Performance of mixed and agroforestry systems: Deliverable 5.1

Simon Moakes, Philipp Oggiano

Research output: Book/ReportCommissioned report

22 Downloads (Pure)

Abstract

Introduction: With increasing pressure on agriculture to reduce its environmental impacts, it has been hypothesised that mixed farming and agro-forestry systems (MiFAS), either at a farm or landscape scale, could be an option to mitigate issues of nutrient excess, the import of synthetic fertilisers or feed production. Traditionally, mixed farming was practised to provide nutrients to crops through rotation breaks and to feed livestock, however the use of synthetic fertilisers and economies of scale has led to increased specialisation.

It has been proposed that the re-integration of cropping and livestock could close nutrient cycles and reduce imports of external feed or fertiliser nutrients, improve soil quality through returns of organic matter, as well as potential socio-economic benefits; although difficulties may occur, such as a loss of more profitable crops. A further potential climate change mitigation measure is the use of agroforestry within specialised agricultural systems to provide shading, soil stabilisation, drought resistant browse material for direct grazing or cutting, as well as directly offsetting GHG emissions through sequestration of carbon through biomass and or soil organic carbon (SOC). Task 5.1 therefore aimed to assess farm survey data (Task 2.3a) from existing MiFAS (Mixed and Agroforestry Systems), to assess their environmental and economic performance, as well as link to Task 5.2 to include coverage of potential labour issues.

Methodology: The assessment of the MiFAS within Task 5.1 employed a quantitative approach to undertake an LCA-based assessment of farm practices to generate results across a wide range of environmental and economic indicators through use of the FarmLCA model that includes both Lifecycle Impact Assessment (LCIA) and the German KTBL standard costs database. As per the guidance of ISO14040 and ISO14044 (ISO, 2006), we followed the recommended four steps to conduct an LCA, including a goal and scope definition phase. Our main goal was to assess the performance of MiFAS, based on data collected through farm surveys in multiple European countries. We also aimed to improve methodologies for assessing these systems and then to make a comparison so that conclusions and recommendations could be made. In terms of scope, we would conduct single farm assessments, with flexible boundaries either at the farm gate or at a smaller scale for some systems. Interactions, such as exchanges of straw or manures were treated as external inputs or outputs to the farm boundary, as these exchanges were assessed within WP3 (D3.4).

The second step, the inventory analysis phase, comprised collection of data for each farm and was linked to Tasks 2.2 and 2.3a where the on-farm data collection was developed and supervised. The survey was conducted in each network and were undertaken either as an on-farm interview or in some countries farm management data was already accessible electronically, with a follow up interview with the farmer. In general, data collection started in autumn 2022 and final queries were completed in autumn 2023.

Collected data was utilised from 9 networks for assessment through Task 5.1. For Denmark, NW02 was organised around reducing nutrient excesses and involved exchanges of manures with biogas plants and other farms, as well as returns of digestates. Whilst in Scotland, NW04 was focussed on the trialling of winter grazing of cereals by sheep, as well as other material exchanges such as straw and manures, but the network also included mixed farms with beef cattle. The German networks comprised NW5, focussed on peatland restoration of former intensively managed land, whilst the second German network NW06, comprised three farms developing agroforestry. In Switzerland NW07 comprised farms with high-stemmed fruit trees as an agroforestry system with grazing livestock and or crops. In France two networks comprised, NW09 with a focus on outdoor pig production in an agroforestry/woodland setting, and the second French network NW10 located in the SW region, included a range of mixed, arable and specialist livestock farms collaborating with partners to improve exchanges of materials and nutrients. In the east of Europe, the Romanian NW11 comprised farms collaborating to develop agri-tourism within the region of diverse small farms, whilst in Poland NW13, a large single farm comprised a biodynamic mixed farm, linking dairy and arable production.

Following data collection, data validation was critical due to the wide range of systems studied and we focused on agronomic data, such as yield, fertiliser use, animal herd and rationing values with

queries passed back to data enumerators for clarification. The assessment tool also included plausibility checks for nutrient requirements for crops or animals of a certain yield or animal type.

Each farm was modelled using the FarmLCA tool, which comprises LCA and economics modules and allows the individual nature of each farm and the management of their crops and livestock to be included. Subsequent sub-models estimating relevant emissions, such as enteric methane, nitrous oxides, or other pollutants, were based on methodologies recommended by IPCC and EMEP. The combination of on-farm data for external inputs together with outputs from the sub-models a farm specific LCA inventory was created. This is used to then calculate the impacts for different crops and livestock systems, which can be reported at various levels. Where multiple products were produced by the same plot or livestock, allocation was conducted as per ISO14040 and 14044.

The assessment of MiFAS required additional model adaptations, as by their nature, MiFAS may produce multiple products from the same land area concurrently, such as apple trees with pasture beneath them, or arable crops that provide winter grazing for sheep. The main issues included allocation of impacts between the co-products, especially when an input is not attributable to a single output. We therefore undertook specific methodological adaptations for assessing MiFAS within the MIXED project including assessing soil carbon changes and biomass with agroforestry (AF) systems. Quantifying SOC changes is scientifically challenging, and we therefore adapted the IPCC Tier 2 steady state method to include a wider variety of land uses including permanent pasture, orchards, and other trees. For agroforestry, we included a module for biomass carbon calculations and adopted a Tier 1 approach according to the meta-review by Cardinael et al. (2018), to estimate the changes in above and below ground carbon for the first 20 years following land use change. Furthermore, to increase accuracy in AF calculations we used a spatial allocation method to differentiate between trees and pasture and to account for tree size and planting density. For the economic analysis, we utilised an adapted database from the German KTBL database which allows for farms from multiple countries to be grouped together without issues of costs differing due to local situations. Gross margin calculations (partial net margins) are reported based on outputs minus inputs, which include labour.

Due to the wide diversity of farm systems within and between networks, we adopted a statistically based two-step clustering approach to group the farms from across all the networks into farm system type groups for comparison. We found that using binary variables related to the presence of AF, livestock, together with proportions of farms with permanent grassland and field crops generated four groups for comparison. To enable a statistical assessment of farm system differences, we also utilised the non-parametric Kruskall Wallis test to enable robust assessment within data that violated the normal assumptions of one-way ANOVAs.

Results: For the impact assessment phase, we presented results for each farm per network, followed by results per farm system type, as well as presented at an enterprise level for a limited number of crops and livestock. For the analysis, we adopted a number of performance indicators to characterise the farm systems, assessed their use of nitrogen as a main agricultural nutrient, estimated potential changes in carbon, environmental indicators as part of the LCIA and economic indicators for sales, costs and partial net margins.

Farm Networks: The networks assessed provided a wide variety of farm systems across a wide geographical area. The data provided about the farm systems included full farm systems through to specific areas of farms that focussed on a particular topic, e.g. agroforestry. The farms also ranged from very diverse, complex systems with crops, livestock and agroforestry through to large highly specialised units, which were included within the MIXED project due to their participation in landscape scale collaborations. These specialised systems also proved extremely valuable as comparators to the MiFAS type systems. Whilst farms within some of the networks had a common theme, such as the French NW09, others were highly diverse, such as French NW10. Livestock types were also diverse, covering all sectors except broiler chickens.

The Danish NW02 farms produced a wide range of arable crops, as well as some farms with large herds of intensively managed cattle or pig systems. Despite the transfer of nutrients via biogas plants, all of the farms had high nitrogen inputs, especially when nitrogen within feed is accounted for (up to 551kg N ha-1). The intensity of farm systems therefore resulted in some very high GHG

emissions, especially when livestock were kept. Whilst in Scotland NW04, most of the farms comprised cropping with or without livestock, and emissions were greatest on farms with livestock. The German NW5 farms showed that whilst emissions from the peat land have declined, the current utilisation, such as extensive beef production, was found to have very high emissions, because of the underlying peat, as well as slow growth rates. The German NW06 farms were diverse, including a free-range egg system which was very reliant on external feeds, resulting in high nitrogen related emissions, whilst the other two systems were less integrated and planted more like tree hedgerows. Whilst new agroforestry showed a potential for new carbon storage, the long-term aspects were unclear, so carbon storage on a 100-year basis was unclear.

In Switzerland, despite the NW07 farms having larger high-stemmed fruit trees as an agroforestry system, the Tier 1 methodology means that beyond 20 years of age the biomass carbon was assumed to be at equilibrium. However, the Swiss farms did demonstrate the improved circularity from using livestock manures as the primary fertiliser source, with low external nitrogen sourcing. However, when livestock are maintained with and feed imported, emissions increase, though offsetting emissions through biomass storage in new trees can be partially effective. In France NW09, despite the woodland setting for the raising of pigs, the high feed imports and stocking densities, combined with slower growth rates caused high emissions. Although comprising a significant level of woodland, the trees were generally older (around 70 years), therefore biomass carbon was assumed to be in equilibrium as per the Tier 1 guidelines. The French NW10 was a mixture of farm types, and the specialist livestock farms were generally very extensive, whilst the cropping farms were quite intensive, and emissions depended largely on their intensity and the presence of livestock. For the Romanian NW11 farms obtaining high quality data suitable for conducting an LCA was problematic, and therefore a single typical farm for the region was constructed comprising a high intensity of livestock, feeds purchased and diverse fruit trees on pasture or in orchards. The high density of livestock resulted in high emissions within this system. In Poland NW13 as a single very large biodynamic mixed farm had few external inputs, but limitations to its crop yields are a severe handicap to economic performance, as well as causing some higher-than-expected product impacts.

The interpretation phase assessed all phases of the analysis, including input data from the farms, methodological challenges, results at the network, system type and enterprise level, as well as making general conclusions from the work undertaken. We found that with such a diverse range of farms in the dataset it was difficult to come to clear conclusions about the performance of different farm system types, therefore, a single farm dataset was formed, and farms were grouped into four system types, integrated cropping and livestock (ICL), specialist arable (SA), specialist livestock (SL) and integrated cropping/livestock and agroforestry (ICLF). Ideally, we would also have liked to compare organic and conventional systems, but the dataset was too small to undertake any valid comparison.

In terms of characteristics. we found that the ICL and ICLF farm clusters were larger than the specialist systems, highlighting the focus of the farm networks. Farm areas were much greater for the ICL, SA and SL systems, whilst the ICL and SA types both had a high proportion of field cropping. However, we also observed that the more integrated system had a reasonable proportion of temporary forages, with a little grassland. The SL was dominated by permanent grassland, with similar livestock numbers for both ICL and SL, though livestock stocking density was greatest for ICLF, probably because of the French pig systems.

The main nitrogen indicators all showed significant differences between the four farm system types, whilst fertiliser application of nitrogen was lower on SA systems. Nitrogen self-sufficiency and the proportion of nitrogen applied as organic manures was always lower on SA farms, intermediate for ICL and higher on the ICLF and SL farms, as may be expected with higher livestock levels. However, nitrogen export as products was lower on SL, with ICL and SA the highest because of the higher N exported per hectare of cropland.

Whilst we found differences in revenue and costs between the farm systems, overall, there was no significant difference between the farm system types. However, when comparing environmental impacts, all environmental indicators showed significant differences between systems. For greenhouse gas (GHG) emissions we found that per hectare, the SA farms had lower emissions,

with SL at an intermediate level and the two integrated systems showing the greatest impacts. Using the alternative functional unit of per kg of nitrogen exported, the results showed the greatest emissions for the SL system, likely in part due to the low productivity extensive systems, whilst the integrated systems were at an intermediate level.

In terms of fossil and nuclear energy (FNE) use, SL farms were lowest per hectare, but again, when assessed by kilogram of N exported, became the highest energy user. The cropping systems showed the greatest energy use per hectare, but SA farms were the lowest per kg nitrogen exported. In terms of mineral resource use, the SA and SL farm types had lower use per hectare, whilst per kg of N exported, SA farms had the lowest impacts, ICL was intermediate with the SL and ICLF farms the largest resource users.

Considering acidification impacts, both indicators (FA and TA) showed SL farms to have low impacts reflecting the far lower levels of N inputs per hectare, whilst for impacts per kg N exported, SA systems showed lowest impacts due to high N outputs compared to the livestock centric ICLF and SL systems. Eutrophication (FEU and MEU) results per hectare reflected the low Phosphorus inputs of the SA and SL systems, whilst for MEU, the SL system was lowest per hectare but greatest per kg N exported. The integrated systems were intermediate for both functional units.

When we assessed data at an enterprise level, we found wheat and beef to be present in many networks. In total we found 36 wheat crops, and results of comparing the underlying farming system indicated very different management between the farm types. The highest levels of mineral nitrogen were used on ICL and SA farm types who also achieved the highest yields. This probably explains why the GHGs and energy use, were lower for the ICLF and SL farm types, however due to heterogeneity within the data, for most of the environmental impact indicators there were no significant differences.

Beef animals were reared on 21 farms within the networks and included animals from both dairy and suckler cows. We found that stocking density was highest on the ICLF and ICL farms, whilst rations were not significantly different, with all systems receiving a high median level of forage. However, the environmental impacts were significantly different between farm types, with the SL farm types showing the highest impacts. Contribution analysis highlighted the greater impacts of the SL system for most impact categories, with greater GHGs likely because of enteric emissions and the greater emissions embedded within the transferred in-stock, such as weaned calves from generally higher GHG suckler cow systems.

Changes in the soil carbon were generally very small, probably due to reporting of only the passive soil pool as the more active soil pools are short term and therefore inappropriate to report within the 100 year GHG basis (GWP100). Soil carbon changes were also more limited due to the single time frame of the detailed data collection, preventing more consideration of specific management changes that may have affected SOC. One factor that became apparent within the modelling, was that in the absence of fundamental system changes, the temperature effect on soil C degradation is already apparent. As temperature increases, we see greater SOC loss under the same management and as the model uses a 20-year period for assessing SOC, the increasing temperature within the climate datasets shows SOC is generally being lost in the carbon dynamic tables.

The biomass modelling was entirely new for the project and the Tier 1 method, together with adaptations for tree size and planting density provided some insight into the potential of agroforestry. We found that there was a great difference in tree biomass potential carbon storage depending on the age structure of the trees, partly as a direct result of the modelling assumptions, i.e. no additional storage in AF systems after 20 years as most AF systems are built around early maturing trees, like fruit, nut or short-rotation coppice (SRC) trees. Furthermore, whilst the initial planting of AF trees adds new above and belowground biomass carbon storage, this is potentially at the cost of soil carbon initially and it may take up to 30 years before an increase in SOC is observed (e.g. Paul et al., 2022), however, the ecosystem services of AF go beyond carbon storage and still represents a viable climate change mitigation option.

In conclusion, we were able to assess a very diverse range of farm systems in varying geographical locations to at least partly, answer the question of whether MiFAS systems provide environmental

and potentially economic benefits. The answer is sometimes and depending on the indicator and functional unit applied. The ICL and ICLF systems, as well as the SL were more self-sufficient in nitrogen supply, but SA farms had better external nitrogen utilisation. In terms of GHGs, the SA farms emitted the least at both per hectare and per kg nitrogen exported from the farm, with SL emitting the highest and the ICL and ICLF farms at an intermediate level. For the other environmental indicators, the SL farms were usually the lowest per hectare because of their extensive characteristics, whilst for the per kg nitrogen FU, SA farms were lowest and SL the highest. Economically, all farm types showed a net loss, with the low input SL farms showing the smallest loss and ICL the greatest, though these differences were not significant.

However, these results are influenced by the farms within each type, and there were clear trade-offs between per area and per product impacts. The results also showed that the impacts are very related to the specific situation on the farm and that strategies such as agroforestry alone will not solve issues, but a whole farm approach to reducing impacts through reduction and efficient use of fertilisers and feeds, combined with additional strategies will have the greatest impact. Some of the ICLF systems were situated with existing woodlands and due to its age, new carbon sequestration was unlikely, whilst the system was also supported by considerable external feed inputs, therefore the system does not appear to be a solution from an LCA impact perspective. However, the more extensive versions of this systems provided direct benefits as well as other factors such as welfare which may be much improved compared to intensive indoor production.

The results from this analysis should be viewed with caution as the systems assessed were only representative within a range of networks available within the MIXED project. Farms had specific management strategies, which may provide considerable benefits either at a local or even wider spread adoption, such as winter grazing of cereals by sheep, exchanges between farms, as well as agroforestry. However, the results could be strongly influenced by certain aspects and generalisations should not be made. From a policy perspective, the results point to variation in impacts due to the specifics of a production system and farms and policies must find a balance between productivity whilst minimising external inputs, with the potential to add agroforestry for additional benefits.

Methodologically, whilst LCA remains a good option for assessing environmental impacts, there is still much work to be undertaken to allow farm LCA assessments to fully understand the complexities of the systems. Furthermore, other ecosystem services and societal aspects are still absent from this study and most LCAs, including biodiversity and animal welfare as two major topics. Increasing crop and forage diversity, agroforestry and a more diverse landscape are all likely positives for ecosystem services, but their assessment remains challenging at a wider scale.
Original languageEnglish
Commissioning bodyHorizon 2020
Number of pages68
Publication statusPublished - 31 Mar 2024

Keywords

  • Mixed farming
  • agroforestry
  • agricultural systems

Fingerprint

Dive into the research topics of 'Performance of mixed and agroforestry systems: Deliverable 5.1'. Together they form a unique fingerprint.

Cite this