equation modeling I examine how
farmers' reactions to actual or threatened farm disruptions relate to
place-specific factors such as attachment and result in specific man-
agement behaviors.
2. Theory
2.1. Place attachment
Place attachment is the formation of emotional and sentimental
bonds between people and places resulting from accumulated bio-
graphical experience, length of residence, community sentiment, cul-
turally shared meaning, and geography (Gieryn, 2000; Tuan, 1977).
There is a history of both qualitative and quantitative work associated
with place attachment, with a predominant focus on creating measures
and identifying its constituent parts (Trentelman, 2009). What these
parts are, however, is still a topic of some debate and confusion stem-
ming from different lenses employed by the multiple disciplines in-
volved in place attachment research (Ardoin, Shuh and Gould 2012).
Environmental psychologists have historically divided place at-
tachment into two constituent parts; place dependence and place
identity (Trentelman, 2009). Place dependence evaluates the utility of a
place for people's needs and wants compared to other places while place
identity refers to how people feel a place to be a part of them
(Trentelman, 2009). In more recent studies additional dimensions of
place attachment have been suggested and tested empirically. These
have included biophysical/environmental attachment (attachment to
natural aspects of place), sociocultural attachment (attachment to the
community and strong relationships with community members), psy-
chological attachment (identification with place), economic attachment
(dependence on the institutions in a place for an individual's liveli-
hood), family bonding (a familial attachment to place), and friend
bonding (an attachment to friends within a specific place) (Ardoin,
Schuh, & Gould, 2012; Beckley, 2003, pp. 105–126; Brehm, Eisenhauer
& Kannich, 2006; Cross, Keske, Lacy, Hoag, & Bastian, 2011; Raymond,
Brown, & Weber, 2010; Scannell & Gifford, 2010). Even when basic
dimensions are agreed upon, they are often labeled in different ways –
for instance Scannell and Gifford (2010b) break place attachment into
natural and civic dimensions while Larson, De Freitas, and Hicks (2013)
refer to environmental, social and economic dimensions. The primary
components of these definitions are the same, but become divided and
specified in different ways through the use of different terminology.
Obviously there is some confusion over the dimensions of place
attachment. Despite this, there is a general agreement that it can be
defined as an affective relation to place, and that this relationship can
come about given time and/or intensity of interactions (Trentelman,
2009). Additionally, social and natural aspects of place attachment
seem to have been generally accepted across studies. However there are
those who have advocated against dividing place attachment into
multiple dimensions, instead arguing that it is a fundamentally holistic
concept that should not be broken into parts (e.g. Williams & Stewart,
1998).
A relevant issue associated with the definitional confusion for place
attachment is disparate findings regarding the effects of place
56
Journal of Environmental Psychology 60 (2018) 55–62
attachment on farmer behavior. Gosling and Williams (2010) found
that place attachment predicted environmentally friendly behavior in-
tentions, but not actual behavior on farms. In contrast, Lincoln and
Ardoin (2015) found that sense of place and related attachment were
very strong predictors of farmer's sustainable practices in South Kona,
Hawaii. Similarly, Marshall et al. (2012) in their study of Australian
peanut farmers conclude that, “… for change events that do not involve
moving locations, place attachment is likely to be a positive influence
on adaptation” (p. 7). Given the limited number of studies and the
disparate nature of these findings, relationships between place attach-
ment and farm management need further testing.
2.2. Place disruption
Place disruptions are theorized to trigger a greater sense of place
attachment because the threat of losing a place makes these affects
extant (Anton & Lawrence, 2014). In general, place disruption appears
to increase the intention to prepare for a disaster and sometimes actual
preparedness behaviors (Anton & Lawrence, 2014; Paton, Burgelt, &
Prior, 2008). However, there are exceptions in which disruption in-
teracts with high attachment to produce avoidance and lessen pre-
paredness (De Dominicis, Fornara, Cancellieri, Twigger-Ross, &
Bonaiuto, 2015). I extend this research to the socio-environmental
problem of herbicide resistant weeds – an environmental and social
disruption co-created by humans and nature. Specifically, I focus on
how these disruptions impact attachment and how this may condition
reactive and preventative management behaviors.
The disruption framework I employ is suggested by Mihaylov and
Perkins (2014). This framework builds on Devine-Wright's five stages of
psychological responses to place change. These are: 1) become aware
that place change is imminent, 2) interpret implications, 3) evaluate
change as positive or negative, 4) decide how to cope/respond and 5)
decide what can be done and act. Mihaylov and Perkins adapt this
model to create a schematic that can be adapted for use in structural
equation modeling. This schematic proposes that environmental dis-
ruption, place attachment dimensions, and belief in collective efficacy
all impact interpretive processes, i.e. step two in Devine-Wright's fra-
mework. These interpretive processes iteratively influence place-based
social interactions and social capital, as well as having a unidirectional
impact on citizen participation and community response to the dis-
ruption, i.e. steps 3 and 4 in Devine-Wright's framework. Additionally,
interpretive processes are moderated by social interactions and citizen
participation to impact community responses – the final step in the
framework. This model helps account for differing responses to place
disruptions and effects of attachment as seen in the literature.
3. Methods
3.1. Survey
During winter 2015 and spring 2016 an interdisciplinary research
team, including the author, implemented a national survey to gain a
better understanding of weed management practices among farmers
across the United States. Questions were developed to reflect the in-
terests and expertise of all team members, including sociologists, weed
scientists, agricultural economists, and extension educators. The sample
included approximately 9000 farmers from 28 states, comprising five
major row crop production regions. Corn and soybean farmers were
targeted as the primary respondents due to their crops' significance to
U.S. agriculture and documented problems with HR weeds (Heap,
2016; USDA, 2018). The full mixed-mode survey conducted via elec-
tronic and regular mail is available from the author.
In total, 839 usable surveys were returned for a response rate of 9.3
percent – just below the national average farmer response rate of about
10–20% (Pennings, Irwin, & Good, 2002). This was largely a result of
the chosen sampling frame. The research team chose a marketing
K. Dentzman
company that maintains a list of farmer mailing and email addresses;
unfortunately it appears that this list was not maintained to remove
individuals who were no longer farming. However those surveys that
were returned were nearly all complete; there was very little mea-
surement error. This is an indication that our survey instrument was
appropriate for the population, despite the fact that there was some
difficulty reaching said population.
Survey respondents were distributed across 28 states, with 41 per-
cent coming from Arkansas, Iowa, Illinois, Minnesota, Nebraska and
Texas. The research team compared demographic characteristics of
respondents with that of Census and USDA information bases.
Statistical (t) tests suggested that, within these regions, respondents
were skewed toward older white men who operated slightly larger
farms. While this does reflect a good deal of the actual farmed acreage
in the U.S., it still must be taken into account in the interpretation of
results. Small and minority farmers are underrepresented and this
survey does not reflect their opinions and attitudes.
Specific to place attachment, the survey included a question set of
11 place attachment items drawn from studies by Stokols and Shumaker
(1981), Proshansky (1978) Raymond et al. (2010), Ardoin et al. (2012)
and Scannell and Gifford (2010). These dimensions included place
identity, place dependence, environmental attachment, economic de-
pendence, community attachment, and family bonding/attachment
(full question wording is available in the online Appendix). Pertinent
information from a set of farmer focus groups (see Dentzman, 2018)
guided the selection of these dimensions for inclusion on the survey as
well as the actual question wording and labels for the dimensions.
Additionally, a question indicating non-attachment to place was added
in order to account for individuals that felt specifically unattached.
3.2. Analysis
Two primary types of data analysis were used for this article: ex-
ploratory factor analysis (EFA) and structural equation modeling
(SEM). I first employed EFA to construct the latent variables used in the
subsequent SEM model (see Table 2 in the online appendix). EFA is
particularly useful for measuring variables that are not directly ob-
servable, such as place attachment, and has been advocated and used
extensively in this literature (e.g. Ardoin et al., 2012; Manzo and
Devine-Wright, 2013). It may also be used, as it is here, to identify the
latent variables used in SEM analysis.
SEM—a type of path analysis—was chosen because it enables the
testing of multiple sequential dependencies involving latent variables
(Duncan, 1966; McDonald & Ho, 2002). In particular, it enables an
interpretation of the correlation of exogenous and endogenous vari-
ables with both direct and indirect effects (Duncan, 1966). While SEM
analysis does not imply causality, it does allow us to interpret and ex-
plicate a specific causal scheme for comparison with theoretical re-
lationships (Duncan, 1966; Land, 1969; McDonald & Ho, 2002). There
are four main categories of variables in the SEM model I use: 1) an
initial disruption variable, 2) place attachment variables 3) mediating
variables, and 4) an outcome variable. The EFA construction of each is
described below.
For the initial disruption variable my model used the presence of the
HR weed Palmer amaranth on a farmer's farm. Although respondents
were asked about a variety of HR weeds, Palmer amaranth constitutes
one of the most common, disruptive and feared HR weeds to date
(Inman et al., 2016) and was chosen to represent larger herbicide re-
sistance problems that are affecting the U.S. This variable is not latent.
Therefore one question ‘Is herbicide resistant Palmer amaranth present
on any of the land in your farming operation’ was used for this variable.
The full model was also tested with an alternative disruption variable:
the presence of HR Palmer amanrath in the respondent's state. This was
not a self-reported question from the survey, but rather a categorical
variable created from descriptive statistics reporting the percentage of
farms infested with HR Palmer amaranth in each state represented by
57
Journal of Environmental Psychology 60 (2018) 55–62
our sample (J. Norsworthy, personal communication, June 16, 2016).
The second category of variables in the model are the place at-
tachment variables. Given that place attachment is an underlying
construct with multiple dimensions, I used EFA to construct these latent
variables. All the place attachment questions used in the survey were
run together in an EFA in order to see where they clustered together
and if different place attachment dimensions were identifiable.
The third variable category is mediation variables, which in
Mihaylov and Perkins' model include collective efficacy, social action/
capital, and interpretation. As these concepts are not directly measur-
able they were constructed using an EFA of measurement variables (see
online Appendix). While collective efficacy and social action/capital
had clear and straightforward measurement variables, the latent vari-
able ‘interpretations’ required theoretical clarification. Measurement
variables for this construct were chosen based on a close reading of the
HR weed literature and previous results from focus group interviews
with corn and soybean growers (see Dentzman, 2018). These provided
insight into the different ways farmers interpret HR weeds, prompting
the inclusion of measurement variables for interpretations including
‘technology will fix the problem’, ‘HR weeds are a serious problem not
easily solved’, ‘incentive programs will fix the problem’, and ‘the
structure of agriculture prevents fixes to the problem’. For specific
question wordings, see Table 1 in online Appendix.
The fourth and final variable category in the disruption model is the
outcome category. For my model, farmers' weed management behaviors
are the outcomes of interest. While individual weed management
practices are easily measured, which practices cluster together and
what weed management principles they represent required an EFA.
Measurement variables representing practices related to herbicide use
and IWM were therefore run in an EFA to determine what typologies of
weed management respondents were engaging in. See Table 1 in online
Appendix for question wording.
After defining the variables using EFA, the full EFA scores were used
to run an SEM. The SEM model was specified as follows. Mihaylov and
Perkins' (2014) place disruption model and Devine-Wright's (20009)
five stage time-sequence responses to place disruption were used as
theoretical grounding for the presence of every directed path—that is,
every presumed direct effect of one variable on another. All non-di-
rected paths were omitted resulting in a recursive and immediately
identifiable model (McDonald & Ho, 2002). Thus it must be recognized
that the results may be fundamentally altered by the discovery and
addition of relevant variables in future studies (McDonald, 2002). I also
condensed two two-way paths in Mihaylov and Perkins' model into one
one-way path. The survey did not include sufficient variables to dif-
ferentiate social action and social capital—therefore these variables
were combined into one ‘social elements’ variable labeled ‘neighbor/
citizen participation’ (see Table 2 in online Appendix). This variable,
given its reconfiguration, is not clearly instantaneously reciprocal with
interpretive processes. Therefore its two-way relationship was modified
to a one-way relationship in which interpretive processes precede
‘neighbor/citizen participation’.
The majority of variables in the model were normally distributed;
where exceptions existed the variables were recoded to eliminate out-
liers and condense categories. Missing data was minimal (∼10%);
therefore any cases with missing data were deleted. The final path
model appears in Fig. 2 through 4; note that although control variables
and measurement variables were included, they do not appear in the
path model due to space constraints. Control variables included region,
age, gender, years on farm, proportion of acres rented, and total
number of acres.
Eight iterations of the defined disruption model were run in two sets
of four. In the first set, the models were run using ‘Palmer amaranth on
farm’ as the disruption variable. In the second set, the models were run
using ‘Palmer amaranth in state’ as the disruption variable. Within each
set, four different models were run; one for each of the weed man-
agement outcome variables determined by EFA. The collective efficacy,
K. Dentzman
Fig. 1. Disruption model for basic herbicide usage [*p < 0.05, **p < 0.00].
neighboring participation, and place attachment variables remained the
same in each of the eight models. Additional alternative models were
specified, run, and analyzed; the model reported in this article had the
best measures of fit. Due to space the alternative models are not re-
ported here; however the author may be contacted to access these
models.
4. Results
4.1. Descriptive statistics and latent variables
In order to test which dimensions of place attachment are relevant
to this sample of farmers, I ran an exploratory factor analysis (Table 1).
Two primary factors were identified. The first was composed of the
place attachment items ‘farm identity’, ‘farm family bonding’, ‘farm
environment identity’, ‘farm job identity’, and ‘farm dependence’. These
variables had factor loadings ranging from 0.50 to 0.70. I term this
factor ‘holistic farm attachment’, as the variables involved represent a
Fig. 2. Disruption model for integrated herbicide usage [*p < 0.05, **p < 0.0].
58
Journal of Environmental Psychology 60 (2018) 55–62
holistic way of being attached to the farm. The second factor is com-
posed of the items that capture ‘community ties’, ‘NOT attached to
farm’, and ‘discuss farming with neighbors’. The factor loadings range
from 0.41 to 0.51. Interestingly, ‘community ties’ and ‘discuss farming
with neighbors’ also had high factor loadings on the first factor—ho-
listic attachment. However, they were also highly tied to the dimension
‘not attached to farm’. I chose to accept this as a second factor because
of theoretical and empirical evidence that community attachment exists
as separable from place attachment, and can have very different im-
pacts on resource management (Larson et al., 2013; Trentelman, 2009).
I term this second factor ‘community attachment’, as it represents
farmers who are not particularly attached to their farm but are closely
involved in and attached to their community. The resulting two-factor
solution meets requirements for identifiability following the in-
dependent clusters basis (McDonald & Ho, 2002). Economic depen-
dence and environmental attachment items were dropped because they
did not strongly load onto any factor; additionally it is suspected that
‘farm dependence’ and ‘farm environment identity’ may have accounted
K. Dentzman Journal of Environmental Psychology 60 (2018) 55–62

Fig. 3. Disruption model for simple IWM usage [*p < 0.05, **p < 0.00].

for these dimensions of attachment. Table 1

Following the EFA for place attachment, I performed factor analyses Exploratory factor analyses of place attachment.
for the other latent variables in my model (see Table 2 in the online Holistic Place Attachment Factor 1 Factor 2
Appendix). These were ‘collective efficacy’, ‘neighboring/citizen parti-
cipation’, ‘weed management’, and ‘interpretations’. All of these had Farm identity 0.68 −0.12
acceptable factor loadings of greater than 0.30 on all dimensions Attached to farm genealogically 0.65 −0.36
Identify with environment on farm 0.65 0.1
(DeVellis, 2012). Additionally, all factors met the requirement of in-
Identify with farming job 0.55 −0.21
dependent clusters for identifability except for ‘Basic Integrated Weed Farm land contributes to success 0.52 −0.12
Management’, which met the independent cluster basis (McDonald & Community Place Attachment
Ho, 2002). ‘Collective efficacy’ and ‘neighboring/citizen participation’ Community ties 0.61 0.59
Not attached to farm −0.52 0.52
each clustered onto one single factor. ‘Weed management’ clustered
Discuss farming with neighbors 0.57 0.61
onto four factors – basic herbicide use, complex herbicide use, basic
IWM, and complex IWM. ‘Interpretations’ clustered onto five factors;
spread, incentive, management, new technology, and concern. A spread financial incentives for farmers to control their weeds. A management
interpretation indicates that respondents think HR weeds are a serious interpretation indicates that the respondent believes modern farming
problem that is highly mobile and easily transferable from farm to farm practices are at least partly at fault for the development and spread of
and region to region. An incentive interpretation indicates that re- HR weeds. A new technology interpretation indicates that the
spondents think HR weeds can be adequately addressed through

Fig. 4. Disruption model for complex IWM usage [*p < 0.05, **p < 0.00.

59
K. Dentzman
respondent believes that new technological innovations, especially new
herbicide modes of action, will solve resistance problems. Finally, a
concern interpretation indicates that the respondent sees HR weeds as
an extremely serious agricultural issue.
4.2. Disruption model findings
Although I tested models with two different disruption variables,
the ‘HR Palmer in state, but not on farm’ variable had no significant
impacts on holistic or community attachment. It also did not influence
belief in collective efficacy. Therefore, it had no direct or indirect im-
pact on farmer's interpretations or actual weed management practices.
This model is not described. The following results are based on the
models run with ‘HR Palmer on farm’ as the disruption variable.
The disruption models that I tested were all within 0.01 of a 0.80
CFI score and a 0.04 RMSEA score, indicating a good fit. Only results
significant at the p < 0.05 level or lower are discussed (see Figures). In
the Mihaylov and Perkins' disruption model, the physical place dis-
ruption is posited to impact place attachment and collective efficacy.
My SEM model supports this—having HR Palmer amaranth on one's
farm increased community attachment by 0.14 and holistic attachment
by 0.08. Having HR Palmer amaranth on-farm also decreased re-
spondents' belief in collective efficacy by −0.08 units.
The two place attachment items had several impacts on the med-
iating variables. Holistic farm attachment had an impact on three dif-
ferent forms of interpretation, while community attachment impacted
two. Specifically, holistic attachment increased a concern interpretation
(0.07) and a spread interpretation (0.14), but decreased an incentive
interpretation (−0.21). Community attachment caused an increase in
concern interpretation (0.04) and new technology interpretation (0.10).
Some mediating variables interacted to influence other mediating
variables. Specifically, an increase in collective efficacy decreased an
incentive interpretation (−2.21), a management interpretation
(−0.98), a concern interpretation (−0.031), a new technology inter-
pretation (−0.45) and a spread interpretation (−0.90). Additionally,
two interpretations affected neighboring/citizen participation. A man-
agement interpretation had a negative impact (−0.15), while a concern
interpretation had a strong positive impact (1.10).
Looking at how the interpretation variables influenced actual
management practices, I considered their impact on each of the four
outcome variables (basic herbicide use, integrated herbicide use, basic
IWM, and complex IWM). Basic herbicide use (Fig. 1) was impacted by
just one interpretation – a new technology interpretation. This in-
creased the intensity of basic herbicide use by 0.13 on a 3-point scale.
Integrated herbicide use (Fig. 2), simple IWM use (Fig. 3), and complex
IWM use (Fig. 4) were not impacted by any of the interpretations.
However, both simple and complex IWM use were increased by the
mediating variable ‘neighbor/citizen participation’. Therefore, two
primary chains of direct and indirect effects emerge; one for the basic
herbicide use outcome and one for the simple/complex IWM outcomes.
In the first chain—with basic herbicide use as the outcome variable
(Fig. 1)—the intensity of herbicide use was predicted by farmer's in-
terpretation that new technology would come along to solve herbicide
resistance problems. This interpretation was positively impacted by
community attachment and negatively by collective efficacy. In turn,
community attachment was increased by the presence of HR Palmer
amaranth on-farm while belief in collective efficacy was decreased.
Therefore, having HR Palmer amaranth on-farm indirectly increased basic
herbicide use through community attachment and indirectly decreased
basic herbicide use through collective efficacy beliefs.
In the second chain – with simple/complex IWM use as the outcome
variables (Figs. 3 and 4) – the intensity of IWM practices was not pre-
dicted by any of the five interpretations, but was increased by the
mediating neighboring/citizen participation variable. Neighboring/ci-
tizen participation itself was increased by a concern interpretation. This
concern interpretation, was increased by holistic place attachment and
60
Journal of Environmental Psychology 60 (2018) 55–62
decreased by a belief in collective efficacy. Therefore, having HR Palmer
amaranth indirectly impacts simple and complex IWM management
practices positively through place attachment and negatively through
belief in collective efficacy.
5. Discussion
Using an EFA I found that U.S. corn and soybean farmers' place
attachment divided into a holistic farm-based place attachment and a
non-farm community attachment. This supports a more holistic view of
place attachment—that is, that it cannot be divided into multiple
smaller dimensions such as environmental, economic, and family at-
tachment (e.g. Williams & Stewart, 1998). It also suggests that, for
farmers at least, place and community attachment may be considered
distinctive kinds of attachment. This is not to discount the literature
from which I drew the different dimensions of place attachment. Ra-
ther, it indicates that for the present sample of U.S. farmers their
identity, dependence, farmland, job, and family legacy are so tied to
place that it is impossible to consider them independently. This fits with
Beckley’s (2003) assertion that farmers have a particularly strong place
attachment due to being so actively involved with their land. Farmers
that are less attached to their farmland also seem to be more attached to
their local community.
I next investigated whether these two kinds of place attach-
ment—holistic and community—were influenced by HR Palmer amar-
anth on a farmers' land. This was supported by the results of the SEM
model. Specifically, within the model, the presence of Palmer amaranth
was related to a statistically significant increase in both types of at-
tachment. However, the threat of herbicide resistance does not appear
to have the same effect. This indicates that, although it may be of
concern, state-level herbicide resistance alone might not be enough of a
threat to trigger place attachment. However, it must also be noted that
the variable ‘HR Palmer amaranth in state’ was not self-reported by
farmers, and therefore individual respondents may have been more or
less aware of this issue. The initial interpretation, however, is supported
by findings from Bonny (2016) and Llewellyn et al. (2004) showing that
farmers tend to adopt IWM practices only after herbicide resistance
appears on their own farms.
Within the SEM model, holistic farm attachment and community
attachment influenced various interpretations. Specifically, the model
suggests that the presence of HR weeds may result in strong feelings of
attachment, causing farmers to interpret herbicide resistance as a ser-
ious concern that is likely to spread, while decreasing their belief that
financial incentives would be sufficient to deal with herbicide re-
sistance. Additionally, the model supports the reading that farmers with
a high level of community attachment interpret herbicide resistance as
a serious concern, but also believe that new technological innovations
will be enough to combat it. It is theoretically significant that, within
the model, community attachment increases the interpretation of
technological fixes, whereas holistic farm attachment does not. This
may be a situation in which farmers who are not attached to their
specific farm think technology is the larger picture answer.
Farmers' weed management behaviors were also impact in several
ways within the model. For one, farmers' basic herbicide use was in-
creased by an interpretation that new technologies would solve current
herbicide resistance problems. This technology interpretation is, as
mentioned above, increased by community attachment. Therefore from
this model I posit that higher community attachment triggered by HR
weeds on a farm encourages farmers to believe in technological solu-
tions and rely on basic herbicide weed management.
Both simple and complex IWM use were increased by farmers'
neighboring behaviors (i.e. discussing and getting weed management
information from neighboring growers). This emphasizes that commu-
nity-level conversations may be extremely important for encouraging
farmers to use basic and increasingly complex integrated weed man-
agement solutions to herbicide resistance. Additionally, neighboring
K. Dentzman
practices were increased by an increasing concern interpretation, which
itself was a result of triggered place attachment. Thus one can read a
pattern in which herbicide resistant weeds on-farm—when filtered
through place attachment and a concern interpretation—increase
neighboring behavior and IWM control practices.
There are some relevant limitations to these findings. The survey
response was somewhat limited and biased towards male farmers with
larger operations making more money as compared to the average for
the USDA Census of Agriculture. Corn and soybean growers were also
the main respondents. Therefore the results reported above may be
different for female operators, different crops, smaller more diverse
farms, and farms on which off-farm income is more important. Weed
management can be very different on such farms, and motivations may
differ from those of our survey respondents.
6. Conclusions
Within an SEM model, holistic farm attachment and community
attachment were increased by the presence of HR Palmer amaranth on
farmers' land. This indicates that herbicide resistance constitutes a place
disruption that may trigger latent feelings of attachment, an inter-
pretation that is supported by the literature on HR weeds. HR weeds fit
the definition of a place disruption in that they threaten place distinc-
tiveness, the self-efficacy of farmers, control over their farmland, and
the continuity of their farm through future generations (Anton &
Lawrence, 2016; Devine-Wright, 2009). Instead of responding to this
threat by detaching from a place, my model suggests that farmers re-
spond through an increased recognition of their attachment to both
their farm and community.
This increased attachment had several impacts on how farmers in-
terpret herbicide resistance within the model and their resulting weed
management practices. These findings point to several paths that may
be effective in helping farmers move towards a more sustainable in-
tegrated weed management plan. First, the SEM model suggests that it
will be important to educate farmers—and in particular those with low
holistic attachment—on the likelihood of new herbicide development
and new technological solutions. A new herbicide mode of action has
not been discovered for over 20 years (Livingston et al., 2015) and is
not expected in the near future. Other solutions, such as crops resistant
to three or more herbicides (Pates, 2016), are short-term solutions and
will exacerbate the herbicide resistance problem in the long-term.
Secondly, one interpretation of the model suggests that there needs
to be a focus on increasing farmers' interactions with their neighbors
and other farmers. In particular, there may be a need to direct this at
farmers with high holistic place attachment and low community at-
tachment. This is especially relevant given that the survey also showed
nearly 45% of farmers have never discussed herbicide resistance with
their neighbors. This is corroborated by findings from Dentzman and
Jussaume (2017). In their focus groups of corn and soybean growers,
they found that growers were very reticent to discuss weed manage-
ment with neighbors as it contrasted with their ideologies on farmers'
individualism and autonomy. Taken together these findings provide
strong evidence that farmers have difficulty communicating with their
neighbors and other farmers about herbicide resistance—however
when they do so it may have benefits in terms of IWM usage. Pro-
gramming that builds on the concern triggered by place disruption may
help bring growers together for these necessary discussions.
As previously mentioned, SEM does not necessarily imply causality.
Rather, interpretations can be made from patterns in the model com-
bined with existing theoretical and empirical evidence. Therefore while
I suggest that HR weeds represent a place disruption impacting place
attachment, various interpretations, and subsequent weed management
practices, this is one possible interpretation of a complex phenomenon.
Although my model provided a good fit as measured by several in-
dicators, it is possible that other relevant factors could be discovered
and added to the model resulting in significant changes to existing
61
Journal of Environmental Psychology 60 (2018) 55–62
patterns. It is thus necessary to further probe the results of this study.
Future research needs include replicating this model with different data
sets, testing alternative configurations including any newly relevant
variables, and scrutinizing the patterns found in this study using further
quantitative and qualitative methods.
Acknowledgements
Primary funding for this research was provided by a USDA-AFRI
grant (#122422) on “Integrating Human Behavioral and Agronomic
Practices to Improve Food Security by Reducing the Risk and
Consequences of Herbicide-Resistant Weeds.”
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.jenvp.2018.10.006.
References
Anton, C. E., & Lawrence, C. (2014). Home is where the heart is: The effect of place of
residence on place attachment and community participation. Journal of Environmental
Psychology, 40, 451–461.
Anton, C. E., & Lawrence, C. (2016). The relationship between place attachment, the
theory of planned behaviour and residents' response to place change. Journal of
Environmental Psychology, 47, 145–154.
Ardoin, N. M., Schuh, J. S., & Gould, R. K. (2012). Exploring the dimensions of place: A
confirmatory factor analysis of data from three ecoregional sites. Environmental
Education Research, 18(5), 583–607.
Barman, K. K., Singh, V. P., Dubey, R. P., Singh, P. K., Dixit, A., & Sharma, A. R. (2014).
Challenges and opportunities in weed management under a changing agricultural
scenario. Recent advances in weed management (pp. 365–390). New York: Springer.
Beckley, T. M. (2003). The relative importance of sociocultural and ecological factors in at-
tachment to place. United States Department of Agriculture Forest Service General
Technical Report PNW.
Bonny, S. (2016). Genetically modified herbicide-tolerant crops, weeds, and herbicides:
Overview and impact. Environmental Management, 57(1), 31–48.
Brehm, J. M., Eisenhauer, B. W., & Krannich, R. S. (2006). Community attachments as
predictors of local environmental concern: The case for multiple dimensions of at-
tachment. American Behavioral Scientist, 50(2), 142–165.
Cross, J. E., Keske, C. M., Lacy, M. G., Hoag, D. L., & Bastian, C. T. (2011). Adoption of
conservation easements among agricultural landowners in Colorado and Wyoming:
The role of economic dependence and sense of place. Landscape and Urban Planning,
101(1), 75–83.
De Dominicis, S., Fornara, F., Cancellieri, U. G., Twigger-Ross, C., & Bonaiuto, M. (2015).
We are at risk, and so what? Place attachment, environmental risk perceptions and
preventive coping behaviours. Journal of Environmental Psychology, 43, 66–78.
Dentzman, K. (2018). ‘I would say that might be all it is, is hope’: The framing of herbicide
resistance and how farmers explain their faith in herbicides. Journal of Rural Studies,
57, 118–127.
Dentzman, K., & Jussaume, R. (2017). The ideology of US agriculture: How are integrated
management approaches envisioned? Society & Natural Resources, 1–17.
DeVellis, R. F. (2012). Scale development: Theory and applications, 26. Sage publications.
Devine‐Wright, P. (2009). Rethinking NIMBYism: The role of place attachment and place
identity in explaining place‐protective action. Journal of Community & Applied Social
Psychology, 19(6), 426–441.
Duncan, O. T. (1966). Path analysis: Sociological examples. American Journal of Sociology,
72(1), 1–16.
Ervin, D., & Jussaume, R. (2014). Integrating social science into managing herbicide-
resistant weeds and associated environmental impacts. Weed Science, 62(2), 403–414.
Gieryn, T. F. (2000). A space for place in sociology. Annual Review of Sociology, 26(1),
463–496.
Gosling, E., & Williams, K. J. (2010). Connectedness to nature, place attachment and
conservation behaviour: Testing connectedness theory among farmers. Journal of
Environmental Psychology, 30(3), 298–304.
Heap, I. (November 2, 2016). The international survey of herbicide resistant weeds. Online.
Available at: www.weedscience.org.
Inman, M. D., Jordan, D. L., York, A. C., Jennings, K. M., Monks, D. W., Everman, W. J., &
Soteres, J. K. (2016). Long-term management of Palmer amaranth (Amaranthus
palmeri) in dicamba-tolerant cotton. Weed Science, 64(1), 161–169.
Larson, S., De Freitas, D. M., & Hicks, C. C. (2013). Sense of place as a determinant of
people's attitudes towards the environment: Implications for natural resources
management and planning in the Great Barrier Reef, Australia. Journal of
Environmental Management, 117, 226–234.
Lincoln, N. K., & Ardoin, N. M. (2016). Cultivating values: Environmental values and
sense of place as correlates of sustainable agricultural practices. Agriculture and
Human Values, 33(2), 389–401.
Livingston, M., Fernandez-Cornejo, J., Unger, J., Osteen, C., Schimmelpfennig, D., Park,
T., et al. (2015). The economics of glyphosate resistance management in corn and soybean
production. Economic Research Report No. ERR-184. US Department of Agriculture.
K. Dentzman
Llewellyn, R. S., Lindner, R. K., Pannell, D. J., & Powles, S. B. (2004). Grain grower
perceptions and use of integrated weed management. Australian Journal of
Experimental Agriculture, 44(10), 993–1001.
Manzo, L. C., & Devine-Wright, P. (Eds.). (2013). Place attachment: Advances in theory,
methods and applications. Routledge.
Marshall, N. A., Park, S. E., Adger, W. N., Brown, K., & Howden, S. M. (2012).
Transformational capacity and the influence of place and identity. Environmental
Research Letters, 7(3), 034022.
McDonald, R. P., & Ho, M. H. R. (2002). Principles and practice in reporting structural
equation analyses. Psychological Methods, 7(1), 64–82.
Mihaylov, N., & Perkins, D. D. (2014). Community place attachment and its role in social
capital development. Place attachment: Advances in theory, methods and application
(pp. 61–74). .
Norsworthy, J. K., Ward, S. M., Shaw, D. R., Llewellyn, R. S., Nichols, R. L., Webster, T.
M., & Witt, W. W. (2012). Reducing the risks of herbicide resistance: Best manage-
ment practices and recommendations. Weed Science, 60(1), 31–62.
Pates, M. (Sept 21 2016). Exec: High-tech ag will lead to next Green Revolution. Agweek.
http://www.agweek.com/news/4120366-exec-high-tech-ag-will-lead-next-green-
revolution.
Paton, D., Burgelt, P. T., & Prior, T. (2008). Living with bushfire risk: Social and en-
vironmental influences on preparedness. Australian Journal of Emergency Management,
23(3), 41.
62
Journal of Environmental Psychology 60 (2018) 55–62
Proshansky, H. M. (1978). The city and self-identity. Environment and Behavior, 10(2),
147–169.
Raymond, C. M., Brown, G., & Weber, D. (2010). The measurement of place attachment:
Personal, community, and environmental connections. Journal of Environmental
Psychology, 30(4), 422–434.
Scannell, L., & Gifford, R. (2010). The relations between natural and civic place attach-
ment and pro-environmental behavior. Journal of Environmental Psychology, 30(3),
289–297.
Stokols, D., & Shumaker, S. A. (1981). People in places: A transactional view of settings.
Cognition, social behavior, and the environment, 441–488.
Trentelman, C. K. (2009). Place attachment and community attachment: A primer
grounded in the lived experience of a community sociologist. Society & Natural
Resources, 22(3), 191–210.
Tuan, Y. F. (1977). Space and place: The perspective of experience. U of Minnesota Press.
United States Department of Agriculture (2018). U.S. farmers expect to plant more soybean
acreage than corn acreage. United States Department of Agriculture National
Agricultural Statistics Service. Online. Available at: https://www.nass.usda.gov/
Newsroom/2018/03_29_2018.php.
Webster, T. M., & Sosnoskie, L. M. (2010). Loss of glyphosate efficacy: A changing weed
spectrum in Georgia cotton. Weed Science, 58(1), 73–79.
Williams, D. R., & Stewart, S. I. (1998). Sense of place: An elusive concept that is finding a
home in ecosystem management. Journal of Forestry, 96(5), 18–23.