On July 25, 2019, we performed a scientific literature search on Web of Science for years 1969 to 2019 with the following combination of keywords in the title or the abstract: [(hedgerow^* OR “living fence^*” OR “woody field margin” OR “woody edge” OR “woody buffer strip”) AND (benefit^* OR ^*diversity OR function^* OR service^*)] and these words in Spanish, namely, [(“cerca^* viva^*” OR seto^* OR “borde leñoso” OR “margen leñoso” OR “franja leñosa”) AND (beneficio^* OR ^*diversidad OR funcion^* OR servicio^*)]. This search identified 985 studies that might report quantitative evidence of levels of biodiversity and/or ecosystem services rendered by hedgerows.

One co-author read the abstracts of all 985 studies. After reading these abstracts, 294, 314, and 377 articles were deemed as useful, possibly useful, or non-useful, respectively, in terms of including quantitative information related to biodiversity or ecosystem services in hedgerows and paired control ecosystems. The 608 useful and possibly useful studies were carefully revised to identify and extract relevant data; this task was performed by the same author in the interest of consistency for data gathering. We finally extracted data from 294 studies (detailed PRISMA flow diagram in Supplementary Figure 1).

A database was built to systematize the extracted information of the following variables for each article: article's identifier, reference of the study, response variables (indicators), units of response variables, effect sign (direct or inverse), taxa (see below), type of ecosystem service (see below), sample size, response value at farmland with hedgerows and control ecosystems (namely, farmland without hedgerows or nearby natural habitat), variation and type of variation (variance, standard deviation (SD), standard error, interquartile range, confidence interval) of the response value, farmland type (crop or pasture), crop type (herbaceous or woody), natural habitat type (forest/woodland or grassland/meadow), country, climate type (tropical or temperate), focus of the spatial scale (field or landscape), overall effect of hedgerows on biodiversity and ecosystem services (positive, neutral, or negative), and type of inference for this overall effect.

Once completed, another coauthor independently revised the database to double-check the procedure and detect possible inconsistencies, errors, and noticeable weird information. This revision corrected the signs of many relationships and excluded 124 studies: 81 of them because they did not report the variability of response variables (i.e., a robust effect size measure could not be used with these articles) and 43 because they referred to non-woody margins. Appendix 1 lists the 170 primary studies that provided data that were finally analyzed. The response variables were linked to the taxonomic groups (biodiversity) plants, invertebrates, and vertebrates, and one of the four categories of ecosystem services according to the Millennium Ecosystem Assessment (MEA, 2005), namely, supporting, regulating, provisioning, and cultural ecosystem services. Farmland type, crop type, natural habitat type, climate type, and focus of the spatial scale were used as context factors (moderators) of hedgerow effects.

We analyzed the database following standard methodologies in meta-analyses (Koricheva et al., 2013). We split the database into four datasets for further analyses. The first and second datasets allowed to compare the biodiversity levels between farmland with hedgerows and farmland without hedgerows (Data File 1; R script BioHEFA) and between farmland with hedgerows and natural habitat (Data File 2; R script BioHENH). The third and fourth datasets allowed to compare the ecosystem services between farmland with hedgerows and farmland without hedgerows (Data File 3; R script ESHEFA) and between farmland with hedgerows and natural habitat (Data File 4; R script ESHENH). As using the broad ecosystem-service categories might result in lumping together cases that clearly have opposite signs, which will level out the overall response and reduce the likelihood of significant findings, we performed supplementary analyses on the ecosystem services with, at least, three primary studies and ten observations (Table 1; Data Files 5, 6; R scripts ESHEFA_specific, ESHENH_specific).

Standardized mean difference (SMD) was computed as a robust estimate of effect size of farmland with hedgerows vs. control ecosystems using the escalc function in metafor R package (Viechtbauer, 2010). The sign of SMD was changed if an effect in each observation was based on an inverse relationship (e.g., invasive species and di-services). The SMD is commonly used to estimate the magnitude of the effect by comparing an experimental group and a control group (Hedges and Olkin, 1986; Gurevitch et al., 2001).

A null mixed model adjusted for meta-analysis was fit (Borenstein et al., 2009) to assess whether farmland with hedgerows showed different biodiversity or ecosystem services levels from either farmland without hedgerows or natural habitat. By using a mixed model, we assumed that each case study was a random sample of a larger overall population and considered the primary study (article) as a nesting factor to avoid violating the assumption that effect sizes are independent from each other. We evaluated the extent of heterogeneity by using the statistic I² (Higgins and Thompson, 2002), which estimates the percentage of variability due to heterogeneity rather than due to the sampling error.

We also evaluated if moderators explained a significant part of the heterogeneity in the true effect by using simple meta-regressions. Moderators were farmland type (crop or pasture), crop type (herbaceous or woody), natural habitat type (forest/woodland or grassland/meadow), climate type (tropical or temperate), and focus of the spatial scale (field or landscape). Finally, we evaluated if the difference between the effect sizes of factor levels was significantly different from zero by conducting an omnibus test based on a chi-square distribution with the number of coefficients tested as degrees of freedom and following default settings. All statistical analyses were performed in metafor R package (R Core Team, 2021).

Data filtering retained 835 case studies from 170 primary studies conducted in 39 countries of all continents but Antarctica (Figure 1; Data File 7; script Figure 1). Most records (70% of observations and 74% of primary studies) were related to biodiversity (abundance = 38 and 41%; diversity = 31 and 32%; plants = 35 and 25%; invertebrates = 21 and 23%; vertebrates = 14 and 16%), followed by supporting services (16%, 20%), regulating services (13%, 26%), provisioning services (5%, 15%), and cultural services (<1%, excluded for further analyses due to low sample size). The number of observations and primary studies was as follows: crop = 64 and 56%, pasture = 4 and 7%, herbaceous crop = 54 and 42%, woody crop = 17 and 16%, forest/woodland habitat = 29 and 24%, grassland/meadow habitat = 3 and 5%, tropical climate = 22 and 21%, temperate climate = 78 and 72%, field scale = 67 and 49%, and landscape scale = 33 and 29% (see Table 2 for absolute numbers).

The effect of farmland with hedgerows as compared to farmland without hedgerows on biodiversity was positive, negative, and neutral in 243, 107, and 13 observations, respectively (N = 363 observations from 72 primary studies). Meta-analysis showed significantly higher biodiversity levels in farmland with hedgerows than in farmland without hedgerows (Figure 2; estimated effect size = 0.64 ± 0.18, p < 0.01). Heterogeneity was very high (I² 95% CI: 99.6–99.8%), suggesting that most heterogeneity not explained by sampling error and thus justifying the inclusion of moderators in the model.

Farmland with hedgerows was equally beneficial for both abundance and diversity metrics (Figure 2; Q_M = 3.24, p = 0.07). Type of organism explained a significant part of the heterogeneity of effect sizes (Figure 2); while plants were more positively affected by hedgerows than invertebrates (Q_M = 7.69, p = 0.01), enhancement of vertebrate biodiversity was similar to that of plants (Q_M = 0.96, p = 0.33) and invertebrates (Q_M = 3.19, p = 0.07).

Hedgerows enhanced biodiversity levels in both cropped farmland and pastures, with no significant differences between these two farmland types (Figure 2; Q_M = 0.13, p = 0.72). However, hedgerows did not show a positive effect on biodiversity for crop types; hedgerows enhanced biodiversity on herbaceous crops, but the biodiversity of woody crops was higher without hedgerows (Figure 2; Q_M = 186.88, p < 0.01). Farmland with hedgerows supported higher biodiversity than farmland without hedgerows in temperate regions only (Figure 2; estimated effect size: temperate = 0.78 ± 0.20, p < 0.01; tropical = −0.02 ± 0.42, p = 0.97), and the difference between tropical and temperate regions was not significant (Figure 2; Q_M = 2.91, p = 0.09). Likewise, both field and landscape studies were equally positively benefitted by hedgerows (Figure 2; Q_M = 0.61, p = 0.43), but hedgerows effects at the landscape scale were barely significant.

The effect of farmland with hedgerows as compared to natural habitat on biodiversity was negative, positive, and neutral in 116, 92, and 9 observations, respectively (N = 217 observations from 43 primary studies). Consequently, meta-analysis revealed no significant differences between farmland with hedgerows and natural habitat in terms of biodiversity (Figure 3; estimated effect size = 0.11 ± 0.14, p = 0.43; I² 95% CI: 87.70–95.16%).

There were significant differences in the effect of hedgerows between biodiversity metrics, organism types, and natural habitat type (Figure 3). Abundance of organisms did not differ between farmland with hedgerows and natural habitat, but diversity was higher in farmland with hedgerows (Figure 3; Q_M = 128.78, p < 0.01). Only invertebrates exhibited higher biodiversity levels in farmland with hedgerows than in natural habitat (Figure 3; estimated effect size = 0.39 ± 0.17, p = 0.02). Biodiversity differed between natural habitat types; farmland with hedgerows had a higher level than grasslands/meadows but similar to forest/woodlands (Figure 3; Q_M = 5.39, p = 0.02). Neither climate type (Q_M = 3.38, p = 0.07) nor focus of spatial scale (Q_M = 0.66, p = 0.41) influenced the magnitude of effect size (Figure 3).

The effect on ecosystem services of farmland with hedgerows as compared to farmland without hedgerows was positive, negative, and neutral in 127, 65, and 12 observations, respectively (N = 204 observations from 43 primary studies). Hedgerows did not enhance overall ecosystem service levels (Figure 4; estimated effect size = 0.30 ± 0.33, p = 0.36; I² 95% CI: 95.75–98.63%). However, meta-analysis revealed significant differences when comparing ecosystem services and farmland types (Figure 4). Farmland with hedgerows showed positive effects on provisioning services (estimated effect size: 1.44 ± 0.38, p < 0.01), but not on regulating (estimated effect size: −0.22 ± 0.35, p = 0.54) or supporting services (estimated effect size: −0.08 ± 0.35, p = 0.82). In addition, farmland with hedgerows provided more supporting than regulating services (Q_M = 5.61, p = 0.02). Crops showed higher ecosystem service levels than pastures (Figure 4; Q_M = 22.06, p < 0.01). The magnitude of effect size did not vary between crop types (Figure 4; Q_M = 0.09, p = 0.76), climate type (Figure 4; Q_M = 0.40, p = 0.53), and the spatial scale (Figure 4; Q_M = 0.13, p = 0.72). The complementary analyses on specific ecosystem services confirmed the pattern observed for the broad ecosystem-service categories (Supplementary Figure 2; Data File 5; R script ESHEFA_specific).

The effect of farmland with hedgerows on ecosystem services as compared to natural habitat was negative, neutral, and positive in 32, 11, and 8 observations, respectively (N = 51 observations from 11 primary studies). Meta-analysis showed that natural habitat provided higher levels of ecosystem services than farmland with hedgerows (Figure 5; estimated effect size: −0.49 ± 0.20, p = 0.02; I² 95% CI: 58.20–94.84%).

The magnitude of the effect size was similar between regulating and supporting services (Figure 5; Q_M = 0.21, p = 0.64), tropical and temperate climates (Figure 5; Q_M = 0.36, p = 0.55) and field and landscape scales (Figure 5; Q_M = 1.39, p = 0.24). There was insufficient statistical power to test the differences in effects size related to natural habitat types and provisioning services.

The complementary analyses for specific ecosystem services (soil formation, nitrogen cycling, and pest regulation) showed that natural habitats exhibited greater pest regulation services than farmland with hedgerows (Supplementary Figure 3; Data File 6; R script ESHENH_specific). These ecosystem services depended on climate type (hedgerows were more beneficial in temperate than in tropical regions; estimated effect size: temperate = −0.29 ± 0.06, p < 0.01; tropical = −1.00 ± 1.13, p < 0.01; Q_M = 23.37, p < 0.01).

Hedgerows enhanced the biodiversity levels in farmland, and both metrics (abundance and diversity) and farmland types (cropland and pastureland) studied here were similarly benefitted (Figure 2). Plants were the most positively affected taxa by hedgerows in agreement with previous studies. For instance, Aavik et al. (2008) found that semi-natural elements in the landscape enhanced plant species richness both at the field and at the landscape scales, and Evans et al. (2011) showed that uncultivated semi-natural habitats such as mature hedgerows are more species rich and have higher biomass than cropped habitat.

Hedgerows around herbaceous crops provide woody habitat otherwise absent or very scarce in agricultural landscapes. However, woody crops contain part of the hedgerows' functional diversity (i.e., the cultivated shrubs or trees) that may explain why herbaceous crops are more positively affected in terms of biodiversity than woody crops. For instance, the cultivated shrubs and trees facilitate the establishment of non-cropped perennial plants and other species (Oliveira et al., 2015). Castro-Caro et al. (2015) also reported that hedges, and ground covers to a lesser extent, efficiently increased the abundance and richness of passerine communities in olive groves, particularly of insectivorous birds. Other vertebrate species benefit from the presence of hedgerows. For instance, Verboom and Huitema (1997) suggested that landscapes with a high density of linear elements have a surplus value for serotine bats because bats use hedgerows as orientation clues, foraging habitat, and shelter from wind and predators.

The plant species that are found only in non-cropped habitat are primarily native and perennials, whereas a high proportion of the species found only in cultivated fields are annual and introduced species (Jobin et al., 1996). More exotic plants in farmland without hedgerows than in farmland with hedgerows have also been reported by Boutin et al. (2008). Plant species richness and composition varied substantially among the five landscapes studied by Freemark et al. (2002), ranging from a row-crop monoculture landscape to a diverse mosaic of cropped and non-cropped habitat landscape. The more complex the hedgerows, the species-richer farmland is (Klimek et al., 2014). Established shrubs may attract seed-dispersing wildlife and create proper environments for plant recruitment, generating reforestation feedback (Suárez-Esteban et al., 2013). In addition, trees planted into living fences generally have greater survival, relative growth, and height compared to those planted in open pasture (Love et al., 2009).

The difference in biodiversity levels between farmland with hedgerows and farmland without hedgerows was maximum in temperate regions (Figure 2). Such an extra benefit of hedgerows in temperate regions with respect to the tropics may be due to more ameliorated microclimate conditions for plant recruitment and wildlife animals (Ceresa et al., 2012; Burgess et al., 2015). For instance, the spider Nephila clavipes persists in highly fragmented tropical farmland landscapes thanks to the prevalence of relatively stable conditions (Pinkus-Rendón et al., 2006), whereas butterflies have been found to be favored by intermediate levels of landscape complexity (Dainese et al., 2015). The influence of drivers other than climate (e.g., hedgerow complexity) may explain why our analysis exhibited non-significant effects of hedgerows on invertebrate biodiversity (Figure 2).

Hedgerows were more beneficial at the field than at the landscape scale (Figure 2) since there is often some natural habitat other than hedgerows to support biodiversity at the landscape scale. The landscape context is important to the delivery of ecosystem services (Dainese et al., 2017). For instance, Fritz and Merriam (1996) concluded that habitat qualities that result from the architectural differences between hedgerows and forest may constrain plant movement. In landscapes of fragmented forest, hedgerows may provide corridors for woodland plants (Fritz and Merriam, 1996). However, non-fragmented (i.e., well-connected) forests do not need hedgerows for connectivity.

Ecosystem service levels in farmland with hedgerows were higher, albeit statistically insignificant, than in farmland without hedgerows. Our synthesis failed to highlight that hedgerows, in general, increased supporting or regulating services; however, hedgerows resulted in a higher amount of provisioning services – the ones that are sought in farmland habitat (Figure 4).

The complementary analyses confirmed the lack of significant differences for individual supporting and regulation services between farmland with hedgerows and farmland without hedgerows. This finding agrees with Albrecht et al. (2021) who concluded inconsistent and highly variable effects of hedgerows on crop pollination and yield and non-significant effects on pest control. Many agricultural systems use hedgerows as a supplementary source for cropping (for instance, intercropping of annual crops between Inga edulis contour hedgerows; Alegre and Rao, 1996). Hauser (2008) found that hedgerows dominated by cassava Manihot esculenta increased groundnut and maize production. In their meta-analysis, Van Vooren et al. (2017) reported that, up to 20 times the hedgerow height, yield is increased by an average of 6% as compared with farmland without hedgerows.

Conserving and restoring hedgerows may enhance some supporting services such as habitat creation (linked to biodiversity levels) and a range of regulating services. Shrubs and trees of hedgerows usually have well-developed root systems compared to herbaceous crops, which increases soil organic matter and sequesters carbon. Boutin's (2006) analyses of soil properties showed that organic matter and nutrient levels are often higher in hedgerows than in open fields. A hedgerow network protects landscapes from soil organic carbon erosion by runoff (Lacoste et al., 2016). Legume shrubs, which naturally fertilize soils, are common in hedgerow composition (Lemage et al., 2021). The architecture of some species such as the broom Retama sphaerocarpa provides shelter for many species (Matías et al., 2011). Franco (1998) and Van Vooren et al. (2017) reported that hedgerows retain 85 and 90% of suspended solids, 73 and 67% of total nitrogen, and 70 and 73% of total phosphorus in runoff, respectively.

Pastures were less benefitted by hedgerows than crop fields in terms of ecosystem services (Figure 4). Crop fields usually support a more intensive management scheme (tillage, fertilization, pesticide application) than pastures, where livestock fertilize the soil on crop fields. Heavy machinery compact soils and reduce supporting and regulating services (Power, 2010; The Research Council of Norway, 2011). Thus, hedgerows (i.e., unploughed, unfertilized features) around crop fields may become a reservoir for ecosystem services on top of biodiversity (Speir, 2015). In short, hedgerows play a relevant role in biodiversity conservation and climate change mitigation (Montgomery et al., 2020).

In farmland with hedgerows, abundance and diversity levels were similar and higher, respectively, than in natural habitat (Figure 3). At the population level, seedlings and juveniles are, usually, proportionally less abundant in hedgerows than in forests (Wehling and Diekmann, 2010), and the establishment of new populations of forest species in hedgerows is lower than in forests (Schmucki and De Blois, 2009). However, hedgerows represent a strong edge effect (forest and open fields) that may favor diversity (Litza and Diekmann, 2019).

We found a higher diversity of invertebrates in farmland with hedgerows than in natural habitat, particularly in grasslands and meadows. While these herbaceous habitats have seasonal food availability and fluctuant microclimate conditions, farmland with hedgerows can feed invertebrates along the year with complementary food resources and offer overwintering refuges. Thus, Jahnová et al. (2016) recorded more diverse ground-dwelling beetle communities in meadow margins than inside the meadows. Hedgerows help in ensuring a continuous supply of nectar on farmland; for instance, Timberlake et al. (2019) suggested that semi-natural habitat maintains complementary flowering phenology. The higher diversity of invertebrates in farmland with hedgerows than in natural habitat might be taxon specific. The finding of this study contrasts with Berwaerts et al. (1998) who found that the genetic diversity of speckled wood butterflies (Pararge aegeria L.) of large continuous woodland was significantly higher than in a highly fragmented area with very small woodland fragments and hedgerows scattered in an intensive agricultural landscape. Similarly, Buddle and Higgins (2020) suggested that lower spider diversity in hedgerows than in riparian forests may be due in part to the invasive honeysuckle shrub Lonicera maackii, which reduces hedgerow complexity on the ground layer.

Farmland with hedgerows provided higher amounts of ecosystem services than farmland without hedgerows, but lower amounts than natural habitat (Figure 5), thus supporting our H2 as other studies did. Hannon and Sisk (2009) found that hedgerow herbaceous flora is intermediate to that of crop fields and woodlands; consequently, regulating services such as pollination provided by natural habitat were higher than those provided by farmland with hedgerows. Byrne and Del Barco-Trillo (2019) found greater densities of bumblebees in unmanaged grasslands than in farmland with hedgerows. In this study, regulating services were particularly greater in natural habitat than in farmland with hedgerows at the landscape scale, whereas this effect is diluted at the field scale. However, we were unable to disentangle what moderators drove such a scale effect due to shortage of data.