Saturday 30 October 2021

Estimating the benefits of agroforestry to European wildlife.

The term agroforestry is used to denote practices in which the cultivation of trees is integrated either with the rearing of livestock (in which case it is called silvopasturalism) or other plant crops (silvoarablism). This is a traditional practice across much of Europe, where methods such as grazing livestock in orchards are very widespread, with newer methods being developed more recently, such as short-rotation coppicing being carried out alongside rows of other crops. Systems in which productive trees are grown around the edges of fields are also sometimes considered to be agroforestry, although in these cases the trees are managed separately to the other produce, and may be under separate ownership.

Pigs grazing in an open Oak forest system in Spain, a system known as a 'dehesa'. Álvarez (2016).

Europe has suffered particularly severe losses of biodiversity compared to other parts of the world, and this is particularly severe in areas where intensive agriculture is prevalent. Agroforestry promotes a more diverse landscape than arable monoculture, potentially resulting in higher biodiversity. Quantifying the benefits of this could potentially lead to the system being more heavily prioritised under the European Common Agricultural Policy or any successor system.

Agroforestry systems have been well studied in tropical environments, where the evidence suggests that the system offers significant advantages in biodiversity preservation over intensive monocultural systems, but nevertheless tends to lead to reduced biodiversity compared to both primary and secondary forests. The system is less well studied in temperate regions, with most studies tending to concentrate on single groups of Animals, such as Birds or Insects. This leaves the benefits of such systems in Europe somewhat unclear, particularly as the definitions of agroforestry can vary, leading to differences in what systems are included in studies, making comparisons between studies difficult. 

Hazel short rotation coppice system alongside crops in Suffolk, UK. Smith et al. (2014).

In a paper published in the journal BMC Ecology and Evolution on 23 October 2021, Anne‑Christine Mupepele of Nature Conservation and Landscape Ecology and Biometry and Environmental System Analysis at the University of Freiburg, and Matteo Keller and Carsten Dormann, also of Environmental System Analysis at the University of Freiburg, present the results of a meta-analysis which combined results from a number of studies of agroforestry systems across Europe.

Mupepele et al. sought to answer three questions, 'What is the effect of agroforestry on biodiversity relative to forests, pastures, cropland or abandoned, shrub-encroached agroforestry?', 'Is the effect of agroforestry on biodiversity influenced by environmental variables, specifically the kind of agroforestry system (silvopasture or silvoarable), sampling method, the specific measure of biodiversity, sampling year, country, climate and the reference used?' and 'How strong and robust is the underlying evidence of these results?'

To which end they located 1411 previous studies of agroforestry systems in Europe, 50 of which were eventually included in the study, representing 69 individual agroforestry sites. Each of these had a direct comparison of a type of agroforestry (silvoarable or silvopastoral) to forests, cropland, pasture, and/or abandoned agroforestry systems.

Map of Europe with the number of effect sites per country. Mupepele et al. (2021).

The studies included in the analysis covered sites across Europe where agroforestry systems have been studied between 1984 and 2019. The majority of these sites were caried out in Iberia and the Mediterranean region, with twelve studies from Spain, eight from Portugal, five from Italy, one from France and one from Turkey. Temperate central Europe was represented by six studies from the UK, four from Romania, two each from France, Germany, and Switzerland, and one each from Belgium and northern Italy. The northern boreal region was represented by four studies from Sweden and two from Finland.

Thirty six of the included studies looked at silvopastoral systems, with thirty six studies looking at 52 sites, while silvoarable systems were the subject of thirteen studies looking at seventeen sites. The biodiversity of agroforestry was most commonly compared to that of pasture (23 sites), or forests (21 sites), then abandoned agroforestry systems (thirteen sites) and cropland (12 sites).

Sheep grazing in a plantation of Pine and Eucalyptus in Spain. Monica Pelliccia/Mongabay.

The different studies measured biodiversity in different ways, and concentrated on different groups. In order to make a comparison between these diverse studies, Mupepele et al. divided the measured wildlife into five groups, Arthropods, Birds, Bats, Plants, and 'Fungi plus Lichens and Bryophytes', Most of the included studies measured biodiversity at the 'species richness level', although other measures were used.

Mupepelele et al.'s results showed no overall benefit for biodiversity compared to the average derived from all systems. However, silvoarable systems were found to host considerably more biodiversity than other croplands, although they generally hosted less biodiversity than forests. Silvopastoral systems produced less clear results, with measures often producing conflicting results in different studies (i.e. one study might show higher Avian biodiversity in a silvopastoral system than a forest, while another showed the reverse.

Birds and Artropods were typically found at higher levels of diversity in agroforestry envoronments than other systems, Where the original group sorted Arthropods into different groups (e.g. Bees, Beetles and Spiders', then this biodiversity increassed, although this was across all environments, with no change in the beneficial effect of agroforestry.

Cereal crops grown alongside trees in Bedfordshire, UK. Agroforestry Research Trust.

Mupepele et al. note that the quality of the studies they were referencing varied somewhat, with some using replicated experimentation with clear controls, whilst others were more observational in nature. To compensate for this, they tried applying a statistical weighting method that gave more value to the more statistically strong studies, but found this made no difference to the overall result. They also carried out funnel plot and Egger’s regression tests for undetected biases in their data, but did not find bias was a problem.

A previous  meta-analysis led by Mario Torralba of the Department of Geosciences and Natural Resource Management at the University of Copenhagen found that agroforestry had a much stronger impact on biodiversity, which caused Mupepele et al. to consider the differences between their findings and that of the earlier study. They note that Torralba et al.'s study was published in 2016, and contained the results from two studies published in 2015 on the benefits of agroforestry in Mediterranean ecosystems, both of which produced very strong positive results, and that if these were excluded from Torrialba et al.'s data then the result was closer to that of Mupepele et al. who included several post 2015 studies with less clear results.

Properly done, meta-analyses can provide a powerful tool for understanding ecological systems in ways not possible from individual studies or unsystematic literature searches. However, the robustness of these results is dependent on the methods used to analyse the data, and in particular the use of weighting to take into account the quality of the studies being referenced. This needs to be done carefully, as failure to apply the right weighting can often lead to very different results. This said, applying weighting to Mupepele et al.'s results resulted in no significant change in the outcome of the study, which strongly supports the robustness of their findings. 

The application of repeated meta-analyses to the same data set can reveal changes over time, as new studies add to the overall picture, dampening the results from atypical studies that might have a profound impact on a smaller data-set. By building a cumulative model in which data were added in chronological order, Mupepele et al. were able to demonstrate that the impact of agroforestry upon biodiversity remained essentially unchanged over time, despite the presence of some anomalous data. They do, however, note that silvoarable systems make up a relatively small proportion of the whole, and that the addition of a higher proportion of studies of these systems in future might change the results of the meta-analysis.

Merino Sheep under a Cork Oak in a montado silvopastoral system in Portugal. European Agroforestry Foundation.

The ability to reproduce results is an important principle in science, but can be difficult in fields like ecology, which look at complex natural systems, no two of which are ever completely the same. Mupepele et al.'s results differed strongly from the earlier results of Torralba et al., resulting in their drawing different conclusions; Torralba et al. concluded that agroforestry has a general positive impact upon biodiversity, while Mupepele et al. concluded that this benefit was only clear when agroforestry was compared to croplands, despite both studies having used much of the same data. Mupepele et al. note that Torralba et al. included hedgerows and woody riparian buffers to agricultural land as agroforestry, while Mupepele et al. excluded them on the basis that they are not emplaced for silvicultural purposes (i.e. the trees used in these settings are grown for their value as boundaries, not as a crop in themselves). Neither did Torralba et al. include data from studies which suggested agroforestry had a negative impact on biodiversity. Mupepele et al. believe that scientists should be very clear about what data they are including in meta-analyses, the criteria for choosing this data, and the reasons to do so, in order to help policy-makers judge the significance of different studies. 

Mupepele et al. conclude that silvoarable systems produce an increase in biodiversity compared to conventional croplands, particularly with regard to Birds and Arthropods, but that this increase is not large, and there was no overall positive benefit of agroforestry to all other settings. Notably, silvopasturalism showed no clear benefit over either forestry or conventional pasturelands. Where previous studies have produced enthusiastic support for agroforestry, and strongly suggested these systems are linked to a significant increase in biodiversity, Mupepele take a more cautious approach, noting that relatively few studies find an unqualified link between agroforestry and increased biodiversity, and that literature reviews and meta-analyses need to be careful to include both the positive and negative impacts of systems when drawing on data from multiple studies. Nevertheless, they do conclude that agroforestry can have a positive impact on biodiversity under some circumstances, as well as providing carbon sequestration and other ecosystem services, and that a better understanding of how these systems work could lead to more informed future decisions by policy makers.

See also...

Follow Sciency Thoughts on Facebook.

Follow Sciency Thoughts on Twitter