Crude Protein as an Indicator of Pasture Availability and Quality: A Validation of Two Complementary Sensors

Abstract

This study evaluated the possibility of using two complementary electronic sensors (rising plate meter (RPM) and active optical sensor (AOS)) to obtain a global indicator, pasture crude protein (CP) in kg ha⁻¹. This parameter simultaneously integrates two essential dimensions: pasture dry matter availability (dry matter (DM) in kg ha⁻¹) measured by RPM, and pasture quality (measured by AOS), and supports management decisions, particularly those related to the stocking rates, supplementation, or rotation of animals between grazing parks. The experimental work was carried out on a dryland biodiverse and representative pasture, and consisted of sensor measurements, followed by the collection of a total of 144 pasture samples, distributed between three dates of the pasture vegetative cycle of 2023/2024 (Autumn—December 2023; Winter—February 2024; and Spring—May 2024). These samples were subjected to laboratory reference analysis to determine DM and CP. Sensor measurements (compressed height (H_RPM) in the case of RPM, and normalized difference vegetation index (NDVI) in the case of AOS) and the results of reference laboratory analysis were used to develop prediction models. The best correlations between CP (kg ha⁻¹) and “H_RPM × NDVI” were obtained in the initial and intermediate phases of the cycle (autumn: R² = 0.86 and LCC = 0.80; and Winter; R² = 0.74 and LCC = 0.81). In the later phase of the cycle (spring), the accuracy of the forecasting model decreased dramatically (R² = 0.28 and LCC = 0.42), a trend that accompanies the decrease in the pasture moisture content (PMC) and CP. The results of this study show not only the importance of extending the database to other pasture types in order to enhance the process of feed supplement determination, but also the potential for the research and development of proximal and remote sensing tools to support pasture monitoring and animal production management.

1. Introduction

In extensive livestock systems of Iberian Peninsula, pastures are the main feed source of ruminants, especially sheep and beef cattle [1]. Pasture availability, expressed in terms of dry matter per unit area (DM, in kg ha⁻¹), and pasture quality, expressed in terms of crude protein (CP, in %), are two key parameters for decision-makers in animal production systems, particularly grazing strategies in terms of the stocking rate management, supplementation, or rotation of animals between grazing parks [2].

Frequent and accurate measurements or estimations of pasture availability and quality are fundamental for maximizing grass production and utilization in pasture-based farms through better management decisions [3]. There is significant potential for improving the availability and accuracy of grassland information by means of precision agriculture (PA) technologies. However, the application of technologies to grassland farming is a challenge of great complexity, especially due to the greater diversity within grassland in terms of the spatial variation of soil and pasture characteristics, and the highly temporal dynamics of grass species [4]. Nevertheless, there are many studies published over the last twenty years on the evaluation of technologies for estimating and monitoring pasture availability [2,5,6,7,8], some of which use proximal or contact sensors. The rising plate meter (RPM) is the most frequently sensor used for pasture DM monitoring in several countries, like Chile and New Zealand [7], Australia [8,9], or Ireland [3,10]. The RPM records a combined measure of pasture height and density, referred to as pasture compressed height (H_RPM), using a weighted disc attached to a scaled staff that is dropped onto the pasture, [3] and, based on this, estimates the DM [11]. The manufacturer of “Jenquip EC20” RPM, for example, proposes a generic (annual) equation (Equation (1)), adjustable to the evolution of the vegetative cycle (seasonal variation) on a monthly basis [9]. Bluetooth-enabled plate meters, which streamline and automate aspects of data collection to generate pasture budgets, are commercially available and can be used in association with internet-based grassland management decision support tools (DST) [11].

$$\mathrm{DM}={\mathrm{H}}_{\mathrm{RPM}}\times 140+500$$

(1)

where DM is the dry matter yield (in kg ha⁻¹) and H_RPM is the compressed height (in cm) measured by the RPM.

According to Murphy et al. [12], the important parameters for evaluating pasture quality in grazing systems are DM, dry matter digestibility, metabolizable energy, organic matter digestibility, crude protein (CP), and water-soluble carbohydrates. For Lugassi et al. [13], chemical composition, water content, and nutrient concentration are widely used as indicators of pasture nutritive value. The CP and neutral detergent fiber (NDF) content stand out among the most common chemical parameters for the purpose of evaluating pasture quality [13,14]. High CP content and low fiber content are characteristics of pastures of high quality [13,14,15].

The conventional measurement methods for pasture quality determination require the cutting of herbage samples and laboratory processing, with high costs in terms of time and chemicals, and can take several days to complete [3] and, consequently, miss the opportunity to intervene [14]. The need to reduce the cost and time associated with these laboratory determinations has led to the development of many studies to evaluate technologies that allow for the rapid pasture quality estimation. The most usual technologies are based on near-infrared (NIRS) spectroscopy or optical sensors. The successful application of NIR spectroscopy in pastures has taken place both in conventional lab-based determinations, with bench-top equipment and physical pre-processing of the pasture samples [16,17,18,19], as well as in situ (on field) and non-destructive determinations, with portable (hand-held) NIR spectrometers [20,21,22]. This NIR equipment (bench-top or portable) is, however, expensive and requires a complex multivariate and modeling data analysis (chemometrics) [12], inaccessible to farmers as a means of supporting extensive animal production [18,19].

Access to information on pasture quality through hyperspectral and multispectral optical sensors has been greatly developed in the last decade through remote sensing supported by platforms installed mainly on drones [23] and satellites [24]. These advancements based on RS can offer moderate-to-high resolution pasture mapping, and highly accurate pasture quality prediction, but there are some limitations. The frequent occurrence of clouds [8] during the vegetative period of dryland pastures in the Mediterranean region (autumn, winter, and spring) and the predominance of agro-silvo-pastoral systems, where pastures are developed under tree canopy (cork oaks and holm oaks in particular in the Alentejo region, Southern of Portugal), open up good prospects for proximal or contact sensors, which overcome the difficulties inherent to the use of RS approaches [6].

There are also a range of commercial optical spectral technologies used for the ground-based measurement of pasture quality (e.g., “GreenSeeker” or “OptRx”). These portable optical sensors are more affordable in terms of the acquisition costs and have more user-friendly outputs than portable “Micro-NIR” sensors. Optical sensors obtain vegetation indexes through calculating the ratio of different spectra of light [25]. NDVI (normalized difference vegetation index) is one of the most common of these indexes [8,12,25]. NDVI estimates the quality of vegetation based on the ratio of red and NIR light wavelengths that are absorbed by pasture photosynthesis [12]. The multispectral “OptRx” (Ag Leader, Ames, IA, USA) sensor, for example, emits radiation in a modulated frequency and receives back the radiance in the different channels. With this sensor, it is possible to obtain the following radiance bands: (i) Red (670 nm); (ii) Red Edge (728 nm); and (iii) NIR (775 nm). NDVI can be calculated with the Red and the NIR bands, using Equation (2).

$$\mathrm{NDVI}={\displaystyle \frac{\mathrm{NIR}-\mathrm{RED}}{\mathrm{NIR}+\mathrm{RED}}}$$

(2)

NDVI is sensitive to changes in plant yield/maturity, which are the drivers of nutrient concentration changes and strongly correlated with pasture CP (positive correlation) and NDF (negative correlation) [26]. The study of Serrano et al. [6], carried out on pastures in the Montado ecosystem in Southern Portugal, showed the potential of the “OptRx” sensor with a strong coefficient of determination (R² = 0.75) between NDVI measured by the “OptRx” sensor and CP determined in reference laboratory analyses.

Pasture availability is usually expressed in terms of kg of DM per unit area (DM, in kg ha⁻¹), and pasture quality, for example, in terms of CP content, is usually expressed in percentage of DM (CP, in % DM). As mentioned above, these are two key parameters for decision-makers in animal production systems. Pasture availability and quality are parameters that have a specific pattern throughout the vegetative cycle of the dryland pasture in Mediterranean climate conditions, although with inter-annual variations depending on rainfall distribution: (i) DM tends to increase from the first rains (usually in September/October) until the peak of autumn production (usually in November or early December); CP is usually maximum at this stage when the plants are young (approximately 25–30%, depending on the type of pasture); (ii) a period of relative vegetative dormancy follows between mid-December and mid-February, corresponding to reduced pasture growth; CP shows a downward trend; (iii) from March until the middle or end of May, the gradual rise in air temperature, as long as it is accompanied by rainfall, leads to an exuberant peak of pasture production (spring production peak); at this stage, CP shows a more noticeable drop, of up to 10–12%; (iv) when high temperatures are accompanied by little or no rainfall, which often happens between the end of May and the beginning of June. The pasture dries out abruptly, losing both availability (DM) and, fundamentally, quality; CP reaches minimum values of around 5–6%. However, the exact timing of this pattern varies from year to year as a result of climatic irregularities.

It is in this complex scenario of unpredictability that the farmer has to manage the stocking and animal feed, and thus the availability of tools that allow for quick access to relevant information is fundamental for ensuring the economic and environmental sustainability of these extensive production systems. A good understanding of the dynamic pasture growth cycle, biomass accumulation, and nutritional value is essential for their efficient management [7]. On the one hand, DM is an indicator that helps adjust systems and establish pasture management criteria [27], but it is also very important to know the quality of the pasture in order to assess whether it covers the nutritional requirements of animals, which in turn helps determine more accurately the stocking rate per hectare [28].

Based on this assumption, this study evaluated two complementary sensors (the rising plate meter (RPM) and an active optical sensor (AOS)) to obtain a global indicator, CP in kg ha⁻¹, that simultaneously integrates two essential dimensions, pasture availability (related with the compressed height measured by RPM) and pasture quality (related by NDVI measured by optical sensor).

2. Materials and Methods

2.1. Experimental Field

The determinations (sensor measurements and pasture sampling) were carried out during the 2023/2024 growing season on “Eco-SPAA” field (Figure 1), at Mitra farm (University of Évora; 38°53.10 N; 8°01.10 W). This is a pluriannual and biodiverse dryland pasture (mixture of various botanical species, legumes, grasses, composite and other spontaneous species), grazed by sheep and integrated into the Montado ecosystem, under holm oak, improved by the regular application (approximately every 2 years since 2015) of dolomitic limestone and binary fertilizer (nitrogen and phosphorus). In this field, 48 sampling areas have been georeferenced within the framework of an experimental study about sheep grazing systems (continuous versus deferred), started in 2021. The sampling scheme of this study was identical to that of a previous study, consisting of 48 georeferenced sampling points (12 in each of the four grazing parks), defined by a botanical expert according to the floristic composition of the pasture in the previous growing season.

In this pasture, representative of the typical dryland pastures in the region, the predominating species are Erodium mochatum, Diplotaxis catholica, Trifolium repens and other spontaneous species which are native or adapted to the region’s climate (Figure 1). The evolution of the pasture in these dryland conditions is determined by the distribution of rainfall and temperature, and conditioned by grazing. Figure 2 shows the temperature and precipitation data (monthly mean temperature and rainfall) of the meteorologic station of Évora (about 10 km from the Mitra farm) between July 2023 and June 2024.

2.2. Equipment

In this work, the following equipment was used:

-

An electronic stud to measure pasture height (Figure 3a);

-

A “bluetooth-enabled Jenquip EC20 electronic platemeter”, RPM (Jenquip, 21 Darragh Road, Feilding, New Zealand; Figure 3b) to measure pasture compressed height (H_RPM);

-

A proximal and active optical sensor “OptRx” manufactured by Ag Leader (2202 South Riverside Drive, Ames, IA 50010, USA; Figure 3c);

-

A Trimble GNSS, GeoExplorer 6000 series receiver, model 88951, with sub-metric precision (GmbH, Am Prime Parc 11, 65479 Raunheim, Germany: Figure 3d);

-

Quadrats of 0.50 m by 0.50 m area to delimit pasture sampling areas (Figure 4a);

-

Portable electric grass shears (Figure 4b) and plastic bags for storing the pasture samples.

2.3. Field Measurements, Pasture Sample Collection, and Analyses

Measurements were carried out between December 2023 and May 2024. Field tests (sensor measurements and pasture collection) were carried out in these 48 areas at 3 dates: I-06DEC2023 (Autumn), II-29FEB2024 (Winter), and III-10MAY2024 (Spring).

The measurements in the 48 sampling areas were made according to the following procedure:

-

An operator, walking slowly, performed measurements within the sampling area with the “OptRx” sensor (associated with the Trimble GNSS receiver), placed about 0.5 m above the pasture. The NDVI values were registered for a 2 min period, so, on average, at each sampling date, 120 NDVI measurements were taken (one per second).

-

Once the optical sensor operator had finished each site, a second operator placed three sampling rings (0.5 m × 0.5 m, corresponding to a sampling sub-area of 3 × 0.25 m²) and measured pasture height (H) with an electronic stud (3 measurements for each sub-sampling area).

-

Then, the same operator took 3 measurements of pasture compressed height (H_RPM).

-

After the measurements with the sensors, the pasture of each sub-sampling area was cut and collected in a plastic bag, identified with the respective code.

On each date, these 48 composite pasture samples were transported to the laboratory where they were weighed, dried in an oven, and weighed again to obtain the pasture availability (in terms of green and dry matter mass per unit area, respectively, GM and DM, in kg ha⁻¹) and PMC (in percentage). The dried samples were analyzed to determine pasture CP content (in % of dry matter) using standard methods [29].

2.4. Data Analysis

The data obtained in each date were organized in an Excel spreadsheet to calculate descriptive statistics parameter (mean, standard deviation, and range).

Crude protein (CP), expressed in kg ha⁻¹, was calculated for each sampling area using Equation (3):

CP (kg ha⁻¹) = DM (kg ha⁻¹) × CP (%)/100

(3)

The pasture availability and quality index (PAQI) was calculated based o Equation (4):

PAQI = H_RPM (mm) × NDVI

(4)

Regression analysis was used to evaluate the relationship between variables: H and H_RPM; H and DM; H_RPM and DM; PAQI and DM and PAQI and CP (in kg ha⁻¹). The coefficient of determination (R²) was used to evaluate the goodness of fit for the regression models. Moreover, Lin’s concordance correlation (LCC) coefficient was used as the measure of each model precision and accuracy. LCC, unlike the Pearson correlation coefficient which only assesses linear relationships, accounts for precision, i.e., how well the values covary, considering their standard deviations, and accuracy, i.e., how close the predicted values are to the observed values in terms of their means. Consequently, LCC provides a more comprehensive measure of the overall agreement [30].

The georeferenced information collected from the “Eco-SPAA” field on the 3 dates (I, II, and III) was processed using the ARCGIS v. 10.5 GIS software. Geostatistical analysis (ordinary kriging considering stationarity within the local neighborhood) with the Geostatistical Analyst extension was used to obtain the respective maps representing the spatial variability of the variables of interest throughout the experimental field.

3. Results

3.1. High Pasture Spatial and Temporal Variability

The database of this work is summarized in

Table 1. Descriptive statistical (mean, standard deviation, coefficient of variation, and range) of pasture and sensor measurements.

Date		Date 1 (06DEC2023)		Date 2 (29FEB2024)		Date 3 (10MAY2024)
Pasture Parameter	n	Mean ± SD (CV, %)	Range	Mean ± SD (CV, %)	Range	Mean ± SD (CV, %)	Range
GM (kg ha−1)	48	6254 ± 4274 (68.3)	1423–21,090	13256 ± 7054 (53.2)	3867–27,917	8245 ± 4759 (57.7)	2580–25,037
DM (kg ha−1)	48	571 ± 257 (45.0)	237–1703	1834 ± 745 (40.6)	580–3460	2353 ± 1108 (47.1)	933–5400
PMC (%)	48	89.0 ± 3.3 (3.7)	79.7–93.4	84.6 ± 3.2 (3.8)	75.2–90.5	69.9 ± 6.9 (9.8)	51.9–80.2
CP (%)	48	24.3 ± 5.3 (21.7)	13.5–34.8	16.4 ± 3.6 (21.9)	9.0–23.5	11.3 ± 1.8 (15.7)	8.1–15.4
CP (kg ha−1)	48	140.6 ± 75.7 (53.9)	38.8–386.1	305.0 ± 153.3 (50.2)	74.0–626.6	260.6 ± 119.4 (45.8)	108.2–650.7
NDVI	2880	0.817 ± 0.057 (7.0)	0.663–0.898	0.762 ± 0.058 (7.6)	0.605–0.840	0.537 ± 0.098 (18.2)	0.313–0.765
H (mm)	144	105.3 ± 71.2 (67.6)	14.0–400.0	163.5 ± 83.5 (51.1)	30.0–420.0	249.6 ± 140.9 (56.5)	60.0–660.0
HRPM (mm)	144	53.1 ± 35.4 (66.7)	14.0–232.0	103.6 ± 55.3 (53.4)	22.0–244.0	77.8 ± 42.2 (54.2)	22.0–240.0
PAQI	144	44.8 ± 25.1 (56.1)	13.7–108.4	79.2 ± 38.3 (48.4)	21.2–154.5	41.3 ± 20.9 (50.6)	9.1–95.0

SD—standard deviation; CV—coefficient of variation; GM—green matter; DM—dry matter; PMC—pasture moisture content; CP—crude protein; NDVI—normalized difference vegetation index; H—height; H_RPM—compressed height, measured by the rising plate meter (RPM); PAQI—pasture availability and quality index.

. The spatial variability of the various pasture parameters measured by sensors (H, HRPM, NDVI) or obtained via laboratory analysis (GM, DM, PMC and CP) is reflected in the high CV, especially the GM, DM, and CP (in kg ha⁻¹) of H and H_RPM (of the order of 40–70%). On the other hand, temporal variability (between measurement dates) is also very significant, with average DM increasing from 571 kg ha⁻¹ in December, to 1834 kg ha⁻¹ in February and 2353 kg ha⁻¹ in May, while the PMC and CP show an inverse evolution (PMC: 89%, 85% and 70%; CP: 24%, 16% and 11%; respectively, in December, February, and May).

The evolution of pasture availability and quality throughout the vegetative cycle is determined by meteorological conditions (especially temperature and precipitation) and conditioned by grazing management (continuous, deferred, stocking rate, among other aspects). Figure 4 shows a relatively balanced distribution of precipitation in autumn and winter, which provided conditions for high pasture productivity.

3.2. Relationship between Variables

The expected relationship between pasture height (H, measured with an electronic stud) and pasture compressed height (H_RPM, measured with the RPM) was assessed. Figure 5 shows that this relationship was significant on all three monitoring dates (R² between 0.67 and 0.78, and LCC between 0.81 and 0.88). The H/H_RPM ratio tends to increase toward the end of the growing season (spring).

The estimation of pasture availability (DM) from height (H) or compressed height (H_RPM), shown in Figure 6, was also significant on all monitoring dates, with a tendency for R² and LCC to drop sharply at the final phase of the growing season (spring). The results also showed greater accuracy of the estimation based on compressed height (H_RPM) than that based on the height measured by the electronic stud (H). This difference was consistent across all assessment dates.

The use of the PAQI, which combines the component height (by H_RPM) and the component quality (by NDVI), helped improve the accuracy of DM estimation (Figure 7), particularly in the initial phase (Autumn) and the intermediate phase (Winter) of the pasture’s vegetative cycle: the R² increased from 0.70 to 0.75 (Figure 6d and Figure 7a, respectively), and from 0.82 to 0.85 (Figure 6e and Figure 7b, respectively).

The same index (PAQI) also showed interesting accuracy in estimating CP in terms of kg ha⁻¹ (Figure 7), especially in the initial phase (Autumn) and the intermediate phase (winter) of the pasture’s vegetative cycle (Figure 7d,e). The PAQI index and the CP parameter simultaneously represent pasture availability and quality.

The association of a GNSS receiver with proximal or contact sensors allows for the georeferencing of the field information. This geographic georeferencing, associated with geostatistical mapping (interpolation by kriging in this case), has led to the establishment of maps showing the spatial pattern of pasture parameters. Figure 8, Figure 9 and Figure 10 show the maps relating DM with HRPM, DM with PAQI, and CP with PAQI, respectively. These maps reflect the relationship between these parameters identified by the respective coefficients of determination, so that they capture not only the spatial variability, but also their temporal evolution. The maps also clearly show that the variability patterns are more accurately captured in the initial (autumn) and intermediate (winter) phases of the pasture’s vegetative cycle, with a significant drop in the final phase (spring).

4. Discussion

4.1. Pasture Spatial and Temporal Variability

The spatial variability of productivity and quality is a common pattern in dryland pastures integrated into the Montado ecosystem [1]. In addition to the biodiversity of species and botanical families, the presence of trees and grazing animals accentuates this variability [4]. This variability is due to a number of factors, including soil, environment, seasonal changes in sward morphology, and grazing management (e.g., stocking rates, herbage allowance, selective grazing, dung pats) [12].

In our study, spatial variability is further promoted by the logistics management of the experimental field in the context of a research project on sheep grazing and pH amendment. The field was divided into four sub-plots of 1 ha each, with two treatments: (i) application versus non-application of dolomitic limestone; and (ii) two different grazing intensities (stocking rates of 1 LU ha⁻¹ and 2 LU ha⁻¹). Each one of these factors has a significant impact on the pasture [1]. For this reason, the high CVs found in this study in pasture height, availability, and CP content (in the order of 40–70%—Table 1) are reflected in the maps of Figure 8, Figure 9 and Figure 10. This spatial variability was also referred to in various works involving dryland pastures [15,31]. Murphy et al. [12] present an overview of the most recent research pertaining to the development of precision grass measurement technologies and indicate that the values of mean sward heterogeneity in terms of pre-grazing grass quantity in temperate grasslands varies between 25 and 46%. The high CV of measurements within fields, resulting from the interaction between sensors and the heterogeneity of pasture vertical profile, are a considerable source of error of estimation [12]. The factors reported to affect this interaction include grass species, season, and grazing intensity [3,12,32].

Heterogeneity increases the difficulty of monitoring pasture DM yield and quality, and requires a large number of samples to account for spatial variation within grazed pastures. Murphy et al. [12] suggest that sample locations should be randomly selected and spatially balanced throughout a pasture. For the purpose of this work, the sampling scheme was established with the georeferencing of 48 sampling points (12 in each of the four grazing parks), defined by a botanical expert according to the floristic composition of the pasture in the previous growing season. These points were maintained on all pasture sampling dates (autumn, winter, and spring). In order to further eliminate operation bias, which is a significant source of measurement error [12], the field sampling team (measured with the sensors and taking pasture samples) was also maintained constant throughout the experimental period.

The temporal variability (between measurement dates) was also very significant in our study. The average DM increased from 571 kg ha⁻¹ in December, to 1834 kg ha⁻¹ in February and 2353 kg ha⁻¹ in May, while the PMC and CP showed an inverse evolution (PMC: 89%, 85% and 70%; CP: 24%, 16% and 11%, respectively, in December, February, and May). In dryland pastures, the evolution of availability and quality throughout the vegetative cycle (temporal dynamic) is determined by meteorological conditions and is conditioned by grazing management [1]. In the agricultural year under study (2023/2024) rainfall occurred evenly during the months of Autumn and Winter (approximately 300 mm of accumulated rainfall in autumn—between September and December, and also in winter—between December and March), which ensured both the first peak of production in late autumn and the maximum accumulated production in spring, phases which greatly determine the productive potential of the pasture [1]. In our study, pasture measurements were taken between December and May to ensure the representability of these determinant phases of pasture vegetative cycle.

The inherent spatial and temporal variability shows, on the one hand, that pasture monitoring is a complex challenge and, on the other, justifies the use of new technologies in grassland farming [4]. The benefits of precision technologies have, however, been relatively slow to be realized in extensive livestock systems [33]. The expeditious collection of information using sensors associated with GNSS technology allows the use of geostatistical procedures, the representation of relevant parameters in the form of maps for spatial analysis and precision agriculture applications [12]. Gargiulo et al. [9] reinforced the interest in integrating pasture measurement sensors with GNSS receivers and mapping tools, providing new opportunities for the regular and efficient collection, and easy processing of large amounts of data. Reducing measurement time and effort is vital not only in saving time and cost for farmers, but also encouraging more farmers to measure grass growth and development on a regular basis [32]. A New Zealand-based study conducted by Beukes et al. [34] found that conducting regular grass measurements can improve farm profits by up to 15% through reduced feed imports and improved grassland management.

4.2. Relationship between Variables

Farmers need information on pasture availability and quality to support decision-making in terms of grazing management (grazing and resting areas, animal stocking rates, etc.) and animal feed supplementation [6,28]. It is common, however, to use the estimation of pasture DM as the indicator of choice for biomass availability to understand the spatio-temporal changes of forage resources in pasture ecosystems and support grazing management decisions [35]. Accurately predicting the biomass available for grazing within pastures is essential for allocating the correct quantity and area of grass to the herd on a daily basis to meet their dietary requirements [3]. According to Alckmin et al. [36], pasture management is highly dependent on accurate biomass estimation.

Most research is focused on testing, calibrating, and evaluating rapid technological tools for estimating pasture DM; there are no established on-farm methods which a farmer can use to estimate pasture quality [12]. Pasture qualitative analysis methods for determining CP or fiber, among others, are typically laboratory-based. Thus, in recent years, several studies based on near infrared spectroscopy (NIRS) have also been carried out to monitor pasture quality [12,18,19].

In our study, the estimation of DM based on compressed height obtained by the rising plate meter (H_RPM) was significant at all monitoring dates, with better accuracy in February (R² = 0.82; LCC = 0.89) and a tendency to drop sharply in the final phase of the growing season (spring). The seasonal variation in the relationship between H_RPM and DM has been well documented in the literature. Several studies have highlighted that variations in the accuracy of biomass prediction can be caused by seasonal, management, and sward characteristic factors [3]. Compressed height is a measure of the combined height and density of the pasture [3]. The relationships between H_RPM and biomass are limited in their accuracy and biased due to plant development stages, canopy architecture (erectophile or plagiophile), or canopy density [37], aspects that are particularly marked in the later stages of the growth cycle of dryland pastures. Morphological changes occur in the plant in the transition from the vegetative to the reproductive stages [1,32]. The limited usefulness of the RPM in the second half of the growing season may also be due to an increasing abundance of senesced material [35].

The estimation approach based on pasture availability (e.g., kg of DM ha⁻¹) or quality (CP, in %) can always be considered incomplete. In practice, the ideal would be an approach that simultaneously integrates both components (availability and quality). To this end, in this study, the RPM was used for estimating the availability dimension (DM) based on compressed height (H_RPM), and the “OptRx” optical sensor was used for estimating the quality dimension (CP) through the calculation of NDVI. Several studies have shown a strong correlation between NDVI and CP [6,38]. For example, Pullanagari et al. [38] obtained an R² between 0.65 and 0.83 in pastures in New Zealand.

Pasture availability and quality index (PAQI), obtained by the product of H_RPM by NDVI (Equation (3)), contributed to improving the accuracy of the DM estimation, particularly in the initial phase (autumn) and the intermediate phase (winter) of the pasture’s vegetative cycle: the R² increased from 0.70 to 0.75, and from 0.82 to 0.85, respectively. Several studies have shown the value of a complementary approach between sensors to predict pasture availability or even pasture quality [39]. These studies include complementarity between proximal sensors (PS), between PS and satellite images (RS), between different spectral configurations of satellite images, and between different RS sources [40]. Gargiulo et al. [9], for example, based on research conducted in Australia, reported a clear improvement in the accuracy of the estimation of biomass using vegetation indexes from satellite images calibrated with an RPM (R² of 0.61 to 0.72). All these options help improve the accuracy of pasture availability estimates, which contribute to maximizing the utilization of fresh grass for animal feed throughout the growing season and consequently to the reduction of whole-farm inputs, emissions, and costs, providing an efficient and sustainable approach for grass-based livestock systems [3].

The PAQI also showed interesting accuracy in estimating CP in terms of kg ha⁻¹, especially in the initial phase (autumn; R² = 0.86; LCC = 0.80) and the intermediate phase (winter; R² = 0.74; LCC = 0.81) of the pasture’s vegetative cycle. The PAQI index and the CP parameter simultaneously represent pasture availability and quality, hence their practical potential for use in grazing management and supplementation in extensive animal production systems [40]. There is potential to increase grassland production in a sustainable manner by achieving a more precise measurement of pasture quantity and quality [12].

With reference to the relatively recent review of Murphy et al. [12] about precision technologies for optimizing pasture management, it can be seen that the evaluation of technologies for the proximal monitoring of pasture availability and quality has been more intensive in countries with a strong vocation for animal production, such as Australia, New Zealand, USA, and, in Europe, Ireland, while there is no mention of any work carried out in the Iberian Peninsula. Regarding the simultaneous use of RPM and “OptRx” sensors, we are not actually aware that this equipment has even been tested on dryland pastures in the Mediterranean region in a similar context (pasture characteristics and climatic conditions). Therefore, no data are available to put into perspective the results of this study, particularly the relationship between PAQI and CP in kg ha⁻¹.

Figure 8, Figure 9 and Figure 10 show the maps relating DM with H_RPM, DM with PAQI, and CP with PAQI, respectively. These maps capture not only the spatial variability, but also their temporal evolution. The variability patterns are more accurately captured in the initial (autumn) and intermediate (winter) phases of the pasture’s vegetative cycle, with a significant drop in the final phase (spring). The greatest accuracy obtained in the early phases of the vegetative cycle can help the farmer decide when to stop the feed supplementation that takes place throughout the summer period (a period without precipitation and, therefore, without pasture availability).

This spatial representation reinforced the interest in integrating pasture measurement sensors with GNSS receivers, an approach that contributes to the interpretation and management of the spatial variability of pastures and grazing, allowing farmers to make prompt and more informed pasture management decisions [2].

4.3. Limitations and Perspectives of This Study

The results of this exploratory study in the Mediterranean region, with dryland pastures characteristic of the Montado (in Portugal; “Dehesa” in Spain), provide good prospects for the RPM as an expedient sensor for estimating pasture DM, complemented with the “OptRx” optical sensor. Through this holistic approach, it is possible to improve the accuracy of DM estimation models and also obtain a pasture availability and quality index (PAQI) that strongly correlates with CP. However, there are two limitations to this study: it was only carried out during one vegetative cycle of the pasture and only on one type of pasture, although it is representative of the mixtures (grasses, legumes, and others) of dryland pastures typical of the Montado (or Dehesa). It is now important to extend the study to more pastures, with different characteristics and over various pasture growing seasons, to gather more detail and significance on the seasonal effects [32], an aspect emphasized by the sensor manufacturer itself with the presentation of different coefficients in the generic regression equation throughout the pasture’s vegetative cycle.

Despite the considerable potential to develop holistic pasture measurement systems, including those based on multi-sensors [40], the value of these technologies to farming systems is unclear, so a user-need-driven development of technologies and a focus on how the outputs arising from precision technologies and associated decision support applications are delivered to users is fundamental to maximize their value [33]. It is, therefore, necessary to conduct new studies involving cost–benefit analysis to determine the efficacy of investing in specific precision technologies at the farm level [12,33].

In addition to the interest in further calibration/validation studies of these sensors in the diverse range of heterogeneous pastures that are characteristic of the region, this work opens prospects for studies that leverage these proximal technologies by associating them with RS, allowing farmers to monitor large areas and determine real-time, in situ pasture availability and quality with little-to-no labor input [38] to support farmers’ decision-making.

5. Conclusions

In extensive livestock systems, pastures are the basis of ruminant feed. In these systems, pasture availability and quality are two key parameters for pasture and grazing decision-making and management. The potential of a complementary multi-sensor approach, an RPM (Jenquip EC20), that measures the compressed height (H_RPM), and an active optical sensor (OptRx), that measures the NDVI, is shown in this study for obtaining an index (PAQI) that simultaneously integrates pasture availability and quality, expressed in terms of kg of CP ha⁻¹.

The best correlation between CP (in kg ha⁻¹), measured by laboratory reference analysis, and PAQI was obtained in the early phases of the growing season (autumn: R² = 0.86; LCC = 0.80; and winter: R² = 0.74; LCC = 0.81), which can help farmers decide when to stop the feed supplementation that takes place throughout the summer period.

The seasonality associated with dryland pastures decisively affected the accuracy of the estimate, tending to decrease dramatically toward the end of the pasture vegetative cycle, which justifies carrying out long-term studies to gather more information on the seasonal effects. These studies, in addition to including other types of grasslands (floristic composition), could also replace proximal sensors by remote sensing (satellite images), making grassland monitoring a more practical and easy-to-operate process.

Creating awareness of the benefits of the on-farm use of precision technologies for the frequent and accurate measurement of pasture availability and quality is one of the main strategies for ensuring the economic and environmental sustainability of these extensive production systems.