Enabling Conditions for Local Food Systems to Emerge in Predominately Rural Regions of Portugal—A Food Access Approach

Abstract

Local food studies have stressed the importance of local food systems (LFS) in shortening the linkages between producers and consumers and in promoting resilient territories. Food consumption patterns are mostly studied around rural–urban dynamics, urban food security, and the revitalisation of rural communities, but little is known about the impact of LFS over rural residents and their capacity to access local foods. This paper explores the development of LFS in rural areas, from a food access approach, by characterising the rural landscapes promoting local food consumption. From a mapping of 74 predominately rural municipalities, statistical data of six socio-economic and political variables were collected to depict each municipality. A cluster analysis and Pearson’s correlation test informed us about the factors enabling these networks to emerge. Three clusters were identified: ‘meso-urban’, (N = 5) presenting urban-like characteristics (higher income and education levels, and reduced road infrastructure and small-scale farming); ‘dense’, (N = 26) characterised by high population density, road infrastructure and small-scale farming; and ‘castaway’ (N = 43) with low population density, income, post-secondary education, and expenditure in RD in agriculture. LFS emergence in rural Portugal was strongly determined by the levels of mean income and education levels in rural municipalities, which brought into question concerns regarding rural residents’ capacity to consume local foods. Low physical access, purchasing capacity, and awareness of food issues appeared to compromise the utilisation of these foods by the most socio-economically disfavoured groups. However, other territorial externalities and empirical work not included in this study could further complement our findings and provide a richer picture for the localisation of food systems in rural areas.

1. Introduction

Our food systems have been transformed rapidly and differently across the globe during the last century, reflecting unsustainable methods of producing and consuming food, an increased disconnection between food source and final consumers, and broadened social inequalities [1]. Heightened urbanisation in developed societies and concerns regarding these issues have prompted the re-definition of the linkages between the ‘rural’ and the ‘urban’ in a renewed set of relationships known as alternative food networks [2,3]. The emergence of such networks reflects the interest of individuals in supporting farmers and rural communities, to protect endangered (local) species and varieties, while considering the wellbeing of consumers in urban places [4].

In the mid-1990s, the concept of local food systems (LFS) experienced a surge in popularity, examining the transformation of rural areas, the new dynamics of the agri-food sector and changes in food consumption [5]. Namely, LFS have been praised for counteracting the concentration of power in transnational food supply actors [6] and empowering primary producers as multifunctional service providers for urban and rural groups [7]. The concept of local foods is one that arose as a ‘solution’ to the negative externalities associated with the globalised and industrialised food system [8], based on the principle that shortening the linkages between food production and consumption can have positive impacts in the promotion of more resilient territories [9,10,11]. Shifting food production out of the industrial model and sourcing local produce through new chains has also been deemed to contribute to community health and nutrition, small producers’ livelihoods, and rural development, while tackling the environmental side effects of input-dependent globalised food systems, according to Edwards (2016), cited in [4,12,13,14].

Some authors have, despite these arguments, cautioned against the generalisation of the concept and have called for more holistic and critical approaches to examining the benefits of local food systems across different scales [7,14,15]. In Europe, specifically, a careful analysis of LFS must consider the implications in the shifts in social structures and power dynamics amidst the new rural development paradigm (‘new rurality’), which responds to changing geopolitical food and agricultural relations, according to Rytkönen and Hård (2016), cited by [16,17].

To date, LFS have largely been discussed in Europe in terms of their potential to contribute to small rural businesses and processes of rural development [18], with rural producers dominating the discourse [12], and the latter assumed to be the multifunctional providers of goods and services for urban consumers [7,16]. This is problematic for two main reasons: (i) it situates LFS as part of the “re-negotiation of the rural-urban agri-food relations, where rural areas, among other things, are required to ‘work’ for cities and their suburbs” [16], pg. 2103, thus leaving gaps in knowledge of the impact of LFS in rural areas; and (ii) the continuous adoption of a producer and urban-centred approach to discussing local foods limits insight into the capacity of consumers in non-urban territories to access these foods, and on how local food systems shape rural landscapes [19].

Knowledge about the transformative capacity of LFS in non-urban geographical spaces is also not uniform across the continent, and this subject is under-researched in Portugal [12,20,21]. For a start, each territory holds unique rural development traits that depend on the existing structures and financial capacities [22]. Social and political efforts for local foods in Portugal were heightened after the economic crisis of 2010–2014, within processes toward a national strategy in favour of family farming and food security. However, these efforts have concentrated on the protection of small family farm holdings as individual agents with social and territorial functions, without necessarily highlighting their food production capacity [23]. Some academic debates around local foods’ consumption in Portugal have largely focused on urban centres and food policy constraints [20,24,25].

The lack of a systemic approach to promoting local foods in Portugal hinders the capacity to recognise the real impact of these processes in rural landscapes. A starting point to fill this void is identifying what enables the emergence of local food channels in these areas. For instance, are there territorially specific socio-economic and political conditions that promote or slow down local food production? Once we can point out the driving forces of this trend, it is sensible to discuss any rising concerns regarding the utilisation of local foods. This paper, thus, aims to determine the key characteristics prompting LFS in predominately rural areas in Portugal based on the assumption that LFS’ development presents similar trends in rural and urban areas, and to discuss food access issues evolving from this setup.

The paper is structured as follows. First, we present the food access approach as a useful framework to analyse local foods’ utilisation, or, in other words, from a consumer perspective. Second, we explain the data sources and analytical methods used. Then, we present the results and discuss our findings. Lastly, we provide our final remarks and suggestions for further research.

2. Conceptual Framework

We start from the fact that food activities (production, processing, distribution, retail, and consumption) occur in agri-food systems composed of a set of actors and relationships with specific outcomes [26]. Food systems approaches emerged in response to the issues produced by the promotion of concentrated and unsustainable food production regimes which led to the persistence of food insecurity despite increases in food yields, plus a series of social, economic, and environmental effects impacting modern societies [27].

Like any other food system, local food systems (LFS) exist in unique contexts, in conjunction with other food schemes at different scales and levels. We assume the concept of local food to be foodstuffs that are produced and processed in a defined geographical area relatively close to where they are marketed and consumed [28]. The concept of “food geographies” [29] can help unpack the mosaic of materialities, people, places, spaces, and scales within food systems. It aligns with the notion that agri-food systems are territorialised entities with complex dynamics, circumscribed in a particular geographic space and coordinated by territorial governance [30]. We consider that LFS are, ultimately, expressions of territorial governance comprising new spatially bound relationships between producers and consumers, through which territories can be assessed [31].

Andress and Fitch [19] maintain that the food access concept can help disentangle the interactions between the social, cultural, and physical environments in food systems to assess their impact on food provision and consumption. Food access is hereby examined by qualifying the six dimensions of access, as proposed by Saurman [32], pg. 37: accessibility, availability, affordability, acceptability, accommodation, and awareness. These dimensions were adapted to understand the specific issues of local food consumption in rural areas.

Accessibility concerns the elements facilitating whether, or not, local foods are in a reasonable proximity to the consumer in terms of time and distance. Its relevance stems from the assumption that a weak infrastructure for food distribution in rural areas might pose challenges to maintaining high-quality produce, such as local foods, available at rural food outlets [33]. Availability considers aspects of local food supply and demand. Specifically, it contemplates the capacity of a territory to meet the food needs of the consumers and communities served and recognises that the main contribution of local food systems is the revitalisation of local food production by (re-)connecting the small producers and consumers of that locality [34]. Affordability refers to the capacity that consumers hold to cover the financial costs of local foods. Acceptability considers the receptivity of LFS in a particular area, from both a community and a consumer food environment perspective [35]. Accommodation hints at the suitability and adaptability of LFS to thrive in a specific context, by looking at how well local food outlets accept and adapt to local residents’ needs (i.e., store hours or types of market places), as well as the existing infrastructure for LFS to flourish. Awareness, lastly, indexes the kind and amount of knowledge that residents have on the relevance and means of purchasing local foods.

3. Methods

3.1. Data Collection

We used secondary data from the national mapping of local food initiatives in Portugal made by Hernández [36]. This mapping was a targeted online search carried out from November 2020–March 2021, from which we extracted the list of municipalities identified to host local food initiatives in predominately rural areas (N = 74). Predominately rural areas (PR) in Portugal correspond to administrative and geographical units with less than 100 inhabitants per square kilometre [37]. For these 74 municipalities, we collected further statistical data linked to six selected socio-economic and political indicators (Appendix A). The aim was to first find a comparative language to then group the municipalities in clusters based on similar characteristics.

The data source for indicators pop_dens, income, high_edu, and RD_agri was extracted from the national statistics, at the municipal and NUTS3 level, to obtain a socio-economic picture of each PR. NUTS3 corresponds to the European nomenclature of territorial units for statistics of small regions for specific diagnoses [38]. The other two indicators (road and agri_profile) were engineered. Each indicator was selected to correspond to one of the six food access dimensions discussed above. Table 1 summarises the six indicators guiding data collection.

Accessibility: we used Sanchez-Zamora et al. [39]’s proposed index to understand each municipality’s road infrastructure (road), taking the total length of motorways (main and secondary) in the municipality and dividing it by the total municipality area. This index is deemed relevant because residents in rural communities often have limited access to food resources due to the infrastructure available in that region [19]. A reduced road connecting system might impact those residents who rely on public transportation options or private vehicle usage to carry out food purchases or home food delivery; these may be, for example, the economically disadvantaged, elder groups, and those living far from medium-size rural cities where local food sales tend to take place [34].

Availability: we created the farming orientation index (agri_profile) to grasp the capacity of each region to produce local foods, based on Rivera et al. [40]’s argument that an increased number of small farms (assumed here as holdings < 5 ha) is linked to regional food systems’ development. We divided the number of small farms in each municipality by the utilised agricultural area (UAA) occupied by these small farms to infer this index.

Affordability: the indicator of gross mean income per household (income) served to determine what the purchasing capacity of residents in each municipality might be, for it is argued that wealthier consumers tend to have much greater access to a wider array of healthy and better foods, such as local foods [41].

Acceptability: we measured the population density (pop_dens) in each municipality to assess the impact range of local foods, considering the number of residents living per square meter. This indicator was chosen to examine the argument that LFS often operate in contexts pressured by the intensification of agricultural practices and urbanisation [42].

Accommodation: we looked at the total expenditure in research and development in agriculture to infer the reported investment per NUTS3 region in the agri-food sector, due to the lack of data at the municipal level. Although a more interesting indicator for this paper could be expenditure in innovation, these data were not retrievable statistically for this sector specifically. Therefore, we take the indicator expenditure in research and development in agriculture (RD_agri) to gauge the development of local food production systems, considering that the sector should prioritise a transition towards sustainable food systems in Europe [43].

Awareness: we look at the percentage of the resident population aged 15 and over with post-secondary education (high_edu) to determine the mean literacy level of the population in each municipality. This variable is relevant because raised interest and sensibility about local food issues, available through awareness campaigns, food events and media platforms, can contribute to the internalised food-specific values needed for consumers to purchase local foods [44]. For instance, Hashem et al. [45] discovered that consumers’ awareness of the safety risks linked with pesticide use in agriculture and the industrial food system was related to consumers’ interest in buying local foods. It is likely that increased knowledge and information is associated with the degree of competency and willingness to read food labels and ask food-related questions.

Table 1. The socio-economic and political indicators considered in this study (municipal level); our own elaboration.

Indicator			Description	References	Food Access Dimension
1	road	Road infrastructure density index	Road length (km)/Municipality area (km2)—value	Own based on Sanchez-Zamora et al. [39,46]	Accessibility
2	agri_profile	Farming orientation index	Relevance of small-scale farming (UAA used by farms < 5 ha/UAA in municipality)—value	Own based on Rivera et al. [40,47]	Availability
3	income	Income per household	Gross mean income declared by fiscal household (HH)—thousand euros	[48]	Affordability
4	pop_dens	Population density	Number of residents in each municipality per square meter—inh/km2	[49]	Acceptability
5	RD_agri	Expenditure in research and development in agriculture	Total expenditure in research and development (RD) in agriculture, by NUTS3—thousand euros	[50]	Accommodation
6	high_edu	High education level	Percentage of the resident population aged 15 and over with post-secondary education—%	[51]	Awareness

Table 1. The socio-economic and political indicators considered in this study (municipal level); our own elaboration.

Indicator			Description	References	Food Access Dimension
1	road	Road infrastructure density index	Road length (km)/Municipality area (km2)—value	Own based on Sanchez-Zamora et al. [39,46]	Accessibility
2	agri_profile	Farming orientation index	Relevance of small-scale farming (UAA used by farms < 5 ha/UAA in municipality)—value	Own based on Rivera et al. [40,47]	Availability
3	income	Income per household	Gross mean income declared by fiscal household (HH)—thousand euros	[48]	Affordability
4	pop_dens	Population density	Number of residents in each municipality per square meter—inh/km2	[49]	Acceptability
5	RD_agri	Expenditure in research and development in agriculture	Total expenditure in research and development (RD) in agriculture, by NUTS3—thousand euros	[50]	Accommodation
6	high_edu	High education level	Percentage of the resident population aged 15 and over with post-secondary education—%	[51]	Awareness

3.2. Data Analysis

We used the IBM SPSS Statistics Software (v.28) to run a one-tailed Pearson correlation test to measure the linear relationship between our selected continuous variables (Table 1). The objective was to detect whether these socio-economic and political aspects were related to one another, or not, to explain the emergence of initiatives prompting local foods’ consumption. Correlation coefficients (R) were deemed significant at the 0.05 and 0.01 level for each relationship. For positive correlations, we assumed the two variables increased or decreased together in the same direction, whereas negative correlations implied that the relationship between variables went in opposite direction (namely, when one increased, the other decreased, and vice versa).

For clustering the sample, we used an Excel free template for cluster analysis intended for research and data mining [52], which helped us organise our data into three segments. The cluster structure and set of correlations were further analysed and discussed through the lens of the six dimensions of access proposed by Saurman [32], pg. 37: accessibility, availability, affordability, acceptability, accommodation, and awareness, based on Andress and Fitch [19]’s food access approach.

4. Results

From the diversity across the country, three clusters of rural regions prompting local food consumption were identified.

Table 2. Mean values of the six indicators collected for all PR municipalities by cluster.

Cluster	No. of Municipalities	Percentage in Sample	pop_dens	income	high_edu	RD_agri	road	agri_profile
Cluster A ‘meso-urban’	5	6.8	40.1	19,079	19.95	2832	0.32	0.04
Cluster B ‘dense’	26	35.1	69.2	14,335	12.01	3509	0.66	0.37
Cluster C ‘castaway’	43	58.1	23.2	13,978	10.78	2674	0.38	0.27
TOTAL	74	100%
AVERAGE			40.5	14,448	11.83	2978	0.47	0.29

contains the summary of the mean values of these three clusters. Cluster A (‘meso-urban’, N = 5, 7% in sample) corresponded to the smallest group and included municipalities with urban-like characteristics (e.g., the largest income per HH, highest education level, an average population density, the weakest road infrastructure, and hardly any small-scale farming). Cluster B (‘dense’, N = 26, 35% in sample) included the municipalities with the highest population density, road infrastructure, proportion of small-scale farming, and the largest expenditure in research and development of agriculture, a lower-than-average income per household, and a somewhat medium percentage of the sample attaining high education. Cluster C was the largest cluster (‘castaway’, N = 43, 58% in sample), and was characterised by low-density population municipalities, the lowest mean incomes, the bottommost education levels and expenditure in RD in the agri-food sector, and a shallow road infrastructure and number of small-scale farms.

We observed a particular geographic distribution of the 74 municipalities in the sample across the country, according to the three clusters (see Figure 1). For a start, most PR municipalities were in mainland Portugal (N = 73) except for one in the Autonomous Region of the Azores, in Faial Island. Municipalities in Cluster A were characterised by either hosting a small city or by being nearby an urban centre, which could explain the high income and education levels and the limited presence of small farms in this group. A good example of this phenomenon is the municipality of Santiago do Cacém in NUTS2 Alentejo, which neighbours Portugal’s second largest port, Sines. As opposed to this municipality, most municipalities in Cluster B were in the northern part of Portugal (in NUTS2 Centro and Norte), apart from two: São Brás de Alportel in the south (NUTS2 Algarve) and Horta in one of the insular regions (NUTS2 Autonomous Region of the Azores). The characteristics and clear geographical location of the members in this cluster were not surprising, as small farms and a higher population density are predominant in northern Portugal. Cluster C was the largest group and occurred transversally across the country, but especially in remote and inland areas close to Spain along Portugal’s northern and eastern borders. Municipalities in this subgroup belonged to NUTS2 Alentejo (N = 10), Algarve (N = 3), Centro (N = 17), and Norte (N = 13), in areas lagging economically and demographically. Remoteness might help explain the low population density, post-secondary education, and income levels, as well as the meagre presence of small farms and expenditure in the development of the agricultural sector in the municipalities of this cluster.

Six correlations (four positive and two negative) emerged in our analysis. Their correlation coefficients informed the intensity of these associations. Most correlations were weak (closer to 0), one was somewhat moderate, and another was strong (closer to 1). We present them below in decreasing order, based on the degree of the relationship between the two variables.

1. income—high_edu (R = 0.806): This strong correlation shows that if mean income values increase, so do high education levels (and vice versa). In our sample, this translates to municipalities hosting LFS initiatives with similar income and education levels. Cluster A presented the highest mean values for these two variables, indicating that municipalities labelled as ‘meso-urban’ were characteristic of having higher incomes in tandem with higher levels of education. In opposition, these two variables had directly proportional low values in Cluster C, where municipalities had low income and low education levels in the cluster ‘castaway’.

2. road—pop_dens (R = 0.422): A somewhat moderate relationship emerged between the variables of road infrastructure and population density in our sample, meaning that the two variables behaved similarly sometimes. Cluster B showed the highest mean values for both indicators, from which we can infer that more densely populated municipalities often consisted of a better road infrastructure. However, this connection was not akin in clusters ‘meso-urban’ and ‘castaway’.

3. road—agri_profile (R = 0.299): A weaker positive correlation between these two variables appeared in our sample, with both indicators increasing and decreasing in tandem. Unlike the item above, Cluster B held the highest values for the two variables, whereas Cluster A had the lowest. This implies that municipalities labelled ‘dense’ were well connected through the road system and had a high presence of small farms in their territory when compared to the other two clusters. Municipalities named ‘meso-urban’ were, in opposition, poorly connected and held a meagre percentage of land used for small farming.

4. income—road (R= −0.217): This negative correlation was not straightforward, but hints at a weak probability that if the mean income of a municipality increases, its road infrastructure is likely to be poor (and vice versa). This is true for Cluster A, where municipalities had higher incomes but low road infrastructure indexes. Clusters B and C showed the opposite trend. Municipalities with medium level incomes had well-developed road infrastructure systems (Cluster B), whereas municipalities in Cluster C with the lowest mean income values had a less-than-average road infrastructure index.

5. RD_agri—agri_profile (R= −0.215): The linearity of the relationship between these two variables was weak, but hints at the opposite behaviour of one variable in comparison to the other. Clusters A and C showed that the weight of the expenditure in RD in agriculture and the presence of small farms were not uniform, but somewhat opposite. On the other hand, municipalities in Cluster B (‘dense’) were characterised by having both the highest values of expenditure in the sector and of small farms. This signalled the likelihood that if the investment of a region in research and development in agriculture is high, the number of small food farms may also increase.

6. pop_dens—high_edu (R = 0.200): The positive relationship found between population density and education levels was weak yet informed us that they could behave similarly. The cluster ‘castaway’ showed a clear linearity, with municipalities having low-density populations and a meagre post-secondary education attendance. Conversely, Clusters A and B related the other way around, where the fraction of the population attaining post-secondary education increased in medium size municipalities (Cluster A) but was lower in locations more densely populated (Cluster B). Although this correlation is not a straightforward trend, one thing was clear; low densely populated regions tended to host a reduced number of residents with post-secondary education.

Last, we measured the weight of each variable across all correlations, based on the correlation coefficient of each pairing. In sum, we discovered that variables reveal different degrees of relevance, as follows (in decreasing order): income (1.023), high_edu (1.006), road (0.938), pop_dens (0.622), agri_profile (0.514) and RD_agri (0.215).

Table 3. Relevance of the analytical variables according to the correlation coefficients. Values with (*) correspond to the negative correlations that were turn into positive values to facilitate measurement; our own elaboration.

	pop_dens	income	high_edu	RD_agri	road	agri_profile	SUM
pop_dens			0.200		0.422		0.622
income			0.806		0.217 *		1.023
high_edu	0.200	0.806					1.006
RD_agri						0.215 *	0.215
road	0.422	0.217 *				0.299	0.938
agri_profile				0.215 *	0.299		0.514

presents these results, considering all correlation values in positive to compare the recurrence of each aspect.

Variables income, high_edu and road stood out in association with other variables across all correlations. This means that these aspects, more so than the other socio-economic indicators, were significant in the sampled municipalities. From this, we infer that income, high_edu and road played a relevant role in enabling municipalities to host local food initiatives. This comes as no surprise, for the strongest correlation was identified between gross mean income and post-secondary education levels (income—high_edu), and a weak correlation was visible between income and the road infrastructure index of each municipality (income—road). The role of the existing transportation infrastructure was deemed relevant too, especially in relation to the demographic pressures in the sample (pop_dens—road) and the relevance of small farm production in the municipality (road—agri_profile).

Next in significance came population density (pop_dens) and the relevance of small farm food production in the municipality (agri_profile). The last to emerge was expenditure in research and development in agriculture (RD_agri), which, despite including values at the NUTS3 region instead of at the municipality level, signalled its weak association with small farms in the region (RD_agri—agri_profile).

5. Discussion

The three sets of clusters identified in our study served to organise the municipalities and shed light on three distinct profiles of territories promoting local food systems in rural municipalities of Portugal. This setup presented interesting geographical trends and six significant correlations among the socio-economic and political variables that were considered. This section discusses the impact of such a combination of factors on local food access in these regions, from a consumer perspective, guided by the six dimensions proposed in our conceptual framework (Table 3). It focuses on the relevance of the variables identified in our results, which we assume could help us discuss which conditions might be enabling and/or hindering the emergence of LFS in our sample.

Following the trends among their urban counterparts, our findings showed three aspects were key characteristics of the rural municipalities promoting local foods in Portugal: consumers’ purchasing capacity (income), consumers’ knowledge and sensibility on food issues (high_edu), and consumers’ ease of commute (road). It is important to stress that none of these aspects can be taken in isolation, but instead are part of a bigger setup composing the intrinsic dynamics of each studied territory. Indicators such as the potential market niche for local foods (pop_dens), the relevance of food production by small farms in the area (agri_profile), and expenditure in research and development in agriculture (RD_agri) had less of an impact in determining the three clusters. However, all are discussed together in this section, as they can impact food access. Figure 2 shows the relevance of variables, based on the correlations identified in Table 3.

5.1. Consumers’ Purchasing Capacity

Reasonable prices of local food products have been argued to be important for consumers [16], as alongside being a key factor in determining the viability of local food systems [7]. Municipalities in the cluster ‘meso-urban’ presented the highest mean income values; in other words, they had an economic advantage, and were able to afford more expensive items such as local foods. This condition can be favourable for small food businesses and small producers selling to nearby residents and can enhance the formation of local food networks. The fact that the clusters ‘dense’ and ‘castaway’ consisted of residents with lower purchasing capacity could inform us of two things. First, the economic sustainability of small producers selling locally could be compromised, because producers might resolve to sell in niche markets elsewhere for better returns (e.g., urban centres). Second, only residents with the financial means might be able to afford buying local foods, even if this implies sourcing from somewhere else.

On the one hand, the allocation of funding to supporting alternative forms of producing and consuming food in territories with a low-income population requires creativity. This may be, for instance, through finding ways to help producers supply local markets, and through reducing food production costs so that prices may remain affordable for all consumers. This entails shifting specific resources to help new channels to emerge and be sustained; this could be a challenge harder to overcome in the municipalities of Cluster C (‘castaway’), which are characterised by both the lowest income and expenditure in RD in agriculture. Besides this, efforts must take a systemic approach that contemplates the ease of reaching local foods and societal improvement, as affordability does not only refer to food prices but also to people’s perceptions of worth relative to food cost [19]. Here, one more aspect must be considered: educating consumers about food issues, which sheds light on the strongest correlation in our analysis (income—high_edu). From our results, we discovered that municipalities in cluster ‘castaway’ may resort to additional means to promote LFS other than residents’ literacy and purchasing capacity, as these values turned out to be the lowest. By this, we do not imply that access to knowledge about local foods takes place solely in post-secondary institutions, but that citizens with a greater portfolio of information are more likely to be more prone to asking questions and seeking answers.

On the other hand, our findings confirmed that gross mean income is a key indicator for measuring local food consumption in rural areas of Portugal. The cluster ‘meso-urban’ consisted of municipalities with high-income residents, informing us that local food initiatives in this cluster benefited from this niche market and/or that small food producers can have greater returns in these municipalities. Although the farming orientation (agri_profile) of Cluster A did not appear to particularly favour small food producers, the presence of wealthy residents in these municipalities allowed us to speculate two things. First, the local food production output in these municipalities is either very high in response to demographic pressures [53]; or second, a secondary sector might be relevant for adding value to locally produced foodstuffs.

Results sustained that clustering was primarily formed by means of economic wealth, supporting Brinkley’s argument that farms involved in direct marketing are more prone to be in areas with high median home values [54], pg. 315. Similarly, restriction on purchasing high-value foods has a direct effect on the opportunities of lower income consumers—largely situated in Cluster ‘castaway’—to consume these high-quality products. Due to the low purchasing capacity in this cluster, municipalities might also lack the chance to develop LFS that accompany production, processing, distribution and retailing processes, proving what Forssell and Lankoski [14] argue to be a shortcoming of local foods.

5.2. Consumers’ Knowledge and Sensibility on Food Issues

From the correlation test we ran, we discovered there was a strong connection between gross mean income and post-secondary education levels in our sample. Local foods’ consumption in municipalities with higher incomes also had better educated populations, in what we assigned as ‘meso-urban’ sites. According to Anderson [55], higher educational levels tend to favour awareness and support of initiatives that challenge the mainstream food system.

The reduced number of residents with higher education in clusters ‘dense’ and ‘castaway’ could reflect the limited ability of consumers in these territories to demand alternative food choices, as “awareness is more than knowing that a service exists, it is understanding and using that knowledge” [32], pg. 138. The lower percentage of residents attaining post-secondary education in clusters B and C suggests they might be missing the chance to consume local foods because of their limited awareness about local foods and where they can be purchased or attained. Awareness becomes the outcome of food literacy (the set of information to which residents are exposed) that is ‘content and context specific’ and the result of effective communication. Social media has become an efficient informational tool for local food business promotion in urban cities in Portugal [56], enabling the democratisation of information access. However, social media’s role in informing consumers about food quality and production methods is mostly a private endeavour (meaning that it is largely carried out by businesses to communicate with customers), and access to such channels might be limited in rural areas due to restrictions on internet access and digital literacy.

Efforts to counteract reduced levels of food literacy often stem from the role of local actors in stimulating the procurement of local foods, but aspects such as population pressure might play a significant role. In practice, densely populated territories often equate to being heterogenous and more democratic, because diverse narratives regarding access to resources, participation, identity, etc. can congregate and create space for finding solutions and developing alternatives [57]. Our analysis showed a weak and positive correlation between population density and the percentage of residents attaining post-secondary education (pop_dens—high_edu). From this, we infer that the denser the population is, the more likely it is to be better informed about food issues, and vice versa. Cluster C confirmed this trend, hosting the least-populated municipalities and the lowest percentage of the population with high education. This condition could jeopardise the development of a more diverse food system that contemplates local food venues and viable dietary options.

5.3. Consumers’ Ease of Reaching Local Food Markets

We understand accessibility in terms of the geographic location of the food supply and the physical ease of consumers getting to that location [19]. From our results, we deduced that the most densely populated municipalities had also a well-developed road infrastructure (Cluster B, ‘dense’), with a less clear trend in clusters A and C. Notions of physical and temporal proximity in municipalities of Cluster B might help explain this linkage. One the one hand, local foods are mostly exchanged through direct markets that benefit from the spatial density and proximity concept typical of localised food systems [58], pg. 3, for instance, if a market is within walking distance. On the other hand, proximity can be measured in terms of the travelling time of consumers reaching local food venues. Andress and Fitch [19] argue that individuals with their own transportation often have a significantly easier time shopping, especially for high quality foods. A well-developed road system implies that roads connect adequately the territory, but also that transportation facilities (could) exist to grant access to certain services for residents without their own vehicle.

Hinrichs (2000) had already cautioned that rural populations are especially burdened by a greater variation in spatial access to grocery stores, leading them to focus on facilitating access to “exclusive products and exclusive customers”, cited in [8], pg. 301. This trend could weigh against the opportunity of low income, rural populations to affordably access the means to eat locally produced foods [19]. Food access in rural communities must, however, be considered carefully, as it can be understood in relation to ‘relative rurality’, as McEachern & Warnaby (2006, p. 198) cited in [59].

Wenzing and Gruchmann (2018) argued that demographic characteristics might influence consumer perceptions and preferences for local food [16]. In other words, if residents perceive that local foods are at an acceptable ‘distance’, either physical or temporal, there is a higher chance of them selecting them as food options. Additionally, the short distance between consumers and their food source helps preserve local foods’ inherent attributes, such as flavour, authenticity, and cultural or territorial identity, attributes that tend to be lost in long food supply chains [9,58].

Making local foods available through territorially adequate channels can reduce social and territorial disparities, decrease transport costs in remote areas, increase territorial cohesion, and bring consumers closer to their food source [60]. However, aspects such as local food demand and institutional support in each specific territory must be considered to make these services viable for all citizens.

Discussing local foods’ availability requires considering the overall supply of local foods determined by production, distribution, and retail processes, as well as the interlinked relationships between these activities [61]. Our results indicated that municipalities in cluster ‘dense’ have a better road infrastructure and hold a larger number of small-scale farmers compared to the other two clusters. We assume, therefore, that such municipalities have a larger capacity to supply local foods, since small farms account for a higher share of [local food] production in regions with a higher population density [40]. This larger capacity to produce local foods could be explained by the converging pressure that densely populated territories have on landholding, which can prompt the development of local food production [53].

The positive relationship recognised in municipalities ‘dense’ between physical proximity (enabled through a well-developed road infrastructure) and food availability (a greater number of small food producers) could hint at a greater offer of local foods. The weaker link between these two variables in the clusters ‘meso-urban’ and ‘castaway’ could be indicative of an opposite trend. This is because a limited infrastructure for food distribution (e.g., roads, storage, frequency of delivery) in rural areas might pose limitations to maintaining local food produce availability at rural food outlets [33]. The lowest percentage of small farms and road infrastructure system in the ‘meso-urban’ municipalities could be explained through the diversification and strong market-oriented service sector that these regions have been able to forge over time. Referred to as “consumption countryside”, the evolution in these territories could allow them nowadays to participate in advanced industrial networks and advanced economic markets [62], thus causing less reliance on primary sector activities.

By the same token, disbursement for research and development for the promotion of local foods can play a key role in expanding local food production and enhance business and employment opportunities in the agri-food system [63], with trickle-down effect in the community at large. Our results showed a weak negative relationship between expenditure in RD in agriculture and the weight of small farms in the municipalities of our sample. This might indicate a contradictory signal regarding public support and local food production, especially in Clusters A (‘meso-urban’) and C (‘castaway’), because the lack of a suitable framework that promotes local food production can hinder the capacity of the territory to accommodate its needs (e.g., increased rural food security) and goals (e.g., a more competitive small food production sector).

The cluster ‘dense’ presented the highest expenditure in the sector, along with the highest presence of small farms, which could hypothetically be explained by a higher demand for local foods (pop_density), as well as the ease of connecting small food production and end users (road), both of which are characteristic of the denser municipalities of northern Portugal. The higher public investment in research and development in agriculture in cluster ‘dense’ may be an indication of these municipalities being better suited to develop and adapt infrastructures that can promote local foods’ consumption. Similarly, fund allocation to increase food literacy (e.g., via policy tools, research efforts, capacity building, and awareness campaigns, etc.) can empower consumers (or provide the ‘knowhow’) and promote initiatives (or ‘social devices’) to increase consumer awareness around purchasing locally grown food products [4,7].

6. Conclusions

To date, most academic debates on local food issues have focused on the evolution of alternative food systems as tools to reconnect small producers and urban consumers, revitalise the countryside, and promote urban food security. Knowledge about the transformative capacity of these systems in non-urban geographical spaces and the consumption of local foods by rural residents are under-researched in Europe and Portugal [12,20,21]. To fill this void, this paper sought to identify which aspects enable the emergence of local food channels in 74 predominately rural areas of Portugal, considering key territorially specific socio-economic and political conditions that can affect the promotion of local food production. Through a statistical analysis of these aspects at the municipal level, we discovered sectoral trends and driving forces. The food access approach of Andress and Fitch [19] was adopted for this exploration, with the guidance of the six dimensions of food access suggested by Saurman [32]. The findings facilitated discussion of rising concerns regarding the utilisation of local foods by residents in rural areas, through the adoption of

This study was pertinent for examining what might prompt the emergence of local food systems in rural areas in Portugal and proposed a theoretical ‘toolkit’, or framework, to discuss what might be enabling or hindering consumption of local foods in such areas. The idea was to pinpoint the national trends in the agri-food sector, but it did not attempt to generalise. Our findings showed that specific territorial conditions help define three distinct clusters of predominately rural municipalities promoting local foods consumption in Portugal: ‘meso-urban’, ‘dense’, and ‘castaway’. The aspects gross mean income and the percentage of residents attaining post-secondary education were shown to be directly linked in determining cluster formation, although the relationship between population density and the road infrastructure of municipalities proofed to be determinant in enabling the accessibility and availability of local foods.

The diversity of rural landscapes enabling LFS emergence could partially be explained through the three sets of characteristics identified in this study; however, they described what may be on the ground and, thus, must be confirmed empirically. For instance, the hypotheses that have arisen from our findings on the ability (or lack thereof) to produce local foods and/or for residents to demand these items need to be assessed and qualified. Interviewing residents in one or more of the municipalities analysed in this study would provide knowledge from a consumer perspective and help identify other aspects affecting consumers’ decision-making processes and perspectives on local foods.

Similarly, we argue that other variables could help further knowledge on the drivers of local food initiatives’ emergence in these areas and could be explored in future research; these variables could be non-national funding sources and internal factors (e.g., entrepreneurship capacity, foreign investment, cross-border relationships, etc.). For the latter, we encourage exploring specific case studies to attain in-depth knowledge.