Questionnaire design and data collection

The required information was obtained through a questionnaire applied to technicians or owners of UPs under protected agriculture in the state of Zacatecas. For the design of the questionnaire, the one applied in 2015 by ^{Padilla-Bernal et al. (2018)}. The questionnaire was applied during the months of May to December 2016 at meetings of the Cluster of Protected Agriculture AC and visits to the UPs.

86 questionnaires were applied, only 81 were completely answered and not duplicated, proving useful for the study. The selection criteria of the units of analysis to be surveyed were an area under protected agriculture ≥1 ha, having reported activity in the 2015 agricultural year and the availability of the technician or owner to answer the questions. The number of UPs and their location was obtained from the registers of producers registered in the Cluster of Protected Agriculture, AC, the Tomato Product System and some others identified, through the Chile Product System or in producer meetings.

The UPs surveyed cover an area of 472.6 ha under protected agriculture and 2 665 ha in the open field (most UPs with protected agriculture also grow products in the open field), distributed in 19 municipalities of the state (Calera, Cañitas de Felipe Pescador, Fresnillo, General Enrique Estrada, Gral. Panfilo Natera, Guadalupe, Jeréz, Loreto, Morelos, Ojocaliente, Panuco, Pinos, Tepetongo, Trancoso, Vetagrande, Villa de Cos, Villa Hidalgo, Villanueva and Zacatecas).

The area with production systems under protected agriculture where the studied UPs are located represents 99.5% of the state area cultivated under this modality (^{SIAP-SAGARPA, 2017}). The structure of most of the UPs is scrape and kneed, has passive climate control and cultivated in soil. Only 34 have any certification, of which 15 export their products (Table 1). The maximum degree of studies of 65% of respondents is undergraduate or postgraduate and 47% are older than 50 years.

Data processing

The determination of opinions on sustainability was made through 12 questions, adapting what was proposed by ^{Rankin et al. (2011)}; ^{Hauschildt and Schulze-Ehlersb (2014)} and what is used by ^{Padilla-Bernal et al. (2018)}. The questions were presented on a scale from 1 to 5 where 1= disagree and 5= totally agree. To reduce opinions on sustainability, to have greater clarity in data management, the information was subjected to a principal component analysis (ACP) with Varimax rotation.

Data was processed using SPSS v23. To include an indicator in a factor, a factor loads greater than or equal to 0.5 was considered and to determine the internal consistency and reliability of the factors, an Alpha Cronbach coefficient greater than or equal to 0.70 was considered. This coefficient can take values from 0 to 1, ^{Tavakol and Dennick (2011)} state that a value greater than 0.7 is considered acceptable. In order to group the respondents based on their opinions on sustainability, a hierarchical cluster analysis (ACJ) was made, applying the Ward method and using the agglomeration table and the dendrogram (^{Czillingová et al., 2012}). The profile of the groups was determined using analysis of variance (Anova).

The environmental performance of the UP was obtained taking as a reference what was proposed by ^{Carruthers (2005)}. Unlike ^{Padilla-Bernal et al. (2018)} where questions were asked about the perception of the actions carried out, the questions were asked taking into account agri-environmental practices, as well as the existing infrastructure in the UP. The variables evaluated were water, soil, biodiversity, agrochemicals, pollution, waste management and environmental management of the business.

Each variable had four questions (indicators), with answer options from 1 to 5, where 1 was non-application of agricultural practice or application in a more rudimentary or less environmentally friendly way or where appropriate the non-availability of infrastructure and 5 application of best agricultural practices in a sustainable way or availability of infrastructure aimed at caring for and protecting the environment. The index per variable represents the relationship between the score of the studied variable with respect to the maximum possible. To make all the scores comparable, the maximum score was considered as ten. The environmental performance index (IDA) was obtained as an average of the variables considered.

For the determination of the motivators and barriers for the adoption of an agri-environmental program, the proposal by ^{Padilla-Bernal et al. (2018)}, seven questions were asked in each case, with options from 1 to 5 where 1= not important and 5= totally important. The value of the motivators and barriers were determined as an average per group of indicators.

Opinion on sustainability and its management

The opinions on sustainability, the mean, standard deviation and the factor loads of the ACP are presented in Table 2. Three factors were obtained with eigenvalues greater than the unit that explain 65.52% of the total variance. These were called ‘environmental and regulatory’, ‘organization-oriented and market’ and ‘driven by profitability’. Unlike that obtained by ^{Hauschildt and Schulze-Ehlersb (2014)}; ^{Padilla-Bernal et al. (2018)}, only two factors were statistically significant. The third factor ‘driven by profitability’ was excluded in subsequent analyzes when a strategy to save costs was explained by the indicator alone. The indicator an opportunity to improve the income of the UP by obtaining a factor load greater than 0.5 in two factors is not considered within the third factor.

The opinion on sustainable agriculture is one of the most important factors that contribute to its adoption (^{SAI Platform, 2015}). Pérez-Espejo et al. (2011) point out that the perception of sustainability and the potential implementation of agri-food policies, such as voluntary programs, is essential for the management of environmental problems in developing countries.

Agricultural practices oriented to the care of natural resources and environmental management

Regarding the agricultural practices oriented to the care and protection of the environment, the IDA was calculated by type of market where the UPs market their products. Based on the single-factor Anova, it was found that the UPs that sell their products in national and international markets report higher average IDAs than those in the local, national and local and national markets (Table 3), reporting statistically significant differences between these three groups and the first.

Having found homogeneity of variances (p-value= 0.197) the Tukey procedure was applied (p-value < 0.001). The results suggest that marketing the product in the international market makes a difference in the agricultural practices of environmental care applied in the UPs. However, it should be noted that the average IDA of the UPs studied is 6.3, a value 37.0% lower than the reference index (IDA= 10). The results show that there is still worked to be done to achieve more sustainable UPs.

The highest IDAs per variable were obtained in agrochemicals (8.65) and pollution (7.43). In the agrochemical variable the indicators were considered: storage of agrochemicals in an adequate place; use of doses and proper application of agrochemicals to crops; pest monitoring to decide the application of agrochemicals and use only of agrochemicals authorized by regulatory bodies. In the case of the pollution variable, the indicators were: avoid burning straw and stubble, as well as other materials; use of drinking water for hygiene of equipment and workers, use of iron for preparing solutions for mixing agrochemicals and use of a special area for storage of clothing and protective equipment different from the pesticide store.

The values obtained in the agrochemical and pollution variables are attributed to the wide dissemination of government programs that promote certification in the good use and management of agrochemicals (BUMA) and the reduction of pollution in their application -risk reduction system Primary Vegetable Production Pollution (SRR)- (^{SENASICA, 2016}).

The variables that recorded the lowest values in the IDA were those involving activities with environmental management of the business (3.88) and waste management (4.55). The indicators considered in the business management variable are: application of a training program in environmental aspects for workers, preparation of environmental contingency plans, design of a program to reduce solid and green waste and use of alternative energy for agricultural production or in postharvest management.

The aspects evaluated in the variable waste management were: triple washing and perforation of empty pesticide containers; it send empty agrochemical containers to a confinement center, send plastic waste, pipes, belt to a collection center for recycling, plant waste deposit in a special area for fertilizer production. The values obtained in these variables denote a 50% lower performance than the maximum value of the IDA, suggesting the need for more information and training in the UPs on the mentioned aspects. Most respondents replied that they do not carry out the activity related to the indicator or do not have the spaces required for the development of the activity.

Strategies oriented to the adoption of a comprehensive program of care and protection of the environment (PICPA): motivators and barriers

The most important motivations for the adoption of a PICPA were: to facilitate access to national and international markets (4.5) and avoid damage to workers (4.5) (Table 4) responses similar to those found by ^{Padilla-Bernal et al. (2018)} and consistent with what was stated by ^{Carruthers and Vanclay (2012)} and ^{SAI Platform (2015)}, who consider the market as one of the main factors for the adoption of good agricultural practices and innovation.

The barriers to the adoption of a PICPA, these were determined according to what was proposed by ^{Carruthers (2005)}; ^{Carruthers and Vanclay (2012)}; ^{Merli et al. (2016)}. Respondents on average indicated that one of the most important barriers is not having government support (4.2), followed by additional expenses in monitoring, training and investment in infrastructure and equipment (4) (Table 5). Result similar to that reported by ^{Padilla-Bernal et al. (2018)}, who point out that the acceptance of an agri-environmental program increases as the costs and technical efforts decrease and the benefits are clearly perceived.

Cluster analysis n sustainability opinions

Cluster analysis was performed based on the two statistically significant components obtained from the ACP applied to sustainability opinions. Three groups consisting of 44 (54.3%), 21 (25.9%) and 16 (19.8%) UPs respectively were obtained. The groups are described based on the one-way ANOVA (Table 6). Applying the Levene test, variance homogeneity was found for most of the variables. In these cases, the Tukey procedure was used to identify significant differences between the groups, and in the rest of the cases the Games-Howell was applied (^{Morgan et al., 2011}).

The three groups presented statistically significant differences in the ‘environmentalist and normative’ opinions of sustainability, showing higher average values in group 1 (4.66), followed by group 3 (4.43) and then group 2 (3.53). However, in the opinions on sustainability ‘oriented to the organization and market’, groups 2 and 3 do not show statistically significant differences. On the other hand, the average value of group 1 (4.49) was statistically different from groups 2 and 3. These results suggest the denomination of the three groups as follows: group 1 ‘rational environmentalist’, group 2 ‘environmentalist by rule’ and group 3 ‘potential environmentalist’.

Regarding the IDA, the results of the cluster analysis show statistically significant differences between the agricultural practices applied in groups 1 and 2, while the average value of the IDA of group 3 (6.04) does not show differences with the average values of the groups. 1 and 2 (6.68 and 5.71 respectively). As regards the motivators and barriers to the adoption of a PICPA, the average barrier indices shown in the three groups do not show statistically significant differences. On the other hand, the average index of motivators has statistically significant differences between groups 1 and 2, the opposite being the case between groups 2 and 3. The results of the cluster analysis show that the ‘rational environmentalist’ group presents average values higher than the other two groups in the indexes analyzed, except for the case of the barriers index that was not statistically significant.

Regarding the willingness to adopt a PICPA in the UPs, it was found that 85.2% of the respondents (69 UPs) expressed their acceptance of the adoption of the program, 12.3% said they did not accept (10 UPs), while 2.5% do not know (2 UPs). These values showed statistically significant differences (χ² test). The reasons that the respondents argued for the acceptance of the program are improve the sustainability of the UP, avoid damage to the environment and have better access to national and international markets. Those who do not accept it or doubt its acceptance, the reasons they presented were mainly: negligence of the owner, lack of resources for activities that help take care of the environment, distrust of government programs and lack of information on the subject.

Unlike the groups ‘environmentalist by standard’ and ‘potential environmentalist’, the group ‘rational environmentalist’ presents indicators that show better environmental performance and characteristics that guide the UPs towards the adoption of a PICPA (Table 7). 50% of the group ‘rational environmentalist’ has scratch-protected and protected environment structures, in this group there are 3 (60%) of the 5 UPs analyzed with active climate and 80% of those that grow in hydroponics. 65.9% of this group has some certification or is in the process of obtaining it. Most of the UPs export their products and 61.4% of respondents have higher education.