SDG-2 is composed of eight targets (i.e., specific, measurable, and time-bound outcomes that directly contribute to the achievement of a goal) and 15 indicators (i.e., metrics used to measure progress towards a target, generally based on available data). The first five targets (2.1–2.5), the focus of this study, are directly related to food security and agricultural sustainability. The last three (2a–2c) are market-related measures aimed at increasing agricultural investments and reducing market restriction, distortions and volatility.
Table 1 summarizes our review of targets 2.1–2.5, highlighting whether each of their indicator is (i) conceptually clear, (ii) quantifiable, and (iii) universally relevant. We then recommend improvements ranging from minor textual changes to major content-related modifications and even their replacement. The proposal of new indicators was based on the availability of empirical data and was kept to a minimum given the already extensive list of indicators with which countries are expected to comply. Ten principles from the Sustainable Development Solutions Network for setting up a robust global monitoring indicator framework guided this exercise, including universality, simplicity, and prioritization of well-established data sources (SDSN 2015).
Table 1 Review of the SDG-2 targets and indicators proposed by the United Nations. More details can be found in the Electronic Supplementary Material (ESM S1) Our review revealed weaknesses in the original UN indicators. First, targets and indicators do not always focus on the same groups of people. Second, although indicators were phrased quantitatively, unclear concepts hinder their quantification. Indicator 2.4.1, for example, refers to the percentage of agricultural area under sustainable practices; while percentage is a quantifiable metric, agreeing on what sustainability is, when it is achieved and what it translates into at different scales can be difficult. Third, targets 2.3 (agricultural productivity), 2.4 (sustainability of food production systems), and 2.5 (genetic diversity) are less clearly defined and not always universally relevant. Their framing could lead to a variety of interpretations due to the vagueness of terms such as “sustainable” or “fair”, as well as their lack of specificity regarding the scale of enforcement and monitoring or the boundaries of “food systems”. While the main challenge concerning targets 2.1 (hunger) and 2.2 (malnutrition) is how to achieve them efficiently, targets 2.3–2.5 first require the definition of what they consist of, even prior to answering how to operationalize them.
A major challenge when selecting indicators under a specific SDG is to capture areas of overlap with other SDGs, such as the link between agriculture, nutrition, and public health (most directly relevant to SDGs 2 and 3). Multidimensional indicators that bring these together should be prioritized. For instance, the reduced incidence of non-communicable diseases (NCDs) is a target under SDG-3 (“Good health and well-being”) but calls for agricultural policies conducive of nutritious, healthy diets.
Because of their complexity, multidimensional indicators usually rely on more detailed data and are seldom available at the national level. Indicators measuring the impact of agricultural interventions on nutrition (Herforth et al. 2016; Herforth and Ballard 2016) would be useful in this context, but were designed to be applied locally. Likewise, newer indicators attempting to explore food access and dietary consumption nationally such as the Food Insecurity Experience Scale (FIES) (Ballard et al. 2013) and the Minimum Dietary Diversity for Women (MDD-W) (FAO 2016b) are still undergoing validation.
Besides underscoring the need to invest in global databases and to conduct complementary assessments locally, data limitations elicit the need for a more holistic policy approach when implementing the SDG agenda. While recognizing the relevance of several factors to food security and the limitations of unidimensional indicators (such as anthropometric or biochemical measures of malnutrition), we base our recommendations on the understanding that the simultaneous pursuit of all indicators under SDG-2 (and other SDGs) will naturally prompt integrated solutions among health, food production, nutrition, and other fields. An analogous rationale applies to the SDG agenda, as the achievement of sustainable development relies on the simultaneous and integrated implementation of all SDGs.
Target 2.1
Data on specific groups highlighted in the target (i.e., poor, vulnerable, infants) are limited. Although the Food Insecurity Experience Scale (FIES) has the potential to capture the complexity of actual and perceived food security, FIES data are not yet available for all countries. To overcome part of these data constraints, we propose two new indicators: 2.1.2—“Per capita food supply variability index” and 2.1.3—“Depth of the food deficit”. Both can be monitored through global, readily available databases and are better aligned with the concept of food security and its pillars.
Target 2.2
Although the indicators under target 2.2 can be monitored through globally available databases, three problems arise. First, indicators of target 2.2. do not fully cover the groups highlighted (i.e., adolescent girls, pregnant/lactating women, and elderly). Second, the phrasing is illogical; unless a base year is determined (thus fixing the range of low heights to be avoided), the prevalence will always be the same (2.5% distribution tail, i.e., − 2 SD.) and the indicator never achieved. Third, not all malnutrition conditions defined by the World Health Organization (WHO) are captured, namely: undernutrition, i.e., wasting (low weight-for-height), stunting (low height-for-age), and underweight (low weight-for-age); micronutrient-related malnutrition, i.e., micronutrient deficiencies (a lack of important vitamins and minerals) or micronutrient excess; and overweight, obesity, and diet-related non-communicable diseases (e.g., heart disease, stroke, diabetes, and some cancers).
We suggest the amendment of the original text (no allusion to distribution curve) and four new indicators: 2.2.3 (anemia among pregnant women) covers one more group mentioned in the target; 2.2.4 (protein supply) offers a proxy for food quality in the absence of data on specific micronutrients; and 2.2.5 (share of protein supply from animal sources) and 2.2.6 (obesity) are directly linked with NCDs, an increasing concern in developed and developing countries. The new indicators cover the health-malnutrition nexus more comprehensively and are readily available through the FAO database.
Target 2.3
The text of target 2.3, particularly with regards to doubling agricultural productivity, is not universally applicable. In some countries, the pursuit of agricultural intensification collides with the pursuit of agricultural sustainability. Notwithstanding the aspirational role of SDG targets and the importance of secure and equal access to inputs, knowledge, etc., such abstract concepts cannot be fully captured.
Measuring productivity on a labor basis instead of e.g., land (indicator 2.1.1) may not be adequate in some contexts. Also, it may be hard to tease out variations in agricultural output stemming from changes in labor productivity vs. other inputs (e.g., machinery). The relationship between target 2.3 (agricultural productivity) and indicator 2.1 (labor productivity) is unknown and may not be proportional, posing further obstacles to the calculation of country-specific threshold values. Finally, the proposition of a global definition for small-scale under indicator 2.1.2 may create distortions. Farmers’ income level offers little insight into their living conditions unless compared against a meaningful benchmark.
We suggest the replacement of indicator 2.3.1 by “yield gap” since the latter addresses agricultural intensification relative to a country’s potential yield on a per land basis, offering a benchmark for productivity. As yield gap is a complex concept that involves several factors influencing agricultural productivity (Lobell et al. 2009; van Ittersum et al. 2013), we adopt the definition used in the Global Yield Gap Atlas (GYGA 2018), based on a global protocol with local application. We also suggest the replacement of indicator 2.3.2 by two new indicators related to farmers’ income level independent of scale. The first refers to the share of the rural population below national poverty lines; the second refers to the share of farmers earning less than the national minimum wage. Country-specific reference values (i.e., poverty lines and minimum wages) account for differences in currency exchange rates and purchase power parity, allowing for international comparisons and ensuring locally meaningful results.
Target 2.4
Sustainability has social, environmental, and economic dimensions, thus permeating every SDG and SDG-2 indicator. In this sense, indicators 2.4.2 and 2.4.3 seem embedded into indicator 2.4.1, rendering unclear why emphasis has been placed on irrigation and fertilizer use but no other equally relevant aspect of environmental sustainability such as water productivity, GHGs, and pesticide use in agriculture. The vagueness and context-dependency of the term “sustainable practices” preclude cross-country comparability. Moreover, although the concepts “sustainability” and “resilience” are intimately related (Cabell and Oelofse 2012), the second is overlooked.
The use of irrigation may be sustainable or unsustainable depending on local water availability, water productivity levels, conditions of extraction and withdrawal, criteria for disposal, etc. Thus, indicator 2.4.2 should consider the pressure that irrigation poses on the renewable water resources of each country, complementing SDG-6 (dedicated to water sustainability) and indicator 2.4.3 (which addresses agriculture-related sources of water pollution). Variations in the efficiency of different irrigation systems should also be considered at country-level whenever data are available.
Concerning indicator 2.4.3, the term “eco-friendly” is poorly defined. In some contexts, the volume and form of fertilizer application may be just as relevant for environmental conservation as the type of fertilizers (e.g., too much manure can also lead to leaching). Also, the text refers to the share of households using “eco-friendly” irrespective of their agricultural yields or total fertilizer use, which may be misleading when many small farmers use eco-friendly fertilizers but represent a small share of total food production, or when farmers use eco-friendly fertilizers but that is only a small share of their total fertilizer usage.
We suggest seven new indicators directly related to key elements of agricultural sustainability: share of water withdrawal for agriculture; average water productivity in agriculture; nitrogen use efficiency and average nitrogen surplus (Zhang et al. 2015); GHG emission intensity of food production (Carlson et al. 2016); average carbon content in the topsoil; and pesticide use per area. We also propose the adoption of the Global Adaptation Initiative (GAIN) climate change vulnerability index for food (GAIN 2015), which summarizes a country’s vulnerability to climate change in terms of food production by forecasting the evolution of key elements of food provision (see ESM S1).
Target 2.5
A very small share of plant species is used in agriculture. Wheat, rice and maize alone provide more than half of the energy consumed by humans. This has led to a major biodiversity loss and genetic erosion. Target 2.5, aimed at the conservation of agrobiodiversity (i.e., the diversity of living organisms used in agriculture), is not only relevant for the maintenance of genetic diversity but also diet quality, resilience of production systems, and biodiversity conservation at the farm and landscape scales. Although all indicators proposed by the UN focus on important aspects of the genetic conservation in agriculture, the data to monitor indicator 2.5.1 are largely unavailable. Besides, indicator 2.5.2 could offer a distorted picture of the countries’ efforts to protect local genetic pools since the proportion of breeds cataloged in each of them varies considerably. We suggest the replacement of 2.5.1 by the average number of gaps in ex situ collections of selected crop genepools—a proxy for agricultural genetic resources secured in conservation facilities (Ramirez et al. 2009)—and the amendment of 2.5.2, so that it refers to breeds whose risk of extinction is known.