Public Perception of Biodiversity Landscape Elements and Autonomous Technologies in Small-Scale Production Systems

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/384687258 Public Perception of Biodiversity Landscape Elements and Autonomous Technologies in Small-Scale Production Systems Conference Paper · September 2024 CITATIONS 0 READS 33 3 authors, including: Andreas Gabriel Bayerische Landesanstalt für Landwirtschaft 56 PUBLICATIONS465 CITATIONS SEE PROFILE All content following this page was uploaded by Andreas Gabriel on 07 October 2024. The user has requested enhancement of the downloaded file. 28 29th September 2024, Reading, United Kingdom Harper Adams Business School Harper Adams University University of Reading https://www.harper-adams.ac.uk/research/giate/ Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 12 Public Perception of Biodiversity Landscape Elements and Autonomous Technologies in Small-Scale Production Systems Andreas Gabriel, Johanna Garnitz and Olivia Spykman Bavarian State Research Center for Agriculture, Germany Abstract The perception and evaluation of rural landscapes resulting from human interaction with nature is highly subjective. However, understanding how the non-agricultural population views the impact of an altered landscape image is crucial. This paper explores the German population's perceptions of changes in agricultural landscapes brought about by multi-crop, small-scale field structures (strip intercropping) combined with the introduction of biodiversity landscape elements and field robotics. An online survey was conducted with German residents aged 18 and older (n = 2,022). Preferences and the importance of individual image components were analysed based on four images depicting a field with strip intercropping, featuring various combinations of tractors, robots, and flowering strips. findings reveal that nearly two-thirds of respondents preferred the image featuring a flower strip and a tractor, associating it with concepts such as green, nature, and environment (flowering strip), as well as the traditional image of agriculture (tractor). Among the two images without flower strips, the tractor was preferred over the robot by more than a sixfold margin. Conversely, the image with a robot and flower strips was chosen about as frequently as the image with a tractor but without flower strips. Additionally, the study highlights how socio-demographic characteristics may influence the evaluation of agricultural landscape changes. Two logistic regression models indicate that factors such as age, gender, direct impact preferences of specific landscape components. Overall, the results suggest a preference for landscapes that are both familiar and environmentally oriented. Nevertheless, the use of autonomous technologies and the shift towards small-scale diversified production systems are not broadly rejected. Keywords Autonomous farming technologies; biodiversity; public acceptance; rural landscape; strip intercropping. Presenter Profile Andreas Gabriel is a member of the 'Digital Farming' working group at the Bavarian State Research Center for Agriculture. With extensive experience in empirical social research, his work focuses on investigating the social acceptance and adoption of digital technologies in agricultural practice. * Corresponding Author: Andreas Gabriel, Bavarian State Research Center for Agriculture, Institute of Agricultural Engineering and Animal Husbandry, 94099 Ruhstorf, Germany; email: andre[email protected].de Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 13 Introduction The visual perception of a rural landscape ("landscape image") is an important factor for acceptance of individual features in agricultural structures among both agricultural stakeholders and the general public. This perception is influenced, e.g. by associated farming processes and environmental effects and is strongly shaped by the subjective perspective of the individual (cf. Roth et al., 2011). Therefore, interactions such as the introduction of new production systems and structural elements (e.g., agroforestry systems, flower strips, etc.) or the use of new technologies (e.g., field robots) to promote ecological sustainability must also be discussed and evaluated in terms of its impact on the landscape. In light of current efforts to promote biodiversity-enhancing production systems (FAO, 2023; Ruggeri Laderchi et al., 2024), small-scale diversified crop production systems such as strip intercropping are gaining importance (cf. Alarcón-Segura et al., 2022; Spykman et al., 2023). Strip intercropping refers to the simultaneous cultivation of different crops on the same field in parallel strips (Vandermeer, 1989). If established on a larger scale, this production system has far-reaching impacts on the landscape image compared to conventional farming. It is assumed that the management of such small-scale diversified production systems can be made labour-efficient through automation (e.g., automatic steering systems and section control), or by using autonomous technologies such as field robots or drones (cf. LowenbergDeBoer, 2021; Gackstetter et al., 2023). Particularly, the introduction of autonomous technologies would further change both the aesthetic appearance of the landscape and agricultural practices. Previous research has demonstrated the influence of user experience and knowledge about a Dentzmann and Goldberger (2020) examined images of a biodegradable alternative to conventional polyethylene mulching foil in focus group discussions with farmers. It was found that the evaluation of this alternative was strongly dependent on the experiences of the respondents, with functional knowledge influencing the visual assessment (Dentzmann and Goldberger, 2020). The visual assessment of the landscape image within the professional group is thus also based on knowledge about farming methods, their feasibility, and economic prospects. However, it is not easy to determine how groups that are not familiar with the operational functions of landscape-shaping farming measures will react to changes in the landscape. Positive ecological effects often occur as part of conservation measures associated with "disorder", but these measures do not necessarily diminish a certain preference for "tidy" landscapes and familiar landscape images. In this regard, farmers differ from the nonagricultural society in their perception and evaluation of the landscape (Burton, 2012). In contrast to farmers, the non-agricultural society partly evaluates linearity in landscape images as negative and "unnatural" (Laroche et al., 2018). The aesthetic perception weighs heavier than other evaluation criteria such as agricultural production or conservation. It is postulated that planting natural elements (e.g., bushes) in linear, structured cultivation forms (e.g., straight rows) can evoke feelings of "cultural dissonance" (Laroche et al., 2018). However, the type of landscape image culturally established is relevant in this context. For example, an agroforestry system within traditional orchards generates higher acceptance (e.g., measured in higher willingness to pay) if more than one crop is grown between the tree rows (Alcon et al., 2020), i.e., if more structures are present. However, not only the visual quality of the Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 14 landscape was evaluated, but also the associated ecosystem services and cultural heritage, represented by manual management as opposed to a tractor (Alcon et al., 2020). This approach also points to the complex interplay of visual perception and associated processes for the non-agricultural population. Warren-Kretzschmar and Von Haaren (2014) emphasize the relevance of positive visual evaluation by society as an important aspect besides the ecological benefits of agricultural practice. This likely also generates acceptance for a change in the cultural landscape. In addition to changes in the landscape image through new agricultural systems or structural elements, an impact from the use of technologies in the fields is expected. Although autonomous technologies such as field robots are associated with various benefits, including reduced labour costs (Lowenberg-DeBoer et al., 2021), a survey of farmers showed that concerns about a negative image of "alienated agriculture" in the population can influence the planned acquisition of field robots (Spykman et al., 2021). Previous research on the population suggests that field robots tend to be rated neutral to positive (Pfeiffer et al., 2020). However, Willmes et al. (2022) describe a negative impact on the willingness to pay for food produced with the help of digital technologies. The authors add that this negative impact can be reduced by additional ecological benefits of the technologies. These findings are reflected in a choice experiment on autonomous technologies in weed control, where the method of weed control (mechanical vs. herbicide broadcast and spot-spraying) influenced the decision more than the degree of autonomy of the technologies used (Spykman et al., 2022). However, a joint consideration of autonomous technologies and altered production systems has not yet been undertaken. The aim of this paper is to analyse the perception of the German population regarding new small-scale diversified production systems, the integration of structural (biodiversity) elements such as flower strips, and the use of autonomous technologies such as field robots using an online survey. A special focus is on identifying and evaluating the triggers for potential preferences and the connections to individual visual components. This is done by categorizing short associations provided by survey participants in connection with their preference decisions. Furthermore, this paper also includes a segmentation analysis and illustrates how various sociodemographic characteristics of the population influence the evaluation of agricultural elements such as flower strips or the use of automated technologies. Methods Online survey among the German population A nationwide online survey of the German population aged 18 and older was conducted from mid-September to mid-October 2023. Access to this consumer panel was facilitated through the engagement of a field service provider. The use of a consumer panel allows the separation of personal data and content data, so that research ethics can be assured. The panel enables a pre-stratification of the sample to ensure that participants were representative of the German population in terms of age, gender, size of residential area, and federal state. In addition to various sociodemographic data, information on leisure activities in rural areas, personal connections to agriculture, attitudes towards technology, local food production and sustainable consumption, and knowledge of agriculture, was gathered using established market research methods. After the final data validation, the survey sample comprised 2,022 usable and completed data sets. Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 15 Analysis of preferences, motives, and short associations In a question set regarding visual evaluation, participants were asked to assess various aspects of a landscape image with strip intercropping using four photomontages. All four image variants were based on an identical strip intercropping image which shows a machine passage. The differences included the use of a field robot instead of a tractor and the presence of flower strip. All four images were photomontages that were deliberately not realistic (slightly divergent size of the machines) but were designed to increase the recognizability of the various components for participants (Table 1). Table 1: Choice of image variants for respondents Note: Image sources: Photomontages, Bavarian State Research Center for Agriculture, 2023. The four images were presented simultaneously to the participants and without randomised arrangements. After selecting their preferred image variant, participants were asked to choose three out of six predetermined image components that influenced their decision, he latter component refers to the order and straightness of the parallel field strips as a structured form of cultivation without any beautiful row of trees in the b response. The selected image components were counted and weighted according to their specified rank rank 1 received a triple weight, rank 2 double weights, and rank 3 single weights. This approach allows for the consideration of all three mentioned image components and a composite ranking. In a follow-up question, survey participants were asked to provide up to three short associations in the form of keywords related to the decisive image component (first rank). These rather spontaneous associations to the picture components shown offer additional insights into the decisive image component and the choice of image variant. While the ranking of predetermined image components served the cognitive evaluation by the participants, the affective and thus emotion-based approach of short associations provides another dimension for determining acceptance (Busch et al., 2019; Pfeiffer et al., 2020; Langer et al., 2022). After data cleaning, a total of 4,872 usable keywords as spontaneous associations for the components of the four images were available. Most of these were related to image 4 (tractor/with flower strip), for which a total of 3,092 keywords were analysed and categorised, manually and in several iterations, into 33 categories. From these, the 16 most frequently mentioned categories (covering 2,995 keywords) were identified and prepared for this contribution. Variants Robot/ no flower strip Tractor/ no flower strip Robot/ with flower strip Tractor/ with flower strip Visualization of the image variants for selection Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 16 Modelling the factors influencing preferences Another goal of this contribution is to identify possible sociodemographic influences on the preference for one of the four image variants. Based on similar studies, it was assumed that personal factors such as age, gender, size of residential area, or living in a specific region (e.g., East Germany with large-structured landscapes) play a role, as may the respondents' direct connection to an agricultural environment (Devlin, 2005; Boogard et al., 2008; Pfeiffer et al., 2020). Additionally, the Green Consumption Value (GCV), which reflects the respondents' tendency towards environmentally friendly shopping behaviour, was used as a valueand attitude-based factor. This was measured using six items (Haws et al., 2014). These six items, presented in a Likert-type scale format, were condensed into an individual standardized factor score through factor analysis and considered as a metric predictor for the selection of the image variant. Since respondents could also choose between the use of a tractor and a field robot in the images shown, the attitude towards technology (ATT) was assessed using nine items in a Likert scale format and condensed into a standardized factor score (Edison and Geissler, 2003). For both scales, negative factor values indicate a stronger manifestation of this characteristic, while positive values indicate a lower manifestation. While the typology of survey participants regarding GCV is right-skewed, indicating that participants' purchasing behaviour is predominantly environmental-conscious according to their statements, attitude towards technology is more evenly distributed, showing a balanced ratio between technologyoriented and tech-averse respondents (Figure 1). Figure 1: Distribution of the factor values for green consumption value (GCV) and the attitude towards technology (ATT) of the respondents (n = 2,022); GVC: median: -0.11; skewness: 0.774; kurtosis: 0.525; ATT: median: -0.09, skewness: 0.339; kurtosis: -0.127 Multivariate regression models determine the relationships between multiple predictor variables and a dependent variable. For binomial and categorical dependent variables, logistic procedures are used to determine the probability of the occurrence or non-occurrence of an event (e.g., selection of an image) based on the values of the included predictor variables (Backhaus et al., 2018). Logistic regression provides information about the transformation of the dependent variable logit (p): (1) where p is the probability that the selection of a particular image is influenced by the -selection. The odds ratio Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 17 represents the ratio of these two probabilities. When incorporating k predictor variables, the model takes the following form: (2) The regression equation provides information about the importance of each predictor based ), allowing for the creation of a hierarchy of the measured variables' effects on group assignment (Backhaus et al., 2018). To capture the overall effects and explanatory contribution of the selected influencing factors ed. For this purpose, the image preference was dummy coded as the dependent variable (Model A: 1 = one of the two images with a robot was chosen; Model B: 1 = one of the two images with a flower strip was chosen). Sociodemographic characteristics included gender (1 = female), age (1 = < 40 years), size of residential area (1 = < 20,000 inhabitants), geographical location (1 = western German states), and educational level (1 = no general higher education entrance qualification). Respondents' statements regarding personal connection to agriculture was included in the modelling either as personal employment in the sector (1 = yes) or through personal contact with agriculture in the circle of friends or acquaintances (1 = yes) (cf. Pfeiffer et al., 2020). The metric factor scores of GCV and ATT were also integrated into the two models as additional independent characteristics. Results Distribution of preferences and selection motives In a central question, participants were asked to evaluate changes in the landscape based on single images, considering both the use of robots instead of tractors and the additional use of flower strip. The overall distribution of the stated preferences indicates that the variant with flower strip in conjunction with fieldwork performed by tractors is preferred (Image 4 in Table 15.3% of the 2,022 respondents, while 14.6% chose the robot in combination with flower strip (Image 3 in Table 1). Only 2.5% of the survey participants favoured image 1 (see Table 1), in which the robot was depicted on the field without flower strip. Figure 2 shows the results of the ranking of the image components that were decisive for the participants' preference choices. For images 3 and 4, which depict the robot and the tractor respectively, the flower strip shown in both images is the primary component (41% and 43%, respectively). This is followed by the technical aspect robot or tractor with 27% and 28%, respectively. also frequently mentioned for both images 3 and 4, with 13% each. For images 1 and 2, which depict the robot or tractor without the flower strip, the focus is primarily on the technological aspect, cited as the reason by 39% for the robot and 40% for the tractor. The second place in the familiar image of traditional agriculture and that the flower strip is perceived as rather disruptive to core fieldwork. Regarding image 4 (tractor with flower strip), some respondents remarked that the flower strip specifically symbolizes nature and animal conservation for Proceedings of the 7th Symposium on Agri-Tech Economics for Sustainable Futures 18 them. For images 1 and 3, which show fieldwork done by a robot, innovation, potential efficiency gains, and novelty were mentioned as distinct motives for selection. Figure 2: Distribution of preferences and selection motives Factors influencing preferences The two binominal logistic regression models examining the influence of sociodemographic characteristics on the selection of images with robots and images with flower strip demonstrate distinct effects. Gender influences the selection of images with the robot (Table 2) as female participants are significantly less likely to choose images with a robot compared to men (Odds Ratio = 0.434). If respondents have personal contacts with acquaintances in the agricultural sector, the likelihood of selecting the robot image is significantly lower. Interestingly, respondents with personal agricultural experience exhibit an opposite, though not statistically significant effect. No additional influence factors, such as the attitude towards technology or origin from western or eastern German states, affect the preference for a field robot compared to a tractor. Predictors Model A Field robot B SE Wald p Odds Ratio 95% CI LL UL Gender (1=female)* -0.835 0.337 6.137 0.013 0.434 0.224 0.840 Age (1=younger than 40 years) 0.067 0.392 0.029 0.864 1.069 0.496 2.304 Education (1=no A-levels and below) 0.100 0.332 0.090 0.764 1.105 0.576 2.118 Size of place of residence (1=less than 20k inhabitants) -0.288 0.342 0.706 0.401 0.750 0.384 1.467 Region (1=Western Germany states) -0.314 0.367 0.730 0.393 0.731 0.356 1.501 Own agricultural experience (1=yes) 0.405 0.670 0.365 0.545 1.499 0.403 5.572 Personal contact with farmers (1=yes)* -2.088 1.043 4.008 0.045 0.124 0.016 0.957 Attitude towards technology (ATT) (negative factor value = higher degree) -0.034 0.141 0.059 0.808 0.966 0.732 1.275 Green Consumption Value (GCV) (negative factor value = higher degree) 0.087 0.178 0.239 0.625 1.091 0.770 1.546 Constant*** -1.398 0.406 11.892 0.000 0.247 Note: assignment classification (contribution of predictor variables) = 90.3% | Source: own survey