Sustainable agriculture: a review of management practices aimend at decreasing environmental impacts of Dutch agricultural systems

Review

Sustainable agriculture: a review of management practices aimed at

decreasing environmental impacts of Dutch agricultural systems

Abstract

Worldwide, agriculture has an impact on the environment. An increasing demand for food in the coming decades might increase the pressure on the majority of the planetary boundaries. Therefore, it is important to study the effect of land management practices on the environmental impact of agricultural systems. The Netherlands, where more than two thirds of the land in managed for agriculture, provide a good case study. In the last decades, Dutch agriculture has been practiced using external resources to optimise conditions for production. In this review, I systematically organise and study literature from 1990-2020 that discusses the effect of selected management practices on the environmental impact of agricultural practices in the Netherlands. The categories addressed in this study, namely greenhouse gasses & energy, nitrogen, biodiversity & ecosystem services and soil, appear as the most well studied topics within the body of literature. In all categories, examples of successful and unsuccessful land management practices were found. I find that in the case of nitrogen management and the implementation of agri-environmental schemes (AES) in Dutch agriculture, management is unsuccessful in decreasing the overall environmental impact. This is due to the compartmentalised approach offered by these management practices. By studying the environmental impacts from a system perspective, it has become clear that this perspective is often lacking from the proposed management practices. Hence, it is important that literature on management practices discusses the system boundary under study, to determine at which scale a solution operates effectively.

Introduction to agriculture

Worldwide, over 38 percent of the habitable land is used for agricultural purposes. An increasing demand for food from a growing population pre-empts an increased pressure on this land, or a possible expansion of the agricultural areal in the coming decades (Godfray et al. 2010). Depending on the agricultural land management, there are certain environmental impacts of all farming activities. The nine planetary boundaries that determine the safe operating space for humans on the planet provide a useful framework for assessing the environmental impact of agriculture. Agriculture is especially relevant to the planetary boundaries, as it is the largest single land use practice on earth.

Date of submission: 8 December 2020 Course:

Literature Review Earth Sciences Period: July-December 2020 Student details: Léon Feenstra (10760636) Keywords: Agricultural land management, planetary boundaries, environmental impacts, functional agrobiodiversity, system.

Agriculture directly relates to six out of the nine planetary boundaries, namely biochemical flows, freshwater use, land-system change, biosphere integrity, the introduction of novel (chemical) entities and climate change (Foley et al. 2011; Steffen et al. 2015). Many of these processes act on a planetary scale, causing agriculture to influence the world far beyond the 38% land covered, through its alternation of resource flows such as water, nitrogen, phosphorous, energy and chemicals (Galloway et al. 2008). Good management is the key to meeting the increasing demand for food within the limits determined by the planet’s functioning (Tilman et al. 2011). At the same time, keeping the change of these planetary systems within the identified boundaries determines the future capacity of the natural world to provide the ecosystem services and stability that agriculture itself relies on.

In anticipation of these increased pressures on the planetary systems mentioned above, lessons can be learned from the Netherlands, a country where historically, agriculture has been the most dominant land use, covering close to two thirds of the inhabitable land (Compendium Voor de Leefomgeving, 2020). Dutch agricultural practices in the twenty first century are the result of the geographic setting of the country and the historical processes of approximately the last two millennia. In terms of the geography, climate and soil set the limits for what types of agricultural land use were possible historically. The Netherlands combine a fertile delta with higher situated Pleistocene sands. Before 1950, soils determine the agricultural practices: on peat soils with high groundwater tables, pastures were combined with sheep and dairy farming. On fertile clay soils, arable crops were grown, while coastal sandy soils were used for flower bulb production (Brusse, 2009). From the 1950s on however, the fruits of the Green Revolution uncoupled the natural environment and the type of agricultural practice (Bieleman 2008).

After experiencing the ‘hunger winter’ in 1945, calls came for the increase of agricultural production, to prevent famine in the Netherlands. Even more importantly, the country aimed for an increase in welfare of the inefficient small scale farmers in the Netherlands. Agricultural policy moved from the national to the European level, where the Dutch farmer and social politician Sicco Mansholt started to push for the upscaling and intensification of agricultural practices in the newly formed ‘Common Agricultural Policy (CAP)’. Upscaling of the small scale Dutch mixed farms was done through land consolidation efforts, that decreased the diversity of the landscape. This facilitated the use of bigger machinery, that required investments from farmers who had financial resilience, giving advantage to bigger farms. On these farms, artificial fertilisers, pesticides, drainage technologies and high yielding breeds of crops and animals were used. This led to less diversity in terms of production systems, specialisation in farms, uniformity in crops with regard to their ripening, growth, genes and production and simplification of the landscape (Bieleman 2008).

Modern Dutch farms are characterised by a high production per hectare. This is realised through the high uniformity, intensity, large inputs of fertiliser, pesticides and energy and the large scale of agricultural practices. However, a direct consequence of this is a loss of functional agro-biodiversity and the associated capacity of the natural system to deliver essential ecosystem services (Tscharntke et al. 2005; Cardinale et al. 2012; Erisman et al. 2016). The elements of these systems, such as high yielding crop varieties and cattle breeds, require optimal conditions to maintain their productivity. Thus, the production of the agricultural systems relies upon external inputs rather than functional agrobiodiversity to manage the conditions of uniformity. Erisman et al. (2016) define this as a ‘control model’. From an ecological point of view, these systems have a high stability and a low resilience (Holling 1973; Erisman et al. 2016). Here, “Stability, (…), is the ability of a system to return to an equilibrium state after a temporary disturbance.” Whereas, resilience, according to Holling: “determines the persistence of relationships within a system and is a measure of the ability of these systems to absorb changes of state variables, driving variables, and parameters, and still persist.” By manipulating fluxes of nutrients, water and energy, the control model attempts to minimize changes in states, drivers and parameters, resulting in a stable and reliable level of productivity. When states, drivers and parameters cannot be controlled, productivity is likely to decrease significantly. States in the system may even shift in internal

organisation under large pressure (Scheffer et al. 2001). An alternative to the control model is the adaptation model, as proposed by Erisman et al. (2016). This model is characterised by a high diversity and a high functional agro-biodiversity, leading to an increased resilience and an increased capacity to deal with changes in states, drivers and parameters, while producing a relatively stable output.

Most of the agricultural production in the Netherlands approaches agriculture through the aforementioned control model. Inputs in the form of energy, nutrients and chemicals are used to maintain productivity. However, these inputs also affect the safe operating space of humans on the planet. The Limits to Growth report (1972), the campaign against DDT (banned in 1973) and acid deposition (from 1967 onwards) are key examples of the realisation that this dependence upon external resources has negative consequences for the environment. Besides these consequences, the control model depends on the stability of higher organisational levels. Agricultural production is vulnerable to changes in resource flows, an example of which is the European oil crisis in 1973. (van Apeldoorn et al. 2011; de Goede, Gremmen, and Blom-Zandstra 2013). Without access to external inputs, farming has to rely more upon the resilience of the system to maintain the internal organisation. However this has been undermined by the strive for uniformity, upscaling and a decrease of agro-biodiversity, creating a complex problem.

For more than three decades, scientists have attempted to study possible solutions to this complex problem. A key role is present for agricultural land management, varying from the spatial organisation of the farm to the use of external resources and the extent to which (nutrient) cycles are closed on a farm scale. Here, the main question is how agricultural land management can decrease the environmental impact of Dutch soil bound agricultural systems. For this review, I take a systematic approach at the literature that has dealt with this question in the last three decades. Systematic review

The objective of this study is to collect research that studies the effect of certain management practices on the environmental impact of soil bound agriculture in the Netherlands. Therefore, a systematic review is taken to be the most suitable approach, as this type of review aims to ‘draw together all known knowledge on a topic area’ (Grant & Booth 2009). However, the reviewed material may also represents different approaches to the defined problem, hence the review contains elements of a ‘Qualitative systematic review’ as well (Grant & Booth 2009). Suitable, peer reviewed scientific literature was selected using Web of Science. Figure 1 provides a flow chart of the selection procedure. One main query was used to gather a body of studies ‘top-down. By scanning titles of the first hundred papers in a ‘Web of Science’ category, using the ‘Analyse Results’ tool, the five categories with most relevant studies were selected. In these categories, titles were scanned one by one. Based upon their perceived relation with the scope of the present study, a total of 180 studies was selected. Data on these studies was downloaded and the abstract was read to double check the relevance. Unfindable or irrelevant papers were removed and the remaining studies were grouped into three levels of suitability from green (highly relevant) through yellow and red (least relevant) (figure 1). The remaining ‘highly relevant’ papers were then grouped into four different groups and eight remaining ‘system thinking’ papers. A bottom-up approach is taken for the ordering into categories and the subdivisions within these categories. In other words, the gathered literature determines the management discussed.

Most of the gathered results fall into one of four categories: Greenhouse gasses & Energy; Nitrogen; Biodiversity & Ecosystem Services or Soil. A small amount of the results is taking a multidisciplinary approach or attempts to take a more integrated approach at regarding the effect of agricultural land management on the environmental impact of farming systems, these are grouped together under ‘System’. The gathered records cover a timespan of 30 years, with the majority of the studies falling in the period 2005-2010 (Figure 2). There is no clear trend between the subjectively identified categories over the period. Overall, almost two thirds (64) of the records

originate at Wageningen University and Research (WUR) (Figure 3). The figure seems to suggest an overall dominance of WUR in the categories of GHG and Nitrogen, while records in the category BioES seem to have a varied origin, just as the studies in the category Soil and System.

Not all possible management practices are discussed, or even mentioned, in this review. First and foremost, not all management practices make sense in the Dutch agricultural setting. More importantly, not all sensible management practices might be practiced, or even considered. While it is interesting to investigate these practices, here attention is given to management that is discussed within the selected body of literature. This has as an advantage that these (more well-known) management practices are more likely to be implemented by farmers all over the management spectrum in the Netherlands. Increasing the chance that the overall environmental impact of Dutch soil bound farming systems will decrease.

Dutch agriculture: the main challenges

Agriculture dominates the landscape in the Netherlands. To realise the high productivity per hectare in dairy farming, horticulture and arable farming, Dutch agriculture has been intensified, mechanised, upscaled and homogenised. Using the fruits of the green revolution, agricultural production has increased tremendously over the last seventy years. However, both the organisation of the system, as well as the resource flows passing through, impact key processes. For one, the high consumption of energy and the emission of greenhouse gasses from soils, livestock and

manure causes the agricultural system to be a main driver of climate change change (Steffen et al. 2015; Foley et al. 2011). In the Netherlands, emissions from agricultural sources, excluding emissions from land use change, contribute about fifteen percent of the national emissions on a yearly basis (Emissieregistratie, 2020). In addition, the intensification of the agricultural system has caused significant alteration of the global nitrogen cycle, through the relocation of crops and manure and the use of artificial fertiliser (Galloway et al. 2008; Erisman et al. 2008). In the

Netherlands, emissions of ammonia, nitrates and nitrogen dioxide affect the natural environment, through biodiversity loss, ocean acidification and climate change, to name the most important planetary boundaries (Galloway et al. 2008; Steffen et al. 2015). Agriculture also drives biodiversity loss through the simplification of the landscape and the agricultural system as a whole, growing a small set of crops (Tscharntke et al. 2005; Steffen et al. 2015). The loss of biodiversity increases the chance of losing functional diversity. This erodes the capacity of the system to provide essential ecosystem services. Thus, by shaping the natural environment, agricultural practices can undermine themselves. Finally, with the majority of the agricultural flows passing through the soil at some point, soil health is vital to agriculture (Erisman et al. 2016; Bünemann et al. 2018). The effect of agricultural practices on the soil is therefore the last major category of interest in this study. Greenhouse gas & energy use

Agricultural practices cause the emission of several greenhouse gasses, both directly and indirectly. The most important greenhouse gasses produced in agricultural systems in order of importance are CO2, CH4 and N2O. Here, the effect of agricultural land management practices on the emission of

these greenhouse gasses is reviewed. Later, the impact of these practices on the energy consumption in Dutch soil bound agricultural systems is discussed. Quantifying the environmental impact of farming systems requires a clear idea of how these systems are defined. The definition of the system determines the size of greenhouse gas and energy flows to a great extent. On a local scale, the exact source and sink of certain emissions can be quantified very well. A more broad system definition however, can incorporate non-local emissions from fertiliser and pesticide production. Most studies that focus on local management approaches consider the local system boundary for their farming system. These will be discussed first. Afterwards, studies that address more holistic systems are reviewed.

Nitrous oxide (N2O) emissions are fourth in line after CO2, CH4 and CFC-12 in terms of

their contribution to radiative forcing (IPCC, 2001a; Flechard e.a. 2007). Approximately 75% of the anthropogenic N2O emissions derive from agricultural sources (46% agricultural soils & 29%

cattle and feedlots) (van Groenigen e.a. 2004). Despite the big impact, there is a large uncertainty regarding the actual emissions of N2O from agricultural systems under different climatic conditions

or different management regimes in the Netherlands. The impact of different fertiliser schemes (van Groenigen et al. 2004), groundwater levels (Velthof, Brader, and Oenema 1996; van Beek et al. 2004; 2010; Best and Jacobs 1997), organic matter (Best and Jacobs 1997) and traffic pressure (Vermeulen and Mosquera 2009) on N2O emissions has been studied. Overall, authors attribute a

large part of the variation in N2O emissions to variation in climatic parameters. However,

management practices seem to be relevant as well.

In 1996, Velthof, Brader and Oenema established that artificially fertilised grasslands have significantly higher emissions of N2O than unfertilised grasslands (Table 1). This idea was

supported by a European study that stated that ‘intensive’ management, as compared to extensive management, increased emissions by a factor 4 (Flechard et al. 2007). In line with Velthof et al. (1996) van Beek (2009) stated that N2O emissions increase by a factor 3,5-6 with more intensive

grazing and increasing wetness of the field (Table 1). Another important management factor is related to the total N-input, mostly affected by fertilisation. Fertilisation with slurry had a negative effect on N2O emissions as compared to solid manure (van Groenigen et al. 2004). Moreover,

Velthof et al. (2011) found that, for slurry application, there was a difference in direct and indirect N2O emissions depending on whether slurry was injected (deep or shallow) or surface applied. As

injection leads to increased NH3 emissions, less N is available for denitrification (Velthof and

Mosquera 2011). Van Bruggen et al. (2017) found the same relationship between increased N application and increased N2O emissions and added that organic and conventional grasslands

around Wageningen (The Netherlands) did not differ significantly in terms of their emissions. However, long-term grasslands displayed a significantly lower emission of N2O. This essentially

Study Management practice Effect Velthof, Brader &

Oenema (1997)

Fertilised grassland (Ca ammonium nitrate) Unfertilised grassland

7,3-42 kg (N. ha-1_{. yr}-1₎

0,5-12,9 kg

Flechard et al.

(2007) Intensive instead of extensive management 4x increase Van Beek (2009) Grazing instead of non-grazing 3.5-6x increase

links back to 2004, when Vellinga et al. established that grassland ploughing formed a major source of N2O that was unrecorded due to the current way of N2O accounting used by the IPCC (Vellinga,

van den Pol-van Dasselaar, and Kuikman 2004). That there is an effect of management was corroborated by Vermeulen & Mosquera in 2009, when they tested the effect of seasonally controlled traffic on greenhouse gas emissions (Vermeulen and Mosquera 2009). Less trafficking equated with lower emissions from both N2O and CH4.

In addition to the impact of greenhouse gasses from the nitrogen cycle, the carbon cycle adds to the process of climate change through CO2 and CH4 emissions. Several management

approaches affect the emission of CO2 and CH4. Best et al. (1997) found that raising the water table

in peat grassland around Wageningen from -55 to -40 cm below the sward level significantly decreases CO2 emissions, while simultaneously raising CH4 emissions. Both of these effects almost

exclusively take place in the ditch (banks) that intersect(s) the fields. Additionally, disturbance of grasslands through ploughing or trafficking reinforces emissions of CO2 and CH4 respectively, as

has been shown by Vellinga et al. (2004) and Vermeulen et al. (2008). In fact, decreasing trafficking increased CH4 sinks in organic vegetable cropping. It was already clear from studies into N2O that

emissions have a distinct spatial character. Schrier et al. (2010) found that the type of landscape element plays an important role in whether CH4 emissions take place, rather than the intensity of

management practices.

Bos et al. (2014) took a more integrated, model-based approach towards the effect of management on energy use and carbon based greenhouse gas emissions from Dutch farming systems. In figure 4 and 5 their results are displayed. The authors reported a much higher energy use and a smaller difference between organic and conventional dairy, arable and horticultural systems than fellow authors, such as Thomassen et al. (2008). Overall, their conclusions line up with a field study around Wageningen by van Bruggen et al. (2017) as they reported higher emissions of CO2 from organic grassland. Concurrently, the authors did not report a significant

difference in CH4 emissions for their different grassland treatments. They did find a remarkable

effect of no-till management on the accumulation of C and N in the soil.

Overall, the effect of some management practices is minimal. As reported here, organic, as opposed to conventional management often does not have a significant effect on greenhouse gas emissions and energy use in arable farming systems, grassland systems and horticulture. In dairy farming, energy use is reported to be lower, just as greenhouse gas emissions, quantified as CO2

equivalents. However, sometimes specific measures seem to impact emissions and energy use, e.g. the use of controlled traffic systems, no-till practices and raised water tables. These tend to result in positive or mixed effects. For some processes, such as N2O emissions, it is clear that knowledge

gaps need to be addressed, before solid decisions can be made on what constitutes effective management.

Nitrogen

The productivity of most natural ecosystems is limited by nitrogen (Vitousek et al. 1997). In the last century, the flux of biologically available nitrogen has increased radically, due to the invention of the Haber-Bosch process. This provided a means to produce NH3 from natural gas at high

temperatures (Erisman et al. 2008). However, large fluxes into agricultural systems are not always used optimally, and can exit these systems through the processes of denitrification, ammonia volatilisation and nitrate leaching. Whereas the first is discussed under the chapter ‘greenhouse gas & energy’, the latter two are discussed here. Ammonia and nitrates have the potential to eutrophicate, acidify and fertilise natural ecosystems. This drives biodiversity loss, climate change, ocean acidification and it affects the carbon cycle (Vitousek et al. 1997; Steffen et al. 2015).

In the Netherlands, these processes are noticeable, as the nitrogen fluxes into and out of agricultural systems increased sharply after the 1940s (Brusse 2009). In the 1980s, acid deposition, partly caused by the formation of nitric acid from nitrogen dioxide and ammonia sparked attention. In the Netherlands, emissions of ammonia and nitrogen dioxide found their origin in industry, but also in agriculture. Neeteson (2000) describes four measures that were taken to limit the nitrogen fluxes from agriculture in the Netherlands. Firstly, a ban on the spreading of manure in winter came into effect. Then, farmers were obliged to start covering manure storage facilities and to implement a low-emission slurry application technique: injection. Lastly, the mineral declaration system (MINAS) became available, that could account for yearly N and P surpluses (RIVM, 2002). Based upon this data, farmers were charged with levies if the nitrogen flux leaving the farm was too high. The use of these levies was under discussion, but more importantly, MINAS did not take into account the biological nitrogen fixation, nor the wet and dry deposition of nitrogen onto the farm. Hence, the total fluxes going in and out of the system must be higher than accounted for.

To mitigate damaging effects for the environment, the Netherlands set goals for the decrease of ammonia volatilisation. In 2005, emissions had to be seventy percent lower compared to the level of 1980. Additionally, the European Bird, Habitat and Water directive provided a legislative basis for the management of nitrogen fluxes from agricultural systems. According to the EU-habitat directive, the Netherlands has the obligation to preserve and maintain specific natural habitat and to mitigate damaging anthropogenic effects, such as fertilisation and acidification by nitrogen deposition. All the aforementioned measures helped to decrease the flow of nitrogen from agricultural systems by sixty percent in 2005 (Ministerie van Landbouw Natuurbeheer en Voedselkwaliteit 2019a). Moreover, artificial fertiliser use had been cut, as MINAS helped farmers to realise that they used more artificial fertiliser than necessary (Erisman, n.d.).

Nitrogen management in the Netherlands has been characterised by a very reactive approach. The measures implemented on national level help to partly reach the desired effect, however, the selected measures were chosen for being the most cost effective and requiring the least amount of change to the system (Kros e.a. 2011; Kros e.a. 2013; Ondersteijn e.a. 2003). This causes several problems: first of all, the burden of pollution got ‘swapped’. For example, slurry injection decreases ammonia volatilisation, but in almost all cases, leaching (and denitrification) increases, thus the measure is essentially tailored at meeting a particular environmental regulation and not at improving nitrogen management (Sonneveld et al. 2008; Kros et al. 2011; Velthof and Mosquera 2011). According to the nitrates directive, the nitrogen content in water has to stay below 50 mg NO3- per litre (European Parliament, 2016). It is reportedly hard to monitor this, but due to the increased leaching caused by this manure management, the trend might be negative. Second, farmers are obliged to implement certain measures, such as injection of manure. Some farmers do not want to inject slurry, due to the perceived loss of soil structure. Sonneveld et al. (2008) investigated the ammonia volatilisation on a farm that used low protein feed. They found that this measure, in combination with water application during manuring to facilitate infiltration, was capable of meeting legal limits just as well. This relates to the third problem, none of the currently implemented measures is actually capable of decreasing emissions beyond the threshold values set for Natura 2000 areas (Kros et al. 2011). It seems that a more thorough set of measures is needed

to decrease the flux of nitrogen leaving Dutch farming systems. Low protein feeding and organic farming have been proposed as being cost efficient and effective measures for reducing this flux. However, several authors report that they consider these to be unfeasible to implement, due to the level of change required in Dutch farming systems (Kros e.a. 2011; Kros e.a. 2013; Kuipers en Mandersloot 1999; Ondersteijn e.a. 2003; Sonneveld e.a. 2008).

Table 2: Effect of nitrogen management in dairy cattle farming.

Management Study Trade-off

Manure pit covering Neeteson (2000)

Ban on winter spreading

Neeteson (2000) More strict growing season, less diversity.

Air scrubbers Kros et al. (2011) estimate a

reduction of 70% in NH3 emission. Increased N2O emissions, leaching and drainage.

Reduced grazing Kros et al. (2011) Increased housing emissions, animal welfare.

Low-emission stables Kros et al. (2011) estimate a

reduction of 40% in NH3 emission. Cost-inefficient.

Reduced protein feeding

Kros et al. (2011), Sonneveld et al. (2008)

Knowledge intensive.

Low emission application of manure (injection)

Kros et al. (2011) estimate a 65% reduction in NH3 emission from soil. Neeteson (2000)

Increased denitrification and leaching.

Organic cattle farms Kros et al. (2011) Authors state that this is unlikely to be implemented.

Wet application Sonneveld et al. (2008) Equally effective as injection. Unsurprisingly, pragmatic choices are made concerning nitrogen management. However, it is clear that the objectives of the policy were not met. Moreover, some measures have led to pollution swapping. Emissions in the stable have increased, due to shorter grazing periods, not to mention the effect of specific measures on soil structure, groundwater quality, soil life and animal welfare (Kros et al. 2011). Hence, the focus of researchers shifted to improving the Nitrogen Use Efficiency (NUE) on Dutch farms. The NUE decreased with increasing inputs from the 1950s (Erisman et al. 2008). In an attempt to increase NUE, several researchers investigated which scenarios or management approach proved to be the best strategy to improve NUE (Groot, Rossing, en Lantinga 2006; De Visser e.a. 2001). These papers observe the same thing as Oenema et al. (2012): an increased production per hectare, partly through decreasing grazing time, is the most efficient way to reach a higher NUE. As a consequence, in a 2012 comparison of NUE in six European countries, Dalgaard et al. find that the Dutch ‘Friese Wold’ area has the highest NUE of all participating countries. Despite that, the authors do not that that the flux of N leaving the farming system is still among the largest of all participating countries.

With environmental standards not being met, the Dutch ‘Programmatic Approach Nitrogen’ (PAS) was launched in 2015. This legislative system allowed permits to be handed out to projects that increased the burden of N on Dutch Natura 2000 areas by relying on a future decrease of N emissions within the area. Several authors expressed that the objective basis for this approach was narrow (Schoukens 2017; Ministerie van Landbouw Natuurbeheer en Voedselkwaliteit 2019). In 2019, this approach was deemed contrary to the European habitat directive by the highest Dutch court for administrative law, sparking an immediate nitrogen crisis, as no new permits could be released for projects that impacted Natura 2000 areas through nitrogen emissions.

Overall, several approaches have been proposed and implemented to decrease the flux of N from Dutch farming systems. While some approaches have been effective in decreasing

volatilisation, pollution swapping might occur, leading to more leaching. Even though leaching is highly point specific, there have been studies that indicated that slurry injection increased leaching on a field scale (Oenema et al. 2010; Kros et al. 2011). A proposed solution has been to increase the levels of easily digestible organic matter, so that bacteria will use excess nitrogen in the soil. Radersma and Smit (2011) found that the addition of paper pulp in combination with plant remains of fodder beet significantly decreased leaching, while also decreasing deni

trification. This adds to the idea that the compartmentalised and reactive approach to nitrogen management in the Netherlands does not help the problems caused by the anthropogenic flux of nitrogen.

Biodiversity & Ecosystem Services

With agriculture covering more than two thirds of the Netherlands, it provides most of the habitat in the country. However, with the onset of large scale monocultures and a simplification of the landscape, habitats have declined in quality and quantity. The associated biodiversity decline has been going on for decades in the Netherlands (Hallmann et al. 2017). Here, I review papers that study the effect of different agricultural land management practices on biodiversity. Two main themes are addressed. Firstly, the effectiveness of European agri-environmental schemes (AES) is discussed and secondly, the effect of different management approaches on the provision of ecosystem services is reviewed. The main ecosystem services of interest are pollination and pest control, as these rely on the availability of agrobiodiversity (Erisman et al. 2016).

Traditionally, biodiversity preservation in farming systems has focused on meadow bird species, which are key indicator species for the integrity of (semi-natural) ecosystems on Dutch farms. As part of a European AES, farmers in the Netherlands receive a reward for facilitating meadow bird breeding by raising groundwater tables, postponing mowing in grassland until June or limiting fertiliser application (Wiggers et al. 2016). These AES have not lead to sustainable meadow bird populations. For particular species, significant negative effects have been reported (Breeuwer et al. 2009; Kleijn et al. 2004). Breeuwer et al. (2009) and Kleijn et al. (2004) explain their findings by stating that the AES do not guarantee ‘the minimal ecological requirements’ for the species.

Taking on a more holistic approach, Kragten et al. (2008) compared conventional and organic farming systems and found that skylarks (Alauda arvensis) have seven times more nests on organic farms (Kragten, Trimbos, & de Snoo 2008). However, the effect was ambiguous. In a bigger study, four species seemed to prefer organic farms, while three species were found more on conventionally managed farms (Kragten and de Snoo 2008). Several authors do report that the AES approach is too narrow. Onrust et al. (2019) showed that earthworm abundance and availability for meadow bird chicks is influenced by the level of the groundwater table. Even plant diversity, stimulated by AES focused on increasing botanical diversity, is connected to meadow bird breeding success. Botanical field margins along the fields seem to increase the abundance of flying insects, which are part of the meadow bird food chain as well (Wiggers et al. 2016).

Next to meadow bird preservation, the creation of botanical diversity in semi-natural habitat on Dutch farmland is a popular AES as well. On farms that participate in this AES, the amount of semi-natural habitat (5.6%) is twice as high as on non-participant farms (2.4%), regardless of whether these are managed in an organic, integrated or conventional manner (Manhoudt and de Snoo 2003). There is, however, a qualitative difference between semi-natural habitat on farms with organic and conventional management. The former are reported to have twice the species richness on ditch banks as compared to the conventional farms (Manhoudt, Visser, and de Snoo 2007). Here again, a narrow approach on either the percentual amount of farmland dedicated to semi-natural habitat, or the mere absence of some pesticides on organic farms, decreases the effectiveness of the AES. As an example, Manhoudt et al. (2007) found that despite the increased species richness on organic farms, there were not significantly more species related to a low nutrient environment, based upon the Ellenberg nitrogen values of the vegetation.

Hence, the actual diversity remains limited, as the species composition is dominated by high nitrogen input. Only on ‘ecologically’ managed experimental farms, did ditch banks show decreased amounts of nitrogen through the vegetation composition. This difference was explained by the fact that these ditch banks were protected with ‘field margins’ where no fertilisation occurred, thus increasing the physical distance from the field, where nitrogen and pesticide inputs are high (Manhoudt, Visser, and de Snoo 2007).

While the meadow bird and plant diversity AES are tied together as part of the same ecosystem, they are not necessarily integrated in practice. Moreover, they do not concern the role of insects either, even though these are part of the functional agrobiodiversity (Erisman et al., 2016). This is surprising with regards to the importance of insects for meadow birds, but also because of the provision of ecosystem services such as pollination and pest control on farms. The schemes affect insect communities nonetheless and not always in a positive way, as shown by Tanis et al. (2020). They find that grasslands managed for meadow birds, as opposed to hay meadows and herb-rich grasslands, negatively impact pollinator abundance. In another study, researchers reported the clear absence of a positive effect of AES in the Netherlands on pollinator abundance and diversity, as opposed to Switzerland, where a significant positive effect was found (Kohler et al. 2007). Batary et al. (2010) showed that ‘the richness of insect-pollinated plants was a good predictor of bee species richness’. Kohler et al. (2007) therefore hypothesize that the large scale and monotonous landscape in Dutch agriculture, negatively affect pollinator abundance. This has been confirmed in both national and international studies (Bukovinszky et al. 2017; Le Féon et al. 2010).

Thus, improving landscape structure might be the management approach needed to improve conditions for pollinators. For insects contributing to natural pest control on farms, it has been shown that a loss of landscape complexity reduces the abundance of insect predators, regardless of whether a crop is sprayed or not. The authors compared organic and conventional farms and found that the ‘agroecological infrastructure’ seems to drive insect abundance and diversity (Booij and Noorlander 1992).This has been studied in more detail by Bianchi et al. (2008). In an experiment set up to establish predation in an agricultural field, the authors tried to correlate the spatial environment to the rate of predation. By measuring predation and quantifying the surrounding landscape in selected spatial circles, the authors manage to connect the abundance of specific landscape elements on particular spatial scales to predation rates. In the study, it was found that amongst other factors, the total surface area of forest in circles with a radius of 1, 2 and 10 kilometres from the experiment, correlated with predation rates on brussels sprouts (Figure 6). This effectively means that predators are affected by landscape changes on these scales as well. Using a combination of ‘robust’ and ‘fine’ elements in a landscape, natural enemy populations can thrive in agricultural landscapes (Steingröver, Geertsema, and van Wingerden 2010). In experimental settings, target larvae were reduced by up to 94% on single plants in the vicinity of robust landscape elements (Bianchi, Goedhart, and Baveco 2008).

Ecology comprehends the study of the living and its environment on every scale, from microscopic to continental. In the Netherlands, much of the natural land has been replaced by intensive agriculture. The decrease of landscape complexity and the creation of similar conditions everywhere is practical for agricultural production, but it undermines biodiversity, that relies on varied landscapes and gradients, for example in nutrient and water availability. Besides the intrinsic value of biodiversity, there is a loss of functional agrobiodiversity, providing ecosystem services, that hampers agricultural productivity in the long term. Management practices that have been oriented at restoring or safeguarding the biodiversity of the Dutch landscape, have so far not been very effective, as meadow bird populations continue to decline, and plant species richness in field margins is dominated

Figure 6: An example of how a model can quantify landscape characteristics on different spatial scales from a point of interest (Bianchi, 2020)

by plants thriving on nutrient rich conditions (Breeuwer et al. 2009; Manhoudt, Visser, and de Snoo 2007).

The focus on red list species of the AES in the Netherlands shifts the focus from the underlying loss of insects (>75% in summer over 27 years) and ultimately makes the attempts to restore key species futile. Regardless of the overall management practices on the farm, whether it is conventional, integrated or organic, ecosystem services are pressured, especially since ‘integrated farming’ has shifted its practices more and more towards conventional farming (Tamis & van den Brink 1999). The effect of organic management, although significantly different, is shown to be mixed, absent or marginal, despite the fact that, for example, no pesticides are used (Manhoudt, Visser, and de Snoo 2007; Manhoudt and de Snoo 2003; Kragten and de Snoo 2008; Kragten, Trimbos, and de Snoo 2008; Yasrebi-de Kom, Biesmeijer, and Aguirre‐Gutiérrez 2019).

With agriculture taking up more than 66% of the land surface in the Netherlands, the management practiced here affects the entire Dutch biosphere. Often, environmental impacts transgress the spatial boundary of the farm. Examples of these include pollution of groundwater, ammonia volatilisation, the release of chemicals and climate change. All these factors influence nature on the farm and beyond the boundary. It is clear that the current management practices, oriented at mitigating the effects of these negative influences on key species, is ineffective. Often, they represent a compartmentalised approach, that is not capable of dealing with all the drivers of change.

Soil

Agricultural crops depend on soils for water, nutrients, anchorage and disease suppression capacity. The capacity of a soil to provide these functions, is defined as soil health, as this depends on the capacity of the soil to function as a vital living ecosystem (Doran and Zeiss 2000).These soil based ecosystem services occur as ‘emergent system properties such as the self-organization of soils, e.g. feedbacks between soil organisms and soil structure (…), and the adaptability to changing conditions’ (Bünemann et al., 2018). Studying the health of soil systems is complex. Here, studies have been divided into several groups, namely those that focus on earthworms, other soil life, soil organic matter (SOM) and ecosystem services. These are the main themes that are addressed in the literature selected for this review.

Earthworms

The abundance and diversity of earthworms is generally taken as an indicator of soil biological activity. Earthworms play an important role in nutrient cycling, through the displacement and decomposition of organic material, but also in the carbon cycle, by the production of stable soil organic matter and in the capacity of a soil to regulate water, as they influence the infiltration rate of the soil. Earthworms can be classified according to their diet, but it is more common to classify them according to their spatial position in a soil into epigeic, endogeic and anecic species (Onrust and Piersma 2019). These groups represent a share of the functional biodiversity in the soil. An increase of all species of earthworms leads to increased infiltration capacity, nutrient cycling and might positively affect carbon storage. Furthermore, earthworms are an important part of the food web, serving as food for indicator species such as meadow birds.

That agricultural land management impacts earthworm abundance is clear, as shown by Jongmans, Pulleman and Marinissen (2001) and Pulleman et al. (2005). They found significantly higher numbers of worms under permanent pasture, as compared to conventional and organic arable land. The authors make the connection to the role of earthworms in the farming system, when they show that increased numbers of earthworms lead to better incorporation of organic material in the soil. Another factor that influences earthworm abundance seems to be the type of fertiliser used. In a study on Dutch dairy farms, earthworm abundance and ‘ecotypes’ were compared in the field and in a growth experiment with farmyard manure, slurry or hay as food source. The authors found that the manure applied in the field, either solid farmyard manure, slurry

or a mix of both, significantly impacted the abundance of detritivores. On the contrary, geophages were not affected by manure management (Onrust and Piersma 2019). In another study by Onrust (2019), an additional correlation was found between earthworm abundance and groundwater level. Here, it was hypothesised that slurry injection might lead to desiccation in the topsoil, negatively affecting detritivores, that are important for nutrient cycling.

In order to better understand which factors affect earthworm populations, two studies focused on the differences between field margins and arable fields. These field margins are often not sprayed, tilled or fertilised, hence larger populations of earthworms are expected. Even though these studies found higher numbers, Crittenden et al. (2015) found no gradient from the margin into the field. It might be that the system needs more time to adapt. However, in another study, it was found that this ‘landscape effect’ of spill over from source to sink areas, might not apply as the effect of management overrules this (Frazão et al. 2017). In other words, arable land is such a large sink, that no significant gradient can be established. Yet, Crittenden et al. (2015) reported that in another study, after thirteen years and no tillage, a clear gradient appeared from field margins into arable land (Figure 7). It is important to mention that this does not necessarily concern all species of earthworms. It seems that some species are more heavily impacted by management such as plowing, compaction due to machinery or manuring, as was partly shown by Onrust and Piersma.

Sink: Managed field (tilled, sprayed, fertilised) Source: Field margin

Crittenden et al. (2015) (Mouldboard ploughing)

Nuutinen et al. (2011) (no tillage, inoculated worms spreading from field margins after 13 years)

Frazão et al. (shape of curve in arable field determined by management)

Figure 7: The figure provides an overview of relationships that studies found in earthworm abundance

and density between fields and field margins. The relationship between earthworm communities in the sink (field) and the source (field margin) is visualised by the line. Whether gradients can establish depends upon management factors. For each study, the most important factors for the experimental setup are mentioned in brackets. The study of Frazão compared the effect of management to the effect of landscape and soil properties, they found that management is the (negative) driver of earthworm species richness and community composition.

Soil life

In addition to earthworms, other soil fauna, such as mites and nematodes, as well as bacteria and fungi, all impact the capacity of a soil to provide ecosystem services. In general, research finds that conventional systems, as opposed to either integrated or organic management, support less complex soil food webs (de Groot e.a. 2016; Didden e.a. 1994). Changes in management might lead to shifts in soil communities. Nematode communities reportedly shifted composition when management changed from ‘conventional/integrated’ to ‘organic’ (Quist et al. 2016). Differences were also found between the communities supported by conventional and integrated systems (Mulder et al. 2003). Within these management systems, researchers have attempted to pinpoint what exactly drives this diversity. Several studies have opted for effects from factors such as soil organic matter, soil moisture and crop type. De Vries et al. investigated this for fungi under several types of nitrogen management. They found that fungal biomass increased significantly with

increased age of a grassland. However, they could not confirm their hypothesis that fungi dominated grasslands were related to more sustainable management, even though there are studies that indicate this (de Vries et al. 2006; 2007). Lastly, also soil mite communities seemed to respond to increasing age. Communities of mites diversify when arable land shifts to grassland, and continues to increase for 23 years, according to a chrono sequence used by the authors. Moreover, groups that were important for nutrient cycling and soil formation increased rapidly with this conversion, indicating the potential supply of ecosystem services with increasingly diverse mite communities (de Groot e.a. 2016). Old grasslands are uncommon in Dutch agriculture, because of the perceived decrease in productivity after a certain time and the relatively large share of clover in these fields. Hence, it is advised to replace permanent grassland approximately every ten years. Contrary to that advice, Iepema et al. showed that grasslands of respectively 5-15 years old and >20 years old, did not differ in terms of productivity. Moreover, soil chemical quality was significantly better in ‘old’ pastures (Iepema et al. 2020). Likewise, a study on 36 years of permanent grassland, arable or ley arable cropping show that permanent grassland had better established soil functions and more diverse soil life (van Eekeren et al. 2008).

Ecosystem Services and SOM

The soil biota facilitate several provisioning ecosystem services. Organic farms reportedly support an overall larger diversity of soil biota, providing a generally higher level of disease suppression. In a study by Postma (2008) this coincided occasionally with the occurrence of antagonistic bacteria, that combat pathogens directly. The composition of the soil biota was mainly determined by biotic soil parameters. Next to the disease suppression capacity of a soil, there are other ecosystem services. Deru et al. (2018) list ‘maintenance of biodiversity’, ‘climate regulation’, ‘water regulation’ and ‘soil fertility’ and link these to management (dairy or semi-natural grassland) and soil ecology. Overall, they reserve a primary role for soil biological activity. Several other studies find either no relation with soil biological activity, which they attribute to a lack of diversity between their samples, or they find a primary role for soil organic matter (van Eekeren et al. 2010; Rutgers et al. 2012).

Soil organic matter is often used as an indicator for soil quality or soil health. As was shown by Rutgers et al. (2012), it is a suitable indicator for the presence of several ecosystem services. A series of papers was published on a comparative investigation into SOM levels on conventional arable (CA), organic arable (OA) and permanent pasture (PP) (Pulleman e.a. 2005; Nierop, Pulleman, en Marinissen 2001; Pulleman e.a. 2003). In their first study from 2000, the authors found no significant differences in SOM between OA and CA, except for the top 4 cm (Pulleman e.a. 2000). Overall, PP plots had SOM in a less humified state. The authors explained this partly through the abundance of earthworms in grass plots, even though this was not actively monitored. In 2003, the authors did investigate earthworm activity and aggregate stability. Now, they found a higher aggregate stability and earthworm activity in organic arable plots. However, these soils also seemed more vulnerable to compaction. In 2005 they established that organic matter under organic arable land had a more stable nature. Additionally, they found that organic arable land displays a higher earthworm activity, and that the SOM clearly originated from farmyard manure. As the effects were small, the authors hypothesize that tillage, which is done on both farms, might reduce the positive impact of organic farming systems.

Earthworms, soil life, ecosystem services and soil organic matter all mutually influence each other. From these results, it is clear that the capacity of the soil to provide the required services, depends partly upon soil biological activity. However, earthworms and soil biota also create feedback mechanisms towards the soil structure and the amount and type of soil organic matter. These factors likewise influence the provisioning of ecosystem services. Essentially, these ecosystem services that agriculture benefits from belong to the ‘emergent properties’ of healthy soil ecosystems. This makes it especially complex to study the effect of management on the eventual

goal: soil based ecosystem services. That becomes clear from the studies discussed here, which for the majority use multivariate analyses, due to the large amount of drivers that affect these systems. Discussion

In this systematic review, studies were collected that discussed the effect of management practices on the environmental impact of Dutch soil bound agriculture. The body of literature was used to identify the main themes that have been addressed by research in the past three decades. These themes, Greenhouse gasses & Energy, Nitrogen, Biodiversity & Ecosystem Services and Soil, covered a large range of the environmental impacts of Dutch agriculture. Changing this environmental impact poses a dual challenge for Dutch agricultural systems. For decades, the system has focused on ‘avoiding exposure’. By using external resources, to create constancy in the environment, production conditions were kept optimal (de Goede, Gremmen, and Blom-Zandstra 2013). Inherently, this ‘control model’ has a high degree of control on the field level, but is very vulnerable to change on higher organisation levels (van Apeldoorn et al. 2011). Alternatively, agricultural output can be kept in place through the capacity of the system to ‘recover from disturbance’ (de Goede, Gremmen, and Blom-Zandstra 2013). This ‘adaptive system’ can maintain its organisational structure after a disturbance. Hence, following Holling (1973) and his theory on ecological systems, it has a higher resilience. Without external resources to create the required constancy, farming systems have to rely upon their capacity to continue production under more varying circumstances, to be more resilient.

The research studied here, has shown a tendency to focus on the decrease of environmental impacts from soil bound agricultural systems in the Netherlands. Especially in the case of nitrogen in the Netherlands, the approach is ‘problem oriented’. The demand for improvement is provided by the EU-Bird and Habitat directives. Management practices have been used to meet the demands from these directives. Farmers use slit injection, produce more liquid manure and keep their dairy cattle in the stable. This decreases volatilisation, but it causes the burden of pollution to be swapped. Measures that are intended to decrease the total flow of nitrogen through the system, namely low-protein feeding and organic agriculture, are considered impractical or impossible. The fact that Dutch dairy farms have the highest nitrogen use efficiency of all European farms should serve as an indication that management should not be focused on efficiency, but rather on the total size of the flow, which is arguably transgressing the planetary boundary.

In terms of greenhouse gasses and energy use, studies mainly focus on decreasing emissions. Agriculture contributes to climate change in the Netherlands by emitting approximately 14% of all the greenhouse gasses (RIVM, 2020). The emission of N2O has shown a marked decline in thirty years, while the emissions of CO2 and especially CH4, have decreased only marginally. Research into management practices that can decrease these emissions take different approaches. Many of the field based emissions have to be studied in the perspective of climate and landscape variability. Compared to this variability, management effects are rather small, according to several authors (van Beek et al. 2010; van Groenigen et al. 2004; Best and Jacobs 1997; Velthof, Brader, and Oenema 1996; Vermeulen and Mosquera 2009). Interestingly, nitrogen management is strongly connected to the performance of Dutch agricultural systems in terms of emissions. Nevertheless, a holistically different management approach, such as organic instead of conventional management, does lead to a marked decrease or increase in emissions, depending on the sector. Oftentimes, the system boundary determines the outcome of the full comparison, underlining the importance of a system based approach for this field of study.

Concerning biodiversity and ecosystem services, a similar approach can be seen as in the nitrogen problematic. The main planetary boundary of concern in this category is that of biosphere integrity. Due to large pressures on habitat in the Netherlands, partly from agricultural sources, insects, birds and plants are under pressure (Hallmann et al. 2017; Breeuwer et al. 2009; Batáry et al. 2010). The ‘problem oriented’ approach towards this concern, proves to be ineffective, mainly

due to the compartmentalised approach (Deru et al. 2018; Tanis et al. 2020). It is clear that an array of measures needs to be taken to safeguard biosphere integrity and the ecosystem services that flow from these systems. An integrated approach should incorporate at least water dynamics, fertilisation, nitrogen pollution, pesticides and landscape diversity. Currently, all management seems to be largely ineffective, due to this lacking system approach.

Lastly, the category soil displays the role of soil agrobiodiversity in the provisioning of ecosystem services. The studies in this category approach the problem posed by the research question differently. Instead of starting from a narrow problem definition, the challenge for most papers is to find management practices that increase soil health: the capacity of a soil to function as a vital living ecosystem (Doran and Zeiss 2000). In the majority of the studies in this category, multivariate analyses are used to consider the drivers of change in soil health. Some authors manage to establish relationships between soil parameters and ecosystem services, such as is the case for Rutgers et al. (2012). They connect soil organic matter levels to the presence of several ecosystem services. Many of these studies however, struggle to make clear relationships, which reflects the complexity of the problem. Clearly, more knowledge is needed into the mechanistic relationships between the soil biological activity and the soil health.

Studying environmental impacts of Dutch agriculture per category has its limitations. The environmental impacts, translated into planetary boundaries, are not completely separated from each other. Several feedback mechanisms exist between these boundaries, hence also between the categories studied here. Nevertheless, the division is a pragmatic choice, as the goal of this study is to identify which management practices have been studied in the Netherlands.

From the onset of this review, the main question has been considering the need for a form of agriculture that decreases the pressure on the planetary boundaries, while safeguarding the production of food for a growing population in the coming decades. The Netherlands serve as the prime case study for answering this question, as for decades, exactly this challenge has been a key focus point. However, as has been made clear, this challenge is not straightforward. As the current production model has shaped everything, from the landscape to the crop types and breeds, the characteristically high productivity only occurs under stable conditions, a control model. The resilience of the system is low. Therefore, management practices that decrease the environmental impact, should also promote the resilience of the system. As large external inputs of nutrients, energy and chemicals affect several planetary boundaries, these have been decreased and continue to be decreased in Dutch agriculture. A share of the studies in this review attempts to keep the decrease of external inputs to a minimum, by investigating the minimum efforts required to meet the necessary regulations. On the contrary, some studies aim at increasing the internal resilience by strengthening (soil based) ecosystem services.

From this review, it becomes clear that the compartmentalised approach taken by several studies is unsuccessful. The loss of biodiversity and the accompanied ecosystem services has not been altered in the last thirty years, partly because the Dutch have not succeeded in altering the influence of biochemical flows on the environment. Additionally, goals set for the decrease of pollution from nitrogen have not been made, even though progress has been large in the first decade from 1990.

Conclusion

Due to the scale and the intensity of agricultural practices in the Netherlands, agriculture is a main driver in the Dutch contribution to the approach of the planetary boundaries (Steffen et al. 2015). This is especially clear in the case of the Dutch nitrogen crisis, which displays the intimate connection between the nitrogen flow, biosphere integrity and agriculture in the country. In response to the crisis, the ‘advisory committee for the nitrogen problem’, presented its first temporary advice titled ‘Not everything is possible’ to the Dutch ministry for Agriculture, Nature and Food Quality in September 2019 (Ministerie van Landbouw Natuurbeheer en Voedselkwaliteit

2019a). Essentially, The Netherlands have reached the planetary boundary for biochemical flows, specifically nitrogen. In light of the worldwide increasing demand for agricultural produce, the management of the agricultural areal determines whether on a world scale, we run up to this and other planetary boundaries as well in the coming decades.

By reviewing a set of studies covering the environmental impacts of management practices in the last three decades, an attempt was made to answer the main question: How do agricultural land management practices influence the environmental impact of Dutch soil-bound agriculture? This question attempts to cover three decades of research into management opportunities that can lower the burden of agricultural production on the natural environment in the Netherlands. Taking into account the need for sufficient food production in the coming decades, the challenge posed to Dutch soil bound agriculture was reasoned to be two-fold. On the one hand, environmental impacts of agricultural practices have to decrease. On the other hand, agricultural systems have to be able to continue their main function of food production in a more stochastic environment, relying more upon internal resilience.

Four main categories of environmental impacts were studied, namely Greenhouse gasses & energy, Nitrogen, Biodiversity & Ecosystem Services and Soil. Over all these categories, it was found that some studies tend to focused on one side of the challenge, namely decreasing the environmental impact of soil bound agricultural systems in the Netherlands. Dairy, arable and horticultural farms use management practices to follow legislation and limit emissions, regardless of the side-effects. In some cases, pollution avoided somewhere in the system gets swapped, as is the case with ammonia volatilisation. In other cases, management practices are not ineffective, such as is the case in meadow bird protection using AES.

Another group of studies focused on the connection between certain management practices and their effect on the functional agro-biodiversity, ecosystem services and the resilience of the farming system. Especially in the category soil, this approach was omnipresent. Here, many studies take into account the complexity of the matter. However, these studies often attempt to find the few driving forces that can explain the variability in the system, in order to find the leverage points of the system. This approach is complex and not always successful.

It is clear that no silver bullets exist in management practices for Dutch agriculture, due to the complexity of the system and the difficulty of studying this complexity. What has become clear as well, is that the internal resilience needed for the sustainable production of food drives on diversity. Diverse farming systems can have diverse sources of income, while diverse landscapes offer habitat to more biodiversity. The higher this diversity, the higher the redundancy in functional agro-biodiversity. This underlies the provision of ecosystem services, that provide the balancing feedback loops maintaining the internal order of the system.

For this study, the term ‘environmental impact’ is used. The framework of the planetary boundaries is used to compile all relevant environmental impacts of Dutch soil bound agriculture. This more broad, system oriented approach allows the reader to see the connectivity between these problems. Through this lens, it becomes clear that the compartmentalised approach featured in many agri-environmental schemes, or in the approach to counter nitrogen volatilisation problems, is highly ineffective, or even contra-effective. This system perspective onto the environmental problems of Dutch agriculture highlight the need for a system perspective in studying the solutions. Only a few studies featured here, properly discussed the consequences of the system boundary they set for their particular study. It may be clear that authors have to justify the system boundary that they use, so that it is clear on what scale, certain management practices are bound to have positive effects.

Literature

van Apeldoorn, Dirk F. van, Kasper Kok, Marthijn P.W. Sonneveld, and Tom (A.) Veldkamp. 2011. ‘Panarchy Rules: Rethinking Resilience of Agroecosystems, Evidence from Dutch Dairy-Farming’. Ecology and Society 16 (1).

Batáry, P., A. Báldi, M. Sárospataki, F. Kohler, J. Verhulst, E. Knop, F. Herzog, and D. Kleijn. 2010. ‘Effect of Conservation Management on Bees and Insect-Pollinated Grassland Plant

Communities in Three European Countries’. Agriculture, Ecosystems & Environment 136 (1): 35–39. Beek, C. L. van, E. W. J. Hummelink, G.L. Velthof, and O. Oenema. 2004. ‘Denitrification Rates in

Relation to Groundwater Level in a Peat Soil under Grassland’. Biology and Fertility of Soils 39 (5): 329–36.

Beek, C. L. van, M. Pleijter, C. M. J. Jacobs, G. L. Velthof, J. W. van Groenigen, and P. J. Kuikman. 2010. ‘Emissions of N2O from Fertilized and Grazed Grassland on Organic Soil in Relation to

Groundwater Level’. Nutrient Cycling in Agroecosystems 86 (3): 331–40.

Best, E. P. H., and F. H. H. Jacobs. 1997. ‘The Influence of Raised Water Table Levels on Carbon Dioxide and Methane Production in Ditch-Dissected Peat Grasslands in the Netherlands’. Ecological Engineering 8 (2): 129–44.

Bianchi, F. J. J. A., P. W. Goedhart, and J. M. Baveco. 2008. ‘Enhanced Pest Control in Cabbage Crops near Forest in The Netherlands’. Landscape Ecology 23 (5): 595–602.

Bieleman, J. 2008. Boeren in Nederland. Geschiedenis van de landbouw 1500-2000. Amsterdam: Boom.

Booij, C. J. H., and J. Noorlander. 1992. ‘Farming Systems and Insect Predators’. Agriculture, Ecosystems & Environment, Biotic Diversity in Agroecosystems, 40 (1): 125–35.

Breeuwer, A., F. Berendse, F. Willems, R. Foppen, W. Teunissen, H. Schekkerman, and P. Goedhart. 2009. ‘Do Meadow Birds Profit from Agri-Environment Schemes in Dutch Agricultural Landscapes?’ Biological Conservation 142 (12): 2949–53.

‘Broeikasgassen - Emissieregistratie’. n.d. Accessed 12 November 2020.

http://www.emissieregistratie.nl/erpubliek/erpub/international/broeikasgassen.aspx.

Brusse, P. 2009. ‘J. Bieleman, Boeren in Nederland. Geschiedenis van de Landbouw 1500-2000’. Book review. BMGN - Low Countries Historical Review. Igitur. 1 January 2009.

Bukovinszky, T., J. Verheijen, S. Zwerver, E. Klop, J. C. Biesmeijer, F. L. Wäckers, H. H. T. Prins, and D. Kleijn. 2017. ‘Exploring the Relationships between Landscape Complexity, Wild Bee Species Richness and Reproduction, and Pollination Services along a Complexity Gradient in the Netherlands’. Biological Conservation 214 (October): 312–19.

Bünemann, E. K. n.d. ‘Soil Quality – A Critical Review | Elsevier Enhanced Reader’. Accessed 2 August 2020.

Bünemann, E. K., G. Bongiorno, Z. Bai, R. E. Creamer, G. De Deyn, R. de Goede, L. Fleskens, et al. 2018. ‘Soil Quality – A Critical Review’. Soil Biology and Biochemistry 120: 105–25.

Cardinale, B. J., J. E. Duffy, A. Gonzalez, D. U. Hooper, C. Perrings, P. Venail, A. Narwani, et al. 2012. ‘Biodiversity Loss and Its Impact on Humanity’. Nature 486 (7401): 59–67.

Crittenden, S. J., E. Huerta, R. G. M. de Goede, and M. M. Pulleman. 2015. ‘Earthworm Assemblages as Affected by Field Margin Strips and Tillage Intensity: An on-Farm Approach’. European Journal of Soil Biology 66 (January): 49–56.

De Visser, P. H. B., H. Van Keulen, E. A. Lantinga, and H. M. J. Udo. 2001. ‘Efficient Resource Management in Dairy Farming on Peat and Heavy Clay Soils’. NJAS - Wageningen Journal of Life Sciences 49 (2): 255–76.

Deru, J. G. C., J. Bloem, R. de Goede, H. Keidel, H. Kloen, M. Rutgers, J. van den Akker, L. Brussaard, and N. van Eekeren. 2018. ‘Soil Ecology and Ecosystem Services of Dairy and Semi-Natural Grasslands on Peat’. Applied Soil Ecology 125 (April): 26–34.

Didden, W. A. M., J. C. Y. Marinissen, M. J. Vreeken-Buijs, S. L. G. E. Burgers, R. de Fluiter, M. Geurs, and L. Brussaard. 1994. ‘Soil Meso- and Macrofauna in Two Agricultural Systems: Factors Affecting Population Dynamics and Evaluation of Their Role in Carbon and Nitrogen Dynamics’. Agriculture, Ecosystems & Environment 51: 171–86.

Doran, J. W., and M. R. Zeiss. 2000. ‘Soil Health and Sustainability: Managing the Biotic Component of Soil Quality’. Applied Soil Ecology, Special issue: Managing the Biotic component of Soil Quality, 15 (1): 3–11.

Eekeren, N. van, H. de Boer, M. Hanegraaf, J. Bokhorst, D. Nierop, J. Bloem, T. Schouten, R. de Goede, and L. Brussaard. 2010. ‘Ecosystem Services in Grassland Associated with Biotic and Abiotic Soil Parameters’. Soil Biology and Biochemistry 42 (9): 1491–1504.

Eekeren, N. van, L. Bommelé, J. Bloem, T. Schouten, M. Rutgers, R. de Goede, D. Reheul, and L. Brussaard. 2008. ‘Soil Biological Quality after 36 Years of Ley-Arable Cropping, Permanent Grassland and Permanent Arable Cropping’. Applied Soil Ecology 40 (3): 432–46.