In the last years, the use of biostimulants has been constantly increasing for sustainable agriculture, because these substances enhance nutrient use efficiency, reduce fertilizers consumption, stimulate plant development and growth (Kunicki et al., 2010; Calvo et al., 2014; Halpern et al., 2015; Le Mire et al., 2016), and counteract stress factors, eventually enhancing crop quality and yield (Ziosi et al., 2013; Van Oosten et al., 2017). The interest in this sector is evidenced by the significant increase of research papers focused on it and by its economical relevance. The market of biostimulant products is projected to increase by 12% annually (Calvo et al., 2014), reaching $2,524.02 million by 2019 (Povero et al., 2016). Biostimulants are generally made of raw organic materials containing bioactive compounds. Their nature is heterogeneous, since they include humic acids, protein hydrolysates, plant growth-promoting Rhizobacteria and fungi, and extracts from seaweeds and higher plant species (Ertani et al., 2013, 2016; du Jardin, 2015). For this reason, also their chemical composition is highly heterogeneous, including mineral elements, vitamins, amino acids, chitin, chitosan, and poly- and oligosaccharides, and therefore it is partly unknown. Moreover, their chemical complexity and the wide range of molecules present make very difficult to understand which are the most active compounds (Brown and Saa, 2015; Bulgari et al., 2015; du Jardin, 2015; Posmyk and Szafrańska, 2016; Yakhin et al., 2017).

Under a biochemical point of view, the increased plant growth induced by biostimulants can be associated with an increase in amino acid levels and enhanced protein biosynthesis, as well as in carbohydrate concentration in leaves (Abdalla, 2013). Higher sugar levels in plants treated with biostimulants have been found in several species, associated with higher chlorophyll accumulation, net photosynthesis (Abbas and Akladious, 2013; Abdalla, 2013), and quantum efficiency of photosystem II (Ferrini and Nicese, 2002; Amanda et al., 2009; Ertani et al., 2012). In lettuce, biostimulant application at the nursery level positively affects plant growth by increasing the development of shoots and roots, strongly stimulating root growth and increasing leaf area (Vernieri et al., 2002; Amanda et al., 2009). Moreover, a positive role on yield and quality of head lettuce has been reported, with particular regard to the reduction of nitrate that is an important issue for human health (Shehata et al., 2016).

Biostimulants are capable to counteract plant stresses that are usually related to a burst in ethylene synthesis and sensitivity, and eventually affect produce quality by altering or accelerating tissue senescence (Saltveit, 1999). The activation of stress responses in plants is often accompanied by the synthesis of secondary metabolites (Mazid et al., 2011; Ramakrishna and Ravishankar, 2011) that often act as antioxidants scavenging in vivo and in vitro (Cook and Samman, 1996) the free radicals (Halliwell, 2008) produced in stress-induced oxidative reactions (Sharma et al., 2012), and counteracting the free radical-induced damages to cell components. In animal systems, and particularly in humans, several studies have shown that a diet rich in antioxidants from plant-derived foods may prevent the onset of a wide range of chronic-degenerative diseases (Manach et al., 2004; Vauzour et al., 2010; Martin et al., 2013; Bertoia et al., 2016).

The largest group of bioactive beneficial secondary plant metabolites is represented by phenolic compounds, ubiquitous in all tissues of higher plants, where they play an important role providing the plant with specific adaptations to changing environmental conditions and eliciting defense mechanisms (Caretto et al., 2015 and references therein). Phenolic compounds are synthesized in the phenylpropanoid pathway, that produces an array of different substances including, amongst others, phenolic acids and flavonoids (El Gharras, 2009), reported to possess powerful antioxidant activity in vitro (Kähkönen et al., 1999). The first committed enzyme in the phenylpropanoid biosynthetic pathway is phenylalanine ammonia lyase (PAL; E.C. 4.3.1.5), that catalyzes the non-oxidative deamination of phenylalanine to trans-cinnamic acid (CA), common substrate for the biosynthesis of various phenylpropanoid compounds (Ferrer et al., 2008), a step that represents a crucial link between primary and secondary metabolism. PAL activity is positively related with increased production of phenylpropanoids (Vogt, 2010). Vegetal extracts derived from red grape, blueberry fruits and hawthorn leaves or from oak, affect PAL activity and expression of PAL genes as well as the levels of polyphenols in maize or grape, respectively, when applied as biostimulant (Ertani et al., 2011, 2016; Pardo-García et al., 2014). To our knowledge, studies on PAL in lettuce have dealt mainly with its involvement in postharvest disorders (Ke and Saltveit, 1986, 1989), tissue browning (Campos et al., 2004), and pigment biosynthesis under different temperature regimes (Chon et al., 2012). In the last decade, the availability of relatively inexpensive and high-performance optical systems, digital cameras and associated software technologies has boosted the development of phenotyping systems, i.e., of semi-automatic or automatic devices enabling repeated and non-invasive measurement of macroscopic plant’s parameters related to growth, architecture features or to main tissue components (Fiorani and Schurr, 2013). These systems are typically based on RGB color cameras to compute leaf area, biomass volume or to count/quantify specific plant organs, but they can also include the use of VIS-NIR hyperspectral cameras, to estimate the levels of main tissue components as chlorophylls, anthocyanins, and water, of fluorescence imaging devices, to map the efficiency of photosystems, or of thermal infrared cameras, to evaluate foliage temperature and leaf water status (Li et al., 2014; Fahlgren et al., 2015).

Early applications of color imaging to monitor lettuce growth were aimed to investigate the possible use of sensed plant-projected area extracted from top-view images to identify nutrient stress in hydroponic cultivation (Giacomelli et al., 1998). More recently, similar approaches have been applied to lettuce growth-rate data extracted from greenhouse imaging, to be used as state-variable to feedback control of nutrient solution in hydroponic system (Jung et al., 2015) or for other crop management operations (Kim et al., 2013). Bumgarner et al. (2012), by conducting an extensive study on imaging of lettuce plants grown in different environments, concluded that a top-view approach is an accurate method to indirectly measure lettuce biomass during the early stages of growth, while on canopy closure the correlation is weakened by occlusions in plant’s top-view due to leaves overlapping. This limitation, also reported by Jung et al. (2015) for lettuce plants at advanced stages of development, is encountered with any plant with erectophile architecture (Stewart et al., 2007; Tackenberg, 2007), and it has been addressed by deploying different approaches, as: by side-view imaging configurations, or, rarely, in combination with top-view (Pereyra-Irujo et al., 2012); by the use of three-dimensional (3D) measuring instrumentation such as LIDAR (Friedli et al., 2016), stereoscopic or multi-view cameras (Rose et al., 2015; Golbach et al., 2016) or time-of-flight (ToF) cameras (Chéné et al., 2012).

Low-cost 3D imaging sensors are emerging as an alternative to expensive 3D measurement systems, especially interesting for experiments involving small-scale, custom-made phenotyping hardware (Azzari et al., 2013; Paulus et al., 2014). The Microsoft Kinect V1 is a popular example of such a device, able to acquire at real time rate (30 frames per second) RGB color images aligned and synchronized with a depth D images. It can operate under indoor (or protected) illumination conditions, in a recommended distance range of 1–3.5 m, with a nominal depth error of ±10 mm (Livingston et al., 2012). A phenomic approach has recently been described in a study dealing with the evaluation of the efficacy of the biostimulant Megafol^® in reducing drought-stress related damage in tomato plants (Petrozza et al., 2014). Recently, an innovative method of selection and characterization of plant biostimulant matrices, involving a combination of technology, processes, and know-how, has been proposed (Povero et al., 2016).

Borage (Borago officinalis L.; Boraginaceae) is native to the Mediterranean region (Baubaire and Simon, 1987). The beneficial properties of chemicals extracted from different organs of this plant are widely acknowledged (for a review, see Asadi-Samani et al., 2014), and they are used in traditional medicine (Krishnaiah et al., 2011; Asadi-Samani et al., 2014), food preservation (Ciriano et al., 2009; Aliakbarlu and Tajik, 2012), and even for packaging purposes (Gómez-Estaca et al., 2009). The antioxidant properties of borage extracts from defatted seeds, leaves or flowers can mainly be ascribed to the presence of phenolic compounds (Wettasinghe et al., 2001; Aliakbarlu and Tajik, 2012). Borage leaves, that represent more than 60% of the plant matter, are considered also a low-cost crop by-product by some food processing industries (Garcia-Herreros et al., 2010).

On these premises, it appeared worthwhile to explore the possibility of using borage as a cheap source of biostimulants. Aim of the present work was to study the efficacy on lettuce plants (Lactuca sativa ‘Longifolia’) of foliar treatments with raw aqueous extracts obtained from leaves or flowers of Borago officinalis L. For this reason, a holistic approach has been adopted, including both traditional and innovative investigation techniques. Within the framework of this multidisciplinary study, a non-invasive measurement setup based on Kinect devices was implemented to evaluate plant growth (biomass) during time. Leaf functionality and stress responses were monitored by non-destructive measurements of chlorophyll a fluorescence, gas exchanges and by ethylene determination. Quality parameters such as concentrations of sugars, nitrates, and photosynthetic pigments, together with those of representative phenylpropanoid compounds (total phenols, flavonoids) and PAL activity and PAL-like polypeptide levels, were assessed by traditional biochemical methodologies. Eventually, a preliminary trial was set up in order to observe the effects of borage extracts integrated into packing films during storage of lettuce leaves.

Biostimulants act at different levels, increasing plant growth, photosynthetic and metabolic activities and nutrient absorption (Bulgari et al., 2015; Yakhin et al., 2017). The production of a potential biostimulant begins with raw material characterization followed by the study of plant responses (Povero et al., 2016). The effect of raw material extracts has to be evaluated under normal or stress conditions by investigating the physiological and biochemical processes that are activated after treatments. Successful biostimulant candidates should increase biomass and yield or counteract the negative effect of different stresses. Their use is nowadays becoming a common practice in crop production to improve productivity and yield. In leafy vegetables, the biostimulant Actiwave^® (containing betaine, alginic acid, and caidrine) applied as an additional component to the nutrient solution increases yield of rocket grown in a floating system, even with reduced nutrient concentrations (Vernieri et al., 2005); its effect was confirmed when administered as a spray on baby leaf lettuce grown in plastic tunnel (Amanda et al., 2009). An extract of brown marine algae was reported to increase growth of spinach in vitro (Fan et al., 2013). Dudaš et al. (2016) observed, on winter production of lettuce ‘Four Seasons,’ that the plant head mass was by 30% higher after treatment with Bio-algeen S-90 compared to control plants. Sternecker and Balas (2014) observed in lettuce ‘Mathilda’ a head weight increase of 31% upon use of a biostimulant composed by a mixture of extracts from 21 plant species associated with Lactobacillus and yeast. The increment that we observed in Romaine lettuce growth is consistent, in spite of the high variability of results, with the cited literature reports, and supports the hypothesized role of borage extracts in stimulating the biomass of treated plants.

Imaging methods have been successfully used for non-invasive estimation of plant growth (Tackenberg, 2007) also after biostimulant treatments (Povero et al., 2016), but literature concerning their application to lettuce is scanty and relies upon a top-view imaging of plant heads. Due to leaf overlap after canopy closure, this approach revealed a generally weak correlation between image-based and destructive measurements of biomass, as growth progressed. For ‘Outredgeous,’ Bumgarner et al. (2012) observed a decrease in correlation coefficient from r = 0.87 at 10 days after sowing to r = 0.22 at harvest (28 days). Similarly, Jung et al. (2015) reported a sharp decrease in the correlation when lettuce heads have a fresh weight above 25 g, even if they found an overall RMSEC of less than 5 g when estimating the biomass of samples with fresh weight ranging up to 70 g. In this study, the adopted multi-angle, side view approach enabled to define a linear model capable to estimate the lettuce biomass through image data with a RMSEC = 2.2 g, for fresh weight values up to 155 g. This model was successfully applied to monitor non-invasively the growth of lettuce heads as affected by borage extracts application, and the multiple side view approach allowed capturing the subtle effects of the treatments during plant growth. The application of multi-angle, side view imaging, instead of classical top-view approach, allowed obtaining a fair correlation (R²= 0.74) with the destructive harvest data even for plants at advanced growth (i.e., commercial harvest stage). It must be noted that the multi-angle approach used in this work can be successfully applied when conducting phenomic studies, but it does not appear suitable for on-the-go measurements in field or in greenhouse, where top-view imaging setup is the best option thanks to its much simpler implementation.

In lettuce, ethylene production is extremely low compared to other plant tissues (Burg, 2004). In lettuce ‘Acephala’ values of ethylene production lower than 10 μL kg^–1 h^–1 are reported (Diaz et al., 2007). Concerning Romaine lettuce, to our knowledge, only scanty literature is available about ethylene production. Regarding the effects of borage extracts described in the present work, it is interesting to notice that, despite the high variability of the results, possibly due to the extremely low levels of ethylene emission, in three out of the five experimental conditions a decrease in ethylene production was induced by borage extracts, particularly evident upon FEs administration, suggesting a healthier physiological status in the FE-treated plants. This result could be explained considering the antioxidant activity due to the presence of radical-scavenging components reported for crude B. officinalis extracts (Bandoniene and Murkovic, 2002; Bandoniene et al., 2005), that may play a role in counteracting the effects of potential stress factors and the related ethylene production.

Biostimulants enhance plant growth and total photosynthesis determining higher dry matter accumulation in vegetable and ornamental crops (Khan et al., 2009; Bulgari et al., 2015; Massa et al., 2016). Chlorophylls (also important for the visual appearance of the produce) and carotenoids (photoprotective molecules whose amount is related to that of chlorophyll) are involved in fundamental photochemical processes tightly associated with crop biomass production. Moreover, carotenoids and chlorophylls play an important role in preventing various human chronic-degenerative diseases associated with oxidative stress (Znidarcic et al., 2011), contributing to the nutraceutical quality of plant produce (Yuan et al., 2015). Biostimulant treatments are often able to increase leaf pigments concentration. In rocket, treatments with a Moringa oleifera extract increased chlorophyll and carotenoids levels (Abdalla, 2013); similar results were obtained with the biostimulant Actiwave^® (Vernieri et al., 2005). The commercial product ONE^® had positive, dose-dependent effects, on the chlorophylls levels of lettuce and endive (Bulgari et al., 2014). Consistently, borage extracts (in particular 1 mL L^–1 LE) slightly increased the chlorophyll and carotenoids levels compared to controls.

Leaf functionality is also described by gas exchange analysis or estimated by chlorophyll a fluorescence. These non-destructive methods can be applied to evaluate the health status of the photosynthetic apparatus or the different responses of plant tissues to stress factors or experimental treatments (Murchie and Lawson, 2013). In the present work, a positive effect of borage FEs may be suggested by the higher values of PI, a general index of the leaf health status. Moreover, the higher number of reaction centers and lower rates of energy dissipation confirmed the hypothesis of a direct positive effect of the treatment on PSII efficiency. In lettuce, significant changes in the Fv/Fm ratio (a good indicator of leaf stress), usually observed after a mid- or long-term exposure to a specific treatment or stressful condition (Stepień and Kłbus, 2006), are considered an index of irreversible photoinhibition of PSII reaction centers (Dias et al., 2014). In our material, the Fv/Fm values did not show any significant change, suggesting a general positive effect of extracts on leaf functionality.

Biostimulant applications in coriander under cold stress were able to increase the Fv/Fm ratio, the transpiration, and stomatal conductance rates, but reduced intercellular carbon dioxide concentration (Pokluda et al., 2016), suggesting that biostimulants may accelerate the adaptation to chilling. The chlorophyll a fluorescence-derived parameters have been used for evaluating the vitality of transplant-sensitive tree species after transplanting, and the effects ohe effects of biostimulants application, that increased leaf functionality as shown by higher values of PI (Fraser and Percival, 2003).

Our results showed that the highest doses of both types of borage extracts increased the net photosynthesis as revealed by gas exchange analysis. In strawberry, Actiwave^® increased the photosynthetic activity by 27% compared with control (Spinelli et al., 2010). Consistent results were found in ornamental plants treated with a municipal biowaste: hibiscus plants showed an increase of net photosynthesis by 24% (Massa et al., 2016) and similar findings were observed in Euphorbia × lomi (Fascella et al., 2015).

Biostimulants improve the primary metabolism of plants, increasing the levels of free amino acids, proteins, carbohydrates, pigments, and various enzymes as reported by Yakhin et al. (2017).

In our material, the leaf sucrose levels were not affected by any borage extract treatment, suggesting that neither the nutritional nor the sensorial quality of the produce were significantly altered. However, the tissue levels of total sugars were diminished by all treatments, whereas the levels of some secondary metabolites, like total phenolics and flavonoids increased, as well as the antioxidant capacity. The opposite changes in the levels of total sugars and phenylpropanoid compounds would contribute to the health-related characteristics of the produce, at the same time maintaining a high level of chemical defense capability (Neilson et al., 2013) and, in turn, a better performance in terms of plant growth. A better general status of the plants treated with borage extracts, as well as potential higher resistance to stress factors for the presence of phenolic substances, is also suggested by the higher levels of total soluble proteins, indicative of the bulk of metabolic activity (Veerasamy et al., 2007 and references therein). The observed tendency to lower ethylene production is consistent with this view. Several primary metabolites (like free amino acids, sugars or other molecules not immediately required for growth and development) are precursors of secondary compounds, among which polyphenols (Mazid et al., 2011). In particular, the deamination of phenylalanine to trans-CA catalyzed by PAL links the primary metabolism to the production of a wide variety of secondary phenolic compounds, that serve diverse functions in plants, including protection against biotic and abiotic stresses, cellular signaling, mechanical support (MacDonald and D’Cunha, 2007). Our results on in vitro PAL activity and levels of PAL-like polypeptides are, in general, coherent with the results on total phenolics and flavonoids concentrations, even if we could not observe a tight correlation between the cited parameters. This result might be attributed to the very complex regulation of this enzyme, that involves several steps, from PAL (iso)genes transcription to assembling of the functional protein, and to enzyme turnover and mechanisms of activity regulation (phosphorylation–dephosphorylation); also feedback control by the levels of total phenolics/flavonoids is reported to regulate PAL protein turnover and catalytic activity (Zhang and Liu, 2015).

High dietary nitrate intake is hazardous for health, since in the human organism nitrate is reduced to nitrite that can react with the free amines deriving from protein digestion and form carcinogenic nitrosamines. For this reason, the European Union has posed limits in nitrate concentrations of commercialized leafy produce (Cavaiuolo and Ferrante, 2014). Nitrate accumulation in leafy vegetables is affected by several environmental factors like light intensity, photoperiod, and temperature (Lillo, 1994 and references therein). Biostimulants reduce the nitrate levels in several species of leafy vegetables (Vernieri et al., 2005; Liu and Lee, 2012; Dudaš et al., 2016). The borage treatments applied in the present work did not significantly affect nitrate in the lettuce cultivar used. Nitrate concentration shows considerable variations in different lettuce cultivars (from 26 mg kg^–1 to more than 2500 mg kg^–1; Cometti et al., 2011). In our material, the nitrate levels were lower than approximately 250 mg kg^–1 FW, possibly explaining the observed lack of effect of borage treatments, similar to what observed for Actiwave^®-treated baby leaf lettuce (Amanda et al., 2009). It should also be stressed that the effect of biostimulants on nitrate levels in leaves can be different depending on the species/cultivar and it is affected, in addition to environmental factors, by dose and time of application (Kunicki et al., 2010).

The effect of borage extracts, containing themselves bioactive molecules or possibly releasing volatile compounds (VOCs), during postharvest of lettuce leaves, was also evaluated in a preliminary trial. The visual appearance (chlorophyll) of the produce and leaf carotenoids levels, both known to be affected by storage conditions (Bolin and Huxsoll, 1991; Bergquist et al., 2007; Agüero et al., 2008) were assessed. Borage extracts, administered either as in vivo treatment to plants or applied as coating on packaging films, exerted a positive effect on the photosynthetic pigments, preventing their degradation, and even inducing their increase during storage. In particular, 10 mL L^–1 FE proved capable to induce a significant increase in total chlorophylls and carotenoids after 7 days of storage as compared to the controls. However, further experiments will be necessary in order to investigate the mechanism of action of the bioactive compounds potentially present in borage extracts, when incorporated into a primary packaging. Moreover, a study of the release kinetics of such active molecules in the package headspace may also help to understand whether the effects of these treatments is linked to the production of active VOCs and/or is due to a direct contact between the produce and the inner surface of the packaging.

Taken as a whole, the multidisciplinary approach used in this work demonstrated that borage extracts do indeed exert biostimulant effects on lettuce plants. This result suggests a possible exploitation of borage extracts in vegetable production of different species, as well as in their commercialization, to improve quality and nutraceutical properties and thus adding value to produce. Moreover, the phenomic approach adopted proved capable to estimate with good accuracy plant growth and represented, in general, a fast and reliable method for the non-destructive screening of the efficacy of experimental treatments, integrating well the biochemical–physiological approach. Aspects related to both primary and secondary metabolism were enhanced, suggesting a potential ability of these extracts to counteract possible stress factors. In particular, FEs proved more effective than LEs on the plant physiological and biochemical parameters considered. These results may be validated at molecular level by studying the transcriptional profiles using high-throughput technological tools like microarrays and RNA-seq. The molecular mechanisms elicited by crude plant extracts acting as biostimulants were recently studied, in Arabidopsis thaliana, through a microarray-based genomic approach (Santaniello et al., 2013). A transcriptional profiling of phenylpropanoid pathway genes in Arabidopsis thaliana as affected by application of microbial products has been recently published (Ali and McNear, 2014). These additional research activities will also allow describing more completely the efficacy of borage extracts in preserving and enhance crop performance. Moreover, the effects of borage extracts and surfactants could be explored in future works in order to take into consideration other factors that can affect the effectiveness of foliar treatments (Fernandez and Brown, 2013).

Our results appear suitable to be fruitfully included in a larger, integrated framework where different approaches are systematically combined (Povero et al., 2016) in order to study at different levels the potential positive effects of various natural extracts on plant performance and biochemical–physiological parameters related to qualitative features, eventually characterizing and validating these extracts as new biostimulant products exploitable in the field.

RB, substantial contribution to the experimental work, interpretation of data, and drafting of the manuscript; SM and NN, substantial contribution to the biochemical work, interpretation of data, drafting and critical revision of the manuscript; GC, elaboration and critical interpretation of data, drafting of the manuscript; SF designed and carried out the experiments on the coated plastic films; AC, EF, and RO, integration of the experimental setup for multi-view imaging of plants, imagery data acquisition and processing, and drafting of the manuscript; IM and AS carried out the ethylene production study and analysis of the data, and participated in the experimental design, critically reading the manuscript drafts; AF, experimental design and coordination of the work, interpretation of data, drafting and critical revision of the manuscript. All authors read and approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.