Increasing food production while reducing the environmental impacts of agricultural activities will be the major challenges for farmers and the research community in the coming decades (Calicioglu et al., 2019). Farming processes that rely on use the current production models, based on intensive use of inputs and monocultures, will require changes to build more sustainable production models (Carvalho et al., 2018).

The challenge of enhancing the sustainability of production systems requires a comprehensive understanding of their fundamental principles, which extend beyond the environmental dimension. As early as 1999, Gibon had already postulated that sustainability should be addressed through economic, social, and environmental perspectives (Gibon et al., 1999). For example, the efficient use of energy derived from solar radiation has been identified as a key factor in sustainable agriculture (Heitschmidt et al., 1996).

According to Oltjen and Beckett (1996), sustainability lies in the use of animal products that are inedible to humans. In sustainable intensification, principles and practices aim to increase the efficiency of natural resources are postulated, either by increasing synergies among different farming processes or by optimizing the use of these resources through reduced losses (Garnett et al., 2013; Godfray, 2015; Adegbeye et al., 2020).

In Brazil, with 193,93 million head of cattle and 160,54 million hectares of pastures, beef production is predominantly pastoral, only 19.86% of slaughtered animals originating from feedlots (ABIEC, 2023). This characteristic of beef cattle production allows pastures to mitigate the effects of GHG emissions from agricultural activities (Vasconcelos et al., 2018). The Brazilian Pampa biome (BPB) is different from others because of its climate, which determines the abundance of fauna and flora, and livestock plays an important role in the exploitation of its resources. Several studies have highlighted the relevance of this region’s productive potential and ecosystem services (Pillar et al., 2015; Carvalho et al., 2019). According to MapBiomas (2025), among Brazil’s biomes, the Pampa has undergone the highest proportion of degradation relative to its total area. Between 1985 and 2020, approximately 21.04% of its native vegetation cover was lost, leading primarily to the degradation of native grasslands, climate imbalances, and biodiversity loss. Since the 2000s, the impacts of expanding crop areas over livestock areas and their effects on the beef supply chain have been discussed (Barcellos et al., 2004). Many other themes have been studied in the BPB (Marques et al., 2011; Ruviaro et al., 2015; Ruviaro et al., 2016; Freitas et al., 2019; Lampert et al., 2020). However, studies analyzing the three dimensions of sustainability in farming systems of the BPB are scarce.

Sustainability assessment tools have been developed by different global initiatives for supply chains. Among them, the Sustainability Assessment of Food and Agriculture Systems (SAFA) guidelines (FAO, 2013, 2014) establish indicators to measure sustainability at different levels of the agricultural supply chains. Although their full application creates difficulties for sustainability assessments in production systems, developers encourage adaptations to enhance accuracy and robustness in the evaluations. Similarly, the analytic hierarchy process (Saaty, 1990) allows the perception of experts to be identified and quantified. Therefore, we hypothesize that a simple, practical, and robust sustainability indicator can effectively assess and differentiate levels of sustainability across various beef cattle production systems. The main objective of this study was to establish a simple and practical index, capable of measuring and evaluating sustainability in beef cattle systems, with low adoption costs.

The study was carried out in the area covering the Brazilian Pampa biome (BPB) in Rio Grande do Sul, Brazil. Five regions (Figure 1) representing this biome were considered: Western border (WB), Northwestern border (NB), Campanha (C), Central (Ce), and South (S).

Figure 1

The BPB occupies the southern half of Rio Grande do Sul, with an area of approximately 176,000 km², which is equivalent in size to Uruguay. This biome extends to the savannas, where the relief is low and herbaceous species are predominant (IBGE, 2004a; Boldrini et al., 2015). Natural grasslands within this biome support livestock activities, despite the strong participation of agriculture, with irrigated rice farming as a consolidated agricultural activity (Reis and Saibro, 2004). Moreover, beef cattle production and natural grasslands comprise the sociocultural identity of this region and are an embodiment of the Gaúcho (Brum Neto, 2008).

Municipalities were selected in the study area based on the number of rural establishments (Table 1), cattle herds (Table 2), and land utilization (Table 3). Owing to the geographic dispersion of these regions, soil differences within the BPB were also considered (Streck et al., 2008). The selected municipalities to represent these regions were Uruguaiana and Alegrete (WB), São Borja, Itacurubi, and Santiago (NB), Bagé and Dom Pedrito (C), Santa Vitória do Palmar and Jaguarão (S), and Rio Pardo and Cachoeira do Sul (Ce).

Table 1

	Rural establishments (2017)
	Total area	< 500 ha	500-1000 ha	> 1000 ha
	Hectares (1000)		Participation (%)
Rio Grande do Sul	21,684.6	51.4	15.3	33.3
Pampa biome municipalities	15,910.5	46.8	17.6	35.6
Analyzed regions total	3,553.7	26.4	20.8	52.8
Western border	1,245.8	23.1	21.1	55.7
Alegrete	726.4	27.2	22.0	50.8
Uruguaiana	519.5	17.4	20.0	62.6
Campanha	801.2	20.1	21.2	58.6
Bagé	353.5	23.3	22.9	53.8
Dom Pedrito	447.7	17.6	19.9	62.4
South	446.4	24.5	18.7	56.8
Santa Vitória do Palmar	284.6	21.4	17.9	60.7
Jaguarão	161.8	29.9	20.1	50.0
Central	483.5	43.0	20.0	36.9
Rio Pardo	176.8	47.6	16.5	35.9
Cachoeira do Sul	306.7	40.4	22.1	37.5
Northwestern border	576.8	29.9	21.7	48.4
São Borja	293.0	20.7	21.3	58.0
Itacurubi	85.3	36.6	24.6	38.8
Santiago	198.5	40.7	20.9	38.4

Source: Adapted from IBGE (2006, 2017, 2017) (Censo Agropecuário 2006; Censo Agropecuário 2017).

Table 2

	Cattle herd (2017)	Evolution (2006-2017)
	Heads (1000 heads)	Percentual (%)
Rio Grande do Sul	11,456.9	1.1
Pampa biome municipalities	9,063.4	1.4
Analyzed regions	3,669.2	0.8
Western border	900.2	23.9
Alegrete	576.7	17.1
Uruguaiana	323.5	38.2
Campanha	542.2	4.7
Bagé	243.1	17.0
Dom Pedrito	299.2	−3.5
South	232.7	−16.5
Santa Vitória do Palmar	141.4	−17.1
Jaguarão	91.4	−15.7
Central	227.1	−18.8
Rio Pardo	89.8	−21.7
Cachoeira do Sul	137.2	−16.9
Northwestern border	382.5	5.4
São Borja	137.9	6.0
Itacurubi	92.9	4.5
Santiago	151.6	−11.9

Source: Adapted from IBGE (2006, 2017) (Censo Agropecuário 2006; Censo Agropecuário 2017).

Table 3

	Land use (2017)			Evolution (2006-2017)
	Temporary crop	Natural grassland	Cultivated pasture	Temporary crop	Natural grassland	Cultivated pasture
Area (1000 hectares)				Percentual (%)
Rio Grande do Sul	7,622.07	7,541.25	1,635.51	19.1	−8.8	32.2
Pampa Biome municipalities	5,915.14	6,583.38	1,374.16	24.0	−7.3	33.4
Total analized regions	936.0	1888.0	429.9	45.2	−2.9	36.5
Western border	206.08	814.57	137.05	39.8	13.6	68.3
Alegrete	107.65	500.92	76.08	36.9	5.0	55.8
Uruguaiana	98.43	313.64	60.97	43.2	30.8	86.9
Campanha	200.53	410.87	112.83	104.2	−2.3	5.0
Bagé	60.04	185.74	57.01	133.9	5.7	19.8
Dom Pedrito	140.50	225.13	55.83	93.7	−8.1	−6.8
South	167.78	231.97	51.29	84.8	−13.8	−0.2
Santa Vitória do Palmar	100.40	144.47	36.87	59.7	−17.6	36.0
Jaguarão	67.38	87.50	14.43	141.4	−6.6	−40.6
Central	196.91	140.31	49.43	26.0	−30.0	12.9
Rio Pardo	63.17	50.19	25.69	25.7	−29.8	89.8
Cachoeira do Sul	133.74	90.13	23.73	26.1	−30.2	−21.5
Northwestern border	164.73	290.28	79.30	8.5	−13.7	156.9
São Borja	121.63	107.98	35.23	4.1	−6.5	69.8
Itacurubi	10.11	74.76	7.80	17.8	4.6	49.5
Santiago	32.99	107.55	36.27	25.1	−28.1	640.5

Source: Adapted from IBGE (2006, 2017) (Censo Agropecuário 2006; Censo Agropecuário 2017).

It was observed that rural establishments larger than 500 ha represent more than 50% (calculated to be 73.6%) of the total area of establishments (Table 1) and are characterized as livestock production and commercial agricultural regions (Andreatta, 2009).

These regions are also characterized by the presence of the largest effective cattle herds, representing almost a quarter of the cattle herd of the Rio Grande do Sul state and more than one-third of that in the BPB municipalities (IBGE, 2004b). The largest cattle herds are observed in the WB and C regions, which are the traditional beef cattle producing regions (Table 2).

Land use was marked by the presence of natural grasslands. Despite a decrease in their share in agricultural land use, natural grasslands are still a fundamental resource for farming in the BPB. In the last two decades, increasing agricultural activities in the biome caused an increase of the number of temporary crops and cultivated pastures, especially soybean (MapBiomas, 2021).

In the last decade, all regions showed an increase in temporary crop areas, mainly C and S with increases of 104.2% and 84.8%, respectively (Table 3).

After the selection of dimensions, themes, and indicators, experts were consulted to rank priority ratings according to the AHP (Saaty, 1990). The AHP is used in different areas (Kroenke and Hein, 2011; Ribeiro and Alves, 2016; Da Silva et al., 2019). Recently, Dabkiene et al. (2021) used AHP to create an index to assess agri-environmental situations in Eastern European farms based on weighted indicators. In this process, paired comparisons occur at different hierarchical levels. Experts evaluate and grant values on a scale of 1 to 9, where 1 corresponds to equivalent priority between parameters (i.e., no priority) and 9 corresponds to top priority for one of the two parameters (i.e., absolute priority) (Saaty, 1990).

For the AHP, an electronic form was developed in Google Forms. In the initial stages of development, the form was sent to a few experts for a pretest to improve and validate it. Subsequently, the form was sent to 28 experts in the selected regions for further evaluation, which made it possible to establish priority levels for selected dimensions, themes, and indicators. The experts were selected according to the following criteria: at least 10 years of experience in agricultural activities and a degree of lato sensu specialization.

Experts were invited from a) teaching and research institutions, b) beef production systems (producers and consultants), and c) retailers and service provider companies, to participate in the study. Among the 28 experts invited 39% of them had lato sensu degree, 29% had master degree and 32% PhD degree. About the experience in the farming sector, 42.9% of experts had at least 25 years of experience, and 39.3% were working at beef production systems, while 32.1% in teaching and research institutions.

Experts evaluated the priorities that would be given to dimensions, themes, and indicators for the index setup while answering the electronic form. Results of these comparisons defined the local priorities, which are assessed at the same hierarchical level. The product of the local priority of a hierarchy and that of a lower hierarchy determines the global priority of the lower hierarchy. Using equations (1) and (2), we calculated the global priority of themes and indicators, respectively.

in which WT_i = global priority of i-th theme; LD_i = local priority of i-th dimension; LT_i = local priority of i-th theme; LI_i = local priority of i-th indicator; and WI_i = global priority of i-th indicator.

In his proposal, Saaty (1990) established parameters to validate the priority values given to components of each hierarchy in the rating matrix. The method of Vargas (2010) was used as a basis for the calculations, and the geometric mean method (Bajwa et al., 2008) was used to calculate the major auto-value of the matrix (λ max). From here, we obtained the consistency ratio (CR) to consolidate the rating matrix, in which consistency values up to 0.1, considered acceptable. For the calculation of CR, consistency index (CI) and random index (RI) were used. From equation (3), we obtained the CI:

in which CI = consistency index; λ max = major auto-value of matrix; and n = number of parameters compared.

CR was calculated from equation (4), in which CI was from equation (3), and RI was a function of the number of parameters compared (Table 6) (Saaty, 1990).

Table 6

Indicators	Axis 1	Axis 2
Water management	0.093	0.752
Soil physic structure	−0.123	0.405
Soil organic matter	0.439	0.314
Land cover and use changes	0.147	0.430
Key species abundance and diversity	0.439	0.313
Production diversity	0.325	0.205
Animal health	0.274	0.097
Appropriate animal husbandry	0.307	0.047
Net revenue	−0.748	−0.092
Production costs	−0.359	0.052
Product diversification	−0.626	0.061
Net cash flow	−0.437	−0.085
Traceability system	−0.142	−0.411
Certificated production	0.022	−0.056
Wage level	−0.111	−0.407
Capacity development	0.633	−0.323
Employment relationship	0.060	−0.503
Child labor	0.377	−0.591
Safety and health trainings	0.557	−0.258
Health coverage and access to medical care	0.054	−0.327

in which CR = consistency ratio; CI = consistency index; and RI = random index.

This analysis must be conducted to control subjectivity in the priority rating so that the ratings are consistent for evaluation. Otherwise, the ratings should be adjusted by experts to maintain consistency. With an increase in the elements of comparison, there is an increase in the probabilities of inconsistent ratings.

The sustainability index (Figure 2) was established by the sum of products of global priorities of indicators from the assessment of farming systems:

Figure 2 NCF - net cash flow; PCo - production costs; CeP - certificated production; TSy - traceability system; PDf - product diversification; WMa - water management; SPS - soil physic structure; SOM - soil organic matter; ANH - appropriate animal husbandry; AHe - animal health; PDv - production diversity; LCC - land cover and use changes; KSD - key species abundance and diversity; SHT- safety and health trainings; ERe - employment relationship; HCMC - health coverage and access to medical care; CDe - capacity development; WLe - wage level; CLb - child labor.

in which iSus = sustainability index; w_i = global priority for i-th indicator; and VI_i = assessment value for i-th indicator.

Therefore, the final value of iSus will always be between 1 score (parameters established by the indicator are not identified) and 5 score (parameters established by the indicator are fully identified) as a function of evaluation performed in farms.

Correlation of dimensions, themes, and indicators were performed using the Pandas Profiling function, “Pandas” package (Reback et al., 2020) in the Jupyter notebook software. Averages of the dimensions and themes were compared by performing a Kruskal–Wallis test with a Dunn post-hoc. Principal component analysis (PCA) was conducted to understand the interrelationships among indicators. PERMANOVA was also performed to verify whether the perceptions of experts were related to their regions of activity, as well as for a comparison of averages using t-test to analyze differences between average priorities of indicators. The analyses were conducted in RStudio Desktop software, considering a significance value of 95%.

The ratings of 28 experts from different regions (six from WB, seven from NB, six from C, five from Ce, and four from S) were analyzed. The ratings did not exceed the CR value of 0.1, maintaining consistency.

Several studies have used SAFA guidelines (Kassem et al., 2017; Pérez-Lombardini et al., 2021). This framework was independently developed by the United Nations and was advantageous due to its flexibility and credibility (Bonisoli et al., 2019). Establishing a sustainability index (collection of indicators) is based on the selection of indicators provide assumptions closest to reality. Defining the scope of analysis makes it more connected (Gasso et al., 2014) with the traits and needs of the region. In this case, the expert’s perceptions obtained through AHP enabled the construction of an index with a greater sensitivity towards local reality.

Marchand et al. (2014) defined the full sustainability assessment (FSA) and rapid sustainability assessment (RSA) tools. The proposed index falls within the RSA. In such cases, it was reported that this type of tool allows analyses to be simpler and based on readily available information. In addition, RSA allows a larger number of farm systems to be evaluated, encouraging farmers to adopt sustainable principles and practices (Marchand et al., 2014). Hence, by identifying specific problems and concerns of farmers, they can be encouraged to adopt evaluations with FSA (Marchand et al., 2014). Thus, using the index allows for a more objective and simplified analysis. This objectivity acts both as a limitation in using the tool, and as a more agile and cost-effective way to carry out farming. However, this objectivity is also a limitation, as it does not encompass all the factors involved in sustainability. This is because, in this study, we sought to simplify the SAFA guidelines. The use of the proposed index allows the assessment of a broad group of farm systems, enabling a regional diagnosis, in addition to promoting the adoption of sustainable practices and principles.

The methodology used was able to produce a sustainability index (iSus) for the BPB within a variety of proposed themes. Additionally, tools for rapid analysis promote farmers’ incentives to adopt sustainable practices, with deeper and more complete sustainability analysis for their farm systems. The use of this index will allow sustainability to be measured in the BPB, and the studied factors will be very important in these assessments.

Establishing public policies and practices that promote the sustainability of agricultural production, either by expanding programs to finance sustainable practices, or by creating tax benefit projects for farmers who apply sustainable practices, should be premised on the assessment of these production systems, and this study contributes to this sense, by proposing a sustainability index objective and applicable to the BPB.

We believe that this study can inspire future analyses that consider other biomes and production systems, with specific indicators and practices for different situations, with new balancing and hierarchical analyses of processes, in order to adapt sustainability assessment indexes, making this type of assessment actually feasible.