An integrative approach to silvopastoral system design: perspectives, potentials and principles

ABSTRACT

Silvopastoral systems have complex impacts on a diverse range of outcomes, making it essential to design these systems using an integrative approach to maximise positive impacts to farms. This paper comprises firstly a systematic review of global silvopastoral processes, and secondly stakeholder-driven synthesis of key opportunities and challenges for future silvopastoralism situated in the context of New Zealand. The systematic review demonstrated that although under-researched, livestock interactions can have overriding influences on the system, and that the traditional functional traits that are typically deemed important for selection (N₂-fixing trees v non N₂-fixing trees, evergreen v deciduous) do not show consistent positive impacts on the agroecological environment. From the New Zealand silvopastoral participatory case study, including the stakeholder workshop, we synthesised 5 key principles that should be considered in future system designs. These were: (1) silvopastoral systems are complex and require holistic management; (2) the views, values and experiences of local people are deeply connected to silvopastoral system design; (3) spatial heterogeneity in environmental and social conditions requires locally specific decisions; (4) understanding of ecological processes must underpin all management decisions; and (5) the complexity and spatial heterogeneity present in silvopastoral systems requires high-resolution data and tools.

KEYWORDS:

• Agroforestry
• tree-pasture
• wooded pasture
• diversity
• biodiversity

Introduction

Silvopastoralism is an integrative and holistic land management practice that has great potential for reducing landscape degradation and the environmental impact of farms (Jose and Dollinger 2019; Zhu et al. 2020; Plieninger et al. 2021; Nair et al. 2022). Silvopastoralism is a type of agroforestry, where agroforestry is defined as an agricultural system where woody perennials (typically a tree) interacts biologically with a crop that is managed for forage, annual or perennial crop production (Somarriba 1992). Silvopasture specifically describes the integration of woody perennials into pasture-livestock systems (Nair 2012; Nair et al. 2022).

The many positive impacts of silvopastoral systems have been well established globally, as outlined in previous review articles (Jose 2009; Jose and Dollinger 2019; Pent 2020; Zhu et al. 2020; Plieninger et al. 2021; Smith et al. 2022) and summarised in Table 1. However, silvopastoral systems do not have universally positive effects (Table 1). For example, many silvopastoral systems show increased pasture growth compared to treeless pasture (e.g., Callaway et al. 1991; Peri 2005; Moreno 2008; Rivest et al. 2013), while other systems show decreased pasture production (e.g., Guevara-Escobar et al. 2007; Benavides et al. 2009; Rivest et al. 2013; Seddaiu et al. 2018). Silvopastoral systems must therefore be designed to maximise the benefits while minimising the negative effects of these systems to ensure they become beneficial land management solutions (Rhoades 1996; Casals et al. 2014; Jose et al. 2019).

Table 1. Review of the positive and negative impacts of silvopastoral systems globally for a range of outcomes.

Outcome	Positive impacts of silvopastoral systems	Negative impacts of silvopastoral systems	References
Pasture production	Some systems can increase pasture production compared to open pasture Processes that impact this outcome include: evapotranspiration reductions via shade and wind reductions; buffering of winter temperature lows; litterfall and litter decomposition; root decomposition; Mycorrhiza fungi interactions; increased soil biota diversity and richness; increased microbial activity; livestock excreta deposition under canopies.	Some systems can decrease pasture production compared to open pasture. Processes that impact this outcome include: photosynthetically active radiation (PAR) interception and excessive shading; lower average yearly temperature under canopy; competition for water via tree water use; nutrient uptake by the tree; livestock overgrazing and camping in the tree-pasture environment.	Bahamonde et al. (2009); Belsky et al. (1989); Benavides et al. (2009); Callaway et al. (1991); Douglas et al. (2001); Fernández-Moya et al. (2011); He et al. (2003); Howlett et al. (2011); Mackay-Smith et al. (2022b); Moreno et al. (2005); Peri (2005); Rivest et al. (2013); Roos and Allsopp (1997); Rossetti et al. (2015); Seddaiu et al. (2018)
Slope stability	Hydrological mechanisms: Interception of rainfall and snow by canopies of vegetation, thus promoting evaporation and reducing water available for infiltration. Root systems extract water from the soil for physiological purposes (via transpiration), leading to lower soil moisture levels. Mechanical mechanisms: Individual strong woody roots anchor the lower soil mantle into the more stable substrate. Strong roots tie across planes of weakness along the flanks of potential landslides. Roots provide a membrane of reinforcement to the soil mantle, increasing soil shear strength. Roots of woody vegetation anchor into firm strata, providing support to the upslope soil mantle through buttressing and arching.	Hydrological mechanisms: Roots, stems, and organic litter increase ground surface roughness and soil's infiltration capacity. Depletion of soil moisture may cause desiccation cracks, resulting in higher infiltration capacity and short-circuiting of infiltrating water to a deeper failure plane. Mechanical mechanisms: Weight of trees (surcharge) increases the normal and downhill force components. Wind transmits dynamic forces to the soil mantle via the tree bole.	Phillips and Marden (2005); Schmidt et al. (2001); Sidle and Ochiai (2006)
Discharge rates and nutrient losses	Silvopastoral systems have typically been shown to reduce surface runoff and associated sediment and nutrient losses compared to paired treeless pastoral areas. Processes that impact this outcome include: rainfall interception and throughfall reduction; tree water and nutrient use; maintenance of a good ground cover; soil quality improvements increasing infiltration; root systems increasing infiltration.	Silvopastoral systems can also increase surface runoff and sediment and nutrient losses. Processes that impact this outcome include: trees encouraging more pasture grazing under trees; stemflow facilitating surface runoff; competition between tree and pasture; tall trees increasing the erosive power of rain.	Bambo et al. (2009); Chen et al. (2018); de Aguiar et al. (2010); Grewal et al. (1994); Levia et al. (2017); Li et al. (2019); López-Díaz et al. (2011); Mackay-Smith et al. (2022a); Michel et al. (2007); Nair et al. (2007); Nanko et al. (2006); Wiersum (1985); Zhou et al. (2016); Zhu et al. (2020)
Greenhouse gas emissions	Silvopastoral systems generally sequester more carbon (both above and below ground) than open pasture.	Higher density tree cover can result in higher methane and nitrous oxide emissions if soil moisture is elevated.	Aryal et al. (2019); Baah-Acheamfour et al. (2016); Baah-Acheamfour et al. (2020); Cardoso et al. (2022); Dold et al. (2019); Haile et al. (2010); Hoosbeek et al. (2018); Lorenz and Lal (2014); López-Santiago et al. (2019)
Livestock welfare	Silvopastoral trees reduce heat stress in cattle and sheep, anc can result in lower respiratory rate, lower body temperature, lower water intake, longer grazing times and higher milk production. Little to no research on cold stress.	Excessive livestock camping under trees, or systems with a higher minimum night time temperature under trees, could increase the prevalence of parasites in livestock.	Blackshaw and Blackshaw (1994); Cloete et al. (2021); Dada et al. (2021); De et al. (2020); Jackson et al. (1986); Hawke (1991); Pent et al. (2021); Skonieski et al. (2021); Vieira et al. (2020); Yamamoto et al. (2007)
Biodiversity conservation	Silvopastures are generally beneficial to bird and insect populations compared to treeless pasture. Land-use history likely an important driver of local biodiversity as well as current land-use type. Systems can harbour high floral diversity. Even if floral diversity reduces under trees, the vegetation structure generally changes which adds to the overall diversity of the flora. Native trees should be a better habitat for native birds and soil fauna.	Studies have also shown silvopastures to have a neutral impact, as well as a negative impact, on biodiversity.	Fischer et al. (2010); Gómez-Cifuentes et al. (2020); Lencinas et al. (2015, 2008); Mackay-Smith et al. (2023); Mupepele et al. (2021); Peri and Ormaechea (2013); Rossetti et al. (2015); Santos-Reis and Correia (1999); Seddaiu et al. (2018); Tölgyesi et al. (2018); Torralba et al. (2016)
Cultural values and aesthetics	Silvopastoral systems can become important cultural practices in terms of the maintenance of traditional knowledge, recreation and eco-tourism services, and cultural heritage, such as in the dehesa system in Spain. Native trees will likely hold more cultural value than exotic trees. Different trees will have different cultural values depending on the location, this is highly location specific and often depends on use.	Farmers that have not typically planted trees may hold less value for silvopastoral systems than treeless pasture.	Borremans et al. (2016); Brown et al. (2012); Garrido et al. (2017); Moreno et al. (2018); Plieninger et al. (2021); Wilkens et al. (2022); Wossink and Swinton (2007)
Income diversification	Trees can provide additional income opportunities through providing fruit, nuts, high quality timber, post lumber and firewood. Either add extra income or a supplementary source to increase resilience. Carbon sequestration by trees can also generate an income if captured in a trading scheme.	High value timber and fruit/ nut trees have typically higher establishment costs Sequestered carbon can become a liability if the scheme requires that any tree loss requires offsetting.	Bruck et al. (2019); Ares et al. (2006); Frey et al. (2012); Holderieath et al. (2012); Klopfenstein et al. (1997); Montagnini and Nair (2016); Orefice et al. (2017); Pizarro et al. (2020); Rodriguez-Rigueiro et al. (2021)
Supplementary forage	Supplementary forage provides a diversified diet that can be more resilient, improve digestion, alter methane emissions and reduce parasite load.	Toxins can restrict use of some trees.	Franzel et al. (2014); Orefice et al. (2017); Vandermeulen et al. (2018)
Management	Management practices can become important cultural identifiers and are a form of traditional knowledge.	There is a cost to establishing trees. Trees can be difficult to protect against livestock damage, and take a long time to influence the system. Trees may also require management (pruning, felling etc.).	Charlton et al. (2007); Garrido et al. (2017); Mackay-Smith et al. (2021); Peri et al. (2016); Thorrold et al. (1997)

Silvopastoral landscapes are complex systems involving numerous ecological components and mechanisms of human management (Mancera et al. 2018; Venturi et al. 2021). Even without considering trees, pastoral agriculture depends on complex interactions between livestock, forage plants, soils, microbes, and humans (Crawford et al. 2005; Robinson 2009; Bryant and Snow 2010; Dean et al. 2021). Integrating trees into the system greatly increases this complexity through adding new interactions and linkages, many of which can modify existing biophysical processes within the system (Figure 1) (van Mil et al. 2014; Mancera et al. 2018).

Figure 1. Ecological interactions between system components with increasing complexity for pasture, livestock and tree interactions in a pastoral system. The direction of the arrow indicates how the component the arrow comes from impacts the component the arrow goes towards. For instance, for number 7, the microclimate impacts livestock through changes to temperature, humidity, PAR and wind.

The impacts of trees in silvopastoral systems are difficult to predict, because these impacts emerge from the system as the net effect of numerous interdependent mechanisms and complicated feedback patterns (Salt 1979; O’Sullivan 2004). Understanding these emergent properties of silvopastoral systems requires a holistic approach, using a systems ecology framing to integrate the known ecological and social interactions in the system (Odum 1977; Salt 1979; Jørgensen 2012). Nevertheless, the majority of silvopastoral research does not take this holistic, systems approach (Plieninger et al. 2021). For example, in a recent review of silvopastoral research in the European dehesa and montado systems, Plieninger et al. (2021, p. 9) concluded that the majority of past research ‘treated the dehesa system in an unintegrated way […] when in fact they are highly interconnected and interdependent’. Furthermore, applied guidance for silvopastoral managers frequently focuses on design to maximise individual benefits, rather than multifunctionality (Wilkinson 1999; Charlton et al. 2007; Mackay-Smith et al. 2021).

Land managers are faced with numerous options when designing silvopastoral systems. For example, determining (1) which tree species are suited to the soil and microclimate conditions of their land; (2) which tree species provide the key benefits to address a particular aim; (3) where in the landscape trees should be planted to gain the maximum benefit; (4) which type of livestock and at what stocking density is most appropriate; (5) and which pasture species are present or needed. Selecting the right combination of tree, livestock, pasture, and management options is challenging due to the many interdependencies and difficult to predict feedbacks between these separate decisions (Jose 2009; Jose et al. 2019; Plieninger et al. 2021).

In this study, we propose an integrative framework for silvopastoral system design that aims to be holistic in delivering multifunctional silvopastoral landscapes. This requires an understanding of key processes that govern the interactions between trees, pasture, and livestock, and a systems ecology lens that considers the multiple ecological and social interactions of these systems. To achieve this overall aim, the paper has three major objectives:

1. Review the key processes through which silvopastoral trees impact a range of outcomes using a systematic review of studies that have compared the impact of different tree species on silvopastoral outcomes at the same study site. By comparing different tree species at the same study site, this systematic review aims to identify how contrasting tree attributes impact silvopastoral outcomes, and evaluate the relative importance of different silvopastoral processes with respect to the outcome that they impact. The conclusions of this objective will provide advice to land managers as to which tree attributes may be important for impacting different outcomes in silvopastoral systems.

2. Use the New Zealand (NZ) silvopastoral context as a case study to characterise the system complexity and demonstrate how future silvopastoral systems can be designed to maximise their positive impacts for improved land management. Since silvopastoral systems are social-ecological systems (Plieninger and Huntsinger 2018), local perspectives should be incorporated into the design process so that new systems are both valued and adopted.

3. Drawing on the systematic review of silvopastoral tree processes, and the in-depth case study of NZ’s silvopastoral systems, we develop 5 key principles for successful future silvopastoral design.

Review of key agroecological processes for silvopastoral system design

A review of silvopastoral system agroecological processes was undertaken to determine the importance of (i) specific tree attributes and (ii) biophysical processes within the system that promote particular outcomes (Figure 1). Studies were systematically selected that compare the impact of different tree species on a range of silvopastoral outcomes for a given site. By comparing two species at the same site, it could be ascertained how different tree attributes have differing impacts on the silvopasture environment, which can be used to provide recommendations as to which tree attributes are important for positively impacting the silvopastoral agro-ecological environment.

It was important to compare different tree species at the same site because varying climate, topography, aspect, livestock type, grazing management, fertility management, livestock behaviour and tree spatial design makes it very difficult to compare the impact of tree attributes on the agro-ecological outcomes from presence/absence studies of single tree species on different farms. We think the most appropriate way of comparing the impact of different tree attributes on agro-ecological outcomes is to review studies that compare different tree species, with differing tree attributes, at the same site with consistent underlying farm parameters where compounding farm and climatic variables have been controlled for.

Review methods

The literature search was undertaken on the Web of Science Core Collection database on 12th September 2022. The following keywords were searched in titles, keywords and abstracts: ‘silvopastoral’, ‘silvopasture’, ‘silvopastoralism’, ‘scattered tree* AND pasture’, ‘scattered tree* AND grass’, ‘wood pasture’, ‘wooded pasture’, ‘space-planted tree*’ and ‘tree pasture’. The search was limited to English articles and no date limit was used. This resulted in 1629 articles. Titles and abstracts were read and any studies that compared multiple tree species at the same site on any silvopastoral outcome were recorded in a database. The studies did not have to compare silvopasture treatments with treeless/open pasture, but had to compare the impact of different tree species at the same site. The impact of differing tree attributes on silvopasture outcomes could then compared, resulting in tree attribute recommendations for silvopasture design. This resulted in 165 articles, which were categorised based on the silvopastoral outcomes they measured.

The second screening step focused on the method section of these papers and removed articles that did not qualify for the following criteria:

• Studies were required to be designed at the field/paddock scale since it is the scale at which the agroecological processes function (Figure 1).

• Studies that reported results from stands that had a density lower than 10 × 10 m or a row spacing of less than 10 m were removed because tree species with different densities can impact the agroecological environment differently (Rozados-Lorenzo et al. 2007). As we were interested in learning how different tree attributes impact different silvopasture processes, by only selecting systems that had a density higher than 10 × 10 m or row spacing higher than 10 m, we could limit density as a confounding factor in the review.

• Studies that measured isolated trees were assumed to have a density greater than 10 × 10 m, and so were included in the study.

• Studies that had a between-row spacing of greater than 10 m but a within-row spacing of less than 10 m were and compared the impact of trees on silvopasture outcomes between the rows were included in the study.

• Studies that reported results from stands that were less than 5 years old were removed because these silvopastoral trees would have likely had minimal impact on the agroecological environment.

• Any studies that only measured tree establishment (tree growth or survival) or fodder production were removed because this review was specifically interested in the agroecological processes within the system, and not outcomes that are related to tree management.

• Studies that did not study silvopasture processes or outcomes in-situ were removed.

• Full papers had to be available, or else they were removed.

Finally, cited references were screened in the original list of papers and included if they met the selection criteria. This yielded 42 articles (see Appendix 1 for the full list of studies).

Despite not technically being silvopastoral systems, several studies yielded in this search were of tree comparisons in savanna grasslands. These were retained in the review because of their relevancy to the research question. ‘Savanna’ was not included in the original search because this addition yielded 17,396 articles, too many to screen with the available resources.

A qualitative and quantitative approach were used to review the studies. The studies were first quantitatively grouped into studies of ‘processes’ and ‘outcomes’ (Figure 2). The main findings of the articles were then discussed qualitatively because of the variety of outcomes and processes that were included in the review. A quantitative comparison of silvopastoral systems in different regions is challenging because of the multitude of factors which influence silvopasture outcomes, such as climate, land use history, pasture and livestock management, tree age, tree density and tree management.

Figure 2. The number of articles on specific silvopasture outcomes (left) and processes (right).

Silvopastoral system process review

Processes and outcomes

Twenty-five papers measured soil (soil chemical properties, soil moisture or soil physical properties) outcomes in the reviewed studies, which was the most researched outcome, followed by pasture (24) and greenhouse gas emissions (21) (Figure 2). The silvopasture processes that were researched the most were N₂-fixing (14) and microclimate (10) (Figure 2).

N₂-fixing trees

Of the reviewed systems, 8 studied systems compared the impact of N₂-fixing and non N₂-fixing trees on soil N (Belsky et al. 1989, 1993a, 1993b; Rhoades et al. 1998; Menezes and Salcedo 1999; Menezes et al. 2002; Tripathi et al. 2005; Treydt et al. 2007; Casals et al. 2014; Dubeux et al. 2014; Cá et al. 2022). Three of these systems found increases of soil N or greater soil N mineralisation under N₂-fixing trees compared to non N₂-fixing tree systems (Rhoades et al. 1998; Menezes and Salcedo 1999; Menezes et al. 2002; Cá et al. 2022). However, 5 of these system comparisons did not report greater soil N under the N₂-fixing trees (Belsky et al. 1989, 1993a, 1993b; Treydt et al. 2007; Casals et al. 2014; Dubeux et al. 2014). Some of the studies hypothesised other processes had overriding impacts on the agroecological environment over whether trees were N₂-fixing or not, such as livestock activity (Belsky et al. 1989, 1993b; Dubeux et al. 2014), litter quantity (Casals et al. 2014) and impacts on soil moisture (Menezes et al. 2002).

Although not included in the systematic search, a previous meta-analysis has measured the impact of different functional groups (deciduous trees, Eucalyptus, N₂-fixing trees and evergreen oaks) on pasture production (Rivest et al. 2013) and so provides an important comparison for this study. This study reported that N₂-fixing trees were the only functional group to significantly increase pasture production compared to open pasture controls, deciduous trees and evergreen oaks had an overall neutral impact on pasture production, and Eucalyptus trees negatively impacting pasture production. Nevertheless, the N₂-fixing results by Rivest et al. (2013) are misleading. For example, three of the N₂-fixing studies reviewed by Rivest et al. (2013) compared N₂-fixing trees and non N₂-fixing trees. Treydt et al. (2007) found no difference between the impact of N₂-fixing trees and non N₂-fixing trees on pasture production. Belsky et al. (1989) measured a similar grass response between of N₂-fixing trees and non N₂-fixing trees. Moreover, despite Belsky et al. (1993a) reporting significantly more grass under acacia (Acacia tortilis (Forssk.) Hayne) trees compared to under baobab (Adansonia digitata L.) trees, this was suggested to be because of greater elephant utilisation under this species and not because the acacia trees were N₂-fixers.

There were two other English studies reviewed by Rivest et al. (2013) in the N₂-fixing group that were not a comparison of different species at the same site and so were not included in the systematic search. Ludwig et al. (2001) found no difference between pasture production under acacia (Aacia tortilis) trees and open pasture, and although Durr and Rangel (2002) did find 132% more grass growth under N₂-Fixing Samanea saman (Jacq.) Merr. trees compared to open pasture, the authors could not confirm why the trees increased pasture production, and suggested nutrient transfer by cattle could have confounded this result. These examples indicate that it is not clear that the N₂-fixing attribute was responsible for the positive impacts of the N₂-fixing group on pasture production reported by Rivest et al. (2013).

The results of this review and a look into the papers reviewed by Rivest et al. (2013) suggest that N₂-fixing should not be relied upon as a key tree attribute when selecting silvopastoral trees to improve soil condition, and that other factors can have an overriding influence on the agroecological environment compared to whether the tree fixes N or not.

Litterfall dynamics

Litterfall dynamics were frequently hypothesised to be an important driver of soil fertility outcomes in the reviewed studies (Casals et al. 2014; Martínez et al. 2014; Lima et al. 2020; Cá et al. 2022). Nevertheless, only two of the reviewed studies compared litterfall dynamics and silvopastoral outcomes (Menezes et al. 2002; Cá et al. 2022). Cá et al. (2022) measured higher total organic carbon (TOC) sequestration in the top 10 cm of soil in a high tree diversity silvopastoral system with higher litterfall rates, however, Menezes et al. (2002) provided evidence that impacts to soil moisture had much greater effects on the system than differing litterfall dynamics between species.

Although the authors did not directly measure litterfall, Casals et al. (2014) hypothesised that litter quantity had much more impact on soil fertility and carbon than litter quality. This was because the impact of different tree species on these soil variables was not related to whether the trees were N₂-fixers or not, nor to whether the trees were evergreen or deciduous. Moreover, measured livestock activity preferences were similar between tree species.

Litterfall dynamics have also been studied in the gliricidia (Gliricidia sepium (Jacq.) Kunth) and mimosa (Mimosa caesalpiniifolia Benth.) silvopastoral systems in Brazil, with litterfall rates being slightly greater in the mimosa system (10,790 kg ha⁻¹ compared to 10,420 kg ha⁻¹) (Apolinário et al. 2016a). However, N concentration of the leaves was greater in the gliricidia system (21.5 kg ha⁻¹ compared to 18.8 kg ha⁻¹). Moreover, the gliricidia system litterfall had a lower C/N ratio and lignin concentration, so it was suggested by the authors that the leaves of this species will decompose faster. This has been confirmed by other studies not included in the systematic review (Apolinário et al. 2016b; Herrera et al. 2020).

It was also reported that pasture production (Gomes da Silva et al. 2021b) and livestock productivity (Gomes da Silva et al. 2021a) were greater in the gliricidia system compared to the mimosa system, however, the authors suggested that this was because of significantly higher soil moisture in this treatment (mimosa: 16.2%; gliricidia: 17.2%) impacting pasture production, and not differences in litterfall decomposition rates (Gomes da Silva et al. 2021b).

Livestock activity

Despite not measuring livestock behaviour, two of the reviewed systems suggested that livestock activity was the likely reason for different soil and pasture outcomes in silvopastoral systems. Belsky et al. (1993a) suggested elephant damage to be the main reason for the disparity between pasture production results under acacia and baobab trees at their higher rainfall site. Furthermore, Dubeux et al. (2014) hypothesised that livestock activity had overriding influences on soil organic matter because they found 64% greater organic matter levels under two non N₂-fixing compared to two N₂-fixing trees at a 50% canopy site in the Itambé Experimental Station in Brazil.

Nevertheless, only two of the reviewed studies quantified livestock preferences (where livestock spend preferential time in the paddock) under different tree species (Wilson 2002; Casals et al. 2014) and have contrasting results. Casals et al. (2014) reported no differences in livestock activity between species, whereas Wilson (2002) reported nearly 20 times more livestock dung on transects around Eucalyptus melliodora A.Cunn. ex S.Schauer trees compared to E. nova-anglica H.Deane & Maiden trees. Clearly livestock preferences vary between different silvopastoral systems, and even between different tree species.

Microclimate

Seven of the reviewed studies measured processes in black walnut (Juglans nigra L.) silvopastures compared to honey locust (Gleditsia triacanthos L.) systems in Virginia (Buergler et al. 2005, 2006; DeBruyne et al. 2011; Fannon et al. 2017; Pent et al. 2020a; Pent and Fike 2019; Poudel et al. 2022). Black walnut silvopastures have tended to reduce annual pasture production compared to honey locust systems (Fannon et al. 2017; Pent and Fike 2019; Pent et al. 2020a; Poudel et al. 2022). Lack of photosynthetically active radiation (PAR) is suggested as the primary cause of pasture production differences, being attributed to the large canopies, leaf shape and canopy density of the black walnut system. This was quantified by Poudel et al. (2022) who reported PAR and temperature were on average 49% and 1.1°C less in the black walnut silvopastures.

Reduced temperatures and increased PAR interception have also been related to improved livestock welfare in the black walnut system, in terms of animal thermal comfort indexes (Pezzopane et al. 2019) and hair cortisol levels (Poudel et al. 2022), and greater grazing time (Pent et al. 2020a). Moreover, despite the black walnut system having reduced pasture growth and a lower stock rate compared to the honey locust and open pasture systems (which had similar stock rates), these positive impacts to grazing time meant that the black walnut system had equivalent total lamb productivity to the honey locust and open pasture systems (Pent et al. 2020a, 2020b).

Microclimate has also been shown to enhance arthropod populations under specific trees, and different arthropod species are adapted to microclimate conditions under different trees (Tripathi et al. 2005).

Finally, three of the reviewed studies found evidence that deciduous trees had improved outcomes compared to evergreen trees (Frost and McDougald 1989; Ratliff et al. 1991; Rusch et al. 2014; Plevich et al. 2019). For instance, Rusch et al. (2014) measured significantly lower pasture production from July to August under an evergreen tree compared to two other deciduous trees, and the evergreen tree also had the greatest crown density in the dry and rainy season. There was no evidence that soil moisture was a function of canopy density, and so the authors suggested that the differing impacts on pasture productivity were because of the evergreen crown intercepting more PAR. A similar conclusion was found by Plevich et al. (2019) as pasture production was 23% and 70% higher under Quercus robur L. (deciduous) between April and October compared to Pinus elliottii Engelm. (evergreen) and Eucalyptus viminalis Labill. (evergreen) trees, respectively. However, this finding was not universal, and other studies found evergreen trees to have similar or improved outcomes compared to deciduous trees (Belsky et al. 1989, 1993a, 1993b; Casals et al. 2014; Castillo et al. 2020). This indicates that in these studies there are other factors that had a greater impact on pasture and soil outcomes compared to competition for PAR in winter by evergreen trees, such as the impact of evergreen/deciduous trees on livestock activity (livestock camping) and litterfall activity as described above.

Tree water competition

Studies in two systems attributed pasture and livestock production reductions to impacts on soil moisture (Menezes et al. 2002; Gomes da Silva et al. 2021a, 2021b). Menezes et al. (2002) found lower soil moisture under isolated Prosopis juliflora (Sw.) DC. trees (14 years old) earlier in the season compared to a grass control, but no difference was found between the grass control and soil moisture under isolated Ziziphus joazeiro Mart. trees (~50 years old). This could have been because of greater competition for water under P. juliflora trees. Alternatively, it could have been because of evapotranspiration reductions under Z.joazeiro because Z. joazeiro and P. juliflora tree canopies intercepted 65–70% and 20–30% solar radiation, respectively, and soil temperature was statistically similar between open pasture soil and soil under P. juliflora, but was significantly less under Z. joazeiro trees.

Other tree groups

Native silvopastoral trees were shown to increase the diversity of dung beetle species and functional groups compared to exotic Pinus taeda silvopastoral systems (Gómez-Cifuentes et al. 2019, 2020).

Two studies in Californian silvopastures have found digger pine (Pinus sabiniana Douglas) trees to almost always negatively impact pasture production compared to pasture under blue oak (Quercus douglasii Hook. & Arn.) and interior live oak (Quercus wislizeni A.DC.) trees in rangelands with varying productivity levels (Frost and McDougald 1989; Ratliff et al. 1991).

Summary of findings and recommendations

Published evidence found the following attributes were beneficial for improving the silvopastoral environment:

1. Livestock activity can have overriding influences on the agroecological environment, and it is important that the tree species and the spatial design of the system facilitates livestock grazing under tree canopies, but is not conducive to overgrazing.

2. Higher litter quality increases litter decomposition rates, but slower decomposing litter may also be important in forming stable soil organic matter in the tropics.

3. Maximising litter quantity can improve soil outcomes – although this must be balanced against leaf smother (Eason 1991).

4. Minimising PAR interception by the tree canopy is important for maximising pasture production, however, this must be balanced with the potential of increased PAR interception resulting in improvements to livestock performance.

5. Deciduous trees can have improved pasture production outcomes, however, this effect is not universal.

In the reviewed studies, there was a distinctive lack of studies measuring processes (Figure 2) within silvopasture, with most studies focused on outcomes (Figure 2). Moreover, there were no studies that measured a full range of processes in a silvopastoral system or related these to silvopasture outcomes. Moreover, none of the reviewed studies that directly compared tree water use between tree species. It will be important for future research to cover a broader range of processes and compare trees with different functional traits, to provide more compelling evidence as to which tree functional traits are important for improving given combinations of outcomes.

An important finding is the evidence that livestock can have overriding influences on the silvopastoral environment, and the review provides evidence that livestock activity is an essential consideration when comparing outcomes between systems. More work is required to measure livestock preferences for different tree species, in what situations livestock preferences exist, and how the impact of livestock activity as a process compares to direct tree processes like litter decomposition or competition for water.

N₂-fixation is generally seen as an important tree attribute in silvopastoral tree selection (Nair 1993; Rivest et al. 2013). However, this review indicates that N₂-fixing trees do not necessarily result in improved soil outcomes and other processes, such as livestock activity and litter quantity, can have overriding impacts on the agroecological environment.

There were two other potentially important processes that received no research in past comparison studies: allelopathy, which is when one species produces biochemical that impacts the survival or growth of another organism (Rice 1974), and leaf smother (Eason 1991). Both these processes can negatively impact pasture growth or seed germination (Eason 1991; Ahmed et al. 2008; Halvorson et al. 2017), so more research is required to quantify the importance of these processes with the processes covered in this review (Figure 2).

Finally, only three studies provided evidence that deciduous trees result in improved outcomes compared to evergreen trees. Moreover, in their meta-analysis, Rivest et al. (2013) reported that both deciduous trees (the majority of which were oaks) and evergreen oaks had an overall neutral impact on pasture production. As with N₂-fixing trees, this suggests that there are likely other attributes that can have more influence on the agroecological environment, which may include the size of the trees, the crown architecture or allopathy effects. Furthermore, there are likely other important factors impacting the system besides tree attributes, such as livestock management and behaviour, or how the spatial design of the trees as well as tree architecture impacts livestock activity. It is important that future research spans a greater variety of tree functional traits, and to research how these traits interact with management factors and livestock activity.

Silvopasture in New Zealand: towards integrative design

To provide an example case study of how more integrative design of silvopasture could be developed, we discuss the design of silvopastoral systems in NZ. Silvopastoralism has already been adopted widely in NZ, with poplar (Populus spp.) and willow (Salix spp.) principally being planted in the context of soil erosion (Kemp et al. 2018; Spiekermann et al. 2021; McIvor et al. 2023). However, there remains much potentials for other tree species to provide a range of additional benefits to NZ farms (Mackay-Smith et al. 2021). New Zealand therefore represents an excellent case study to showcase the important of taking a holistic and integrative perspective to silvopasture design.

The New Zealand silvopasture context

Following contact with European traders and settlers in the 1800s, extensive deforestation for pastoral farming resulted in high erosion and sedimentation rates (Goff 1997; Glade 2003; Green 2013; Phillips et al. 2018). Erosion processes in NZ remain very active compared to elsewhere (Jansson 1988), due to a predisposed natural environment with steep slopes, weak sedimentary rocks, and a climate featuring high annual rainfall and relatively frequent high magnitude rainfall events (Hicks et al. 2011, 1996; Basher 2013). High sediment yields can have detrimental outcomes on riverine and estuarine habitats by reducing the diversity, types, and abundance of fauna (Fuller and Death 2018).

Scattered or widely spaced silvopastoral trees have been used as a soil conservation and sediment mitigation tool on erosion-prone sloped pastures in NZ since the 1960s (Wilkinson 1999; Kemp et al. 2018; Spiekermann et al. 2021). The primary silvopastoral species that have been planted in NZ are poplar (Populus spp.) and willow (Salix spp.) (Wilkinson 1999; Kemp et al. 2018; Spiekermann et al. 2021). These trees have been chosen and actively promoted because they can be easily planted as sharpened, coppiced poles in the presence of grazing livestock (Wilkinson 1999; Phillips et al. 2014), and because they perform well in terms of soil erosion (Wilkinson 1999; Douglas et al. 2013; McIvor et al. 2015; Mackay-Smith et al. 2021; Spiekermann et al. 2021).

Nevertheless, there are other tree attributes that will be important for different silvopastoral functions on NZ farms (Mackay-Smith et al. 2021), and silvopastoral trees will likely provide additional value to less erosion-prone pastures in NZ, either in more rolling pastureland or flatter more productive pastures. In short, these systems are not designed considering the complexity of their functions, interactions or potentials, which limits the value silvopastoral systems add to NZ farms.

Stakeholder perspectives on current and future NZ silvopastoral systems

Methods

A broad range of individuals including farmers, landowners, land management advisors, policy advisors, and advocates were openly invited to a workshop to explore current silvopastoral systems used in NZ, and potentials for the future (Table 2). The range in occupations of participants enabled perspectives to be collected, spanning from those active in tree planting to those that advised or advocated the use of silvopastoral systems. By having focus groups dedicated to specific occupations, opinions could be compared between groups, with the intention to identify gaps between those using these systems and those advocating or providing advice.

Table 2. Occupation and participant numbers for the workshop participants.

Occupation	Number of focus groups	Number of participants
Landowner/farmer	3	16
Land management advisors (LMAs)	1	6
Policy advisors	1	5
Advocacy groups	1	7
Total	6	34

Focus groups were held on 11^th October 2022 on a farm in the Wairarapa region of NZ. Individuals were grouped in focus groups based on occupation and there were 34 participants in total (Table 2).

There were two focus group sessions during the day. The first session focused on silvopastoral systems that are currently used in NZ. The questions for the landowner group explored (i) whether they planted silvopastoral trees on their farms, (ii) the reasons they did or did not plant, and (iii) challenges related to the use of these systems. The land management advisors (LMAs), policy advisors and advocacy groups were asked (i) whether they advise farmers to plant silvopastoral trees, (ii) the reasons they do or do not advise farmers to plant, and (iii) challenges related to advising the use of silvopastoral trees.

The second session explored the potential of silvopastoral systems in the future for positively impacting NZ farms. For the landowners, the following questions were asked: (1) What potential benefits of these systems are important to you? (2) What potential barriers are there to planting in the future. For the LMAs, policy advisors and advocacy groups, the following questions were asked: (1) Where do you see the potential for silvopastoral systems to respond to future challengers? (2) Do you think there will be any barriers to adoption of future silvopastoral systems.

Responses were recorded and compiled on flipcharts. Participants were fully anonymised and not recorded. Full consent for this work to be published was provided by all participants at the beginning of the workshop.

Workshop findings – current silvopastoral systems that are used, the reasons for their use and challenges of their use

Only one landowner participant did not plant or manage silvopastoral trees on their farm. Poplar (Populus spp.), followed by willow (Salix spp.) and eucalyptus (Eucalyptus spp.) trees, were the most common silvopastoral trees planted. Other trees that were planted included tagasaste (Cytisus proliferus L.f.), cork oak (Q. suber L.), acacia (Acacia spp.), chestnut (Castanea spp.), Tasmanian blackwood (A. melanoxylon R.Br.), maple (Acer spp.), radiata pine, ginkgo trees (Ginkgo biloba L.) and mānuka (Leptospermum scoparium J.R.Forst. & G.Forst.). Native remnant tree species were also managed as silvopastoral trees, such as totara (Podocarpus totara G.Benn. ex D.Don), kānuka (Kunzea spp.) and southern beech (Nothofagus spp.).

The landowner participants had substantial knowledge of the potential benefits of silvopastoral systems, with soil erosion, shade, shelter and carbon credits being the most commonly stated benefits (Figure 3). The most common limitations associated with these silvopastoral systems included protection against pests, tree survival, and cost of planting, protection and maintenance (Figure 3).

Figure 3. Landowner responses for silvopastoral benefits (left) and limitations (right).

Many of the benefits and limitations of silvopastoral systems stated by the non-landowner groups (Table 3) were similar to those stated by landowners (Figure 3). Additional benefits stated by the policy group included iwi partnership/Te Ao Māori benefits, tourism and contributing to NZ’s clean green image, and ecosystem health. The central challenges for LMAs when advising landowners included difficulties with regulation (competing incentives and regulations), managing farmer expectations and farmer resistance, timescale issues (in terms of a time lag until effectiveness and total costs during a tree’s life), and the cost of providing advice. The advocacy group stated they did not provide much advice on silvopastoral systems as they were not often consulted. A reason suggested for this was that the focus in NZ has been more on pastoralism, as opposed to silvopastoralism.

Table 3. Limitations given by non-landowner groups for the potential of silvopastoral systems to be realised in the future.

Workshop occupation	Benefits	Limitations
Policy	Animal welfare Biodiversity Climate Cultural amenity/outcomes Farm diversification Ecosystem health Shelter Te Ao Māori Water quality	Cost Lack of knowledge and demonstration of benefits Lack of science funding Long-term thinking required/short-term planning common Social resistance Some benefits are intangible
LMA	Aesthetics Bee populations Biodiversity Carbon/Emissions trading scheme Climate Bee populations Resilience of farm system Shade Shelter Slope stability Soil management Summer feed Water quality	Cost ETS legislation important for individual trees Importance of maintaining ground knowledge if too much emphasis on technology Lack of knowledge Lack of supply of trees Protection Undeveloped industry for alternative timber
Advocacy	Carbon farming Ecosystem services Employment potential Erosion Fodder Income diversification Organic addition to the soil Slope stability Water retention	Bad experience with past trees Emissions Extreme weather Lack of interest Lack of knowledge Lack of success stories Need for economic value of land Policy thresholds

Future silvopastoral systems and their potential

The landowners already had considerable knowledge of the future potential of silvopastoral systems, indicated by the wide variety of benefits this group already stated in Figure 3. The potential barriers to adoption were also very similar to the limitations reported in Table 3. The use of silvopastoral systems by landowners, and knowledge of them, was greater than what was anticipated when planning the workshop.

In addition to barriers presented in Table 3, landowner participants raised more intangible barriers to the future adoption of silvopastoral systems in this section, such as narratives and perceptions. For example, one group discussed that the current paradigm is focused on fixing what is broken rather than creating future possibilities. Other comments focused on the importance of managing the tension between the economic and environmental impacts of these systems, and past landowner attitudes that have typically been negative towards trees.

Another important barrier that was discussed was the regulatory environment for NZ’s Emissions Trading Scheme (ETS). There was frustration that credits could not be collected from individual trees that had been planted if they were not in a group of trees greater than 1 ha in size that had a canopy cover greater than 30% (Ministry for Primary Industries 2020). There was also frustration that the regulatory environment was not recognising the environmental work farmers were already doing on their farms.

One of the most important barriers brought up by both the landowners and non-landowner groups was the cost of these systems. A lack and cost of labour was emphasised as an important barrier for these systems, as well as the cost of protection. An important barrier to future adoption discussed in the landowner groups but not in the non-landowner groups (policy group, LMA, advocacy) was the impact of pests on trees (Table 3).

Some landowners and the LMA groups expanded on future steps that may be required to increase the use of silvopastoral systems in NZ. There was an emphasis on scientific research, specifically research that informs tree selection and improves tree cultivar options. Funding was also deemed important to help with planting, pest control and tree management. Improving pest control methods was also suggested, in addition to various answers related to capacity building, both in terms of support for farmers, and capacity building that improves relationships between researchers and farmers.

Limitations

It is likely that the participants who attended the workshop had more knowledge of silvopastoral systems than if participants were randomly sampled in the region because those that attended likely had an interest in trees. Therefore, the number of landowners who used silvopastoralism was most likely an overestimation. Moreover, it is expected that the typical landowner in the region would not have planted the diverse set of trees mentioned, but instead planted just poplars and willows, or no silvopastoral trees. Nevertheless, because the landowners likely had more experience using silvopastoral trees compared to a randomly sampled landowner in the region, these experiences with trees were important for informing on the challenges and potentials of these systems in NZ.

Integrative silvopasture design in New Zealand

Outcomes

While silvopastoral systems in NZ have primarily been implemented for erosion mitigation, there is potential for future design to target a broader range of outcomes. Table 1 presents a broad range of silvopastoral benefits that have been identified globally. Here, we summarise previous silvopasture research into some outcomes that may be particularly relevant for NZ.

The impact of silvopastoral systems on soil erosion and pasture production has been particularly well studied in NZ. Most empirical studies aimed at quantifying the effectiveness of biological erosion control under silvopastoral systems have been carried out in NZ (Douglas et al. 2013). A paired comparison study following a storm event in April 2011 in the Hawke’s Bay region, selected 86 sites with trees (defined as stands of 1–14 trees) and 25 control sites with pasture only (McIvor et al. 2015). Landslide erosion was 78% less on sites with trees compared to the pasture control sites. A similar method was used across 65 sites in the Manawatu and Wairarapa following storms in 2004 and 2006, respectively (Douglas et al. 2013). Trees reduced landslide occurrence by 95% compared to paired pasture control sites (0.4% vs. 7.9% scar area, respectively), and scars occurred on fewer sites with trees than pasture (10 vs. 45).

Since slope instability is caused by multiple interacting factors, Spiekermann et al. (2022a) developed a data-driven approach to improve targeting of erosion control from slope to landscape scales to improve upon the empirical observations of previous research (e.g. Douglas et al. 2013). By modelling future shallow landslide occurrence based on the actual tree cover versus a treeless (pasture only) baseline scenario, the performance of silvopastoralism in terms of reducing landslide erosion was assessed for two farms. The difference in outcomes across the two selected farms (17% versus 43% reductions) demonstrated the importance of targeting tree planting to slopes that are more likely to be sites of future landsliding. Furthermore, Spiekermann et al. (2022b) undertook similar modelling to quantify the effect of silvopastoralism on landslide-derived sediment yields, contrasting the cost-effectiveness of targeted and non-targeted approaches to tree planting, which will be explored below.

The majority of research measuring the impact of silvopastures of pasture production has studied poplar (Populus spp.) and radiata pine (Pinus radiata D.Don) trees, and without exception, every study has found reduced pasture production under low-density systems using these species (< 200 trees ha⁻¹), with reductions ranging from 12% to 94% (Percival and Knowles 1988; Hawke 1991; Gilchrist et al. 1993; Cossens and Cossens 1995; Devkota et al. 1997; Cossens and Hawke 2000; Douglas et al. 2001; Wall 2006; Guevara-Escobar et al. 2007; Benavides et al. 2009). However, a recent study reported that individually spaced, mature kānuka (Kunzea robusta de Lange & Toelken) trees increased pasture production by 108% compared to open pasture at two sites, which was associated with significantly greater Olsen-P, organic matter, Total-N and porosity under the trees. This increase was hypothesised to be because the trees facilitated livestock activity under the canopies, in addition to the potential addition of organic matter by the trees to the soil (Mackay-Smith et al. 2022b). Mackay-Smith et al. (2022b) is the only published study to report increased pasture production under mature silvopastoral trees compared to open pasture in NZ. Further research is required to explain the disparity between the increased pasture production under kānuka trees reported by Mackay-Smith et al. (2022b) and the past poplar research (Benavides et al. 2009; Kemp et al. 2018).

There has only been one study that has compared the impact of fully grown isolated silvopastoral trees on surface runoff and associated nutrient and sediment loss in NZ (Mackay-Smith et al. 2022a). The authors reported 54.0 mm of annual surface runoff under kānuka (K. robusta) trees and 7.5 mm of annual surface runoff in open pasture. Sediment and nutrient losses were 10–100 times greater under the kānuka trees, however, sediment and nitrogen (N) losses were small compared to past NZ studies, although phosphorus (P) losses were relatively high compared to past studies (McDowell and Wilcock 2008; Mackay-Smith et al. 2022a). There was evidence of selective grazing of pasture under the kānuka trees in this study, which likely increased surface runoff and associated sediment and nutrient losses in this treatment (Mackay-Smith et al. 2022a).

A few studies have been conducted in NZ on greenhouse gas (GHG) emissions or soil carbon sequestration in silvopastoral systems. Several NZ studies measured elevated soil carbon concentrations under radiata pine (Saggar et al. 2001; Chang et al. 2002) and kānuka silvopastoral systems (Mackay-Smith et al. 2022b) compared to bare soil or open pasture. These studies, however, were not designed to compare the effect of long-term silvopasture to open pasture on soil carbon stocks at depth and dedicated research focussed on measuring carbon sequestration under silvopastoral systems is needed. Douglas et al. (2020) compared soil carbon under 14-year-old poplars, 16-year-old alders and open pasture. Soil C mass was reported to be 11–18% lower in the pasture-poplar system compared to open pasture, and 6% greater in the pasture-alder compared to open pasture. Because the poplar and alder trees in this study were planted on separate sites, they were not included in the systematic review as part of this paper.

Moreover, animal welfare and productivity has rarely been studied in NZ silvopastoral systems, with previous research focusing on artificial shade (Kendall et al. 2006; Fisher et al. 2008) and widely spaced willow (Salix spp.) trees (Betteridge et al. 2012). The benefits of shelter (tree lines, sheds, rocky out crops etc.) to newborn lamb survival in NZ are clearly established in Pollard (2006), yet none of these studies examine the effect of silvopastoral systems on cold stress in animals.

Earthworm populations have been studied under poplar trees, which were found to reduce or maintain earthworm populations compared to open pasture (Guevara-Escobar et al. 2002), however, no published studies comparing bird or insect biodiversity in silvopastoral systems in NZ, despite there being potential for these systems to harbour greater abundance compared to treeless pastures (Torralba et al. 2016; Gómez-Cifuentes et al. 2020, 2019).

Historically, planting of poplars and willows in NZ was primarily for the purpose of erosion control, but the co-benefit of forage provision (leaves, twigs, bark) has long been appreciated and researched (Kemp et al. 2001). More recent research has highlighted the potential of a broader range of species, including 9 exotic and 10 indigenous tree species, although almost no research has been conducted on indigenous species (Tozer et al. 2021). More research is required on other income opportunities in these systems, especially native high-value timber species and the viability of trees to produce additional tree by-products.

There has been much research and development on the current pole planting method that is used to plant poplar and willow trees in NZ (Wilkinson 1999; Charlton et al. 2007; Phillips et al. 2014). There has been limited research into the development of protection strategies for planting alternative tree seedlings in the presence of livestock.

Finally, there has been little formalised research on the cultural values of silvopastoral systems in New Zealand. Moreover, given that NZ recognises two underpinning worldviews, Te Ao Pākehā (the British Monarchy) and Te Ao Māori, the indigenous Māori worldview (Harcourt et al. 2022), it is important to create a space for Māori to explore how silvopastoralism may fit with their own values and aspirations. We advocate for a kaupapa Māori approach (research conducted by, with and for Māori) so that Māori communities can assess the potential benefits of silvopastoralism.

Despite great potential for silvopasture to positively impact a range of outcomes if designed correctly, the paucity of research into the impacts of silvopastoral systems on cultural values, surface runoff and sediment and nutrient losses, biodiversity conservation, GHG emissions, livestock welfare and diversification options highlights many opportunities for silvopastures to provide a range of additional benefits to NZ farms. Moreover, if research is expanded to a greater breadth of outcomes, this will enable different NZ silvopastoral trees to be more fairly compared and assessed over their full range of known functions (Mackay-Smith et al. 2021), and likely show how a more diverse range of trees species could be used in silvopastoral systems within NZ.

Tree attributes and processes

As was discussed in Section Integrative silvopasture design in New Zealand: Outcomes, a function that has not been utilised in New Zealand is the potential for silvopastoral trees to increase pasture production compared to open pasture. As was explored in the processes review, there are likely more nuanced tree functional traits than whether trees are evergreen or deciduous, or are N₂-fixers or not, that are important for improved pasture production in silvopastoral systems. This becomes clear when comparing various studies of a single tree system comparison between tree pasture and open pasture positions.

For example, Durr and Rangel (2002) reported 132% more pasture production under N₂-fixing deciduous Samanea saman trees in north-east Queensland, Australia. Mackay-Smith et al. (2022b) also found over 100% more pasture production under kānuka trees in NZ compared to open pasture, although kānuka trees are evergreen and non N₂-fixing. Evidently, pasture production can be greater under deciduous or evergreen trees, and N₂-fixing or non N₂-fixing trees.

Pasture production can even vary under the same tree species in similar climates and systems. Gea-Izquierdo et al. (2009) measured ∼3500 kg dry matter (DM) ha⁻¹ under the canopy of holm oak (Quercus ilex L.) trees, and ∼2000kg DM ha⁻¹ outside the canopy in a Spanish site with rainfall of 573 mm. On the contrary, Cubera et al. (2009) reported significantly less pasture production under the canopy of holm oak trees in unfertilised (open areas: 2540 kg DM ha⁻¹; under tree canopy areas: 1390 kg DM ha⁻¹) and fertilised (open areas: 5230 kg DM ha⁻¹; under tree canopy areas: 2080 kg DM ha⁻¹) treatments in a site with 665 mm of rainfall. Clearly, there are likely other tree attributes that are important for ensuring silvopastoral trees maximise pasture production on farms.

A factor that was not measured in these studies was livestock activity (Durr and Rangel 2002; Cubera et al. 2009; Gea-Izquierdo et al. 2009; Mackay-Smith et al. 2022b). Livestock activity was an important process in the studies reviewed in the process review, and both Mackay-Smith et al. (2022b) and Durr and Rangel (2002) suggest that livestock spending preferential time under the measured silvopastoral trees could be having important impacts in the measured positions. The impact of livestock activity could be an integral piece in the puzzle that explains much of the variation seen within broader-scale functional groups generally considered in pasture silvopastoral comparisons (Rivest et al. 2013).

It is unknown whether the livestock camping effects measured by Mackay-Smith et al. (2022b) and Durr and Rangel (2002) provided a net improvement to pasture production at the paddock scale. For example, the livestock effects could have been nutrient re-distribution from open areas of the paddock to tree-pasture areas of the paddock. Alternatively, as the trees studied by Mackay-Smith et al. (2022b) were on 20–25 degree slopes, the re-distribution could have been from flatter areas of the paddock where nutrients were likely in excess, to the tree-pasture areas, resulting in a net increase in pasture production at the paddock scale. Further work is required to understand the relationship between pasture production, tree spatial design and livestock activity at the paddock scale in silvopastoral systems.

Another important factor could be the size of silvopastoral trees. Rivest et al. (2013) reported that eucalyptus trees were the only functional group to significantly reduce pasture production compared to open pasture, and all past work on poplar trees greater than 15 years old in NZ has reported reduced pasture production under their canopies compared to open pasture (Benavides et al. 2009). Both these genera (eucalyptus and poplar) are typically large trees and greater than 30 m when fully mature. This contrasts with the principle silvopastoral trees used in Southern Europe (holm oak and cork oak) and Patagonia, Argentina (Nothofagus antarctica (G.Forst.) Oerst.), that are all between 4 and 15 m (Pulido et al. 2001; Sánchez-González et al. 2005; Gouveia and Freitas 2008; Peri et al. 2016), and past studies in all these systems have reported increased pasture production under these silvopastoral trees in certain situations (Peri 2005; Moreno 2008; Gea-Izquierdo et al. 2009).

These examples show that there are still many unknowns in terms of tree attribute-pasture production relationships, and more research is required so silvopastoral systems can be designed to maximise the productive benefits they add to farms. Furthermore, it is likely that the impact of tree attributes on silvopastoral outcomes will vary depending on regions and farm types, so this must be considered in future research.

Local perspectives

The perspectives presented highlight that silvopastoral systems are already used in NZ, albeit with many limitations. This indicates that if new systems can be designed which overcome some of the limitations presented in Figure 2, they should be adopted. Nevertheless, a key limitation raised in the workshop was the practical realities of planting and managing silvopastoral trees in general. Tree survival is integral, in addition to the cost of establishing and managing the system, and protection against livestock and pest damage (Figure 3). This is likely one reason why poplar and willow have been so successful in NZ, because they can be established as sharpened, coppiced poles in the presence of livestock. If other trees are to be used in NZ, planting methods and survival rates must rival that of poplar and willow pole planting.

If more research can show whether trees with different attributes to poplar consistently increase pasture production compared to open pasture (Mackay-Smith et al. 2022b, 2021), then this could potentially offset the costs of planting and managing silvopastoral trees. Moreover, it will also be important to select silvopastoral trees that require minimal management. One key constraint of poplar plantings is current land management advice to fell and replant the trees after 40 or 50 years (Charlton et al. 2007). More recent varieties (e.g. Populus maximowiczii × nigra) are expected to grow for longer than this, which would reduce the overall management required for each of the trees. Additionally, if other silvopastoral tree species can be found that grow for over 200 years, as it common in the dehesa system in Spain (Plieninger et al. 2021), this may substantially increase the attractiveness of these systems.

Targeting

Figure 4 illustrates the importance of carefully planned silvopastoral design in the context of soil erosion, based on the integration of landslide susceptibility and connectivity modelling (Spiekermann et al. 2022a, 2022b). A large number of widely spaced trees are shown in the aerial photo, which are mainly poplar and willow trees, as well as a labelled eucalyptus grove (shown in the photo). These trees were planted on slopes to reduce landslide erosion following the severe rainfall event of 1977. The two maps identify areas where future landslide-derived sediment is likely to be sourced from, with approximately two-thirds estimated to be sourced from the ‘high’ zone. The maps show the difference with and without trees present. The Eucalyptus grove has been planted primarily in the red zone, which is where landslides would be expected to occur in future – and deliver sediment to adjacent streams – if no trees were present. The second map shows the same landslide susceptibility classification, now including the influence of the trees depicted in the orthophoto. The slope with the eucalyptus grove has mostly changed to green, which indicates it has successfully been stabilised. It provides a contrast to the poplars and willows to the west, which are primarily located on lower slopes or valley bottom where landslides are unlikely to occur with or without trees. Therefore, there is little change to the extent of the red zone in this part of the map and demonstrates the difference between a targeted and non-targeted approach to silvopastoral design. Soil conservation trees not fulfilling their primary function have additional implications due to the possible reduction in pastoral production beneath their canopy, ultimately resulting in an undesirable net negative outcome.

Figure 4. Potential landslide-derived sediment delivery to streams based on modelling landslide susceptibility (Spiekermann et al. 2022a) and landslide connectivity (Spiekermann et al. 2022b) modelling for a small pastoral area in the Wairarapa, NZ. The comparison of a treeless baseline scenario with the actual tree cover (shown in the orthophoto from 2013) demonstrates that the eucalyptus grove (photo insert) has led to a much greater reduction in future sediment delivery compared with the poplars and willows to the west, which were largely planted in areas where landslides are unlikely to occur with or without trees present.

Decision-support tools can be used to develop well designed silvopastoral systems. For example, Spiekermann et al. (2022b) undertook a cost-effectiveness assessment by quantifying the reduction in sediment delivery across fifty farms in the Wairarapa region of NZ, based on contrasting planting options: (1) targeting critical source areas of sediment and (2) a non-targeted approach to tree planting. Results of the cost-effectiveness assessment suggest there is a ten-fold difference between a targeted and non-targeted approach in terms of the ratio of investment to reduction in sediment delivery achieved. In monetary terms, the average cost for a 500 ha farm to achieve a 10% reduction in landslide-derived sediment was estimated at $14,100 for targeted mitigation and $131,000 for non-targeted tree planting.

An additional key finding in integrating landslide susceptibility and connectivity modelling is that, with a shift in focus from soil conservation to freshwater management, the area requiring tree planting halved since many landslide deposits would not reach streams. In total, only 6.5% (2370 ha) of the 36,000 ha of farmland assessed were deemed to be the potential source of approximately two-thirds of landslide-derived sediment across the 50 farms. Due to existing tree cover, this area requiring tree planting has already been reduced to 4.7% (1720 ha). This example illustrates how targeting trees in the pursuit of a desired outcome can be achieved with appropriate decision-support tools.

Key principles for an integrative approach to designing silvopastoral systems

Silvopastoral landscapes are complex systems that require holistic management across all dimensions

In reviewing the processes silvopastoral system literature globally, it is evident that there is a lack of studies assessing the biophysical processes holistically. When discussing why silvopastoral trees may facilitate or compete with pasture production, many studies focus on just tree effects and miss the impacts livestock activity has in the system (Frost and McDougald 1989; Gea-Izquierdo et al. 2009; Rivest et al. 2013; Seddaiu et al. 2018). Nevertheless, considering all known ecological interactions is integral to more fully understand the processes within the system, and how different tree attributes impact these processes and the impact this has on different outcomes.

Moreover, there is a bias in terms of which outcomes have been studied in silvopasture comparison studies (Figure 2), and in past NZ silvopasture research. Researching and considering the impact of silvopastures on a more diverse range of outcomes is integral for these systems to fulfil their potential, and viewing them holistically should facilitate this process.

The views, values and experiences of local people are deeply connected to silvopastoral system design and must be incorporated

Considering local and indigenous values, experiences and perspectives is essential if ecologically sound silvopastures are to be adopted. Every region will have slightly different challenges and desires. Understanding these and considering them when selecting silvopastoral trees should be the cornerstone to integrative silvopasture design. The workshop findings also demonstrate the importance of practical constraints when designing silvopastoral systems, and these must be considered in any proposed system. These can only be understood by speaking to practitioners.

Spatial heterogeneity in environmental and social conditions requires locally specific decisions

The summary of possible silvopastoral outcomes (Table 1) identifies a multitude of possible objectives when designing silvopastoral systems. However, it is likely that a general hierarchy in the prioritisation of outcomes exists and that the order of priorities will vary spatially as a function of limiting environmental factors (e.g. climate and soils), risks (e.g. erosion, heat stress in livestock), and individual preferences (e.g. cultural). Since in general silvopastoral systems are foremost aimed at animal production, economic sustainability is a fundamental priority. Thus, the central desired outcome in a silvopastoral system is generally animal production. However, for a given farm, the priorities in outcomes may vary spatially, giving rise to silvopastoral designs bespoke to the differing limiting factors, risks, and preferences.

Determining the central function prior to planting decisions, and selecting trees that have attributes that are beneficial to this function, are an integral part of designing value-maximising silvopastoral systems. Figure 5 shows a schematic for a pastoral landscape with multiple land-use classifications. This schematic highlights how different tree planting will be appropriate for different areas on the farm, and will vary spatially depending on the function of the trees. For example, to conserve soils and reduce sediment delivery to adjacent streams, steep slopes prone to landslide erosion will require different tree species and planting densities compared to rolling pastures. While there is substantial evidence that poplar, the most commonly planted silvopastoral tree in NZ, negatively impact pasture production (Benavides et al. 2009; Mackay-Smith et al. 2021), the main intended function of poplar tree planting is soil conservation. While a reduction in pastoral production may be acceptable in low producing areas of the farm, to achieve a balance between production and environmental outcomes in medium producing pastoral land will require a change in focus with respect to tree attributes and spatial layout of trees (Figure 5).

Figure 5. Schematic diagram illustrates how silvopastoral design could vary spatially within a farm and how the design relates to desired outcomes. A typical farm will be more spatially diverse than this diagram indicates, and this figure is purely a representation of how there is a relationship between silvopasture outcome, topography and tree attribute/vegetation type on the farm.

Understanding of ecological processes must underpin all management decision

There are many different ecological considerations in silvopastoral system design (Jose et al. 2019). When attempting to positively impact a specific outcome, it is important to understand how different tree attributes impact the agroecological processes within silvopastoral systems (Figure 1). Silvopastoral can then be optimised based on tree attributes so they increase facilitation whilst minimising competition (Jose et al. 2019).

The processes reviewed (Figure 1) did not focus on one specific outcome (Table 1). This is because if silvopastoral trees can improve the agroecological environment, this will positively impact most outcomes. For instance, independent of the process, if silvopastoral trees can improve soil fertility, soil organic matter and soil physical capacity, this will likely have positive impacts on pasture production (Durr and Rangel 2002; Mackay-Smith et al. 2022b), discharge rates and sediment losses (Zhu et al. 2020). Moreover, the addition of a tree will sequester carbon (Guevara-Escobar et al. 2002), improve slope stability (Spiekermann et al. 2022a, 2021), have important livestock health benefits (Blackshaw and Blackshaw 1994), and could benefit biodiversity (Torralba et al. 2016; Gómez-Cifuentes et al. 2020; Mupepele et al. 2021). Taking a processes understanding of the system should help in determining which tree attributes are important for improving the agroecological environment, resulting in integrative improvements to silvopastoral outcomes and the pastoral environment.

The complexity and spatial heterogeneity present in silvopastoral systems requires high-resolution data and tools

Section Targeting illustrated the use of decision support tools in the form of spatial predictions of future landslide-derived sediment delivery to determine where to focus silvopastoral tree planting to mitigate erosion and sediment loss. However, since priorities of outcomes will differ spatially – also due to individual preferences, spatial tools that can cater for these preferences are required to aid land managers in making informed decisions. There is significant further field-based research required before outcomes such as dry matter production can be modelled at the scale of individual trees while accounting for agroecological interactions in the silvopastoral system (Figure 1). Notwithstanding current knowledge gaps, decision support tools (e.g. multi-criteria decision analysis) need to be designed to better integrate technical information and stakeholder values.

Moreover, additional information would result in more informed and possible different decision making (Spiekermann et al. 2015). Decision-support tools can help minimise limitations related to subjectivity in decision-making and help overcome the complexity involved in considering trade-offs between environmental, economic, political and cultural impacts (Huang et al. 2011).

Conclusion

This paper took an integrative approach to silvopastoralism design to account for the ecological and social complexity of these systems. Ecological processes within silvopastoral systems were systematically reviewed. This was followed by an integrative design process using a NZ case study, which generated 5 key principles that should be considered in future silvopastoral system designs globally.

The silvopastoral system process review demonstrated that livestock activity can have substantially more impact than more typically researched processes, such as microclimate influences and N₂ fixation.

More research is required on under-researched areas such as livestock activity, leaf smother and allelopathy effects, and there is a need for studies that measure a greater range of processes in a silvopastoral system which are then related to silvopasture outcomes. Furthermore, it is important that future research spans a greater variety of tree functional traits and silvopasture processes, such as tree size, tree architecture, livestock preferences, tree water use, leaf smother and allelopathy effects.

The NZ case study explored integrative design in four themes: outcomes, processes, local perspectives, and targeting. A NZ case study workshop generated information of the practical constraints of using silvopastoral systems, such as the impact of pests, survival challenges and costs, and any evaluation of new systems must account for these limitations. Furthermore, silvopastoral systems have not fulfilled their potential over a broad range of outcomes in NZ, and more research is required to learn which key functional traits are required for maximising pasture production in these systems. Finally, an example was shown how targeting soil conservation trees is fundamental for improving the cost-effectiveness of these systems and the positive impacts of them to farms.

Following the NZ case study, the authors propose 5 key principles that should be considered in future silvopasture design:

1. Silvopastoral systems are complex and require holistic management.

2. The views, values and experiences of local people are deeply connected to silvopastoral system design.

3. Spatial heterogeneity in environmental and social conditions requires locally specific decisions.

4. Understanding of ecological processes must underpin all management decisions.

5. The complexity and spatial heterogeneity present in silvopastoral systems requires high-resolution data and tools.

Considering these principles will be integral for maximising the positive benefits novel silvopastoral systems provide to pastoral farms.

Acknowledgements

Our thanks to all participants of the silvopastoralism workshop, especially Elizabeth McGruddy, Esther Dijkstra, David Boone, to Richard Tosswill for letting us use his farm for the venue of the workshop, and to Nicolette Faville for designing Figures 1 and 5.

Disclosure statement

No potential conflict of interest was reported by the author(s ).

Additional information

Funding

This work was funded by the NZ Ministry for Business, Innovation and Employment’s Our Land and Water (Toitū te Whenua, Toiora te Wai) National Science Challenge, contract C10X1901, as part of the ‘Maximizing benefits of silvopastoral systems for multifunctional rural landscapes in Aotearoa’ (A Think Piece) project.

Appendix 1.

Table of articles included in the systematic review

Table

Silvopastoral system comparison	Density	Tree age (yr)	Location	Long-term annual rainfall (mm)	Reference
Acacia nigrescens Oliv.; Sclerocarya birrea (A.Rich.) Hochst.; Colophospermum mopane (J.Kirk ex Benth.) J.Léonard; Lonchocarpus capassa Rolfe; Cassia abbreviata Oliv., Lannea schweinfurthii (Engl.) Engl.	15 m apart on average	Not given	Saadani National Park, Tanzania	540	Treydt et al. (2007)
Acacia tortilis (Forssk.) Hayne; Adansonia digitata L.	Isolated trees	Not given	Tsavo National Park, Kenya	450; 750	Belsky et al. (1993a)
Acacia tortilis (Forssk.) Hayne; Adansonia digitata L.	Isolated trees	Not given	Tsavo National Park, Kenya	400	Belsky et al. (1993b)
Acacia tortilis (Forssk.) Hayne; Adansonia digitata L.	Isolated trees	Not given	Tsavo National Park, Kenya	400–500	Belsky et al. (1989)
Albizia saman (Jacq.) F.Muell.; Enterolobium cyclocarpum (Jacq.) Griseb; Tabebuia rosea (Bertol.) Bertero ex A.DC.; Guazuma ulmifolia Lam.	Isolated trees	Mature	Matalagalpa district, Nicaragua	1500	Casals et al. (2014)
Andira inermis (W.Wright) DC; Bellucia pentamera Naudin; Guarea Guidonia (L.) Sleumer; Phragmanthera leonensis (Sprague) Balle; Zygia longifolia (Humb. & Bonpl. Ex Willd.)	Isolated trees	Not given	Caqueta, Colombia	3793	Álvarez et al. (2021)
Brazilian native tree system (Anadenanthera colubrina (Vell.) Brenan); Peltophorum dubium (Spreng.) Taub.; Zeyheria tuberculosa (Vell.) Bureau ex Verl.; Cariniana estrellensis (Raddi) Kuntze; Piptadenia gonoacantha (Mart.) J.F.Macbr; Guazuma ulmifolia Lam.; Croton floribundus (Spreng.); Eucalyptus urograndis (EUCUG) system	Native system: 17 m wide rows and three tree line rows with a spacing of 2.5 m x 2.5 m. Eucalyptus: 15 m wide row; 4 × 4 m spacing between trees.	Native system: 7–10 years; Eucalyptus system: 3–6 years	São Paulo, Brazil	Dry season: 250 mm. Rainy season: 1100 m	Pezzopane et al. (2019)
Caesalpinia eriostachys Benth.; Cordia elaeagnoides A.DC.	Isolated trees or 10 × 10 m	Not given	Jalisco, Mexico	746	Galicia and Garcı́a-Oliva (2004)
Cassia grandis L.f.; Guazuma ulmifolia Lam.; Tabebuia rosea (Bertol.) Bertero ex A.DC.	Isolated trees	Not given	Central Nicaragua	1500	Rusch et al. (2014)
Eucalyptus melliodora A.Cunn. Ex S.Schauer; Eucalyptus blakelyi Maiden; Eucalyptus nova-anglica H.Deane & Maiden	Isolated trees	Not given	New South Wales, Australia	800–900	Wilson (2002)
Ficus insipida Willd.; Ficus colubrinae Standl.; Nectandra ambigens (S.F.Blake) C.K.Allen	Isolated trees	5–30	State of Veracruz, Mexico	4725	Guevara et al. (1992)
Guazuma ulmifolia Lam., Cassia grandis L.f.; Albizia saman (Jacq.) F.Muell	16 × 16 m	13	Cordoba, Colombia	1380	Martínez et al. (2014)
Guazuma ulmifolia Lam.; Crescentia alata Kunth	Isolated trees	Mature	Southwestern Nicaragua	1400	Hoosbeek et al. (2018)
Hybrid Eucalyptus grancam; Hybrid Eucalyptus urograndis	12 × 2 m (rows)	8	São Paulo, Brazil	1100–1500	Sarto et al. (2020)
Inga spp.; Phragmanthera leonensis (Sprague) Balle	Isolated trees	Not given	Pichincha, Ecuador	3198	Rhoades et al. (1998)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	Row spacings of 1.8, 3.7, 7.3 and 14.6 m	7–8	Virginia, USA	1000	Buergler et al. (2005)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	Row spacings of 1.8, 3.7, 7.3 and 14.6 m	7–8	Virginia, USA	1000	Buergler et al. (2006)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	Isolated trees	10–11	Virginia, USA	1000	DeBruyne et al. (2011)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	12 × 4 m (single row)	13–14	Virginia, USA	1000	Fannon et al. (2017)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	12.2 × 12.2 m	10	Virginia, USA	1000	Pent and Fike (2019)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	12.2 × 12.2 m	9–11	Virginia, USA	1000	Pent et al. (2020a)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	12.2 × 12.2 m	9–11	Virginia, USA	1000	Pent et al. (2020b)
Juglans nigra L. (black walnut); Gleditisia triacanthos L. (honey locust)	12.2 × 12.2 m	15	Virginia, USA	1000	Poudel et al. (2022)
Low tree diversity system (Eucalyptus grandis W.Hill ex Maiden); high tree diversity system (Acacia mangium Willd., Acacia angustissima (Mill.) Kuntze, Mimosa artemisiana Heringer & Paula, Leucaena leucocephala (Lam.) de Wit, Eucalyptus grandis)	Triple rows of trees (3 × 3 m) interspersed by 30 m of pasture	22	Coronel Pacheco, Minas Gerais, Brazil	1500	Cá et al. (2022)
Lysiloma acapulcensis (Kunth) Benth. and Vachellia pennatula (Schltdl. & Cham.) Seigler & Ebinger	Isolated trees	Not given	Veracruz, Mexico	909	Avendaño-Yáñez et al. (2020)
Mimosa caesalpiniifolia Benth.; Gliricidia sepium (Jacq.) Kunth	10 × 1 × 0.5 m (double rows)	4–5	Pernambuco, Brazil	1200	Apolinário et al. (2016a)
Mimosa caesalpiniifolia Benth.; Gliricidia sepium (Jacq.) Kunth	15 × 1.0 × 0.5 m (double rows)	6–7	Pernambuco, Brazil	1200	Gomes da Silva et al. (2021a)
Mimosa caesalpiniifolia Benth.; Gliricidia sepium (Jacq.) Kunth	15 × 1.0 × 0.5 m (double rows)	6–7	Pernambuco, Brazil	1200	Gomes da Silva et al. (2021b)
Mimosa caesalpiniifolia Benth.; Gliricidia sepium (Jacq.) Kunth	10 × 1.0 × 0.5 m (double rows)	7–9	Pernambuco, Brazil	1300	Lima at el. (2020)
Mimosa caesalpiniifolia Benth.; Machaerium aculeatum Raddi; Phragmanthera capitata (Spreng.) Balle; Artocarpus integrifolia L.	Isolated trees	Not given	Pernambuco State, Brazil	1200	Dubeux et al. (2014)
Native tree silvopasture; Pinus taeda L. silvopasture	Isolated trees	Native tree not given; Pinus taeda: 8–15	subtropical Atlantic Forest, Argentina	2000	Gómez-Cifuentes et al. (2019)
Native tree silvopasture; Pinus taeda L. silvopasture	Isolated trees	Native tree not given; Pinus taeda: 8–15	subtropical Atlantic Forest, Argentina	2000	Gómez-Cifuentes et al. (2020)
Pinus elliottii Engelm; Eucalyptus viminalis Labill; Quercus robur L.	20 × 2 × 2 m (double rows)	15	Córdoba, Argentina	700–800	Plevich et al. (2019)
Pinus palustris Mill.; Pinus taeda L.; Quercus pagoda Raf.	12 × 2 × 2 × 2 m (triple rows) and 24 × 2 × 2 × 2 m (triple rows)	8–11	North Carolina, USA	1500	Castillo et al. (2020)
Prosopis cineraria (L.) Druce; Acacia nilotica (L.) P.J.H.Hurter & Mabb., Ziziphus nummularia (Burm.f) Wight & Arn., Capparis decidua (Forssk.) Edgew; Acacia senegal (L.) Willd.	Isolated trees	Not given	Jodhpur district, India	100–450	Tripathi et al. (2005)
Prosopis juliflora – legume (maybe); Ziziphus joazeiro – non-legume; Spondias tuberosa – non-legume	Isolated trees or 10 × 10 m	13–14	Pernambuco, Brazil	600	Wick et al. (2000)
Quercus douglasii Hook. & Arn; Quercus wislizeni (A.DC.); Pinus sabiniana Douglas	Isolated trees	Not given	California, USA	500	Frost and McDougald (1989)
Quercus douglasii Hook. & Arn; Quercus wislizeni (A.DC.); Pinus sabiniana Douglas; Ceanothus cuneatus K.Brandegee	Isolated trees	Not given	California, USA	500	Ratliff et al. (1991)
Quercus robur L.; Betula pendula Roth	Isolated trees	Mature	Östra Vätterbranterna, Sweden	656	Jakobsson et al. (2019)
Zeyheria tuberculosa (Vell.) Bureau ex Verl; Eucalytpus grandis W.Hill ex Maiden	15 × 4 m (rows)	Zeyheria tuberculosa: 20; Eucalytpus grandis: 10	Lagoa Santa, Brazil	Not given	Lana et al. (2016)
Ziziphus joazeiro Mart.; Prosopis juliflora (Sw.) DC.; Spondias tuberosa Arruda	10 × 10 m	Ziziphus joazeiro and Spondias tuberosa: > 50; Prosopis juliflora: 14	Custódia, Brazil	740	Menezes and Salcedo (1999)
Ziziphus joazeiro Mart.; Prosopis juliflora (Sw.) DC.	10 × 10 m	Ziziphus joazeiro: > 50; Prosopis juliflora: 14	Custódia, Brazil	740	Menezes et al. (2002)