Ecological Economics 129 (2016) 72–81

Contents lists available at ScienceDirect

Ecological Economics

journal homepage: www.elsevier.com/locate/ecolecon

Pesticide lock-in in small scale Peruvian agriculture

Courtney Hammond Wagner a,⁎, Michael Cox b, José Luis Bazo Robles c
a
Gund Institute for Ecological Economics, Rubenstein School of Natural Resources, University of Vermont, 81 Carrigan Drive, Burlington, VT 05405, United States
b
Environmental Studies Program, Dartmouth College, 6182 Steel Hall, Hanover, NH 03755, United States
c
Instituto Huayuná, Av. Santiago de Surco 4285, Lima 33, Perú

a r t i c l e i n f o a b s t r a c t

Article history: Despite decades of research into the negative impacts of synthetic pesticides, farmers in Latin America continue to
Received 18 November 2015 use pesticides at high levels and at a high cost to social and environmental sustainability. In this paper, we present a
Received in revised form 19 May 2016 case study of pest management strategies in small-scale agriculture, focusing on the unsustainable technological
Accepted 30 May 2016
lock-in of synthetic pesticides. Of the 196 smallholder farmers we surveyed in the coastal Mala and Omas Valleys
Available online xxxx
of Perú, 22% of respondents experienced pesticide poisoning themselves or by an immediate family member.
Keywords:
Additionally, the two most common pesticide categories reported in use are potent neurotoxins. We hypothesized
Pesticides that the farmers in the valleys were locked into synthetic pesticides due to uncertainty, coordination and learning
Agriculture associated with adopting an alternative strategy. Logistic regressions revealed gender (male), consulting an agro-
Lock-in chemical technician, quantity of cultivated land, and apple as a primary crop to be important predictors of synthetic
Technology adoption pesticide use. Our findings suggest that these predictors represent the lock-in of synthetic pesticides through
Sustainable food systems network externalities, learning economies and adaptive expectations. We conclude with opportunities to transition
to sustainable pest management strategies at the local level in Latin American communities through interventions
countering the lock-in of synthetic pesticides.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction America continue to use highly toxic synthetic pesticides and in such
high quantities.
In the 25 years since the release of the World Health Organization's A critical consideration is the challenge for farmers of adopting or
(WHO) report Public Health Impacts of Pesticides Used in Agriculture moving to a new pest management technology in lieu of synthetic
studies have continued to show the negative human health impacts of pesticides. Previous research has shown that the ability for farmers to
pesticide exposure and the pervasiveness of pesticide poisoning in move away from pesticide intensive practices in Latin America is
developing countries (World Health Organization, 1990, Ecobichon, hindered by farmers' lack of access to accurate information on pesticides
2001, Jørs et al., 2006; Dasgupta et al., 2007; Bhat et al., 2010; Lee and their alternatives, the limited markets available for organic produce,
et al., 2010; Cole et al., 2011). Farmers in developing countries tend and inadequate community organization to implement alternatives
to use more pesticides more frequently, and apply more highly toxic (Cole et al., 2011). Additionally, at national and international scales,
varieties than their counterparts in developed countries (Ecobichon, the pesticide industry influences and encourages pesticide adoption
2001). Additionally, it has been shown that while short-term synthetic through low prices, ease of access, and availability of technological
pest management strategies lower costs and boost yields on farms, in support (Galt, 2008).
the long-term, synthetic pesticides can raise farmer costs and lower Efforts have been made both globally and locally to address these
yield (Wilson and Tisdell, 2001). These global trends are born-out barriers to adopting more sustainable alternatives. Sustainable pest
amongst smallholder farmers in countries across Latin America, including management alternatives are those that seek to manage pests in a
Costa Rica, Bolivia, Ecuador, and Colombia among others (Crissman et al., way that socially, economically and environmentally meets farmers'
1998; Jørs et al., 2006; Feola and Binder, 2010a,b, Galt, 2013). The well- needs today without compromising the needs of future generations
known human health and economic risks of pesticide use in agriculture (Brundtland et al., 1987). International efforts such as the Rotterdam
and the existence of alternative pest management techniques highlight Convention and the Food and Agriculture Organization's (FAO) Code
the need for a better understanding of why smallholder farmers in Latin of Conduct on the Distribution and Use of Pesticides strive to ensure
that pesticide products are labeled with accurate information on toxicity
so that countries and farmers can make informed choices (Angelo,
⁎ Corresponding author. 2013). At the local level in many Latin American countries, government
E-mail address: [email protected] (C. Hammond Wagner). and non-government organizations have run programs to increase

http://dx.doi.org/10.1016/j.ecolecon.2016.05.013
0921-8009/© 2016 Elsevier B.V. All rights reserved.
 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81 73

management can be seen as a case of increasing returns due to the posi-

tive feedbacks associated with increased adoption (Cowan and Gunby,
1996). Technologies that demonstrate these increasing returns can
become locked-in such that once adopted, it is difficult to switch to
a superior alternative technique without a high switching cost. In
cases of unsustainable lock-in, or lock-in to an inferior technology,
ongoing costs, in this case, the social, economic and environmental costs
of using pesticides, are higher than that of an alternative technology
(Arthur, 1989).
Synthetic pesticide use can be traced to the expansion of the global
synthetic pesticide market following the Second World War. Adoption
following the war increased rapidly because of synthetic pesticides'
effectiveness in lowering cost and boosting yield (Wilson and Tisdell,
2001). The positive feedbacks associated with synthetic pesticide use
include: 1) economies of scale — a reduction in unit cost with increases
in production and yield, 2) learning economies — cost decreases and
performance increases as individuals gain more experience with apply-
ing pesticides, 3) adaptive expectations — a decrease in uncertainty
about how or whether a pesticide will perform, and 4) network external-
ities — the benefits accompanying networks, infrastructure, associated
technologies and other external structures that form around synthetic
pesticides as adoption of the technology increases (Perkins, 2003, Kallis
Fig. 1. Map of the cities of Mala and Omas on the Peruvian Pacific coast, located about and Norgaard, 2010).
60 km south of Lima, Perú. From the perspective of competing technologies with increasing
returns, the early organosynthetic insecticides developed in the 1940s
were in direct competition with non-synthetic pest control strategies
based on pest biology and ecology (the precursors to today's organic
preventative practices amongst farmers and to build capacity in alterna- and IPM strategies) (Kogan, 1998). Synthetic pesticides, in comparison
tive strategies such as Integrated Pest Management (IPM) and organic with the ecologically-based pest control strategies, boasted high initial
pest management (Bazo Robles et al., 2010; Jørs et al., 2014; Carrión effectiveness in eliminating pests and corresponding high initial boost
Yaguana et al., 2015). However, acute and chronic pesticide poisoning in yield. The first field trials with organochlorine pesticides in Perú in
persists (Jørs et al., 2006). The sustained use of synthetic pesticides and the 1940s, as with many trials across the globe, left farm operators
the difficulty of adopting alternative practices suggest that many amazed at the effectiveness of the products: no insects survived the
small-scale farmers in Latin America may be locked-in to using synthetic application (Herrera, 2010). Synthetic pesticides were and continue to
pesticides. be relatively easier to use and implement than alternative strategies
To explore local-level persistence of pesticide use and lock-in affecting and do not require extensive information about the agricultural ecosys-
small-scale agriculture, this study focuses on two rural agricultural tem (Waterfield and Zilberman, 2012). These attributes of synthetic
communities in the Mala and Omas Valleys along the coast of Perú. In pesticides, along with market pressure from synthetic producers and
the summer of 2013, a survey was implemented in the valleys through government policies to boost supplies and invest in research and devel-
a partnership between the Dartmouth Institute for Health Care Delivery opment, led to their dramatic rate of adoption and entrenchment in the
Science and Instituto Huayuná, a Peruvian non-governmental organiza- market, which continues today (Wilson and Tisdell, 2001).
tion that has been active in issues of community health and sustainable Synthetic pesticides are seen as an important tool globally for
agriculture for over 30 years. In each valley, we examined the predictors increasing production to meet the food demands of a growing popula-
of pesticide use, the level of interest among pesticide users for tion (Waterfield and Zilberman, 2012). However, from a social and
transitioning to organic pest management, and the barriers to environmental perspective, a reliance on synthetic pesticides as the
transitioning. Finally, we investigated whether there is a case of mal- primary pest management technique may be inefficient (Wilson and
adaptive technological lock-in in the use of pesticides in the valleys Tisdell, 2001). Human poisoning is known to occur broadly through
and considered potential pathways for transition to more sustainable acute or chronic exposure during application of pesticides and environ-
pest management practices. mental exposure to pesticide drift or contaminated soil, water or crops.
In the following sections we first examine the phenomenon of tech- In 1990, the World Health Organization estimated that pesticides are
nological lock-in and its application to the adoption of agricultural pest responsible for three million cases of acute poisonings, including
management technologies before introducing the case study site in suicides, and 220,000 deaths (World Health Organization, 1990). Addi-
section three. In section four we describe the study methods followed tionally, certain pesticides may be associated with increased risks of
by results and discussion in sections five and six. Finally in section seven chronic health impacts such as specific cancer types, reproductive and
we acknowledge the limitations of our study and in section eight we other health effects, and neurological effects (Miranda-Contreras et al.,
offer the concluding remarks. 2013; Muñoz-Quezada et al., 2013; Parrón et al., 2014). In many
countries, governments have banned or restricted access to highly toxic
2. Technological Lock-in, Pest Management, and Agricultural pesticides to protect against exposure and poisoning (see Table A.1 in
Innovation Appendix A for the number of countries that have banned pesticides
found in the present study). Synthetic pesticides are also known to have
The theory of technological lock-in suggests that a technology adverse environmental impacts, including loss of beneficial pests, broader
becomes “locked-in,” or entrenched, when that technology gains dom- loss in biodiversity, and soil and water contamination (Wilson and Tisdell,
inance over alternatives due to a historical event or chance that enables 2001, Pimentel et al., 2005).
it to capture a market (Kallis and Norgaard, 2010). The existence of tech- Awareness of the negative human health and environmental risks of
nological lock-in is tied closely to the concept of increasing economic an overreliance on synthetic pesticides arose as early as the late 1950s
returns to adoption (Perkins, 2003). The adoption of synthetic pest and inspired efforts to identify alternative pest management strategies,
74 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81

such as IPM and organic agriculture (Kogan, 1998). IPM as a strategy 3. Study Area
integrates biological and synthetic control, utilizing multiple techniques
and methods to control pest populations, with the goal of benefiting The Mala river basin stretches from the Peruvian Pacific coast, where
farmers (economically), society (minimization of public health risks) the city of Mala is situated on the Panamericana highway, up to its
and the environment (Kogan, 1998, Parsa et al., 2014). The adoption headwaters in the western range of the Andes. Likewise, the Omas
of IPM is associated with decreases in pesticide use and a corresponding River runs parallel to the Mala River, about 20 km to the south.
decrease in the negative externalities of pesticide use (Carrión Yaguana Both basins are situated in a coastal subtropical arid climate, featuring
et al., 2015). Farmers are trained to use target applications of pesticides low levels of precipitation and high humidity in the low altitudes of
only in instances that pose economic threats to the farmer. However, the basins (Alba, 2004).
IPM techniques have low rates of adoption in developing countries The Mala Valley is home to approximately 40,000 inhabitants, the
(Parsa et al., 2014). Compared to a synthetic pesticide strategy, IPM majority located near the coast (Bazo Robles et al., 2010). Omas Valley
is knowledge-intensive and requires a significant amount of time is much less populated than Mala, with a little over 12,000 people
and financial investment to adopt (Waterfield and Zilberman, (Perú Instituto Nacional de Estadística e Informatica, 2012). The city of
2012). Additionally, many programs rely primarily on the appropriate Mala is a commercial agricultural hub for the region, both for local
application of pesticides as the principal management tactic (Kogan, consumption and for transport to markets in Lima and to the south
1998). (Alba, 2004). The primary economic activity in the valleys is agricultural
Alternatively, organic agriculture is defined as a system of production from small-scale farms.
agriculture that focuses on holistic ecosystem management through, Prior to the Peruvian Agrarian Reform in 1969, agriculture in the
among other things, supporting soil health, biodiversity, and biolog- coastal region was dominated by large plantations growing cotton for
ical cycles to promote healthy agro-ecosystems, as opposed to a reli- export in the lower elevations of the valleys, and peaches and grapes
ance on external inputs (The Codex Alimentarius Comission and the at higher elevations. Within the plantation structure small-scale farmers
FAO/WHO Food Standards Programme, 2007). In practice, and for worked the land as sharecroppers (Kay, 1982, Alba, 2004). The Agrarian
organic certification, this typically requires that farmers do not use Reform of 1969 transitioned the plantations to worker-owned Agri-
synthetic pesticides or fertilizers. The adoption of organic agriculture cultural Production Cooperatives. However, in the decade following
is also a knowledge-intensive process and usually takes multiple the Agrarian Reform, farmers abandoned the cooperative model
years to achieve certification. Organic pest management techniques and began dividing parcels up among farm workers. Today, there
allow for additional benefits for farms beyond IPM, including reduction are no remaining agricultural cooperatives in either of the valleys,
in health risks, a price premium for produce, and improved long-term and instead the valley is made up of smallholder farms. In our
farm environmental conditions (Waterfield and Zilberman, 2012). It is sample, the average farm size in the Mala and Omas Valleys was
important to note that some varieties of organic pesticides, although 2.5 ha, as reported in Table 1.
they do not contain synthetic compounds, are still highly toxic for An important remnant in the valleys of both the plantations and
farmers, such as the organic pesticide arsenic. Compared to synthetic cooperatives is the use of synthetic pesticides. The large plantations of
pesticides and IPM, organic agriculture can be seen as a riskier pest the '50s and '60s utilized agro-chemicals early on, primarily persistent
management technology because it entails a more limited amount organic pollutants (Bazo Robles et al., 2010). Interestingly, the large
of pest control strategies (i.e., no synthetic inputs) (Waterfield and cotton plantations in the nearby valley of Cañete were also one of the
Zilberman, 2012). earliest adopters of a form of IPM in the 1950s because of their first-
In Latin America, while efforts have been made to introduce both hand experience with pesticide resistance and pest resurgence with
IPM and organic agriculture across many countries, farmers have been the early organochlorine pesticides (Herrera, 2010). Unfortunately, this
shown to still over-apply pesticides and use pesticides banned in many process of IPM was interrupted and lost in the 1970s with the Agrarian
industrialized countries (Eskenazi et al., 1999, 2007; Bjorling-Poulsen Reform (Herrera, 2010).
et al., 2008; Lesmes-Fabian et al., 2012). In particular organophosphate During the period of the 50s and 60s, fruits grown higher up in
and carbamate pesticides, known to be potent neurotoxins, are still the valleys (peaches and grapes) were plagued by scale insects (of the
commonly used and poisoning incidents occur relatively frequently families Diaspididae, Coccidae, and Pseudococcidae). Production
(Sherwood et al., 2005; Jørs et al., 2006; Pesticide Action Network, became unprofitable because of the impacts of pests and farmers
2015). began to move away from these crops. When the Delicia apple, later
Alternative pest management techniques such as organic and IPM known as the “Delicia de Viscas,” was introduced in the 1960s to the
have a hard time competing with synthetic pesticides because of the higher altitudes in Mala valley and thrived, farmers throughout the
large cost of transitioning (e.g., network co-ordination, technological region, including those at lower altitudes, began installing orchards.
interdependencies, new work practices, skills and patterns of behavior) The cultivation technique introduced and adopted with the Delicia
(Perkins, 2003). Using the language of Liebowitz and Margolis (1995), apple included the use of pesticides and fertilizers, with intense pruning
this could be considered a case of third-degree path dependence, and planting density. By the mid-1990s more than half of the agricultural
where reliance on synthetic pesticides continues to be the dominant land in Mala valley was in apple production.
pest control strategy despite knowledge that it is a relatively inefficient Today, small-scale farms continue the tradition of agro-chemical use
strategy (Wilson and Tisdell, 2001). In this way, the different mecha- in fruit production, with a high proportion still growing apples despite
nisms that contribute to increasing returns to synthetic pesticides can decreases in their profitability. According to agronomists at Instituto
be viewed as a barrier to adoption of a competing, superior technology, Huayuná, yield of apples per hectare has been declining for the last
when this particular inferior technology is locked-in. two decades. The principle pests of economic importance are the San
Trends in synthetic pesticide use in Latin America can reveal the José scale (Quadraspidiotus perniciosus) and the fungus Lasiodiplodia
existence of persistent problems like pesticide poisoning, but do little theobromae, as well as powdery mildew, bull beetles, and the codling
to describe the conditions that influence adoption and technological moth.
lock-in at the farm scale. To better understand farm-level pest man- Aside from the dominate crop of apples, the other main economic
agement decisions, this study investigates the use of pesticides and crops grown in the valleys as found in our study are maize, alfalfa,
occurrence of pesticide poisoning in two small-scale agricultural valleys banana and grapes, which are impacted economically by pests to
in Perú. In the case of pesticide use and persistent poisoning in the two varying degrees. Maize is significantly affected a stalk borer, the fall
valleys, we examine the barriers to adoption of an alternative form of armyworm (Spodoptera frugiperda), and corn earworm (Helicoverpa
sustainable pest management. zea). Grape growers in the region deal with a fungus that causes powdery
 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81 75

Table 1
Variables including descriptive statistics, description of the variable and categories or units.

Variable name Description Categories N (%) Mean ± SD

Omas valley Valley of residence Mala; Omas 137 (73%); 51 (27%) –

age Age of respondent Years – 56.8 ± 15.4
education Education level of farmer Only primary; at least some secondary 72 (38%); 116 (62%) –
education
female gender Gender of respondent Male; female 158 (84%); 30 (16%) –
land Hectares of land cultivated Hectares – 2.5 ± 2.6
land tenure Primary condition of land tenure Bought or inherited; rented; shared 142 (76%); 21 (11%); 25 –
ownership (13%)
pesticide Uses synthetic pesticides No; yes 33 (18%); 155 (82%) –
organic Uses or interested in using organic agricultural techniques No; yes 68 (36%); 120 (64%) –
crop diversity Effective species diversity, range in sample: 1.00 to 8.00 Index of crop diversity – 1.7 ± 1.0
technician Receives information about pesticides from an agro-chemical shop No; yes 72 (38%); 116 (62%) –
huayuna Receives information about pesticides from Instituto Huayuná No; yes 175 (93%); 13 (7%) –
farmers Receives information about pesticides from other farmers No; yes 168 (89%); 20 (11%) –
poison Experienced pesticide poisoning personally or in immediate family No; yes 146 (78%); 42 (22%) –
traps Use of non-synthetic pesticide traps No; yes 166 (88%); 22 (12%) –
apple Apple as primary commercial crop — greater than 50% of cultivated land No; yes 90 (48%); 98 (52%) –
in apple.
registry Keeps records of crops and inputs No; yes 136 (72%); 52 (28%) –
equipment Wears protective equipment when applying pesticides No; yes 93 (49%); 95 (51%) –
storage Stores pesticides in an appropriate location No; yes 53 (28%); 135 (72%) –

mildew of grape (Oidium tuckeri). Bananas have many pests, mainly week period in July and August 2013, with 142 interviews in Mala and
nematodes, but farmers do not tend to apply pesticides against them 54 in Omas. We conducted the sampling for the interviews in multiple
and instead replace the plant when needed. Finally alfalfa, in the region stages and included cluster, purposive, snowball, and convenience-
is plagued at times by the red spider (Tetranychus urticae), but harvesting based elements because a full list of farmers was not available for each
process removes the pest and therefore is not particularly damaging. valley. First, in consultation with community health workers associated
Since the initial IPM efforts in Cañete's cotton plantations, the with Instituto Huayuná, we identified clusters of farmers that lived in
Peruvian government has tried numerous efforts to introduce biological districts and annexes within each valley. Then, within a district or
pest management techniques. In the 1960s, the national Center for annex, we approached local residents to identify farmers to interview
Insect Breeding and Research Tools (CICIU) focused on efforts to intro- and asked interviewees to suggest additional farmers to interview within
duce exotic species for pest control. In the 1990s, these efforts passed the region. Due to our sampling methodology, we cannot claim that the
hands to the National Agrarian Health Service (SENASA) and their sample is statistically representative of farmers living in the study region.
National Biological Control program. Since 2001, SENASA has promoted Prior to the interviews, we developed a questionnaire through our
biological control on many thousands of hectares, however, these collaboration with Instituto Huayuná and several community health
efforts were primarily aimed at the large agricultural productions in workers (see Appendix B for the full questionnaire). Six broad topics
the country (Valdivieso Jara, 2011). were covered in the questionnaire: home attributes, health problems,
Separate from the state's endeavor, Instituto Huayuná implemented land and livestock, agricultural activities, external interactions, and agri-
an IPM program in the Mala and Omas valleys in the 1990s, but neither cultural associations. Additionally, the questionnaire was intended to
the state nor the local effort saw much success in the adoption of the collect data on the types of pesticides used by farmers in an annual
technique by small-scale agricultural producers. Additionally, those farming cycle, including the pesticide names, active ingredients, crop/s
farmers that did attempt to implement IPM in the valleys tended to on which the pesticides were applied, and the quantities applied. We
rely on synthetic pesticide application as the primary pest control also assessed history of acute pesticide poisoning incidents by asking
strategy, rather than the strategy of last resort. Following the IPM respondents to describe any time that they, or someone in their family,
program in the 1990s, a small group of farmers in the valleys, concerned has been poisoned. For each acute pesticide poisoning incident
with the number of severe pesticide poisonings and deaths in the recounted, we recorded the date, the relation (self or other family
community, decided to transition to organic agriculture. Since then, member), the severity of the incident (light or severe), the pesticide
Instituto Huayuná has been supporting farmers in transitioning to responsible and finally whether the respondent sought care at a clinic.
organic agriculture and training farmers in organic techniques through The questionnaire was divided into sections according to the six topics
farmer schools and an experimental organic farm. Because of Instituto above and each section includes a combination of dichotomous,
Huanyuná's focus on promoting organic techniques, the challenge of nominal, ordinal, and open-response questions.
measuring degree of implementation of IPM, and the institutional Interviews were conducted in Spanish by a fluent speaker: either a
difference between organic techniques and synthetic techniques (i.e. local interpreter or a community health worker in the Mala Valley, or
price premiums for crops and certification) that do not exist for IPM, a local farmer in the Omas Valley. The community health workers had
we focus this study on the use of two pest management strategies in previous experience in conducting interviews for a local census. Inter-
the Mala and Omas Valleys: synthetic pest management and organic viewers were trained on the questionnaire, participated in practice
agriculture. interviews, and were compensated for their time. After the completion
of the surveys, we reviewed each of the questionnaire's responses with
4. Methods the interviewers to ensure accurate interpretation of the data and
responses.
4.1. Data Collection This study was approved by the Internal Review board at Dartmouth
College. Before beginning each interview, participants were notified of
To investigate pesticide use, pesticide poisoning, and the adoption of the intent of the study, ensured that all data collected would be kept
alternative pest management strategies in the Mala and Omas Valleys, confidential and asked if they would like to participate. Upon comple-
we completed 196 semi-structured interviews in-situ during a four- tion of the data collection, identifying information was separated from
76 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81

responses both within the survey instruments and the database. Identi- management strategies. Table 2 also lists any economies of increasing
fying information is stored in a separate, secure location to protect the returns to adoption associated with each variable, so that the models
anonymity of the respondents. simultaneously investigated the existence of factors contributing to
technological lock-in. In our analysis we focused on economies of
4.2. Data Analysis learning, adaptive expectations, and network externalities, and not
economies of scale, to draw attention to non-monetized increasing
Following data collection, the data was cleaned, coded, and entered returns to adoption. Model 2 was run with a subset of the data to
into a Microsoft Access Database (2013). In cleaning the data, we investigate conventional pesticide user's interest in organic agricultural
removed observations from family members of the interviewers and techniques. The Model 2 sample subset included only those 155 respon-
incomplete interviews from the database, resulting in 188 unique farm- dents who qualified as a conventional pesticide user according to the
er responses. See Table 1 for a list of all variables included in the analysis variable pesticide.
and a summary of their values in our sample. Finally, we qualitatively coded open-ended responses regarding the
To categorize respondents as conventional pesticide users versus difficulties of using organic pest management strategies according to
non-conventional pesticide users for the binary variable pesticide, we categories identified in the responses to this question. Only a subset of
coded all the pesticides reported in the questionnaire, over 650 unique respondents was asked the question on organic pest management
mentions, as non-synthetic or synthetic pesticide products. We catego- difficulties or barriers. The logic of the questionnaire instructed the
rized general classes of pesticides (i.e., organophosphates, carbamates, interviewer to ask only those who responded in the affirmative to
pyrethroids) through consultations with a local agro-chemical shop in “use or interest in organic techniques,” which resulted in 114 responses
the city of Mala and the Peruvian Ministry of Agriculture's pesticide to this question. In our coding of responses, a single response could fall
registration documents (Servicio Nacional de Sandidad Agraria, 2014). into more than one category and would be categorized as a response in
Then we considered all pest management inputs reported by each each category. The coded responses to this question were not included
respondent and coded a respondent as a conventional pesticide user in either of the logistic regression models. Rather, we utilized these
(value of “1” for variable pesticide) if the respondent reported using at qualitative descriptions in our discussion section to interpret the results
least one conventional pesticide in the past year. We coded a respon- of Model 1.
dent as a non-synthetic pesticide user if they only reported to use
non-synthetic pesticides (value of “0” for variable pesticide). 5. Results
In addition to the pesticide variable, we also calculated effective
species richness per respondent. In our models, we used effective 5.1. Summary Statistics
species richness per farm (the crop diversity variable), to assess the
degree to which crop diversity interacts with pesticide use and interest Table 1 presents the summary statistics for all variables included in
in organic pest management techniques. We drew upon the respondent's both Model 1 and Model 2. With regard to acute pesticide poisoning
reported total cultivated land (land variable in Table 1), and each incidents, 22% of respondents reported that they or their family had
respondent's self-reported crops cultivated, which included crop type ever experienced an acute poisoning incident. It should be noted that
and area of land or number of plants associated with each crop. If cultivat- some respondents reported more than one acute pesticide poisoning
ed land and crop area were reported in number of plants, we translated incident for themselves or for their immediate family members. In
the number of plants into hectares with assistance from a member of total, 50 individual acute pesticide poisoning incidents were reported
Instituto Huayuná. The effective species richness, or eH′, is the exponenti- by the respondents from 1970 to 2013 (the year the survey was fielded).
ation of the Shannon Weiner Diversity index, H′ (Tilman et al., 2001). The Of these 50 incidents, 28 were reported to have occurred since the year
Shannon Weiner Diversity index was calculated by taking the proportion 2000, and 14 of those 28 were reported to have occurred between 2010
of a farmer's land occupied by each crop reported (Pi), multiplying each and 2013.
crop proportion by the natural log of that proportion (Pi ∗ ln(Pi)), and Most respondents, 82%, used conventional synthetic pesticides. Of
then summing this value for all crops listed by a respondent to attain a the 510 unique synthetic pesticides in the data set, the top pesticide
single diversity index value for each respondent (Keylock, 2005). The classes reported in use by respondents were carbamates (active ingredi-
effective species richness index allowed for a comparison between ent Methomyl) and organophosphates (active ingredients Dimethoate,
respondent's crop diversity based purely on frequencies, such that the Methamidophos and Chlorpyrifos). Respondents also reported the use
variable displayed an intuitive doubling property, i.e. twice the number of many other categories of pesticides, including triazoles, cyanides,
of species results in a crop diversity value twice as large (Jost, 2006). pyrethroids, and glyphosate, see Table A.1 in Appendix A for frequency
We analyzed the data using Stata and R statistical software, and we of pesticide categories reported and active ingredients. Only a small
calculated descriptive statistics for all variables (Table 1). We then ran portion of respondents, 12%, reportedly used targeted, or pest-specific,
multiple logistic regression models. In these models, we estimated the non-synthetic traps, as represented by the traps variable.
explanatory power of demographic, agricultural, and informational Sixty-four percent of respondents expressed an interest in using, or
variables on a farmer's use of synthetic pesticides (Model 1), and on a already used, organic agricultural techniques. With regard to pesticide
farmer's interest in organic agriculture (Model 2). For both Model 1 preventative practices, 72% of respondents stored pesticides and
and Model 2 we ran a single logistic regression simultaneously includ- empty pesticide containers in an appropriate location, 51% of respon-
ing all variables with a hypothesized relationship. Our hypotheses for dents wore protective equipment or clothing while applying pesticides,
the expected relationships in each model are shown in Table 2. For and 28% of respondents kept records of pesticide inputs and crop data.
those variables listed in Table 2 with a hypothesized positive relation- Finally, the majority of respondents, or 62%, used agro-chemical shops
ship in Model 1 (“M1” in Table 2), we expected that an increase in the or pesticide technicians as sources of information on pest management
value of the variable was associated with an increased likelihood that techniques, while only 7% and 11% of respondents cited Instituto
the respondent was a conventional pesticide user. Similarly, for those Huayuná and other farmers respectively, as sources of information on
variables listed in Table 2 with a hypothesized positive relationship pest management.
with Model 2 (“M2” in Table 2), we expected that an increase in the
value of the variable was associated with an increased likelihood that 5.2. Regression Analyses
the respondent was interested in organic pest management techniques.
We hypothesized the relationships for Models 1 and 2 using both Latin The results of Model 1, predicting conventional pesticide use are
American literature and global literature on small-scale agricultural pest shown in Table 3 and the correlations between variables in Model 1
 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81 77

Table 2
Expected relationships for Model 1 (M1) with dependent variable “pesticide” and Model 2 (M2) with dependent variable “organic.”

Variable Hypothesesa Explanation Related

economiesb
M1 M2

Omas valley + Mala Valley respondents are situated closer to a commercial center, Mala City, which may increase access to pesticide shops and NE
technicians as compared to Omas (Cowan and Gunby, 1996).
crop diversity − + Crop diversification can reduce vulnerability to climate and market variation, thus it may also allow farmers to take greater risks in LE
implementing new technologies (Cole et al., 2011; McCord et al., 2015).
education − + Increased education has been found to be associated with farmer awareness of the harms of pesticides (Hashemi et al., 2012).
age + − Younger farmers may be more likely to consider pesticides as harmful (Isin and Yildirim, 2007).
female gender − More equal household gender relations, in which women participate in farm business (as represented by female respondents) are
associated with reduced pesticide use (Nkamleu and Adesina, 2000; Cole et al., 2011).
technician + Agro- technicians are likely to promote pesticide use along with information about how to use the product (Wilson and Tisdell, NE and AE
2001; Feola and Binder, 2010a; Sherwood and Paredes, 2014).
land − + Larger farms may have more flexibility and capital to try new technologies which resembles a broader trend for farm conservation AE
behavior suggesting that larger farm operators may be more willing to invest in new technologies (Daloğlu et al., 2014)
poison + A farmer who has experienced poisoning knows first-hand the health risks of pesticide use (Hashemi et al., 2012).
traps − If a respondent uses non-synthetic pest traps the respondent is already using alternative techniques and learning through LE
experience (Cowan and Gunby, 1996).
apple + + Apples are intensive to grow in Mala and Omas and rely on heavily synthetic inputs, however, apple growers have a high degree of NE and LE
exposure to pesticides and potentially the negative impacts of using them.
tenure + − Farmers who do not own the land outright, through inheritance or purchase, are less likely to consider the health and long-term AE
effects of synthetic pesticide use on the land and soil (Nkamleu and Adesina, 2000).
registry + Keeping a registry of pesticide applications implies awareness for quantities and toxicities of pesticides applied (Cowan and Gunby, 1996).
equipment + Wearing pesticide protective equipment is generally a hindrance and is associated with greater understanding of the health risks of
pesticide spraying (Feola and Binder, 2010b).
storage + Storing pesticides in a safe location reflects an awareness for health risks of pesticides (Cowan and Gunby, 1996).
a
“+” represents an increasing likelihood, or an odds ratio greater than one, and “−” represents a decreased likelihood, or an odds ratio of less than one.
b
Under related economies, NE represents Network Externalities, LE represents Learning Economies, and AE represents Adaptive Expectations.

are shown in Table 4. In Table 3, an odds ratio greater than one is indic- more highly educated. The variable land was positively correlated
ative of an increased likelihood that the respondent was a conventional with Omas valley (r = 0.38) and crop diversity (r = 0.27), such that
pesticide user, whereas an odds ratio of less than one is indicative of a we would expect respondents with more land to be more likely to live
decreased likelihood that the respondent was a conventional pesticide in Omas and have more crop diversity on their farm. Finally, the variable
user. apple was negatively correlated with valley (r = −0.30) and positively
According to the results of Model 1 in Table 3, gender (male), having correlated with poison (r = 0.21), such that those respondents who
a technician as an information source, higher quantities of cultivated grew apples were more likely to live in the Mala valley and more likely
land, having apples as a primary crop, a rented state of land tenure, to have experienced pesticide poisoning in their family. The variance
and residing in Omas valley were indicative of conventional pesticide inflation factor for each variable in Model 1 is below 2, indicating that
use. Model 1 appears to have good model fit (chi-square = 52.13, multicollinearity was not an issue for Model 1.
df = 12, p b 0.001) and explanatory power (McFadden's R2 = 0.32). In Model 1, a female respondent was about three times less likely to
The relatively strong correlations between gender and pesticide use conventional pesticides than a male respondent (odds ratio of 0.30,
(r = − 0.18) and technician and pesticide (r = 0.36) shown in CI = (0.10, 0.88)). Additionally, respondents with more cultivated land
Table 4 reflected these findings. Additionally, the correlation matrix in had nearly a 30% increased likelihood of using pesticides per added
Table 4 shows a number of relationships worth noting in the data. The hectare (odds ratio of 1.27, CI = (0.96, 1.68)).
variable crop diversity was positively correlated with valley (r = Also the results of Model 1 suggest that respondents who relied on a
0.28), such that we would expect to see higher crop diversity in Omas pesticide technician or agro-chemical shop for pest management infor-
as compared to Mala. The variable age was negatively correlated with mation were over six times more likely to use conventional pesticides
education (r = − 0.51), such that younger respondents tended to be (odds ratio of 6.51, CI = (2.35, 17.98)) and respondents who grew
apples as a primary crop were two and a half times more likely to use
conventional pesticides (odds ratio of 2.66, CI = (0.94, 7.52)). The
variables technician and apple both show fairly high odds ratios, but
Table 3
wide confidence intervals. As the confidence intervals' lower bounds
Model 1 logistic regression predicting farmer use of synthetic pesticides.
for these variables are near one (apple) and well above one (techni-
Variable OR (95% CI) p cian), and they both have a relatively high upper bound, we interpreted
Omas valley† 2.71 (0.76–9.67) 0.13 these results to suggest a good likelihood that growing apple as a prima-
crop diversity 0.92 (0.57–1.49) 0.73 ry crop and relying on a pesticide technician had positive relationships
education 0.86 (0.27–2.77) 0.80
with the use of conventional pesticides. However, given the wide confi-
age 0.97 (0.94–1.01) 0.20
female gender† 0.30 (0.10–0.88) 0.03 dence intervals, we could only interpret the odds ratio associated with
technician† 6.51 (2.53–17.98) 0.00 these positive relationship as suggestive.
land† 1.27 (0.96–1.68) 0.09 Finally, Model 1's results mildly suggested that the valley in which a
poison 0.59 (0.18–1.98) 0.39 respondent lived and whether the respondent rented their land may be
traps 0.38 (0.11–1.35) 0.14
important in predicting pesticide use. If a respondent lived in Omas, as
apple† 2.66 (0.94–7.52) 0.07
tenure (rent)† 5.61 (0.59–53.20) 0.13 compared to Mala, the respondent was more likely to use conventional
tenure (shared) 1.99 (0.22–18.14) 0.54 pesticides (odds ratio of 2.71, CI = (0.76, 9.67)). Similarly, if a respon-
Variable significance: † = highly suggestive odds ratio and 95% confidence interval; n = dent rented land, as compared to owning land, the respondent was
188; Log Likelihood = −61.27 (df = 13); chi-square = 52.13 (df = 12), p b 0.001; more likely to use conventional pesticides (odds ratio of 5.61, CI =
McFadden's R2 = 0.32. (0.59, 53.20). Both of these variables demonstrated wide confidence
78 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81

Table 4
Correlation matrix for variables in Model 1.

Variable number Variable 1 2 3 4 5 6 7 8 9 10 11 12

1 Omas valley 1.00

2 crop diversity 0.28 1.00
3 education 0.01 0.13 1.00
4 age 0.02 −0.02 −0.51 1.00
5 female gender 0.09 −0.04 0.01 0.02 1.00
6 technician −0.09 −0.09 −0.08 −0.11 −0.10 1.00
7 land 0.38 0.27 0.03 −0.08 −0.07 0.05 1.00
8 poison −0.24 0.03 0.11 −0.06 −0.09 0.13 −0.02 1.00
9 traps −0.04 0.15 0.12 −0.04 0.07 −0.12 0.05 0.00 1.00
10 apple −0.30 −0.33 −0.10 0.02 −0.08 0.21 −0.05 0.21 0.12 1.00
11 land tenure 0.23 0.09 0.16 −0.34 −0.09 0.06 0.16 0.00 −0.10 −0.05 1.00
12 pesticide 0.12 −0.09 0.01 −0.18 −0.18 0.36 0.17 0.01 −0.14 0.17 0.19 1.00

intervals, with the lower and upper bounds spanning one, so we have 6. Discussion
interpreted these relationships as only mildly suggestive.
According to Model 1, crop diversity, education, age, history The results presented above suggest that conventional pesticide use
of pesticide poisoning, and use of non-synthetic pest traps did not in the Mala and Omas Valleys is a case of unsustainable technological
appear to be important predictors of conventional pesticide use in lock-in. Despite the availability of alternative techniques promoted by
the two valleys. Instituto Huayuná, we found pervasive use of highly toxic conventional
The results of Model 2, predicting interest in organic techniques pesticides in the two valleys and persistent acute pesticide poisoning in-
(organic), are shown in Table 5. Table 5 shows that the only variable cidents. Nearly a quarter of respondents had either themselves suffered
found to have power in predicting interest in organic practices amongst an acute pesticide poisoning incident, or had a close family member that
pesticide users was the variable equipment. Model 2 suggests that any suffered one. The 24 reported poisoning events since 2000 suggest that
individual that wore protective equipment while applying pesticides pesticide poisoning was still a common occurrence in these communities.
was about two times more likely to be interested in organic pest man- This number likely represents an underestimation of broader pesticide
agement than those that did not wear protective equipment (odds exposure in the sample because we did not test for chronic exposure
ratio of 2.02, CI = (0.95, 4.30)). As with Model 1, the variance inflation levels via blood testing (Jørs et al., 2006). In this section, we examine
factor for each variable in Model 2 was below 2, indicating that the dynamics of the unsustainable technological lock-in of conventional
multicollinearity was not an issue for Model 2. However, Model 2 ap- pest management strategies in the Mala and Omas valleys as suggested
pears to have low explanatory power (McFadden's R2 = 0.07) and by results of our models.
poor model fit (chi-square = 12.50, df = 10, p = 0.25). For this reason, Model 1, presented in Table 3 above, examined the predictors of
we interpreted the results of Model 2 as inconclusive. conventional pesticide use by farmers in the valleys and suggested
Both those respondents that had an interest in organic tech- factors associated with this technological lock-in. Using the framework
niques and those that already used them reported their perceived of competing technologies and technological lock-in, Model 1's results
difficulties in adopting the techniques. The responses fit into four suggest that economies of network coordination, learning, and adaptive
categories of perceived difficulties with organic techniques (percent expectations are contributing to the lock-in in the Mala and Omas
of respondents): communities (Cowan and Gunby, 1996, Kallis and Norgaard, 2010). As
described in Table 2, we proposed that many of the variables in Models
1. Lack of sufficient information and/or training on using organic method 1 and 2 reflect these different forms of increasing returns to adoption.
(67%) These variables include quantity of cultivated, consulting an agro-
2. Lack of sufficient physical resources (time, energy, funds) (13%) chemical technician, growing apple as a primary crop, living in Omas
3. Perception that organic practices wouldn't be able to address an issue valley, and renting land, which are five of the six variables suggested
that conventional methods can (e.g. a fungus) (11%) by Model 1 as predictors of conventional pesticide use. Additionally,
4. Lack of coordination among farmers – perception that conventional the barriers to transitioning to organic pest management, as described
methods of neighbors will be problematic for organic growers (9%). by respondents (e.g., lack of information and training, lack of physical
resources, perception that organic techniques will not work and lack
of farmer coordination), align with the barriers one would expect to
Table 5
see when attempting to transition away from a technology entrenched
Model 2 logistic regression predicting interest in organic techniques amongst pesticide through increasing returns. Although the results of Model 2 were incon-
users. clusive for predicting interest in organic agricultural techniques
amongst conventional pesticide users, it is important to note that the
Variable OR (95% CI) p
majority of pesticide users did express interest in switching to the alter-
registry 1.41 (0.63–3.16) 0.40
native technology (Pimentel et al., 2005). In the following paragraphs,
equipment† 2.02 (0.95–4.30) 0.07
storage 0.80 (0.33–1.93) 0.62 we discuss the role of network externalities, learning economies, and
land 1.06 (0.92–1.22) 0.44 adaptive expectations in the lock-in to synthetic pesticides in the Mala
age 0.99 (0.96–1.02) 0.39 and Omas Valleys and consider them each individually as barriers to
education 1.10 (0.49–2.47) 0.82 transitioning to organic pest management.
apple 1.24 (0.58–2.65) 0.58
The difficulty of network coordination amongst farmers and
crop diversity 1.24 (0.76–2.02) 0.40
tenure (rent) 1.52 (0.51–4.53) 0.45 neighbors to switch to organic, as noted by synthetic pesticide
tenure (shared) 2.36 (0.75–7.38) 0.14 users, may reflect the strong influence of agro-chemical technicians
Variable significance: † = highly suggestive odds ratio and 95% confidence interval, n = in the region. This is in line with Model 1's results which suggest
155; Log Likelihood = −96.23 (df = 11); chi-square = 12.50 (df = 10), p = 0.25; that respondents who consulted agro-chemical technicians were
McFadden's R2 = 0.07. more likely to use conventional pesticides. Mala City is host to ten
 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81 79

agro-chemical specialty shops, which is more than two times the number train new workers in the techniques required for organic practices.
of shops that were present in the city in the late 1990s. Respondents Both the transition to a more labor-intensive practice and the associated
consistently reported visits and community meetings with agro- learning curve may be a difficult barrier to overcome for apple farmers
chemical technicians. These technicians typically represent a specific interested in organic practices.
agro-chemical brand or line and therefore have strong incentives to We propose that economies of adaptive expectations can be seen in
encourage the use of synthetic pesticides and reinforce the agro- farmers work with agro-chemical technicians, quantity of land cultivated
chemical network. This reflects a similar finding in Ecuador where and variation in land tenure arrangements. A farmer's own past experi-
Sherwood and Paredes (2014) found evidence of agro-chemical vendors ence using synthetic pesticides reduces uncertainty in the current use of
promoting the use of highly toxic varieties of pesticides at higher applica- the technology. In addition, agro-chemical shops and synthetic techni-
tion rates than necessary in order to reach sales quotas. Additionally, cians offer information and services to further reduce uncertainty as
farmers who consistently used pesticides had strong incentives to contin- farmers may be facing a new pest. This may partially explain why in
ue asking and receiving information from agro-chemical technicians as Model 1 we saw farmers who consult an agro-chemical technician as
they continued to purchase pesticide products. In an attempt to break more likely to use conventional pesticides. While these sources of infor-
this cycle of dependence, Instituto Huayuná and two small organic mation may be biased and there may be superior methods for combating
cooperatives in the valleys are working to build an organic market infra- a particular pest in the long term, agro-chemical shops and technicians do
structure that connects to a biofería, or farmer's market, in the Lima serve to assure farmers that use of a pesticide will resolve the farmer's
Region. Currently, however, this network exists at a dramatically smaller issue of concern in the short term. This reduction in the uncertainty
scale than the agro-chemical network. This improvement in coordination with using pesticides creates a barrier to switching to an alternative
amongst organic growers may resolve some of the risk and uncertainty practice where a farmer may not understand exactly how the technology
that currently stands as a barrier to transitioning. is going to work. This is reflected by respondents' perception that organic
The high odds ratio associated with apples in Model 1 also suggests practices would not be able to address an issue that conventional
that network coordination amongst conventional pesticide users may pesticides could.
have been a large barrier for farmers who grew apples as a primary In Table 2 we hypothesized that quantity of land cultivated would be
crop. Parsa et al. (2014) found that with regard to adoption of IPM, negatively associated with conventional pesticide use, however, the
experts in developing countries cited lack of “collective action within a results of Model 1 showed a positive relationship, such that the more
farming community” as the most important obstacle to its adoption. land a respondent cultivates, the more likely they are to use conventional
Apples are a very popular crop in the region, and switching from a pesticides. In hypothesizing this relationship, we expected larger farms to
conventional apple product, with extensive market infrastructure, to have relatively greater flexibility in management techniques due to
an organic apple product with a less established market network, is greater production capacity and access to capital. We related this to a
very risky. The alternative network of the two organic cooperatives is broader trend in conservation agriculture, which may or may not be
striving to create a market for organic produce and avenues to reach appropriate for smallholder agriculture, that sees farm size as represen-
this market. The coupling of a local market for organic produce and tative of farm capacity in that you would expect larger farms to have
new consumer demand for organic produce could greatly reduce greater economies of scale and more capital to invest in adopting new
the risk of transitioning to organic. Developed countries, such as the practices (Prokopy et al., 2008). However, the results of Model 1 suggest
United States have seen dramatic increases in organic agricultural that the opposite may in fact be true in Mala and Omas. An explanation
production driven by consumer demand (Klonsky, 2000). This effort for this result may be the relative size of farms that we are considering
in Mala and Omas, if successful, will improve network coordination in this study (0.04 to 13 ha). Amongst the smallholder farmers in Mala
amongst organic users and may lower the barrier of transitioning from and Omas, it is likely that the larger farms represented market-oriented
conventional pesticides. small farm systems, while the smaller farms practiced more subsistence
With regard to network externalities it is also worth noting that in agriculture. In this case, adaptive expectations related to transitioning
Table 2 we hypothesized that because Omas is physically located further to organic pest management techniques may have been a greater barrier
away from the agro-chemical stores in Mala, respondents from Omas for small market-oriented farms as compared to subsistence-oriented
would be less connected to the Mala City pesticide network and therefore farms.
less likely to use conventional pesticides. Instead, Model 1's results mildly The presence of renters as a mildly suggestive factor in predicting
suggest that respondents in Omas were more likely to use pesticides than pesticide use in Model 1 may also be an example of technological uncer-
those in Mala. Due to the low level of respondents in Omas that did not tainty contributing to unsustainable technological lock-in. We are cau-
use conventional pesticides and the wide range in the confidence interval tious to interpret the odds ratio associated with renting land in Model 1
associated with the odds ratio for Omas valley in Model 1, we are cautious due to the wide confidence interval. However, one potential interpreta-
of interpreting network economies from this result. It is possible that due tion for the mildly suggestive relationship between renting land and
to the ease and frequency with which farmers in Omas accessed Mala city conventional pesticide use is that renters and the landowners they rent
and agricultural wholesalers from Mala visited Omas, physical distance from face split-incentives. Organic practices serve as a long-term invest-
from market centers was not an important factor in pesticide use. ment in the land that may increase the value of the land for the landown-
The learning economies associated with synthetic pesticide use are a er, but the cost of transitioning falls primarily on the renter who may not
barrier for farmers to transition to organic agriculture. As suggested by have guaranteed long-term access to the land. Additionally, farmers who
conventional pesticide users in the study, farmers in the valleys faced rent land in the valley, as opposed to farmers who bought or inherited
uncertainty in the time, energy and funds required for organic tech- land, may have less capital available to invest in alternative agricultural
niques. This barrier may also be reflected in the increased likelihood techniques that may not give an immediate return on investment.
for apple growers to use synthetic pesticides, as seen in Model 1. Therefore, it may be that farmers who primarily rent land may have a
Many farmers in the region used synthetic defoliants, herbicides, and more difficult time transitioning away from synthetic pesticides.
pesticides to reduce the amount of labor required for the intensive Gender was the final variable found important in Model 1 for
process of producing apples in the regional climate. Transitioning to predicting conventional pesticide use. As predicted in Table 2, female
more labor-intensive practices will require investments of time and respondents were less likely to use pesticides than males. This finding
energy and may require additional on-farm workers. It may be difficult supports Orozco et al.'s (2011) finding that amongst Andean Ecuadorian
for farmers to recruit additional workers to the farm because of the small-scale farmers, women's participation in farm decision making and
high opportunity costs associated with on-farm labor (Beckmann more equality in household roles is associated with less use of toxic pes-
and Wesseler, 2003). Furthermore, it may be difficult to find and ticides. We must note that in our study we were not able to differentiate
80 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81

between female farmers that may have been widowed or single, and renters and pesticide-intensive crop growers (such as apple growers) to
female farmers that participated in the household's farming with their introduce programs to reduce the risks of learning, reduce the initial
spouse. The relationship between gender in household equality and financial investment and build infrastructure around organic agriculture
technological lock-in of pesticides would be a fruitful area for future and other sustainable pest management techniques.
research, as it appears to be an important factor in predicting the use
of pesticides. Acknowledgements

7. Study Limitations and Future Research This research was made possible through funding from the Dartmouth
Center for Health Care Delivery Science. We would like to thank the
In this paper we have shown that pesticide use and poisoning are farmers who participated in our survey and the community health
continuing in the Mala and Omas Valleys and that agro-chemical techni- workers and farmers who assisted in the implementation of the survey.
cians, quantity of cultivated land, gender (male), growing apples, and We would also like to express our gratitude to Instituto Huayuná, Dr.
renting land, are important predictors of conventional pesticide use Robert Gougelet, Jennifer Murray and Kate Bradshaw for their support
that support the technological lock-in of conventional pesticides. More and guidance during our fieldwork, and to Keri Bryan Watson for her as-
broadly, our research demonstrates the continuing reliance on highly sistance in creating Fig. 1. Finally, we would like to thank Phoebe Spencer,
toxic synthetic pesticides in smallholder agriculture in Latin America, Michael Wironen and three anonymous reviewers for their thoughtful
and contributes a new case study context to the building literature comments on a previous version of this manuscript.
describing this phenomenon.
We acknowledge a number of limitations to our study including the Appendix A
unrepresentative sample, use of self-report survey methodology, and po-
tential for omitted variables in our regression analyses. Because we can- Table A.1
not claim a representative sample of the Mala and Omas Valleys, there Pesticide categories and active ingredients of unique pesticides recorded.
may be limited generalizability to our results. Additionally, our data is po- Pesticide categorya Unique instances WHO Number of Banned
tentially skewed by participant recall bias due to the use of self-reported and active ingredient recorded in classificationc countries that in
methods for gathering health and farm management data. Finally, we rec- surveyb have bannedd Perúd
ognize that it is possible we overlooked other important variables, such as Carbamate 99
income, in our model selection for Model 1 and Model 2. Future research Carbofuran 2 Ib 46 No
can build upon our study to examine the degree to which the variables Methomyl 96 Ib 11 No
identified as important predictors of pesticide use in the Mala and Omas Carbaryl 1 II 32 No
Organophosphate 99
valleys explain pesticide use and lock-in in other contexts. Dicrotophos 1 Ib 33 No
We also acknowledge that the binary presentation of pest manage- Dimethoate 63 II 4 No
ment techniques as either synthetic-based or organic in the pesticide Azinphos-methyl 1 Ib 36 No
variable is an oversimplification of a broad spectrum of pest control Methamidophos 19 Ib 47 No
Chlorpyrifos 13 II 1 No
strategies that includes varying degrees of IPM. For future research, an
Methyl parathion 2 Ia 26 Yes
exploration of this spectrum of ecologically-based pest management Triazole 84
techniques (including degrees of IPM and organic) would help to deter- Tebuconazole 4 II NA No
mine whether organic pest control is simply one end of a spectrum that Propiconazole 33 II NA No
includes IPM or categorically different from the farmer perspective (due Penconazol 33 III NA No
Bitertanol 1 U 29 No
to certification requirements and price premiums). Triadimenol 12 II NA No
Finally, we would like to note that the scope of our study was explicitly Triadimefon 1 II NA No
a local level analysis of farming in the Mala and Omas Valleys, and Cyanide 79
therefore we acknowledge that we do not address the interactions and Hydrogen cyanide 79 FM 1 No
Pyrethroid 41
influences of regional, national, and international institutions on the ob-
Permethrin 1 II 29 No
served farming behavior and results (Agrawal, 2001). Future research Cypermethrin 32 II NA No
can build upon the local level analysis completed in this study to investi- Alpha-cypermethrin 4 II NA No
gate the interface between Peruvian market dynamics, government Deltamethrin 2 II NA No
regulation and the unsustainable use of synthetic pest management in Chitin synthesis 28
inhibitor
the Mala and Omas Valleys. This broader perspective could shed light Buprofezin 3 III NA No
on market and government interventions that could assist farmers in Lufenuron 25 NA No
transitioning to more sustainable pest control technologies and reduce Glyphosate 18 III 1
negative human and environmental externalities at the farm level. Petroleum-based oil 12
Dithiocarbamate 7
Propineb 2 U NA No
8. Conclusions Mancozeb 5 U 1 No
a
510 individual conventional pesticide products were reported in use in the survey. The
This paper demonstrates the utility of applying technological adoption 467 pesticide counts included in this table represent all categories with more than 5 unique
theory to questions of local level sustainable agriculture and important mentions by respondents. In addition, respondents mentioned 29 individual pesticides from
public health problems in order to identify barriers to transition and 15 other categories, and 14 individual pesticides recorded as “herbicide”, “insecticide”, etc.
b
opportunities for action. Local government, farmers and NGOs in the Because some individual pesticides were reported at the level of the pesticide category,
not all reported active ingredient sum to the total number of individual pesticides reported
Mala and Omas Valleys and across Latin America could strategically ad-
per pesticide category.
dress the technological barriers associated with uncertainty, coordination In the table, pesticide categories are listed in normal font. The active ingredients belong-
and learning to support a transition to more sustainable pest manage- ing to a pesticide category are listed in the rows below each pesticide category in italics.
c
ment strategies. As mentioned previously, Instituto Huayuná has been World Health Organization's recommended toxicity levels (Chemicals and W. H.
working to grow a network of organic farmers, host trainings for farmers, Organization, 2010): Ia — extremely hazardous, Ib — highly hazardous, II — moderately
hazardous, III — slightly hazardous, FM — fumigant not classified, and U — unlikely to present
and create new market opportunities for farmers in the valleys. Further acute hazard in normal use.
interventions can focus on decreasing the influence of agro-chemical d
According to the Pesticide Action Network's Consolidated List of Pesticide Bans 2015
technicians on farmer pest management decisions, and on targeting (Pesticide Action Network, 2015).
 C. Hammond Wagner et al. / Ecological Economics 129 (2016) 72–81 81

Appendix B. Data Collection Instrument Jørs, E., Lander, F., Huici, O., Morant, R.C., Gulis, G., Konradsen, F., 2014. Do Bolivian small
holder farmers improve and retain knowledge to reduce occupational pesticide
poisonings after training on integrated pest management? Environ. Health 13 (1), 75.
Supplementary data to this article can be found online at http://dx. Jost, L., 2006. Entropy and diversity. Oikos 113 (2), 363–375.
doi.org/10.1016/j.ecolecon.2016.05.013. Kallis, G., Norgaard, R.B., 2010. Coevolutionary ecological economics. Ecol. Econ. 69 (4),
690–699.
Kay, C., 1982. Achievements and contradictions of the Peruvian Agrarian Reform. J. Dev.
Stud. 18 (2), 141–170.
References Keylock, C.J., 2005. Simpson diversity and the Shannon-Wiener index as special cases of a
generalized entropy. Oikos 109 (1), 203–207.
Agrawal, A., 2001. Common property institutions and sustainable governance of resources. Klonsky, K., 2000. Forces impacting the production of organic foods. Agric. Hum. Values
World Dev. 29 (10), 1649–1672. 17 (3), 233–243.
Alba, J., 2004. Diálogo de experiencias: atriculando procesos para el desarollo. Los Olivos Kogan, M., 1998. Integrated pest management: historical perspectives and contemporary
141. Instituto de Promocion y Desarrollo, La Molina Vieja, Huayuna. developments. Annu. Rev. Entomol. 43 (1), 243–270.
Angelo, M.J., 2013. The Law and Ecology of Pesticides and Pest Management. Ashgate. Lee, W.J., Cha, E.S., Moon, E.K., 2010. Disease prevalence and mortality among agricultural
Arthur, W.B., 1989. Competing technologies, increasing returns, and lock-in by historical workers in Korea. J. Korean Med. Sci. 25 (Suppl), S112–S118.
events. Econ. J. 99 (394), 116–131. Lesmes-Fabian, C., García-Santos, G., Leuenberger, F., Nuyttens, D., Binder, C.R., 2012. Dermal
Bazo Robles, J.L., Mallqui, E.F., Quiroz, E.K.P., 2010. Reducción de los contaminantes exposure assessment of pesticide use: the case of sprayers in potato farms in the
orgánicos persistentes (COPs), en El Valle de Mala. Institutó Huayuná, Av. San Luis Colombian highlands. Sci. Total Environ. 430, 202–208.
2134, San Borja, Lima 41, Perú. Liebowitz, S.J., Margolis, S.E., 1995. Path dependence, lock-in, and history. J. Law Econ. Org.
Beckmann, V., Wesseler, J., 2003. How labour organization may affect technology adoption: 11, 205.
an analytical framework analysing the case of integrated pest management. Environ. McCord, P.F., Cox, M., Schmitt-Harsh, M., Evans, T., 2015. Crop diversification as a
Dev. Econ. 8, 437–450. smallholder livelihood strategy within semi-arid agricultural systems near Mount
Bhat, A.R., Wani, M.A., Kirmani, A.R., Raina, T.H., 2010. Pesticides and brain cancer linked Kenya. Land Use Policy 42, 738–750.
in orchard farmers of Kashmir. Indian J. Med. Paediatr. Oncol. 31 (4), 110–120. Miranda-Contreras, L., Gómez-Pérez, R., Rojas, G., Cruz, I., Berrueta, L., Salmen, S.,
Bjorling-Poulsen, M., Andersen, H.R., Grandjean, P., 2008. Potential developmental neurotox- Colmenares, M., Barreto, S., Balza, A., Zavala, L., 2013. Occupational exposure to organo-
icity of pesticides used in Europe. Environ. Health 7, 50. phosphate and carbamate pesticides affects sperm chromatin integrity and reproductive
Brundtland, G., Khalid, M., Agnelli, S., Al-Athel, S., Chidzero, B., Fadika, L., Hauff, V., Lang, I., hormone levels among Venezuelan farm workers. J. Occup. Health 55 (3), 195–203.
Shijun, M., de Botero, M.M., 1987. Our common future (\‘Brundtland report\’). Muñoz-Quezada, M.T., Lucero, B.A., Barr, D.B., Steenland, K., Levy, K., Ryan, P.B., Iglesias, V.,
Carrión Yaguana, V., Alwang, J., Norton, G., Barrera, V., 2015. Does IPM have staying power? Alvarado, S., Concha, C., Rojas, E., 2013. Neurodevelopmental effects in children associat-
Revisiting a potato-producing area years after formal training ended. J. Agric. Econ. ed with exposure to organophosphate pesticides: a systematic review. Neurotoxicology
Chemicals, I.-O.P.f.t.S.M.o., W. H. Organization, 2010. WHO Recommended Classification of 39, 158–168.
Pesticides by Hazard and Guidelines to Classification 2009. World Health Organization. Nkamleu, G.B., Adesina, A.A., 2000. Determinants of chemical input use in peri-urban
Cole, D.C., Orozco, T.F., Pradel, W., Suquillo, J., Mera, X., Chacon, A., Prain, G., Wanigaratne, lowland systems: bivariate probit analysis in Cameroon. Agric. Syst. 63, 111–121.
S., Leah, J., 2011. An agriculture and health inter-sectorial research process to reduce Orozco, F.A., Cole, D.C., Ibrahim, S., Wanigaratne, S., 2011. Health promotion outcomes
hazardous pesticide health impacts among smallholder farmers in the Andes. BMC associated with a community-based program to reduce pesticide-related risks
Int. Health Hum. Rights 11 (Suppl. 2), S6. among small farm households. Health Promot. Int. 26 (4), 432–446.
Cowan, R., Gunby, P., 1996. Sprayed to death: path dependence, lock-in and pest control Parrón, T., Requena, M., Hernández, A.F., Alarcón, R., 2014. Environmental exposure to
strategies. Econ. J. 106 (436), 521–542. pesticides and cancer risk in multiple human organ systems. Toxicol. Lett. 230 (2),
Crissman, C.C., Antle, J.M., Capalbo, S.M., 1998. Economic, Environmental, and Health 157–165.
Tradeoffs in Agriculture: Pesticides and the Sustainability of Andean Potato Production. Parsa, S., Morse, S., Bonifacio, A., Chancellor, T.C., Condori, B., Crespo-Pérez, V., Hobbs, S.L.,
International Potato Center. Kroschel, J., Ba, M.N., Rebaudo, F., 2014. Obstacles to integrated pest management
Daloğlu, I., Nassauer, J.I., Riolo, R.L., Scavia, D., 2014. Development of a farmer typology of adoption in developing countries. Proc. Natl. Acad. Sci. 111 (10), 3889–3894.
agricultural conservation behavior in the American Corn Belt. Agric. Syst. 129, 93–102. Perkins, R., 2003. Technological lock-in. Online Encyclopaedia of Ecological Economics.
Dasgupta, S., Meisner, C., Wheeler, D., Xuyen, K., Thi Lam, N., 2007. Pesticide poisoning of Perú Instituto Nacional de Estadística e Informatica, 2012. Población total at 30 de junio,
farm workers-implications of blood test results from Vietnam. Int. J. Hyg. Environ. por grupos quinquenales de edad, según departamento, provincia y distrito. http://
Health 210 (2), 121–132. www.inei.gob.pe/estadisticas/indice-tematico/poblacion-y-vivienda/.
Ecobichon, D.J., 2001. Pesticide use in developing countries. Toxicology 160 (1), 27–33. Pesticide Action Network, 2015. Pesticide database.
Eskenazi, B., Bradman, A., Castorina, R., 1999. Exposures of children to organophosphate Pimentel, D., Hepperly, P., Hanson, J., Douds, D., Seidel, R., 2005. Environmental, energetic,
pesticides and their potential adverse health effects. Environ. Health Perspect. 107 and economic comparisons of organic and conventional farming systems. Bioscience
(Supplement 3), 409–419. 55 (7), 573–582.
Eskenazi, B., Marks, A.R., Bradman, A., Harley, K., Barr, D.B., Johnson, C., Morga, N., Jewell, N.P., Prokopy, L.S., Floress, K., Klotthor-Weinkauf, D., Baumgart-Getz, A., 2008. Determinants of
2007. Organophosphate pesticide exposure and neurodevelopment in young Mexican- agricultural best management practice adoption: evidence from the literature. J. Soil
American children. Environ. Health Perspect. 115 (5), 792–798. Water Conserv. 63 (5), 300–311.
Feola, G., Binder, C.R., 2010a. Identifying and investigating pesticide application types to Servicio Nacional de Sandidad Agraria, 2014. Resolución Directoral N 0005-2014-
promote a more sustainable pesticide use. The case of smallholders in Boyacá, MINAGRI-SENASA-DIAIA. http://www.senasa.gob.pe/senasa/normas-nacionales-
Colombia. Crop. Prot. 29 (6), 612–622. sobre-plaguicidas/.
Feola, G., Binder, C.R., 2010b. Why don't pesticide applicators protect themselves? Exploring Sherwood, S.G., Paredes, M., 2014. Dynamics of perpetuation: the politics of keeping highly
the use of personal protective equipment among Colombian smallholders. Int. J. Occup. toxic pesticides on the market in Ecuador. Nat. Cult. 9 (1), 21–44.
Environ. Health 16 (1), 1–23. Sherwood, S., Cole, D., Crissman, C., Paredes, M., 2005. From pesticides to people: improving
Galt, R.E., 2008. Beyond the circle of poison: significant shifts in the global pesticide complex, ecosystem health in the northern Andes. The Pesticide Detox: Towards a More Sustain-
1976–2008. Glob. Environ. Chang. 18, 786–799. able Agriculture, pp. 147–164.
Galt, R.E., 2013. From homo economicus to complex subjectivities: reconceptualizing The Codex Alimentarius Comission and the FAO/WHO Food Standards Programme,
farmers as pesticide users. Antipode 45 (2), 336–356. 2007d. Organically Produced Foods. third ed. World Health Organization and the
Hashemi, S.M., Hosseini, S.M., Hashemi, M.K., 2012. Farmers' perceptions of safe use of Food and Agriculture Organization of the United Nations, Rome.
pesticides: determinants and training needs. Int. Arch. Occup. Environ. Health 85 (1), Tilman, D., Reich, P.B., Knops, J., Wedin, D., Mielke, T., Lehman, C., 2001. Diversity and
57–66. productivity in a long-term grassland experiment. Science 294 (5543), 843–845.
Herrera, J., 2010. Primera experiencia a nivel mundial del manejo integrado de plagas: el Valdivieso Jara, L., 2011. Historia del Control Biologico en el Perú.
caso del algondonero en el Perú. Rev. Peru. Entomol. 46 (1), 1–8. Waterfield, G., Zilberman, D., 2012. Pest management in food systems: an economic
Isin, S., Yildirim, I., 2007. Fruit-growers' perceptions on the harmful effects of pesticides and perspective. Annu. Rev. Environ. Resour. 37, 223–245.
their reflection on practices: the case of Kemalpasa, Turkey. Crop. Prot. 26 (7), 917–922. Wilson, C., Tisdell, C., 2001. Why farmers continue to use pesticides despite environmental,
Jørs, E., Morant, R.C., Aguilar, G.C., Huici, O., Lander, F., Bælum, J., Konradsen, F., 2006. health and sustainability costs. Ecol. Econ. 39, 449–462.
Occupational pesticide intoxications among farmers in Bolivia: a cross-sectional study. World Health Organization, 1990. Public Health Impact of Pesticides Used in Agriculture.
Environ. Health 5, 10.