To illustrate the approach and dynamics of creating spatial healthscape maps, we used as a case study the Lincoln University Mount Grand Station (LUMGS) high-country station. The station is located close to Lake Hawea in Central Otago, South Island (lat 44°38′01.93"S; long 169°19′42.89"E), New Zealand (Figure 2) and encompasses a total area of 2,131 ha, of which 93% is a steep hill country. From the total area, 1,602 ha are used as a pastoral system, while the remainder is under a conservation area of the New Zealand Department of Conservation. The annual rainfall averages 703 mm and the annual mean temperature is 10.6°C (Maxwell et al., 2016).

The database used in this study was a detailed list provided by Duncan et al. (1997), of plant species found at LUMGS. The list was used for the identification and phytochemical description of plants in LUMGS with nutraceutical, prophylactic, and medicinal potential. In addition to such a database, and as suggested by Torres-Fajardo et al. (2020), other scientific reports, expert knowledge, New Zealand national plants online database, local knowledge, and public online databases were also considered (refer to Tables 1 and 2).

A total of 22 plants were identified and considered as illustrative of the case study for the methodology proposed here in this study. Based on the abundance of LUMGS and spread in New Zealand high country grasslands, 11 of those 22, including grasses, herbs, shrubs, and trees (Table 1), were selected for further chemical analyses and an additional three were indicated by local experts and the station manager, due to their abundance in the area and LUMGS, and because those plants are known to have medicinal properties.

At least three subsamples of leaves and flowers (if present) of the selected plants were manually collected from different locations in the case study area, indicated by local knowledge. Subsamples were pooled together to create representative samples of each plant species at the station and immediately frozen and stored for further chemical (primary and secondary composition) analysis at Lincoln University Riddolls laboratory.

Historical data of LUMGS vegetation cover and composition has been previously reported by Duncan et al. (1997) and was the data source used in our database. Nevertheless, due to the period elapsed since that report, botanical composition analysis was conducted during Summer (December 2020 and January 2021) to update and add detail about the vegetation cover and composition of LUMGS.

According to New Zealand National Land Cover Database (LCDB), and based on vegetation abundance, LUMGS is classified into different vegetation types, including grassland (dominated by grasses, forbs, and herbs), Manuka and/or Kanuka, exotic forest, and indigenous forest. Because the “grassland” area is the most representative part of the animal foodscapes in the station, besides the most abundant class in LUMGS, sampling for botanical composition was restricted to that area. Eighteen transects of 100 m were randomly positioned in the “grassland” areas on LUMGS, and five (1 × 1 m) quadrats were equally distributed over each transect for the visual evaluation and register of the absolute cover percentage of species present on each quadrat (Ellenberg and Mueller-Dombois, 1974).

Sampling points of the botanical composition surveys were georeferenced with a GPS (GPSMAP 76S Garmin, 3 m), indicating the location and name of plant species. Fourteen plants from our database were located during the botanical composition and georeferenced – six from the ones selected for chemical analysis (grasses, herbs, or forbs) and eight other plants from our database that were not selected for further analysis (Table 2). Although Trifolium dubiom, Trifolium glomeratum, and Phormium tenax are known to be present in LUMGS, there were not detected during the botanical composition sampling; and, therefore, were not included in the healthscape maps.

The geographic coordinates of sampling points were used to map plants over a high-resolution image at ArcGIS/Arcmap 10.7.1^® software. The high-resolution base dataset was captured (flighting at 1,550 m) from LUMGS during Summer 2020 in terms of RGBN 16-bit (Red, Green, Blue, NIR) multispectral imagery at 12.5 cm of ground sampling distance. By using advanced image classification methods (machine learning and object/pixel-based classification) at ArcGIS, plant species locations were drawn as polygons over pixels values of the RGBN image to create a training dataset (signature file). By using the train support vector machine classifier tool, the training dataset was used to train a machine-learning model and create a Esri classifier definition (.ecd) file to recognize patterns over the RGBN image in terms of plant species distribution. The support vector machine-learning tool is a supervised learning method based on statistical learning theory (Vapnik, 1998). The “classify raster” tool was used to extrapolate the information from the .ecd file over the areas considered as “grassland” according to LCDB.

Healthscape maps were created following three different approaches. The first one was based on the spatial distribution of plants and the other two were based on the plant composition according to the chemical analysis.

For the first healthscape map, only the spatial distribution of plants was considered. We mapped the plants geolocated from the botanical composition, plus the vegetation classes from LCDB, where shrubs and trees from our database were located (Table 1), according to local knowledge—Coprosma propínqua, Salix spp, Buddleja davidii, Kunzea ericoides, and Rosa rubiginosa.

The other two maps were created based on the potential distribution and location of either plant phenolic compounds that have potential medicinal effects or fatty acids variety, having the spatial distribution of plants and the chemical analysis results as the approach. For the phenolic medicinal effects maps, we choose to illustrate four different medicinal effects from the PSC found in the analyzed plants according to the literature: antioxidant, antimicrobial, and anti-inflammatory. For each of those effects, plants were ranked hierarchically according to their total concentration of all compounds that show that effect. Thus, four different maps were created in ArcGIS/Arcmap 10.7.1^® software to illustrate the location of plants according to the concentration of compounds that have potential medicinal effect. The same process was used for the fatty acids maps, but the plant rank was based on their total concentration of short, medium, long, or very long chain fatty acids, and saturated, monounsaturated, or polyunsaturated fatty acids.

Primary chemical composition of plants (DM, CP, NDF, ADF, Ash, and WSC) is presented in Table 3. The results indicate numerical divergencies for all the parameters analyzed. Protein was on average 13.96%, ranging from 6.80 to 26.65% for Gaultheria depressa and Salix spp., respectively. Buddleja davidii, Hieracium pilosella, Rosa rubiginosa, Salix spp., Taraxacum officinale, and Trifolium dubium had minimum values of NDF (±20.17%) and ADF (±15.10). Anthoxanthum odoratum, Dactylis glomerata and Phormium tenax had high values of NDF (±62.24%), but only Phormium tenax had a high value of ADF (44.18%). Ash ranged from 3.29% (Phormium tenax) to 14.43% (Anisostome flexuosa) and the highest values of WSC were found in Hieracium pilosella and Phormium tenax (±26.34%), while Anisostome flexuosa, Buddleja davidii, Gaultheria depressa, Kunzea ericoides, Rosa rubiginosa, Salix spp., Trifolium dubium and Trifolium glomeratum showed minimum values (±6.08%), with remaining plants (Anthoxanthum odoratum, Coprosma propínqua, Dactylis glomerata, Taraxacum officinale) showing medium values (±11.13%).

Table 4 presents the DPPH activity and total polyphenols content of each of the analyzed plants. In total, 29 polyphenols compounds were found: Gallic acid, Vascalagin, Castalagin, Procyanidin B2, Hydroxybenzoric acid, Caftaric acid, Chlorogenic acid, Catechin, Cyanidin-3-glucoside, Syringic acid, Caffeic acid, Cyanidin-3-rutinoside, Epicatechin, Malvidin-3-O-glucoside, Epigallocatechin gallate, p-Coumaric acid, Ferulic, Sinapic acid, Rutin, Epicatechin gallate, Quercetin-3-glucoside, Apigenin-7-glucoside, Cinnimic acid, Resveratrol, Luteolin, Quercetin, Apigenin, Diosmetin, and Kaempferol. The number of individual compounds in each plant is available in the Supplementary Table S1.

Fatty acids profiles of the selected plants are presented in Table 5. All plants had a high concentration of C18:3 c9,12,15, apart from Anisostome flexuosa, which presented the highest concentration in C18:2 c9,12. No plant contained C7:0 fatty acids, while some fatty acids were found only in one or two plants. For example, C4:0 was found only in Anisostome flexuosa and Kunzea ericoide. Anisostome flexuosa was also the only plant containing C18:2 t9,12 and C20:3 c8,11,14 fatty acids, while C11:1, C12:1, C14:1 t9, and C15:0 were only found in Kunzea ericoides. Dactylis glomerata was the only plant presenting C17:0 iso fatty acid and C24:1 c15 was only found in Taraxacum officinale.

Healthscape maps are presented in Figures 3–5. Figure 3 presents the distribution and location of phytochemical-rich plants with potential medicinal use in LUMGS. Plants from grassland areas were well distributed along all the station, showing a high diversity, with an exemption of Anthoxanthum odoratum, which is more abundant in the bottom and right boundary of grassland areas. Among all the plants in grassland, the highest abundance is noticed for Hieracium pilosella, through the presence of big clusters. Low abundance is noticed for Geranium mole, Anisotome flexuosa, Agrostis capilares, Verbascm Thapsus, Taraxacum magellanicum, Taraxacum officinale, Cirsium vulgare, and Malva neglecta. Other areas of the station, containing shrubs and trees are concentrated in the bottom parts of the station and around it.

Figures 4–6 present the distribution and location of the plant phenolic compounds that have potential medicinal effects and fatty acids variety in terms of chain structure and saturation, respectively, according to concentration in the selected plants. The phenolic maps depict different medicinal effects from different phenolic properties. The highest concentrations of each effect in each map are represented with dark colors, while the lowest concentrations are represented with light colors, according to the location and distribution of the plants containing the specific medicinal effects. Although depicting different effects, the phenolic maps show strong similarities to each other, with slight differences coming from plants with a lower concentration of each effect. The fatty acids maps follow the same scheme, a gradient color variation from dark to light representing the highest concentration to the lowest concentration of fatty acids in terms of chain structure and saturation. Plants showed a high variance of fatty acids profile, thus resulting in very divergent maps.

Healthscape maps depict different perspectives from plants with multiple benefits for animals, humans, and the environment, presenting a new strategy for illustrating and designing heuristically healthier and more holistic pastoral livestock production systems, where plants are seen not only as food source, but also as medicines. For example, ewes during sensitive periods to gastrointestinal parasitism, such as the peripartum could be placed in paddocks with plants rich in anthelmintic properties to have the effects of parasitic burdens attenuated. Animals during the weaning transition could graze plants with properties that enhance their immune system or plants that increase antioxidant activities and reduce oxidative stress to help them go through this challenging period. Sick animals could be separated from the herd to strategically graze plants with specific properties to treat the disease, such as antibacterial properties. Similarly, finishing animals could have diets rich in plants with long-chain polyunsaturated fatty acids to produce higher quality and nutraceutical meat. These examples illustrate some of the benefits of using healthscape maps as a new facilitating management strategy for the inclusion of different plants for different medicinal purposes to create healthier animals.

The mapped plants are rich in PSC with anti-inflammatory, antiseptic, antidiabetic, anti-obesity, antioxidant, antiviral, antimicrobial, anthelmintic, anticancer, antifungal, and antiprotozoal activities that act in the most varied systems in the body (Sharma et al., 2013; Chevallier, 2016; Díaz et al., 2017; Ganeshpurkar and Saluja, 2017; Naveed et al., 2018; Frutos et al., 2019; Mazur et al., 2019), with positive effects extended to prevent and treat animals from diseases (Villalba et al., 2014; Reddy et al., 2020), to enhance welfare (Chauhan et al., 2014; Beck and Gregorini, 2020), and productive efficiency, while reducing environmental impacts (Jayanegara et al., 2012; Peyraud and Delagarde, 2013; Hoste et al., 2015). As some exogenous factors constraint the ability of animals in consuming and choosing a diverse feed, such as paddock fencing and variability of feeding sites, the knowledge of the spatial diversity of plants in terms of medicinal properties composition would facilitate animal access to a particular/desirable site (Zanon et al., 2022), being a new management tool to positively manipulate animals within a phytotherapeutic diverse foodscape. For example, animals can be managed to eat plants that reduce their stress and, thus, enhance their welfare, which is a big concern within livestock production systems (Fisher, 2020). Farmers can also use the maps to manipulate ruminants to graze prophylactic plants that enhance their immunity against disease infections or plants that act as treatments against infectious pathogens and parasitism. This would also reduce the use and dependency on drugs such as antibiotics and vermifuges, which besides being an economic issue in livestock production systems, have created microorganism resistance, thus reducing drugs efficiency, and generating residues in milk and meat which consequences are hazardous to human health (Beyene, 2016). Furthermore, as some compounds have other nutraceutical properties, such as modifying rumen fermentation as aforementioned, animals could be managed to eat plants that reduce CH₄ production and/or N excretion, thus benefiting the environment health.

A phytotherapeutic diverse diet enhances not only ruminant health but also human health (Gregorini et al., 2017). Besides the aforementioned compounds, the plants also present different fatty acids profiles, some of which are essential for humans, as they are not synthesized by the human body and have positive effects in the prevention and treatment of many diseases due to their anticarcinogenic, anti-diabetic, anti-inflammatory, anti-obesity and anti-atherogenic activities (Rajaram, 2014; Blondeau et al., 2015; Yang et al., 2015; Fuke and Nornberg, 2017). Thus, animals consuming medicinal plants can not only self-medicate but also incorporate those compounds in tissues enhancing the quality and nutraceutical composition of meat and dairy that will feed and, therefore, medicate humans (Wrage et al., 2011). For example, animals can be managed to produce meat with potential anticarcinogenic properties to prevent or ameliorate cancer incidence in consumers (Van Vliet et al., 2021), or meat with a greater content of fatty acids that humans are not able to synthesize.

The creation of specific maps identifying fatty acids with different structures can be of great use as their biological functions, metabolism, source, and effects change accordingly. Polyunsaturated fatty acids (PUFAs), for instance, are only produced by plants and phytoplankton and must be included in the animal diet as they are a source of the endogenous synthesis of further fatty acids, such as Eicosapentaenoic acid (EPA; 20:5o-3) and docosahexaenoic acid (DHA; 22:6o-3) that have essential effects on animal health (Rustan and Drevon, 2005). It is also well known that replacing saturated fatty acids (SFAs) for MUFAs or PUFAs in human diets reduces the risk of cardiovascular diseases (Briggs et al., 2017), thus, the healthscape maps could also be used to reduce the presence of specific plants on the ruminant diet, such as plants with a high abundance of SFAs. As animal products are influenced by their diets and thus, influencing human health (Provenza et al., 2019), the fatty acids maps would contribute to the design of animal products with enhanced beneficial fatty acids and reduced undesirable fatty acids toward the enhancement of not only animals but human health as well. All the healthscape maps bring valuable information in terms of plant benefits for animals, humans, and the environment, and could, therefore, be cooperatively used for enhanced multi purposes achievements.

Our methodology is just a starting point. We cited the benefits of only some of the powerful constituents plants may have, but actually, plants contain hundreds or thousands of different constituents that interact in complex ways, which details and benefits are still not understood fully. In addition, it is known that some plant properties effects might be opposite at different doses (Chevallier, 2016). Although animals can select plant properties at self-regulated safe levels of intake (Provenza, 1996), a better knowledge of the most abundant plant properties and their interaction with other plants would facilitate animal decisions during grazing and animal access to specific plants in terms of offering them adequate doses of adequate plants according to their physiology. It is also important to understand the actual effect of plants and arrange plants in different doses on ruminants, as the effects of PSCs are dose dependent (Raubenheimer and Simpson, 2009), which should be subject to further research.

The spatial reference taken in this methodology approach refers to the heterogeneity in terms of plant species across the case study station, which is the reason why we pooled the subsamples of our plants together to represent each specie selected. However, as the concentration and arrangement of PSC, as well as its medicinal and nutritional effects, change across plants and parts of plants (Provenza, 1995; Villalba et al., 2017), keeping subsamples from different locations at the station, or different parts of the same plant separately and statistically comparing their PSC concentration would enrich the spatial information used in the building of the maps. Similarly, we applied the methodology for only one season—Summer. As the concentration of PSC changes with the phenology of plants (Kuhnen et al., 2021), the growth pattern of plants and seasonal influence on their PSC content and effects would further inform management decisions to enhance the particular effects of some dietary PSC on the health and wellbeing of the animals. Therefore, we strongly recommend the inclusion of temporal variability on PSC synthesis for future studies.

We also highlight the importance of understanding the effect of herbivory on plants. Ruminants may induce the development of some swards while depriving the survival of others, due to their highly selective activity, affecting the diversity of plants and, therefore, threatening the stability and resilience of the system (Wrage et al., 2011). If the distribution, composition, and growth pattern of plants along the different seasons is known, grazing can be adjusted for the available plants in space and time according to animal feeding patterns and selectivity to better manipulate the vegetation dynamics and improve the diversity, stability, and resilience of the system. This highlights the need of creating different healthscapes for different environments not only spatially but also in a temporal context.

This study provides a novel approach to the study and management of grasslands by incorporating spatial dimensions to plant chemistry. The methodology presented in this study emerges as a heuristic tool for the design and analysis of modern pastoral livestock production systems to promote health as a whole. As an innovative approach, we caution the reader on the need for proper adjustment of building variables and analyses according to the particular contexts and goals.