Modelling the Benefits and Impacts of Urban Agriculture: Employment, Economy of Scale and Carbon Dioxide Emissions

Abstract

This paper examines the social, economic and environmental potential of Urban Agriculture (UA) based on information from forty distinct locations in each of the two regions: Adelaide, South Australia and Kathmandu Valley, Nepal, representing the diverse developmental background. Modelling is used to estimate equivalent employment, scale appropriateness, and earnings in comparison to labour use from commercial urban farming and gardening style UA, together with carbon dioxide emissions for two vegetable types. The study investigates the influence of distance and production scale under manual to modest mechanisation for urban vegetable production, finding that the interplay between labour use and mechanisation can favour scale-appropriate UA practices with better labour productivity and economic and social advantage. The distribution (assumed to be by car for UA) contributes the largest proportion of emissions, and the production component (even with mechanisation) contributes a relatively small portion per unit of production. We recommend that governments and planners should facilitate scale-appropriate mechanisation through better planning and policy instruments for UA’s sustainability.

1. Introduction

Urban Agriculture (UA), also referred to as urban farming or urban gardening [1], includes diverse forms of production ranging from the uncontrolled environment to controlled environment agriculture [2]. The more common uncontrolled environmental UA practices cover farming in the community, allotments, farms, gardens on housing areas and rooftops, organisational gardens, market-driven vegetable production, and growing food in public parks [3]. Hi-tech farming, including indoor and vertical farms, are examples of controlled environment UA practice, which is less common but proposed within the city for social, economic, environmental [4] and food production motives [1]. Localised food production systems like UA are gaining attention from planners and advocates due to claims of multiple and integrated benefits [5]. According to the Food and Agriculture Organisation of the United Nations, more than 800 million people are engaged in UA activities, contributing 15% of the total food supply in the world [6]. The distinguishing features of UA are ecological, social and economic integration within the urban system. These features are influenced by resources like land, labour, production and generate impacts through food security, environment, economy, health, poverty and culture [7].

Despite several limitations, UA has broader social, economic and environmental roles that can be promoted globally [8]. The broad reasons for expanding UA are in response to rapid urbanisation, concerns about climate change impacts of conventional agriculture, and addressing food security and accessibility. UA is promoted as a source of livelihood, social cohesion, leisure, environmental stewardship, and belongingness [9]. For example, UA can feed some proportion of food to citizens and helps to maintain urban green space [10]. UA is considered a worthwhile alternative to solve climate change and food issues, at least to some extent [11]. Furthermore, UA has the potential to reduce food insecurity through a wide variety of locally produced fruits and vegetables by overcoming transport barriers and provide a source of income [12]. Through direct and indirect multiplier effects, local food production activities can help reduce economic leakage through imports that fuel the local economy [13]. UA is thought to play a critical role in creating a more stable food supply while generating jobs and income to help make cities more resilient to external shocks [14]. UA has drawn attention in research due to changing attitudes towards food production, consumption and distribution [15]. Urban areas are facing challenges like natural disasters, pandemics, scarcity of natural resources, poverty and food security; cities need prompt measures to cope with these challenges [16]. UA can be one of the best avenues as, along with the above advantages, it can provide employment [17,18].

The purposes of UA in the cities may differ; in the global south, UA is intended to feed rapidly growing populations, while in the north, UA is promoted for lifestyle, health, community development and innovation motives [19]. However, the convergence points of UA in both contexts rely on social and environmental sustainability and resilience to change [20]. While UA has fewer direct economic impacts, it can have substantial social and environmental benefits for urban dwellers [21]. UA is a powerful instrument for developing more sustainable food systems by shifting its economic approach towards a social orientation [22]. UA has been proposed as a tool for sustainable, livable and resilient cities in the current state of the fossil fuel crisis and climate change [15]. The expansion and promotion of UA within the city could help decrease supply chain emission problems through intensive land utilisation, bringing production and consumption closer [23]. In addition, UA fosters sustainability by reinventing urban space and promoting community development through local food culture [24]. Therefore, UA should be evaluated in terms of multifunctionality and multiple services and benefits [25].

The challenge of providing healthy food for a growing urban population could be met over time by complementing conventional large-scale farming through UA as a resource-efficient alternative farming method [26]. However, whilst the social, economic, and environmental dimensions of sustainability have different levels of influence in UA, based on a diversity of scales, contexts, and approaches [27], it may have higher environmental and other values than commercial agriculture [28]. Some have contended that the benefits of UA, especially the environmental and social dimensions, have been grossly underestimated [29] and that UA can potentially play multiple roles, including job creation [30,31]. UA can reduce food miles and economic pressure, with estimates that some 100–200 million urban farmers worldwide are already providing fresh horticultural products direct to city markets [32]. Furthermore, UA is claimed to help achieve food security during crisis periods [33]. Urban centres face severe food and nutritional insecurity due to disruption in the food supply chain, physical and economic barriers, and increased food waste due to labour shortage, highlighting the importance of a resilient and locally produced food system [34]. UA is recognised as a significant source of the urban food supply with economic, social and environmental values [6]. UA can also impact food security at the household level, either as a commercial business to earn money or as an opportunity cost to buy foods from production [35].

Food-related GHG emissions are approximately one-quarter of global GHG emissions, a significant global problem from the production and distribution of foods [36]. Therefore, reducing emissions from food is the greatest challenge, and in order to address this, agricultural efficiency must be improved with low carbon emissions using scalable and affordable technologies [36]. Increasing local production and reducing food miles decreases food supply disruption [37]. UA provides multiple services like social cohesion, empowerment, resilience and acts as a sustainable food system [38]. The solutions to sustainability and resilience of food systems for cities can be evaluated through integrated social and ecological analysis [39]. UA’s overall social and environmental contribution depends on the area ultimately under production, ecosystem services and social factors based on local conditions [16].

Kafle et al. [40] showed that, for UA to be scaled up, economic viability needs to be considered, and within that consideration, labour cost—at least in high-income countries—is potentially a very significant factor. For broader economic advancement, raising labour productivity is generally considered part of the strategy to lower food prices [41]. Farmers are under pressure due to the “cost price squeeze” (i.e., balancing production costs and market prices), and to survive under such pressure, farmers need to increase output and/or decrease the cost of production, but the increment in labour (required to boost output) is directly related to cost. Introducing scale-appropriate labour-efficient technologies like mechanisation may play a double role through increased production efficiency and reduced overall labour, thus reducing the pressure of the cost price squeeze [42]. Mechanisation is complementary to labour inputs, thus raising labour productivity [43,44]. Mechanisation also plays a vital role in curbing the problem of farm drudgery, high cultivation and labour cost [45]. Labour shortages, increased wages and sustainable intensification of small-scale farms are the primary factors driving mechanisation [46].

The conventional food system has many challenges in terms of sustainability, in both social and environmental dimensions, due to mass production and long distribution chains. UA is identified as an appropriate localised system of growing food with short supply chains [15]. Closer distribution with reduced food distribution chains is a prominent feature of UA [47]. Kafle et al. [48] recommended further exploratory research on social, economic and environmental nexus to quantify the growth potential of UA. UA’s social and environmental outcomes are influenced by parameters like labour use, scale, level of mechanisation, and, ultimately, greenhouse gas emissions during different stages. As a more localised and diverse food production system, UA can—in principle—bring production closer to the consumer, provide access to cheap and nutrient-rich food, and should, therefore, ultimately help in mitigating adverse effects associated with the long-distance global food supply chain [15].

Some past studies have made and/or repeated unfounded claims about the potential for UA to reduce emissions; others have demonstrated a firm foundation, e.g., Kulak et al. [49], but in the context of specialised case studies that may not be broadly generalisable; others may have a firm foundation but have not addressed the issue of employment potentiality and economic viability of production or explored GHG emissions in the context of mechanisation (which may be vital for economic viability). This study examines the potential socio-economic and environmental problems of current uncontrolled environmental UA in practice with an emphasis on employment, earnings, scale appropriateness and carbon emissions and suggests the key considerations for sustainable UA practice based on different distances and scales. This study uses 40 distinct land parcels of different sizes and distances to represent different scales and locations of potential urban farms to explore their impact on production and distribution. Two cities are used: Adelaide, South Australia and Kathmandu Valley, Nepal, to provide divergent development contexts. This study helps to develop a sustainable UA system based on the interplay between Scale-appropriate mechanisation, an optimal crop mix, labour planning and transport mode that links the product with the market through alternative distribution mechanisms.

2. Materials and Methods

2.1. Study Locations

Adelaide is a growing city and is the capital of South Australia, with a 3260 km² area, 1.3 million people and a large population concentrated in small suburbs [50]. Private and public-led UA is dominant in South Australia [27]. 59% of people with back or front yard gardens are estimated to be engaged in some form of UA in South Australia [51]. Information on land area, cost and distance from the city centre of Adelaide, South Australia (Figure 1) were randomly selected from 40 land parcels advertised for sale on a widely used real estate website https://www.realestate.com.au (accessed on 25 December 2020). Land parcel areas ranged from 154 m² to 2705 m². The farthest distances were up to 57 km in Adelaide, representing a trip from the Adelaide Central Business District (CBD). These distances are taken as indicative of the proximity of the land parcel (production) to the market and allow us to identify any potential relationships that emerge between UA production scale and transport requirements.

The Kathmandu Valley includes three districts, namely Kathmandu, Bhaktapur and Lalitpur, with the entire areas of Bhaktapur and 45% of Kathmandu and 50% of Lalitpur, respectively, with a total area of 665 km² [52]. It is one of the most crowded cities in south Asia, with a population of 2.54 million, growing at 6.5% per year [52]. Urban farming in Kathmandu is sprawling due to its fertile soil, a high level of engagement in food production, and high demand for food from a rapidly growing urban population [53]. Though there is a limited understanding of types, food production practices, and demand and supply scenarios, provincial and local governments have recently become involved in UA support activities in city areas like Kathmandu Valley [48]. Rooftops, backyard UA and farming in vacant lots are typical UA practices observed in Kathmandu Valley [40]. The land area and distance from the CBD (Figure 2) were randomly selected from 40 advertised land parcels listed for sale on real estate websites (https://www.housingnepal.com, accessed on 26 December 2020 and https://hamrobazaar.com, accessed on 26 December 2020). Land parcel areas ranged from 127.2 m² to 7949.1 m². The trip distance was considered from the CBD to the land parcel i.e., up to 19 km. These distances were taken as the assumed trip distance for product distribution based on scale and transport.

For the purpose of this study, the distances from CBD to the assumed productive land parcel in each case study location were categorised into two groups Inner-city (up to 10 km from CBD), Suburban (greater than 10 km from CBD), smaller-plot (up to 400 m²) and larger-plot (greater than 400 m²) based on Kafle et al. [40].

2.2. Study Approach

This study explores the claims of the social and environmental benefits of UA that are made about its benefits and impacts but which remain largely under-researched, namely economic viability and possible employment outcomes (through the understanding of small-scale mechanisation) and environmental benefits (through studying realistic production and distribution impacts). The Full-Time Employment Equivalent (FTE) is used as a measure of how many people may be employed in UA under various scales and production assumptions. This is directly related to labour efficiency and the total economic output. In general, a low FTE model would show higher economic efficiency (i.e., greater labour productivity), while a high FTE corresponds to better employability. UA is claimed to offer urban employment prospects [17,18], but previous studies have called into question its economic viability based on labour costs [40]; therefore, the FTE concept is worthy of investigation.

Figure 3 shows the conceptual framework underpinning the current study. The main variables collectively represent the factors being analysed for social and environmental feasibility along with economic quantification (gross earnings minus labour use), and these depend on the parameters as presented in Figure 3. The four key influencing variables under investigation are crop types, distances, scale and level of mechanisation. The information on land (size and distance), crop types, labour inputs, machine use, vegetable requirements and carbon dioxide emissions to estimate benefits and impacts were collected from different primary and secondary sources, which are mentioned in the methodology and parameter description sections.

2.3. Methodology

The model of social and environmental impact is based on four key calculations: The equivalent full-time employment (indicative of potential job creation), net earnings (indicative of financial viability), economy of scale (indicative of scale-appropriate mechanisation for labour productivity) and greenhouse gas emission analysis (indicative of environmental impact). These four calculations are outlined below:

2.3.1. Full-Time Employment Equivalent Calculation

The FTE to produce recommended vegetable intake under mixed and mid to high-value vegetables is calculated under the base yield scenario and is given by:

$$\mathrm{FTE}=\frac{\frac{{\mathrm{L}}_{\mathrm{req}.}}{{\mathrm{N}}_{\mathrm{con}.}}}{260},$$

(1)

where

FTE = Full-time employment equivalent,

L_req. = Number of days of labour required to feed people,

N_con. = Number of people with vegetable consumption.

260 = Number of effective working days in a year (assuming five days of full-time work in a week).

Owing to the diversity in approaches to UA, the FTE is calculated based on the operating and non-cultivating labour data required to produce urban vegetables using two distinct labour scenarios: a high productivity scenario derived from the Small Plot Intensive (SPIN) farming method (i.e., a highly efficient labour use case for commercial urban farming) [54] and a high labour input derived from the Edible Garden project (i.e., urban gardening) [55].

The number of days of labour required (L_req.) to feed the people is calculated through the following formula:

$${\mathrm{L}}_{\mathrm{req}.}=\frac{{\mathrm{L}}_{\mathrm{hrs}.}}{8},$$

(2)

where

L_hrs. = Total labour hours for production.

The total labour hours (L_hrs.) is divided by 8 assuming the ‘eight-hour work day’ as a universal practice for full-time daily work.

The total labour hour (L_hrs.) is calculated based on the following formula:

$${\mathrm{L}}_{\mathrm{hrs}.}={\mathrm{A} \times \mathrm{L}}_{\mathrm{sqm}.},$$

(3)

where

A = Area available for cultivation (m²),

L_sqm. = Labour input (hours/m²/year).

Likewise, the vegetable availability is calculated based on the area available multiplied by the base yield. The total production (Y_veg.) is assumed as the total available quantity of vegetables available to consumers. The vegetable requirement is calculated based on the recommended 5-cup standard serving of mix-vegetables (one standard serving equivalent to 75 g including cole, root, leafy, fruit vegetables and legumes) [56] and finally, the number of people to feed mixed or mid to high-value crops (N_con.) is calculated as below:

$${\mathrm{N}}_{\mathrm{con}.}=\frac{{\mathrm{Y}}_{\mathrm{veg}.}}{{\mathrm{R}}_{\mathrm{veg}.}},$$

(4)

where

Y_veg. = Vegetable production (kg/year),

R_veg. = Vegetable requirement (kg/person/year).

The total production of vegetables (Y_veg.) is calculated based on the base yield of UA in Adelaide and Kathmandu Valley by assuming total UA production as vegetables only. The calculation is done based on the following method:

$${\mathrm{Y}}_{\mathrm{veg}.}={\mathrm{Y}}_{\mathrm{bas}.} \times \mathrm{A}$$

(5)

where

Y_bas. = Base productivity of UA crops (kg/m²).

2.3.2. Net Earnings Calculation

The net earnings (I_net.) for the intensive labour use (non-mechanised) and mechanised UA under mixed vegetable farming and high to mid-value vegetables is calculated as follows:

$${\mathrm{I}}_{\mathrm{net}.}={\mathrm{I}}_{\mathrm{tot}.}-{ \mathrm{C}}_{\mathrm{tot}.}$$

(6)

where

I_tot. = Total income from the mix or mid to high-value vegetable production in $,

C_tot. = Total labour cost for production ($).

The total income from mixed or mid to high-value crops (I_tot.) is calculated through the following formula:

$${\mathrm{I}}_{\mathrm{tot}.}={\mathrm{Y}}_{\mathrm{veg}.}{ \times \mathrm{P}}_{\mathrm{ave}.}$$

(7)

where

P_ave. = Average retail price of mix or mid to high-value vegetables ($)

The total labour cost (C_tot.) is calculated through the following formula:

$${\mathrm{C}}_{\mathrm{tot}.}={\mathrm{L}}_{\mathrm{req}.}{ \times \mathrm{C}}_{\mathrm{lab}.}$$

(8)

where

C_lab. = Average hourly labour rate ($).

The Nepalese currency was converted to an Australian dollar value for ease of comparison of case study findings (87 Nepalese rupees equivalent to $1AUD based on the typical rate observed in 2022). The model assumes a typical retail price can be achieved for UA produce, either through selling direct to consumers or for self-provisioning (opportunity cost). The average retail price of mixed vegetables (possible value generated from the combination of the root, legumes, fruits and leafy vegetables, including brassicas) and mid to high-value vegetables (possible retail value through the cultivation of tomato, capsicum, beans, coriander, broccoli, lettuce, spinach, basil and celery) was acquired from a review of prices currently promoted by the Adelaide central market in Adelaide (available online: https://shop.adelaidecentralmarket.com.au accessed on 10 June 2022) and the Kalimati Fruit and Vegetable Market in Kathmandu (available online: https://ramropatro.com/vegetable accessed on 10 June 2022 ).

2.3.3. Economy of Scale Calculation

The production cost (C_prod.) relating to the economy of scale for different scales under non-mechanised UA and small-scale mechanisation is calculated using published labour data available from the SPIN farming guide [54] and Edible Gardens Project [55] through the following formula:

$${\mathrm{C}}_{\mathrm{prod}.}=\frac{{\mathrm{C}}_{\mathrm{eqp}.}\times \max \left(\frac{1}{{\mathrm{LS}}_{\mathrm{ass}.}},\frac{{\mathrm{H}}_{\mathrm{ope}. }\times \mathrm{A}}{{\mathrm{R}}_{\mathrm{hrs}.}}\right)+{\mathrm{H}}_{\mathrm{ope}}\times \mathrm{A} \left({\mathrm{C}}_{\mathrm{lab}}.+{\mathrm{C}}_{\mathrm{run}}.+{\mathrm{C}}_{\mathrm{fue}.}\times {\mathrm{F}}_{\mathrm{hrs}}.\right)+{\mathrm{LH}}_{\mathrm{nc}. }\times {\mathrm{C} }_{\mathrm{lab}. }\times \mathrm{A}}{\mathrm{A}}$$

(9)

where

C_eqp. = Equipment cost ($),

LS_ass. = Asset life span(years),

H_ope. = Operating hours (hours/m²),

R_hrs. = Replacement (hours),

C_run. = Running cost ($/hour),

C_fue. = Fuel cost ($/litre),

F_hrs. = Fuel consumption (litre/hour),

LH_nc. = Non-cultivating labour (hours/m²).

The operating hours (H_ope.) for non-mechanised farming, garden tiller or cultivator is estimated using the following formula:

$${\mathrm{H}}_{\mathrm{ope}.}=\frac{1}{{\mathrm{A}}_{\mathrm{til}.}} \times 4+\frac{1}{{\mathrm{A}}_{\mathrm{bed}.}}\times 2$$

(10)

where

A_til. = Area covered during tillage (m²/hour),

A_bed. = area covered during bed preparation (m²/hour).

The multiplication factor of 4 is applied to account for the standard practice of tilling twice before cultivation for summer and winter season vegetables, while factor 2 is applied for bed preparation activities before sowing winter and summer season vegetables.

The fuel consumption (in L/hour) during tillage and bed preparation for a garden tiller and a cultivator was taken from machine specifications for typical small-scale cultivation equipment from Grafton power products (Available online: https://www.graftonpowerproducts.com.au/listing/honda-gx25-mini-4-stroke-engine/ accessed on 15 July 2022).

The rates of cultivation, i.e., tillage area (A_til.) and bed preparation area (A_bed.), are derived from information from Satzewich and Christensen [54]. For this, the assumed velocity of the machine is multiplied by the working width using the following formulas.

$${\mathrm{A}}_{\mathrm{til}.}={\mathrm{V}}_{\mathrm{til}.}{ \times \mathrm{W}}_{\mathrm{wid}. }\times 1000$$

(11)

where

V_til. = assumed velocity for tillage activity (km/hour)

W_wid. = Working width (m)

Similarly,

$${\mathrm{A}}_{\mathrm{bed}.}={\mathrm{V}}_{\mathrm{bed}.}{ \times \mathrm{W}}_{\mathrm{wid}. }\times 1000$$

(12)

where

V_bed. = assumed velocity for bed preparation activity (km/hour).

2.3.4. Carbon Dioxide Emission Calculation

Carbon dioxide is a major source of Green House Gases (GHGs). The distance and area-wise calculation of CO₂ emissions during tilling and distribution were estimated in this study. Other emissions (such as N₂O emissions from soil, agrochemical use, on-farm postharvest activities, embodied emissions from the water supply for irrigation and machine emissions) were assumed to be common to vegetable production at both UA scale and commercial scale, and only the fuel emission from a machine used during production and car emission during the distribution of UA produce were considered. So, the total amount of carbon dioxide emission per kg of production, E_tot. (kg CO₂ equivalent) from production to distribution activities were determined based on the available information using the following method:

$${\mathrm{E}}_{\mathrm{tot}.}=\frac{({\mathrm{E}}_{\mathrm{prod}.+{\mathrm{E}}_{\mathrm{dist}.}})}{{\mathrm{Y}}_{\mathrm{veg}.}},$$

(13)

where

E_prod. = Emissions from fuel during production activities (kg CO₂ equivalent)

E_dist. = Emissions during distribution activities

The total amount of emissions for mechanised and non-mechanised UA during production E_prod. (CO₂ equivalent per kg production) is calculated using the following method:

$${\mathrm{E}}_{\mathrm{prod}.}={\mathrm{E}}_{\mathrm{til}.}+{ \mathrm{E}}_{\mathrm{bed}. },$$

(14)

where

E_til. = Emission during primary tillage,

E_bed. = Emission during bed preparation.

The emission for machine use in primary tillage (E_til.) and bed preparation (E_bed.) for the petrol engine is calculated as below:

$${\mathrm{E}}_{\mathrm{til}.}{\mathrm{and} \mathrm{E}}_{\mathrm{bed}.}=\frac{{\mathrm{A} \times \mathrm{F}}_{\mathrm{hrs}.{ \times \mathrm{H}}_{\mathrm{ope}.}{\times \mathrm{E}}_{\mathrm{fac}.}}}{1000},$$

(15)

where

E_fac. = Emission factor (gram/litre of petrol use)

The amount of emissions during distribution E_dist. for non-mechanised and mechanised UA is calculated through the following formula:

$${\mathrm{E}}_{\mathrm{dist}.}={\mathrm{E}}_{\mathrm{fac}.}{ \times \mathrm{F}}_{\mathrm{con}. }\times 2{ \times \mathrm{D} \times \mathrm{T}}_{\mathrm{num}.}$$

(16)

where

F_con. = Per kilometre fuel consumption by car during distribution (litre/km),

D = Travel distance (km),

T_num. = Number of trips.

The number of trips per year (T_num.) is calculated as follows:

$${\mathrm{T}}_{\mathrm{num}.}=\max \left(52,\mathrm{ROUNDUP}\left(\frac{{\mathrm{Y}}_{\mathrm{veg}.}}{\mathrm{F},0}\right)\right),$$

(17)

where

ROUNPUP implies rounding up to the nearest integer values

F = Maximum freight capacity (kg/trip)

T_num. is taken as the maximum of either 52 (i.e., one trip per week, assumed to be the minimum frequency to bring produce to consumers) or the value calculated by dividing the production quantity by the vehicle capacity (i.e., more than one trip per week). The distribution process was simplified to a straightforward activity of driving, dropping and returning from the point of production.

2.3.5. Parameter Description

The list of values used in the above equations is presented in Table 1.

Table 1. Descriptions and list of values used in calculations.

Symbol and Unit	Description	Values		Sources
		Adelaide	Kathmandu Valley
Lsqm. (hour/m2/year)	Cultivating labour hours for non-mechanised production	0.442	0.442	[54]
	Non-cultivating labour hours	0.146	0.146
	Non-cultivating labour hours	2.31	2.31	[55]
Ybas. (kg/m2)	Base productivity of UA vegetables	2.21	1.95	[55,57]
Rveg. (kg/person/year)	Per person per year requirement of mix or mid to high-value vegetables based on recommendation	136.875 (mix) 54.75 mid to high-value vegetables	136.875 (mix) 54.75 mid to high-value vegetables	[56]
Pave. ($/kg)	The average price of mixed vegetables	6.8	2.0	[58,59]
	The average price of mid to high-value vegetables	10.0	2.86
Clab. ($/hour)	The average hourly labour rate	25.88	0.875	[60,61]
Ceqp. ($)	Cost of
	Manual equipment	20	10	Approximated cost for hand-digging equipment like a spade
	Cost of the garden tiller with bed maker	1000	800	Approximated cost based on Honda mini 4-stroke engine tillers with bed maker
	Cost of the garden cultivator with bed maker	2000	1600	Approximated cost based on single cylinder 4-stroke engine cultivator with bed maker
LSass. (years)	Asset life span manual	1	1	Approximated years
	Asset life span tiller	5	5
	Asset life span cultivator	5	5
Rhou (hours)	Asset replacement hours manual	2000	2000	Approximated replacement hours
	Asset replacement hours garden tiller	2000	2000
	Asset replacement hours cultivator	2000	2000
Cfue. ($/litre)	Fuel cost	1.75	1.95	[62]
Efac. (gram)	Emission factor per litre of petrol use	2392	2392	[63]
Fhrs. (litre)	Garden tiller	0.54	0.54	[64,65]
	Garden cultivator	1.98	1.98
V (km/hour)	Manual	0.26	0.26	Calculated based on the SPIN farming guide [54]
	Garden tiller and cultivator	1.32	1.32
Wwid. (meter)	Manual	0.1	0.1	[64,65]
	Garden tiller	0.28	0.28
	Garden cultivator	0.48	0.48
Fcon. (litre/km)	Car petrol consumption during distribution	0.111	0.074	[66,67]
F (kg/trip)	Maximum freight capacity	200	200	Assuming medium-sized passenger car/small petrol vehicle

Table 1. Descriptions and list of values used in calculations.

Symbol and Unit	Description	Values		Sources
		Adelaide	Kathmandu Valley
Lsqm. (hour/m2/year)	Cultivating labour hours for non-mechanised production	0.442	0.442	[54]
	Non-cultivating labour hours	0.146	0.146
	Non-cultivating labour hours	2.31	2.31	[55]
Ybas. (kg/m2)	Base productivity of UA vegetables	2.21	1.95	[55,57]
Rveg. (kg/person/year)	Per person per year requirement of mix or mid to high-value vegetables based on recommendation	136.875 (mix) 54.75 mid to high-value vegetables	136.875 (mix) 54.75 mid to high-value vegetables	[56]
Pave. ($/kg)	The average price of mixed vegetables	6.8	2.0	[58,59]
	The average price of mid to high-value vegetables	10.0	2.86
Clab. ($/hour)	The average hourly labour rate	25.88	0.875	[60,61]
Ceqp. ($)	Cost of
	Manual equipment	20	10	Approximated cost for hand-digging equipment like a spade
	Cost of the garden tiller with bed maker	1000	800	Approximated cost based on Honda mini 4-stroke engine tillers with bed maker
	Cost of the garden cultivator with bed maker	2000	1600	Approximated cost based on single cylinder 4-stroke engine cultivator with bed maker
LSass. (years)	Asset life span manual	1	1	Approximated years
	Asset life span tiller	5	5
	Asset life span cultivator	5	5
Rhou (hours)	Asset replacement hours manual	2000	2000	Approximated replacement hours
	Asset replacement hours garden tiller	2000	2000
	Asset replacement hours cultivator	2000	2000
Cfue. ($/litre)	Fuel cost	1.75	1.95	[62]
Efac. (gram)	Emission factor per litre of petrol use	2392	2392	[63]
Fhrs. (litre)	Garden tiller	0.54	0.54	[64,65]
	Garden cultivator	1.98	1.98
V (km/hour)	Manual	0.26	0.26	Calculated based on the SPIN farming guide [54]
	Garden tiller and cultivator	1.32	1.32
Wwid. (meter)	Manual	0.1	0.1	[64,65]
	Garden tiller	0.28	0.28
	Garden cultivator	0.48	0.48
Fcon. (litre/km)	Car petrol consumption during distribution	0.111	0.074	[66,67]
F (kg/trip)	Maximum freight capacity	200	200	Assuming medium-sized passenger car/small petrol vehicle

3. Results

3.1. Full-Time Employment (FTE) Equivalent Calculations

The full-time employment calculation in Adelaide and Kathmandu Valley was done under non-mechanised and small-scale mechanisation conditions for mixed and mid to high-value vegetable production scenarios using two labour cases, high-intensity commercial production (from the SPIN farming guide) and low-intensity gardening (from the Edible Gardens project). Results are shown in Figure 4A,B, illustrating the potential employees (farmers) per consumer—i.e., a value of 0.01 would be one farmer per 100 consumers, and 0.1 would be 1 per 10 consumers.

In Adelaide (Figure 4A), the area-wise FTE under non-mechanised UA practices under both contexts reveals a relatively higher level of employment for mixed vegetables compared to mid to high-value vegetables. In the case of mechanisation (i.e., garden tiller and cultivator), there is a negligible difference in FTE between the types of machines used. However, mechanisation (using either machine) substantially reduced the FTE compared to non-mechanised UA, which is much lower in mixed to high-value vegetables.

In Kathmandu Valley (Figure 4B), similar trends of FTE were observed, with a slightly higher FTE than in Adelaide. Despite a slight difference in UA productivity (i.e., 1.95 in Kathmandu versus 2.21 kg/m²/year in Adelaide), there is a much larger range of land parcel area (almost 3-fold) in Kathmandu Valley (127.2 m² to 7949.1 m²) compared to Adelaide (154 m² to 2705 m²) resulting in slightly higher potential FTE in Kathmandu Valley.

The nearly six-fold difference in FTE between the SPIN and Edible Gardens project data is due to the six-fold difference in labour use between those two modes of farming. It is important to note that, in the Edible Gardens project study, a much larger fraction of labour is allocated to non-cultivation gardening activities than in the SPIN case, so the addition of labour-saving technology through mechanised cultivation has only a minor impact on overall labour input. In the SPIN-Farming cases, mechanisation has a more profound impact on FTE due to the proportionally larger reduction in labour.

3.2. Net Earnings Calculation

The net earnings, i.e., the value from mixed and mid to high-value vegetables minus labour cost, were calculated for Adelaide and Kathmandu Valley under non-mechanised and mechanised conditions for urban gardens (i.e., labour use from Edible Gardens) and commercial urban farming (SPIN labour data) contexts (Figure 5A,B).

In Adelaide (Figure 5A), mechanisation substantially helped to improve returns under mid to high-value vegetables in the SPIN farming scenario (nearly 1.5 to 2 times higher than mixed vegetables under non-mechanised conditions). Meanwhile, the Edible Gardens data shows the margins spanning from negative to positive in smaller plots (negative margins to slight gains), and larger plots showed highly negative margins due to high labour input requirements. This result supports the findings of Kafle et al. [40], who investigated the economic viability of UA using similar labour input assumptions from the Edible Gardens case without mechanisation.

In contrast to Adelaide, a low return from mechanisation was observed in Kathmandu Valley under the SPIN farming labour use (almost three times) with much lower margins due to a less competitive market price (Figure 5B). However, the margins are still positive under the Edible Gardens labour use scenario as the hourly labour cost for cultivating crops is nearly 30 times cheaper in Kathmandu Valley ($0.875/hour) compared to Adelaide ($ 25.88/hour).

The intensity of labour use and mechanisation, the scale of cultivation and the selection of crops were identified as key factors influencing the net return of UA in Adelaide and Kathmandu Valley. For commercial UA (i.e., SPIN farming), there is a 2.5-fold difference per square meter labour hour between mechanised and non-mechanised (0.442 hours/m²/year for non-mechanised, 0.1737 hours/m²/year for the garden tiller and 0.1623 hours/m²/year for garden cultivator). While, in the case of gardening (i.e., based on Edible Gardens labour use), there is a slight difference in labour hours (2.61 hours/m²/year for non-mechanised, 2.34 hours/m²/year for the garden tiller and cultivator respectively). The six-fold difference in labour use between the two modes of farming is the prime factor for the high difference between SPIN and Edible Garden type of UA’s net earnings in Adelaide (leading toward highly negative margins in Edible Gardens case), as the labour cost of Australia is the third highest globally [68]. The earnings difference between non-mechanised UA with large and small plots is mainly due to the high labour inputs required to cover a larger area.

3.3. Economy of Scale

Scale appropriateness is a major factor in determining the sustainability of agricultural practice. The possible impact of mechanisation in urban vegetable farming was evaluated using two sources of labour data: ‘SPIN farming’ (representing an idealised case of high-productivity small-scale commercial urban farming) and non-cultivation labour from research on garden labour input from the ‘Edible Gardens’ project (representing non-commercial household gardening) [54,55]. Scenarios were analysed in Adelaide and Kathmandu under non-mechanised, garden tiller, and garden cultivator conditions (Figure 6A,B and Figure 7A,B).

Using the idealised SPIN farming data in Adelaide (Figure 6A,B) suggests that non-mechanised UA may have the lowest production costs for very small scales (up to 30–40 m²), but beyond this scale, garden tillers become economically more efficient. The garden cultivator seemed more efficient than non-mechanised UA from a slightly larger scale (about 50 m²) but did not improve the efficiency of tillers until larger scales (approximately 900 m²), and even then, the differences in production cost between the two machines were minimal.

Using the non-cultivation labour data from the Edible Gardens project in Adelaide gives a similar pattern but different values to SPIN farming, as the non-cultivation labour (in hours/m²) is almost 16 times higher than SPIN farming. Again, non- mechanised UA seemed economic at very small scales (up to 20–30 m²), beyond which tillers became cheaper. The cultivator showed similar results as SPIN farming. Both forms of mechanised UA showed only a minor gain over non-mechanised UA due to the very high proportion of non-cultivation labour in gardening.

Using SPIN farming assumptions, UA in Kathmandu Valley (Figure 7A,B) shows a potential advantage of non-mechanised UA (up to approximately 1000 m²) compared to both forms of mechanisation. Due to cheap labour costs in Kathmandu, the maximum scale for non-mechanised UA is far higher (by approximately 30-fold) than in Adelaide. Beyond this scale, the garden tiller and cultivator became approximately equivalent to non-mechanised UA, but there is no major economic advantage from mechanisation up to the maximum adopted land parcel. A similar pattern emerges using the Edible Gardens labour use data, albeit with a larger overall production cost due to the increased non-cultivation labour input.

3.4. Carbon-Dioxide Emissions Calculation

The distance and area-wise CO₂ equivalent per kg of vegetable production were calculated for Adelaide and Kathmandu Valley. Emissions from the machine used during production and petrol emissions from a car during product distribution have been calculated as these are the dominant points of difference stemming from the different scales at which UA takes place.

The area-wise CO₂ emission analysis for Adelaide and Kathmandu under non-mechanised and mechanised vegetable production in smaller and larger plots is given in Figure 8A,B.

In Adelaide (Figure 8A), the emission difference between non-mechanised and mechanised urban vegetable farming seems small, as the emissions are dominated by vehicle fuel consumption during distribution. Emissions from the smaller plots are higher than larger plots emissions. The fact that there is little difference between non-mechanised and mechanised results points to the dominance of transport-related emissions, with high variability (particularly in smaller plots) due to the combination of different travel distances and varying degrees of utilisation of the vehicle making the weekly car trip, which depends on the volume of production being generated from the different-sized land parcels.

The per kg emissions in Kathmandu Valley (Figure 8B) seemed lower compared to Adelaide due to the very short distribution distance (i.e., 19 km compared to 57 km in Adelaide), with the very high emissions from the smaller plot and the little difference between mechanised and non-mechanised UA. However, compared to Adelaide, the smaller plot total emissions were slightly lower in Kathmandu as most of the smaller plots lie within the inner city, and as transport-related emissions are dominant, the shorter distances give lower emissions overall. The per kg emissions from smaller plots were 3.7 to 5.5 times higher than the larger plots due primarily to differences in the efficiency of utilisation of the car as a mode of transport. The variability of emissions was high in smaller plot UA in Adelaide; the cases where higher emissions arose in smaller plots in both contexts were due to the plots being located in the suburban area, leading to substantially higher emissions from long transport distances compounding the under-utilisation of the vehicle.

The distance-wise CO₂ emission equivalent per kg of production for Adelaide and Kathmandu Valley is analysed in Figure 9A,B. In Adelaide (Figure 9A), the suburban area emissions were almost 3.5 to 3.9 times higher than emissions from plots in the inner city under both non-mechanised and mechanised conditions. Under both distance categories, there was little difference between mechanised and non-mechanised UA emissions, again suggesting the dominance of the adopted transport-related emissions over the adopted cultivation-related emissions. High variability in the suburban context resulted from large variability between plots in terms of both distance and scale, the latter leading to differences in vehicle use.

In Kathmandu Valley (Figure 9B), there is a similarly large difference in emissions between the suburban and inner-city categories (with suburban 3.5 to 4.2 times higher than the inner city). As with Adelaide, the mechanised and non-mechanised UA show a similar number of emissions, but higher variability of emissions has been observed in the suburban context due to variability of land parcels (i.e., a larger range of parcel sizes in suburban areas). The per kg distance-wise emissions were slightly higher in Kathmandu Valley, but high variability was similarly observed in Adelaide.

3.4.1. Comparative Emission Analysis

Emissions from Fuel Use

The CO₂ emissions per kg of vegetable production from fuel use during production and from Adelaide and Kathmandu have been calculated (Figure 10) and compared with the data from the Australian vegetable industry [69]. Their study estimated the CO₂ emissions from fuel and total emissions from commercial vegetable farming. The comparative fuel emissions from small machine use during production (i.e., tillage and bed preparation) activity in UA and the Australian vegetable industry is shown in Figure 10. Though the commercial vegetable industry in Australia uses large machines, almost similar potential fuel emissions from mechanisation have been observed (i.e., 0.02 to 0.04 kg per kg of production). This analysis showed potential for small-scale mechanisation for production activities as the observed potential emissions are in line with commercial vegetable production.

Distribution and Total Emissions

The total CO₂ emissions per kg of vegetable production and distribution from the present study have been compared with a similar study conducted at a global scale by Poore and Nemecek [70], which is presented in Figure 11. The emissions analysis from selected vegetables (tomatoes, brassicas, other vegetables, root vegetables, onions, and leeks) have been included for comparison. The share of distribution as part of total emissions was almost 16% in their study. Their study accounted for emissions from land use change, farm, processing, transport, retail, packaging and losses. Without considering the above factors in our study, the share of emissions due to transport only seems large due to the assumption of a small vehicle (car) for distributing products at weekly intervals, as opposed to larger vehicles with higher product loads. The observed three-fold difference in kg of emissions between Adelaide and Kathmandu Valley is due to the distribution distance (three-fold higher distribution distance in Adelaide), as emission is directly related to trip distances.

4. Discussion

4.1. Full-Time Employment Equivalent and Earnings

Mechanisation and types of UA crops have led to significant differences in projected employment (expressed as FTE) and earning potential. UA under mixed vegetable farming showed better employment ability compared to mid to high-value cultivation under both contexts due to the higher demand (per consumer) for mixed vegetables compared to mid to high-value in daily life to meet dietary requirements, thus demanding higher production that corresponds to more FTE. FTE is directly related to the food demand (i.e., the number of consumers per unit area). Counterintuitively, higher yields may work against improved employment outcomes, as less labour may be required per unit of food produced. Likewise, mechanisation improves labour efficiency and thereby improves net earnings (and thus economic viability) but reduces the potential employment per consumer.

The production efficiency and economy of scale of UA can be improved by mechanisation in the long run [71]. In our study, small-scale mechanisation has significantly reduced FTE, thus creating labour use efficiency and improving net earning potential. Interestingly there was minimal difference in FTE between small tillers and cultivators, but this could change depending on several fixed assumptions in this study, such as lifespan and the ongoing maintenance costs of the different machines.

The cost of labour, productivity, crop mix, and market price are the most critical factors limiting UA viability in Adelaide [40]. The labour cost under two vegetable mixes has been investigated to explore the potential impact on savings under non-mechanised and mechanised urban vegetable production in Adelaide and Kathmandu. The low likelihood of UA viability, particularly in Adelaide, as also estimated by Kafle et al. [40], has been greatly improved, primarily through the switch to SPIN-Farming labour input assumptions. Adding mechanisation further improves viability. However, when we use low-intensity Edible Garden labour assumptions, there is no viable economic case (irrespective of whether we include mechanisation or not). The lower gains in Kathmandu are more related to the low labour cost, making it less advantageous to adopt labour-saving practices (such as mechanisation). However, mechanisation has added economic advantage in both contexts compared to non-mechanised UA. Though the relative economic advantage is different, there may be a long-term positive impact of mechanisation even in low-wage settings where there is an acute labour shortage in agriculture.

4.2. Economy of Scale

Scale appropriateness is one of the major factors to be considered when planning production practices in UA (i.e., non-mechanised and mechanised farming practices). Scale-appropriate mechanisation based on the land–labour–machine nexus helps properly divide and allocate labour according to need [72]. Our study has identified the interplay between the amount of labour and the level of mechanisation as key factors in the viability of UA at different scales. However, there is substantial uncertainty in non-cultivation labour in UA. To this end, we have taken two extremes: The commercial UA (SPIN farming) and urban gardening (Edible Gardens) approaches. We have considered both approaches under non-mechanised and small-scale mechanisation (garden tiller and cultivators) in an attempt to shed light on the optimum scale for technological interventions.

The maximum scale for non-mechanised farming under Adelaide’s high labour cost conditions was very different to that of Kathmandu due to the extremely low labour cost in the latter case. Assuming the SPIN labour scenario, mechanisation significantly improved scale efficiency in Adelaide from relatively small garden scales (approximately 30 m² and above). However, if production practices were to adopt the high level of non-cultivation labour from the Edible Gardens project, the overall impact of mechanisation would be minimal as labour costs remain high for non-mechanised activities. In Kathmandu, under both high and low assumptions for non-cultivation labour, UA seems feasible under non-mechanised conditions, and the garden tiller showed marginal improvements to economic efficiency on larger scales (around 1000 m² or more).

In the context of improving labour use efficiency suggested through this study, further research is required to improve understanding and reduce uncertainty around non-cultivating labour hours, primarily focusing on countries/cities with high labour costs for improving scale and efficiency of production [16].

4.3. Carbon Dioxide Emissions

In this study, we have considered distance and area-wise potential fuel emissions from non-mechanised and mechanised practices during production and petrol emissions from a car for distributing produce. Larger emissions (per kg of produce) were estimated for smaller plot UA compared to larger plots under both contexts due to inefficient distribution (i.e., fixed emissions of the car transport irrespective of the amount of produce distributed). This was exacerbated in suburban areas due to longer distribution distances. This result supports the findings by Hu et al. [73], who found that UA emissions are significantly affected by factors including produce distribution practices.

There have been limited studies on emissions from fossil fuel use in vegetable farming, and no study has been found focusing on the impact of different scales of mechanisation and resultant carbon emissions in UA. In a study on 23 commercial vegetable crops in Australia, the average emissions were 0.46 kg equivalent per kg production per year. The highest proportion is from energy used in electricity for irrigation and postharvest on-farm activities (65%), followed by N₂O emission from fertiliser use (17%) and agrochemicals (10%) [69]. That study found that fossil fuel and on-farm machine emissions amounted to approximately 8% of emissions. We assumed other on-farm emissions as equivalent between UA and commercial farming and calculated only the emissions from fuel during machine use in production and distribution emissions. The fuel emissions from the small machine use during tilling and bed preparation estimated in this study were almost the same as that of commercial vegetable production in Australia.

In a study by Poore and Nemecek [70], vegetables, including tomatoes, root vegetables, brassicas and other vegetables, were calculated to annually emit 0.814 kg equivalent CO₂ per kg of production globally from land use change, farm, processing, transport, retail, packaging and losses. Farm-level production was calculated as approximately 38% of the total (about 0.3 kg CO₂ per kg), with food transport emissions of approximately 0.12 kg CO₂ per kg (about 15% of total emissions). In their analysis, they considered large vehicle transport through sea and road. In contrast, the major share in the current study was from product distribution due to the assumption of transport occurring at least weekly and using a small vehicle driving to and from the CBD. Our study showed a larger share of transport-related emissions due to a small vehicle with presumably low capacity (200 kg/trip), a frequent trip from the point of production to consumption (maximum 52 per year) and assumption of transport as driving, dropping the product and returning to the point of production. Reduction of GHGs emissions compared to our assumption and conventional practices are possible through the adoption of lower emission transport or zero transport by ensuring that suburban UA supplies produce directly to nearby households within a neighbourhood or through self-provisioning at the individual household scale.

Past studies have suggested agricultural GHG emissions may be lowered through improved cultivation and input management practices [74]. In this context, there is huge scope for improving cultural practices like organic farming that may contribute to reduced pesticide use and N₂O emissions from soil which contribute 10% and 17% in total vegetable production emissions, respectively [69]. Researchers such as Venkat [75] have pointed out the scope for reducing GHG emissions through increased soil organic carbon stocks. Adopting these approaches may however impact productivity and crop quality and have flow-on effects for the economic feasibility of UA sites but may be helpful to harness the added advantage of reduced distances and external input use, particularly environmental.

On average, the share of emissions from distribution was far higher in both contexts compared to machine emissions during production, which resulted in an almost negligible difference between non-mechanised and mechanised UA. However, UA has been promoted as a mitigation strategy against food supply disruptions [76]. Our study pinpoints that small vehicle distribution to access markets to sell produce (based on a function of distance and production volume) seems inefficient in terms of emission reduction. To that end, the statement made by Lam [23] on shifting long-distance agriculture into local production reduces reliance on highly price volatile fuels and reduces emissions that may not be justified in the UA context based on the results of our study. Importantly, however, Maassen and Galvin [77] conducted a study in Rosario, Argentina, which showed a CO₂ emission reduction of up to 95% through self-production (i.e., localising UA with zero transport requirement) compared to products imported from a long distance. In a review study done by Kiss et al. [78], it was noted that the environmental advantage of a short supply chain was through reduced carbon emissions, minimal processing and packaging and extensive production methods of local food production. Likewise, a study in Seoul, South Korea, implementing UA over a 51.51 square kilometre area (8% of the total metropolitan area) showed a potential of CO₂ emissions reduction by 11.67 million kg a year, equivalent to planting 20 square kilometres of 20-year-old pine forest [23]. Distribution using electric vehicles could further reduce emissions from the levels [49]. Other opportunities to reduce GHG emissions exist in exploring aggregation of distribution (e.g., multiple small UA farms cooperatively using a single larger vehicle—a van or small truck—to distribute produce) as a practical strategy to reduce the emissions associated with getting produce to market.

Thus, carefully considering production practices and distribution mechanisms (hyper-localised production within the neighbourhood, self-provisioning by households or aggregated distribution in a large vehicle) are needed to harness UA’s potential environmental (GHG emission) benefits compared to conventional farming approaches.

5. Conclusions

This study has demonstrated a method for quantifying social, economic and environmental benefits from urban vegetable farming, focusing on labour use and the impact of small-scale mechanisation on employment, income and carbon dioxide emissions from two samples of 40 parcels of land each from Adelaide, Australia and Kathmandu Valley, and Nepal. The comparative socio-economic and environmental nexus study between highly developed and low-dense cities like Adelaide, Australia and developing and highly populated cities like Kathmandu, Nepal, is done to explore the benefits and impacts of UA practices with a particular focus on employment and scale efficiency through small scale mechansiation and carbon emission.

The labour input from small plot intensive commercial UA and research on garden labour input from the Edible Gardens project (non-commercial household gardening) suggests non-mechanised UA may have the lowest production costs for very small scales (up to 40 m²) in Adelaide, with substantial benefits from mechanisation especially if non-cultivation labour can be minimised through efficient production practices. Meanwhile, in Kathmandu non-mechanised production was shown to be potentially competitive in up to a 30-fold larger area than in Adelaide due to very cheap labour costs, and beyond this, there is only a marginal economic advantage from small-scale mechanisation.

This analysis highlighted the importance of labour-efficient workflow (particularly non-cultivation labour), as well as the potential improvement from scale-appropriate mechanisation as part of the economic viability of UA. This was shown to be particularly relevant in high-wage settings such as Adelaide. From the analysis of the CO₂ emissions and comparing this with studies of conventional commercial vegetable production, the fuel emissions from machine use during production appear comparable between UA and commercial production, but a larger fraction of emissions has emerged from the assumed use of petrol-driven small vehicles driven to distribute produce. Small vehicles (cars) are deemed inappropriate for distributing UA products. Interestingly, Kafle et al. [40] showed the distribution as a minor part of the economic cost. This current study shows that whilst distribution might not impact economic costs; it could undermine the case for environmental benefit (reduced GHG emissions through transport) by adopting UA.

The adoption of labour-saving mechanisation improves efficiency/viability but reduces employment and adds GHG emissions. Likewise, growing higher-value crops improves earnings but employs fewer farmers per consumer. These trade-offs should be considered when UA activities are planned or proposed. Scale-appropriate mechanisation, a proper selection of the optimal crop mix, labour planning and transport mode that links the product with the market through alternative distribution mechanisms (like highly localised, i.e., neighbourhood-scale, farmers’ markets), local food swaps, and farmgate sales can potentially create a substantial social and environmental benefit along with added economic advantages.

A key limitation of this study is that it has excluded controlled environment UA (such as greenhouse cultivation), and future research is recommended to improve the understanding of how such production performs in terms of social and environmental impact. Moreover, past research showed the contribution of fuel emissions as only a relatively small fraction (7%) of total emission, with a much larger proportion of emissions generated from activities like energy use during irrigation, production and postharvest activities, pesticide and fertiliser use, further investigation is needed in the UA context to understand and quantify these emissions, particularly to determine if they are the same, higher or lower than large scale commercial agriculture per unit of produce. A study of the circular economy benefits of Alternative Food Networks (AFNs) in Australia showed the potential to reduce environmental problems [79]. The results of this study suggest that conventional approaches (e.g., distributing to centralised markets) may not deliver the intended outcomes, and future studies may include the impacts of such alternative movements (such as AFNs), which may create overall economic, environmental and social benefits from more localised distribution chains.

6. Policy Recommendations

Based on our study, the following policy measures are recommended for the sustainability of UA in the long run:

• The advantage of mechanisation is dependent on labour costs; if it is a priority for governments to see wage growth, then mechanisation will become advantageous in the future as labour costs increase. Thus, the government could identify ways to support and promote sustainable small-scale mechanisation to improve the viability and sustainability of UA.
• There is considerable scope for improving the current distribution system for UA produce (e.g., shared modes and methods of transport, as well as more localised/decentralised markets), ultimately reducing larger emissions during distribution.
• Planners and policymakers may consider ways for subsidised labour arrangements, especially for high wage setting scenarios, improvement into the current UA production practices with less environmental footprints and production based on market value and demand for better economic viability.