Agricultural ecosystems play a key role in the conservation and availability of sufficient and quality water resources (Aznar-Sánchez et al., 2018), being the main providers of food but also the main consumers of water resources globally (Velasco-Muñoz et al., 2019; Shi et al., 2021). Irrigated agriculture is the primary user of intercepted water for human purposes, reaching a proportion exceeding 70–80% of the total in arid and semi-arid areas (Liu et al., 2017; López-Felices et al., 2020) food production that requires irrigation uses more than 40% of the total and uses only about 17% of the agricultural area (Saeed et al., 2021). The reduced availability of water resources for agri-food production systems increases the complexity of the social, economic and environmental implications in less developed regions (Gambelli et al., 2021). At present and even more so in the near future, irrigated crops can be grown under conditions of reduced water availability (Canaj et al., 2021; Pardo et al., 2022). Insufficient water resources will, therefore, be the norm rather than the exception and the emphasis of irrigation technology will shift from increasing productivity per unit area to maximizing water productivity (Ungureanu et al., 2020; Campana et al., 2022).

In this context, the solution to move toward is the development of knowledge and innovative solutions for the management and distribution of water resources to Mediterranean agro-productive systems, to make them more resilient to climate change, economically and technically efficient, sustainable, and able to contribute to the economic growth and development of the agricultural sector. The final recipient of such innovations is the agri-food chain, which, faced with climate change and water shortages, risks disrupting supplies of raw materials with quantity, but also quality and health standards (Pandya and Sharma, 2021; Dawit et al., 2022). But a second aspect to be emphasized is the benefit to the environment and water resources generated by innovations; indeed, every activity must be aimed at rationalizing the use of water in agriculture and containing the release of contaminants into the environment (Bowmer and Meyer, 2014). Since an increase in water endowments is not imaginable, the innovative solutions to be pursued must concern the integration of purified wastewater with traditional water endowments and, in particular, the improvement of water use efficiency. Improved irrigation efficiency can be achieved through innovation and technological adoption (Iocola et al., 2020; Tong et al., 2022). Water use efficiency depends on the technology and approach to irrigation (Velasco-Muñoz et al., 2019).

We therefore distinguish three main distribution methods in fruit tree plantations: surface (75% of agricultural land), sprinkler (20% of agricultural land) and micro-irrigation (5% of agricultural land) (Adeyemi et al., 2017; Rouzaneh et al., 2021). The application efficiencies of these general categories vary widely (Mateos et al., 2016; Al-Agele et al., 2021b). Micro-irrigation is generally the most efficient (80–90%) and surface irrigation is the least efficient (50–70%). The efficiency of sprinkler irrigation is 55–80% (Loures et al., 2020; Lopriore and Caliandro, 2022). In other words, the more efficient irrigation approach is used less and vice versa (Mateos et al., 2016). Emerging technologies, such as precision agriculture, agri-voltaic systems and technological innovation in irrigation (Loures et al., 2020), can further increase efficiency and productivity in the food-energy-water nexus (Assouline et al., 2015; Zhu et al., 2022). Precision agriculture in particular proves to be a potential technological ally for increasing water use efficiency (Turral et al., 2010; Pino et al., 2017). This type of agriculture includes decision support systems (DSS) (Bonfante et al., 2019; Souza and Rodrigues, 2022) to perform what-if analyses for managing small amounts of irrigation water, monitoring the water status of soil and/or vegetation (with remote sensing, proximal sensing and soil and vegetation sensors) and precision irrigation (Abioye et al., 2020, 2022).

Traditionally, irrigation was considered precise if the same amount of water could be distributed evenly over the entire surface area, without taking into account the spatial variability of the soil and vegetation (Cabarcas et al., 2019). In contrast, precision irrigation pursues the objective of adapting water supplies to actual crop needs on a small scale (Fernández et al., 2019; Jiménez et al., 2022). Precision irrigation is still underused (Al-Agele et al., 2021a; Beyá-Marshall et al., 2022), but farmer interest is high and would be rapidly implemented if supported by adequate technical dissemination entrusted to new professional figures (Bwambale et al., 2022; Silva et al., 2022).

Mediterranean tree crops such as olives, citrus fruits and vines have deep roots in our country, which run deep into our history. They represent an important reality for Italy in the trade balance with foreign countries and for the economies of many parts of the country (Biasi et al., 2012; Sgroi et al., 2015). Although Italian fruit-growing cannot be defined as a homogeneous system, it often suffers from the same structural problems common to other agricultural systems (de Ollas et al., 2019), which are leading to a downsizing of cultivated areas and production, and faces similar challenges (Testa et al., 2015; Medda et al., 2022). In order to do this, awareness of the sector's critical issues must be raised: its future will depend on the strength of the ideas that entrepreneurs and technicians will translate into concrete actions. For the future, it will be essential to transfer useful innovation to farms, both process and product (Pereira et al., 2020), by introducing into common practice the results of the great advances made in the field of sensor technology, for precise and timely monitoring of orchard conditions (Campos et al., 2019; Yildirim et al., 2021).

It will be necessary to take greater advantage of advances in mechanization to simplify cultivation operations and lower production costs, both for canopy and soil management (Sarri et al., 2015; Campos et al., 2019; Kourgialas et al., 2019). Planting systems and forms of cultivation will have to be revisited in order to exploit their potential in passive defense against pathogens and pests and climatic adversities. The success of such agricultural enterprises will also increasingly depend on their ability to collect and exploit the large amount of data that will be generated, especially to control costs and increase the quality of production (Keswani et al., 2019). It is essential to invest in skills creation in a sector characterized by operational processes based more on generational skills and knowledge transfer than on innovation and process optimisation (Wolf et al., 2001). The most present and easy to take up innovation for these crops has always been genetics (de Ollas et al., 2019; Pinto et al., 2021). The varietal landscape of these species of 30 years ago has, in fact, been completely turned upside down and this type of varietal innovation is justified by the need to have “another” product to market (Parra-López et al., 2021). Technological innovation for these crops, on the other hand, has a much harder time penetrating (Kudryashova and Casetti, 2021; Dinelli et al., 2022), and hardly ever seems to do so through processes that guarantee impartiality on the part of those who issue technical advice that has a cost, and therefore must guarantee a return (Sarri et al., 2015; Montanaro et al., 2017). From various works in the literature, it appears to be that age is the factor that most counteracts the fruit grower's propensity to adopt innovation, while the first mitigating factor would appear to be the availability of a person to provide assistance in the event of difficulties (Chen and Liang, 2020).

This study will seek to understand which psychological constructs actually influence the innovation decision-making process of agricultural entrepreneurs growing Mediterranean tree crops, and to do so it will take into account the Theory of Planned Behavior (TPB) through an extended conceptual model where other factors come into play that influence the intention to implement sustainable irrigation innovations in agriculture and thus complement and extend the classical model.

TPB provides a theoretical framework for the systematic study of factors influencing behavioral choices and has been widely used in other studies to analyze behaviors such as leisure choices, driving offenses, shoplifting and fraud (Zhang et al., 2015). The Theory of Planned Behavior (TPB) states that the decision to take a particular action is directly related to the individual's behavioral intentions (Hansson et al., 2012; Gao et al., 2017; Soorani and Ahmadvand, 2019). Intention (I) is, in turn, influenced by three factors (Figure 1):

This study proposes an integration of the Theory of Planned Behavior (TPB) by including additional variables to increase its predictive accuracy (Joao et al., 2015; Rezaei et al., 2018; Tama et al., 2021; Sarkar et al., 2022). This conceptual model considers, in addition to the three classical TPB factors, namely attitude (A), subjective norms (SN) and perceived behavioral control (PBC), three other factors (Hou and Hou, 2019) such as perceived innovation characteristics (PIC), Benefits (B), and Transferability (T) and hypothesizes that all these six elements could directly or indirectly influence innovation intention (Wauters et al., 2010; Müller et al., 2021). In the indirect case, this would be due to the effect of perceived innovation characteristics (PIC), which act as a link between the other factors and intention (Figure 2).

Attitude toward the adoption of an innovation in agriculture refers to an individual's positive or negative evaluation of its implementation (Senger et al., 2017; Tóth et al., 2020). The second determinant of intention in TPB, is the subjective norm, which refers to the perceived social pressure on the person from the peer group, family, society or culture to perform the behavior under consideration (Tóth et al., 2020; Sarkar et al., 2022). The objective is therefore to detect whether the adoption of sustainable innovation in agriculture is conditioned by third parties. The third factor within the TPB model is perceived behavioral control (PBC) which refers to the sense of self-efficacy or ability with respect to a potential behavior (Tóth et al., 2020; Saeedi et al., 2022). The fourth element considered for the conceptual model is defined as characteristics of perceived innovations (PIC) and is taken from the theory of diffusion of innovations (Rogers, 1995). Rogers, in his theory, mentions five characteristics of an innovation that can affect the relative rate of adoption by different members of a social system: relative advantage, compatibility, complexity, observability and trialability. All these factors influence the decision to adopt or not to adopt technological innovation (Figure 3).

Among the additional variables considered, a further determinant of intention is transferability, which makes it possible to define whether the results of the experiment are transferable. Within the extended model, the final determinant of intention is the benefits, which allow the positive effects of innovation on the company to be measured.

Based on this knowledge, we formulated 11 hypotheses:

The survey to detect the propensity to adopt an innovation that favors the sustainability of the production process, i.e., respectful of the environment, animals, health and workers' rights, as well as the resulting economic and environmental benefits, was carried out through a specially designed questionnaire using the “Google Forms” tool and divided into 4 sections. It was disseminated online through the main social media channels between 6 June 2022 and 6 September 2022 with an active survey period of 90 days.

The first section deals with general information about the company, consisting of a series of questions about the company and the entrepreneur. The second section refers to the organizational choices of the surveyed farm manager, in terms of both needs and market. The third section concerns the analysis of the propensity to adopt an innovation. This section consists of a series of pre-defined questions designed to measure the behavior of the entrepreneur. The fourth section concerns the expected results following the adoption of an innovation. This section consists of a series of questions designed to capture key elements (such as perceived innovation characteristics, benefits and transferability of innovations) of the behavioral model. The latter two sections of the questionnaire are those concerning the TPB items in relation to the intention to adopt sustainable innovation and mostly use the 7-point Likert scale, where higher scores indicate greater compliance with the items, except for the benefits which were assessed on a multiple-choice format. Once the planning phase of the questionnaire had been completed and before starting data collection, we moved on to the control phase. At this stage, the necessary checks were carried out to ensure that there were no programming errors (bugs or malfunctions) and that the questionnaire was computerized appropriately to achieve the set research objectives. A total of 200 responses were collected from as many farms, of which 125 were selected as suitable for data analysis.

Previous studies have largely focused on socio-economic characteristics and ignored psychological factors influencing adoption intention (Borges et al., 2019). Instead, in this study, we sought to examine psychological factors by hypothesizing that these could explain greater variation in the dependent variable (the intention to implement innovation) than the socio-economic characteristics of farmers. The study aims to verify whether the TPB variables, together with transferability and benefits, predict the intention in relation to the adoption of sustainable innovation in agriculture. The data were analyzed using the Statistical Package for the Social Sciences (SPSS) version 27. First, we cleaned and checked the data to identify any missing values or irregularities. Secondly, we calculated descriptive statistics (e.g., averages and standard deviations). We checked the quality and adequacy of the measurement model and, through exploratory factor analysis (EFA), attempted to associate the variables with the various latent factors. Next, Pearson correlation coefficients were calculated to assess the correlation between the factors (Adnan et al., 2018). Therefore, we tested the causal relationships between the different factors of the TPB model with integrations by means of a hierarchical regression analysis, where we entered intention as the dependent variable and the TPB constructs as independent variables in the first stage, and then in the second stage we entered PIC as the dependent variable and the remaining constructs as independent variables (Saeedi et al., 2022).

The results described in Table 1, show that the majority of respondents (94.4%) were male, confirming that the role of women is still marginal. Most of the respondents were between 31 and 50 years old (56.8%), while only 14.4% of the respondents were young people under 30. With regard to the legal form of the companies surveyed, it emerges that 84.8% of them are run by professional agricultural entrepreneurs and the form that prevails is that of the “Simple Company”, which accounts for 84% of the companies surveyed. With regard to farm size, we note that 48.80% of the farms cover an area of between 5 and 20 hectares, 41.60% have an area of less than 5 hectares and only 9.60% are identified as large farms with an area of more than 50 hectares. Looking at the production addresses, however, a homogeneous distribution appears, with citrus (19.2%), fruit (19.2%), and olive (17.6%) being most present, and only the mixed fodder/fruit address showing a very low percentage (0.8%) with only one answer. Most of the respondents (33.6%) completed their education, while 16.8% completed only primary education and 17.6% completed secondary education. Finally, with regard to the labor used in the company, 52% of the companies surveyed use family labor, the remainder (48%) use external labor.

The analysis of the questionnaires made it possible to construct Figure 4, which well expresses with a visual element the frequency with which a 'word or phrase' connected with a sustainable irrigation innovation is used by entrepreneurs in relation to other words in an irrigation dataset. In our case, the evaluations proposed in the optimization of water use were related to a number of parameters among the many possible ones and, among them, were considered:

Minimization of water losses can be ensured through different agronomic practices such as (i) mulching, use of (ii) grafted plants and (iii) biostimulants. Mulching, in addition to preventing weed growth, reduces evapotranspiration, improves root growth and the uptake of water and nutrients, and increases water use efficiency. Similarly, the use of grafted plants, due to the better net CO₂ assimilation and transpiration efficiency and the greater development of the root system, is another valid strategy to help reduce irrigation volumes. Not least, the use of microbial and non-microbial bio-stimulants can improve the morphological and physiological characteristics of crops, enhancing their productive performance and contributing to a more virtuous use of water resources.

The arboreal and mixed (arboricultural and other) farms that participated in the survey stated that they adopt an irrigation sustainability strategy partly for ethical reasons and partly to optimize water and energy consumption and achieve adequate levels of economic-productive performance (Figure 5). On the initiatives undertaken, in some cases there has been a prior study of the terrain, the type of planting and the characteristics of the irrigation service (with substantial differences between those who own irrigation investments such as wells, storage tanks, etc.) and those who acquire water from a public body in charge of the purpose (irrigation consortium mainly, which entails irrigation shifts, operating pressure, watering volumes, etc. that are not always dependent on their own will). Innovations in irrigation techniques include sprinkler and micro-irrigation systems. Drip lines are also widely used with micro-sprinklers to combine the positive effects of a drip system with overhead sprinklers. Finally, innovations in irrigation strategies were limited, such as the use of deficit irrigation combined with micro meteorological remote sensing and proximal sensing technologies, probably also due to the lack of adequate expertise.

In order to extract latent variables from the questionnaire items, exploratory factor analysis (EFA) was used. Through the KMO and Bartlett's verification tests, it was possible to confirm the validity of the extended TPB model comprising seven latent factors indicating intention, attitude, subjective norm, PBC, PIC, benefit and transferability. The results show an adequate fit of the model (Kaiser-Meyer-Olkin measure of sampling adequacy = 0.66, Bartlett's test of sphericity with Sign < 0.001). Table 2 shows the number of items considered for the extraction of each latent factor and their standardized factor loadings. Each item corresponds to a question on the questionnaire that was measured by entering a single scale from 1 to 7 differentiated by individual question, where value 1 means Not at all agree/Absolutely unlikely and value 7 means Completely agree/Absolutely likely. Item factor loadings below 0.50 were discarded from the analysis. In order to assess the internal consistency and reliability of the scale, the study estimated Cronbach's alpha coefficients for each factor. Cronbach's alpha to assess internal consistency can be classified as: excellent (α ≥ 0.9), good (0.7 ≤ α < 0.9), acceptable (0.6 ≤ α < 0.7), poor (0.5 ≤ α < 0.6), and unacceptable (α < 0.5). The results show adequate internal consistency of the scale items, as Cronbach's alpha coefficients range from 0.70 to 0.96. In addition, descriptive analyses of the items were conducted and the table shows the mean and standard deviation, with the highest mean value for attitude and the lowest for benefits.

The results of Pearson's correlation coefficient test between the variables are shown in Table 3, which reveals significantly positive correlations between intention and all other variables in the model with the exception of transferability. There is also a good correlation between the variables, with a few exceptions, e.g., aptitude appears to be uncorrelated with PBC and transferability, just as there is no correlation between PBC and transferability itself, which appears to be the most problematic variable in this respect.

In order to test the general relationships between the variables and thus answer the assumptions made, two different linear regressions were conducted. The first was performed in order to understand which variables influence the intention to adopt the innovation and therefore intention was set as the dependent variable and TPB constructs as independent variables. The second regression was aimed at understanding the mediating effect exerted by the characteristics of perceived innovations (PIC) against intention for the other variables, thus setting PIC as the dependent variable. With regard to the first regression, the ANOVA table shows an F-value of 11.43 and a significance level p of < 0.001, the regression model therefore fitted well. The summary table of the model shows that R2 has a value of 0.52, which indicates that 52% of the variance of intention can be explained by attitude (ATT), subjective norm (SN), perceived behavioral control (PBC), perceived innovation characteristics (PIC), benefits (B) and transferability. The results in Table 4 show that intention is strongly determined by attitude (ATT) as the most important variable influencing behavior (B: 0.337, significance level p < 0.001). Subjective Norms (SN) (B: 0.377, significance level p = 0.001) and Perceived Innovations Characteristics (PIC) (B: 0.263, significance level = 0.008) also show a good level of influence toward intention. The remaining factors such as perceived behavioral control (PBC) (B: 0.042, significance level p = 0.647), benefits (B) (B: −0.044, significance level p = 0.651) and transferability (T) (B: −0.077, significance level p = 0.449), as we expected, do not directly influence intention.

The second regression shows an ANOVA table with F equal to 17.84 and a significance level p equal to < 0.001, the regression model therefore fitted well. The summary table of the model shows that the R2 has a value of 0.58, which indicates that 58% of the variance in the characteristics of perceived innovations can be explained by attitude (ATT), subjective norm (SN), perceived behavioral control (PBC), benefits (B), and transferability. The results in Table 5 show that perceived behavioral control (PBC) is directly related to PIC (B: 0.222, significance level p = 0.007), the same applies to benefits (B) (B: 0.211, significance level p = 0.010) and transferability (T) (B: 0.354, significance level p < 0.001).

The results show, therefore, that only attitude (ATT) subjective norms (SN) directly influence the intention to adopt sustainable irrigation innovations in agriculture (I). Other factors such as perceived behavioral control (PBC), benefits (B), and transferability (T), indirectly influence intention, due to the effect of perceived innovation characteristics (PIC), which influences intention (I) by acting as a mediator between PBC, T, B with I, and thus acting on the psychology of the individual.