Journal list menu
Ecological intensification and arbuscular mycorrhizas: a meta-analysis of tillage and cover crop effects
Summary
- Reliance on ecosystem services instead of synthetic, non-renewable inputs is increasingly seen as key to achieving food security in an environmentally sustainable way. This process, known as ecological intensification, will depend in large part on enhancing below-ground biological interactions that facilitate resource use efficiency. Arbuscular mycorrhizas (AM), associations formed between the roots of most terrestrial plant species and a specialized group of soil fungi, provide valuable ecosystem services, but the full magnitude of these services may not be fully realized under conventional intensively managed annual agricultural systems.
- Here, we use meta-analysis to assess how reducing soil disturbance and periods without roots in agricultural systems affect the formation of AM and the diversity and community composition of arbuscular mycorrhizal fungi (AMF). We compiled data from 54 field studies across five continents that measured effects of tillage and/or cover cropping on AMF colonization and/or communities and assessed effects of management and environmental factors on these responses.
- Less intensive tillage and winter cover cropping similarly increased AMF colonization of summer annual cash crop roots by ∼30%. The key variables influencing the change in AMF colonization were the type of cover crop or the type of alternative tillage, suggesting that farmers can optimize combinations of tillage and cover crops that most enhance AM formation, particularly with no-till systems and legume cover crops.
- Richness of AMF taxa increased by 11% in low-intensity vs. conventional tillage regimes. Several studies showed changes in diversity and community composition of AMF with cover cropping, but these responses were not consistent.
- Synthesis and applications. This meta-analysis indicates that less intensive tillage and cover cropping are both viable strategies for enhancing root colonization from indigenous arbuscular mycorrhizal fungi (AMF) across a wide range of soil types and cash crop species, and possibly also shifting AMF community structure, which could in turn increase biologically based resource use in agricultural systems.
Introduction
Steady increases in crop yields since the green revolution have come at substantial environmental cost (Pingali 2012). Any further increases in yields must not further erode the natural capital upon which agriculture relies, especially in times of environmental change, and must minimize negative effects on ecosystem sustainability. Increasing reliance on supporting and regulating ecosystem services instead of synthetic inputs, that is ecological intensification, is increasingly seen as one way of achieving food security in an environmentally sustainable way (Jackson et al. 2012; Bommarco, Kleijn & Potts 2013). Ultimately, this will depend in large part on enhancing below-ground biological interactions that facilitate resource use efficiency (Jackson et al. 2012; Bender, Wagg & van der Heijden 2016).
Crops take up approximately only half of the nutrients in applied inorganic fertilizers, with the remainder at risk of being lost from agroecosystems (Robertson & Vitousek 2009). These losses have widespread and serious consequences for climate change, biodiversity and human health (Erisman et al. 2014). Plants have evolved many traits for optimizing nutrient acquisition and utilization (Hodge 2004; York, Nord & Lynch 2013; Lambers, Martinoia & Renton 2015), including the formation of arbuscular mycorrhizas (AM), associations between the roots of most terrestrial plant species and a specialized group of soil fungi (Smith & Read 2008). Arbuscular mycorrhizas also provide valuable ecosystem services beyond nutrient acquisition, such as increasing nutrient retention (Cavagnaro et al. 2015), plant drought resistance (Augé, Toler & Saxton 2015) and soil structure formation (Rillig & Mummey 2006). But the full magnitude of these services is often not realized in intensively managed, annual agricultural systems (Gosling et al. 2006). High rates of soil disturbance and long periods when roots are not present limit the formation of AMs, due to life-history characteristics of these obligate biotrophs (Smith & Read 2008), which depend on carbon (C) from roots to grow and reproduce. A prior meta-analysis (Lekberg & Koide 2005), including both field and greenhouse studies, showed that reducing soil disturbance and shortening fallow periods does increase colonization of roots by arbuscular mycorrhizal fungi (AMF). But inconsistent results from field studies included in that analysis and many more recent ones (e.g. Gavito & Miller 1998; Duan et al. 2010; White & Weil 2010; Higo et al. 2014) point to a need to assess how the wide variety of conditions [e.g. soil type, crop species, soil phosphorus (P) status] across these studies influence AM responses.
The composition and diversity of AMF communities is increasingly recognized as an important factor in how plants respond to colonization and potential benefits they receive (Verbruggen & Kiers 2010). Similar to plants, AMF species have different traits that make them functionally distinct (Aguilar-Trigueros et al. 2015). For instance, some taxa provide better disease or drought stress resistance while others better enhance nutrient uptake and reduce leaching (Marulanda, Azcón & Ruiz-Lozano 2003; Maherali & Klironomos 2007; Köhl, Lukasiewicz & van der Heijden 2016). Ideally, agricultural management would support a functionally diverse AMF species mixture to increase the multi-functionality of the symbiosis. But intensively managed agricultural systems impose strong filters that limit the community assemblage of AMF species to those that can persist in the face of high rates of disturbance, long fallow periods and often monocultures of plant hosts (Verbruggen & Kiers 2010). Often this selects for ruderal species that invest heavily in reproduction and less in nutrient scavenging or transfer to hosts (Oehl et al. 2003; Verbruggen & Kiers 2010; Chagnon et al. 2013). By changing disturbance regimes and temporal resource availability, low-intensity tillage and cover cropping would be expected to change AMF community composition and potentially enhance diversity if more niche space is created, for example for slower-growing species (Oehl et al. 2009). Conversely, since AMF sometimes prefer specific plant hosts (Johnson et al. 2004), AMF communities measured on the same cash crop may not differ to a great extent in spite of differences in agronomic management.
A number of field studies have examined changes in AMF communities in response to less intensive management, so that meta-analysis is now possible for assessing how such management could optimize this below-ground interaction considered central to ecological intensification. The focus here is on the impacts of two management interventions that reduce soil disturbance and periods without roots – the use of low-intensity tillage regimes and cover crops – and their impacts on AM colonization and the AMF community. We evaluated 278 comparisons from 54 field studies published between 1990 and 2015 (see Methods), spanning five continents. Studies on cover cropping encompassed a range of cover crop groups, cash crops and sampling times, and studies on tillage included different tillage types, cash crops and soil texture.
Materials and methods
Literature Search and Data Collection
We searched the literature in 2015 using ISI Web of Knowledge (available online). Two separate searches were conducted for assessing effects of cover cropping or alternative tillage on AMF colonization rates on cash crop roots or on the AMF community. Although AMF colonization of roots is not necessarily indicative of AM functionality, for example benefits for plant nutrient uptake or productivity, it is the most widely measured attribute of AM and the best indicator available. For cover cropping, the search terms were ‘mycorr*’ AND ‘cover crop*’, which resulted in 108 articles in March 2015. For alternative tillage, the search terms were mycorr* AND (‘conservation till*’ OR ‘no-till*’ OR ‘reduced till*’), which resulted in 239 articles in October 2015. We screened these articles to meet our selection criteria: (i) a field trial comparing (a) bare winter fallow (i.e. unplanted) vs. cover crop(s), or (b) multiple types of tillage, including a comparison between a ‘conventional’ type (usually a mouldboard plough, i.e. soil inversion) and an alternative (e.g. no-till, chisel till); and (ii) data on AMF colonization rates (i.e. per cent root length colonized) on roots of the subsequent annual cash crop. We also examined studies that analysed AMF community composition from soil or root samples (by spore morphological identification or genetic analysis) following incorporation of the cover crop and/or tillage. Only studies with imposed, replicated treatments at one or more sites were included. We expanded our search by checking the reference lists of studies that met our selection criteria.
Multiple comparisons within a single study (e.g. comparing different cover crop species vs. a single winter fallow control) were considered distinct within-study observations to assess the effect of moderator variables. In total, there were 17 papers comprising 93 comparisons for cover cropping and 30 papers comprising 131 comparisons for alternative tillage (Appendix S1, Supporting Information). For effects on the AMF community, there were 15 papers comparing alternative vs. conventional tillage comprising 25 comparisons for AMF abundance, 15 for AMF richness and 13 for AMF diversity (Appendix S1). The focus was on species richness (i.e. the total number of species or taxa present, based on spore taxonomy or genetics) and the Shannon Index as a metric of diversity. Since only five papers reported on AMF community composition in studies comparing cover crops, these papers were evaluated qualitatively in the discussion. Data were extracted from tables and figures (using WebPlotDigitizer; Rohatgi 2015) in publications meeting the selection criteria.
We examined several factors commonly reported across the studies as moderators: the type of cash crop, the sampling stage for roots and soil available P. Categories for type of cash crop were based on which cash crops were commonly included in the selected studies. Maize Zea mays L. was the most common crop in both cover cropping and tillage meta-analyses. Other cash crop categories for cover cropping included the next most common crop, soya bean (Glycine max [L.] Merr.) and all other crops. For alternative tillage, other cash crop categories included small grains (e.g. wheat Triticum aestivum L. and oats Avena sativa L.), legumes (e.g. soya bean and common bean Phaseolus vulgaris L.) and all other crops. Sampling stage for roots was based on the phenological stage of the cash crop, including vegetative, flowering and maturity. Where not stated in the journal articles, we estimated these stages by determining the days after planting for each sampling time and matching with crop development timelines from extension resources available close to the study area or in a similar climate. Soil available P (μg P g−1 soil) was a continuous variable measured in several ways across the studies, most commonly as Olsen, Mehlich III and Bray, or the measurement method was not reported.
Other explanatory variables were specific to either cover cropping or alternative tillage. For cover cropping, non-AMF hosts included species in the Brassicaceae family (e.g. rapeseed Brassica napus L. and radish Raphanus sativus L.) and buckwheat Fagopyrum esculentum Moench, which is considered non-mycorrhizal (Wang & Qiu 2006). Functional groups of cover crops included graminoids, legumes and non-legume dicots. The latter were mostly non-AMF hosts but also included AMF hosts sunflower Helianthus annuus L. and dandelion Taraxacum officinale F.H.Wigg. Categories of weed control included whether or not weeds were controlled (by herbicides or mechanical control) in the winter fallow treatment and whether or not cover crops were terminated with any form of tillage (e.g. by herbicides or mowing and mulching).
Alternative tillage categories were based on the level of disturbance, including no-till, non-inversion (e.g. chisel), shallow inversion (e.g. shallow disking) or ridge tillage. The type of conventional tillage was either deep inversion (mouldboard plough, representing the majority of conventional tillage treatments) or shallow inversion (same as above). Soil texture was divided into light (i.e. high silt and sand content) loam and heavy (i.e. high clay content) (NRCS 1993). We also noted whether or not a cover crop was present prior to tillage.
Data Analysis
where 
and 
are, respectively, the treatment (cover crop or alternative tillage) and control (winter fallow or conventional tillage) mean calculated for that observation. On the log scale, an effect size of 0 means no difference and a positive value means that cover cropping or alternative tillage has a positive effect on AMF colonization of cash crop roots. The variance of response ratios was calculated according to Hedges, Gurevitch & Curtis (1999) using the standard error and number of replicates reported for each individual study. Where standard errors were not presented or could not be calculated, the authors were contacted to request the missing data. When no information was obtained, standard deviations were imputed based on the ratio of standard deviations and means (of either control or treatment groups) from studies that reported both (Lajeunesse 2013; Ellington et al. 2015). The median value of this ratio was used to impute standard deviations for trials that reported only means. A sensitivity analysis assessed the effects of these assumptions and found that almost all results were robust (Appendix S1). We note where particular results were sensitive to the imputed standard deviations.
Response ratios were calculated and analysed using the ‘metafor’ package (Viechtbauer 2010) in r (R Development Core Team 2015) using a mixed effects approach. A publication-level random effect allowed us to account for non-independence of multiple within-study observations (Mengersen, Gurevitch & Koricheva 2013). A model was first run without any moderator variables to assess the overall heterogeneity, and each moderator was subsequently tested one by one as a sole covariate. A categorical moderator variable was considered to have a significant effect on the change in AMF colonization of cash crop roots when the omnibus test of all model coefficients (i.e. including all levels of a categorical variable) was significant (P < 0·05) (Viechtbauer 2010). We used funnel plots to confirm there was no evidence of publication bias (Philibert, Loyce & Makowski 2012). All models were fit using restricted maximum likelihood estimation. To facilitate ease of interpretation, mean log response ratios and upper and lower bounds of 95% confidence intervals around the mean were back-transformed (elnR) and expressed as a per cent change relative to the control.
Results
Field studies spanning five continents (all but Africa and Antarctica; Fig. S1) showed strong positive effects of cover cropping and alternative tillage on AMF colonization of cash crop roots. Cover crops increased colonization of summer cash crop roots by 28·5% (95% CI: 12·1–47·4; Fig. 1) relative to winter fallows. Median colonization rates across all observations were 47 and 37% for cover cropping vs. fallow, respectively (Fig. S2). The change in colonization was greater when the cover crop was an AMF host (30·5 vs. 17·4%), but even non-AMF host cover crops (e.g. radish or rape) significantly increased root colonization (95% CI: 2·2–34·8 for non-AMF hosts and 14·1–49·3 for AMF hosts). Legume cover crops had a greater effect on cash crop root colonization than graminoids or non-legume dicots (Fig. 1). Roots of maize and soya beans, the two most common cash crops in the studies, had similarly higher AMF colonization following a cover crop (95% CI: 16·2–62·8 for maize and 16·5–80·5 for soya beans), but this was not apparent for other cash crops, which encompassed a number of different crop species. The sampling stage of cash crop roots, fallow weed control or prior tillage did not affect the change in colonization of cash crop roots following a cover crop (Fig. 1). Soil available P had a marginally significant (P = 0·08) negative, but weak, effect on the magnitude of the effect size (Fig. 2).
Across all observations, alternative tillage increased colonization of cash crop roots by 27·0% (95% CI: 14·4–41·0) relative to conventional tillage (Fig. 3). Median colonization rates across all observations were 38 and 29% for low-intensity vs. conventional tillage, respectively (Fig. S2). The strongest influence on the magnitude of change was the type of alternative tillage. No-till increased colonization by 30·3% (95% CI: 17·3–44·8), which was similar to shallow inversion and ridge tillage but higher than the 11·2% (95% CI: −1·5 to 25·6) change for non-inversion tillage. Maize and small grain cash crops had less of a change in root colonization than legumes or other cash crops (e.g. sorghum, flax or cotton). The presence of a prior cover crop affected how AMF colonization responded to alternative tillage, increasing colonization by 41·5% (95% CI: 24·0–61·5) compared to 23·8% (95% CI: 11·7–37·2) when no cover crop was present. All cover crops grown in field studies comparing tillage treatments were AM legumes (e.g. hairy vetch Vicia villosa Roth). The sampling stage of roots did not affect the change in colonization. Whereas the overall effect of soil texture was not significant, colonization in heavy (i.e. clayey) soils showed no change from alternative tillage, whereas changes occurred in colonization in light and loam textured soils (Fig. 3). The type of conventional tillage did not affect the change in colonization, although only a small number of trials were shallow inversion (Fig. 3).
Richness of AMF taxa increased by 11·3% (95% CI: 1·0–22·6) in alternative tillage regimes compared to conventional tillage (Table 1). A metric of diversity, the Shannon Index, was not significantly different for AMF taxa in alternative vs. conventional tillage regimes. Within the set of studies reporting effects on AMF community composition or diversity, alternative tillage increased AMF abundance by 60·5%, as measured primarily by spore counts in soil, although the response was highly heterogeneous (95% CI: 15·5–123·0).
| Response ratio estimate | 95% CI lower bound | 95% CI upper bound | Number of observations (n) | Number of studies | |
|---|---|---|---|---|---|
| Abundance | 0·4730 | 0·1441 | 0·8019 | 25 | 9 |
| Richness | 0·1072 | 0·0103 | 0·2040 | 15 | 9 |
| Diversity (Shannon) | 0·0256 | −0·0673 | 0·1186 | 13 | 7 |
Discussion
Although it is often stated that reducing soil disturbance and bare fallows increases AM formation (Gosling et al. 2006; de Vries & Bardgett 2012; Schipanski et al. 2014), a lack of a systematic analysis of results across field studies have precluded decisive conclusions about the relative efficacy of these interventions, and the key management and soil factors that moderate their effect. The results of this meta-analysis show that across replicated field studies from five continents, less intensive tillage and winter cover cropping similarly increased AM formation in summer annual cash crop roots by ~30%. These results suggest that farmers could optimize combinations of tillage and cover crops that most enhance AM formation, particularly with no-till systems and legume cover crops. But importantly, cover crops increased AM formation similarly whether tillage was used or not, suggesting that the continuity of root associations with cover crops is at least as important for AM formation as decreasing disturbance. This is a significant finding, especially for agricultural systems that may rely more heavily on services provided by AM, for instance organic management (Gosling et al. 2006) or low-input systems used by most of the world's farmers (Cardoso & Kuyper 2006). In such systems, tillage is often required for weed control and incorporation of organic matter into the soil (Smukler et al. 2008). When a cover crop, especially a legume, is used in these systems, then AM formation in the cash crop apparently can withstand some tillage. Although AMF colonization rates are widely measured, their relationship with actual functions remains unclear (Lekberg & Koide 2005), so future work that uses innovative approaches like non-AM plant mutants (Watts-Williams & Cavagnaro 2015) will be needed to determine decisively how and when these changes are linked to enhancements in ecosystem services like crop productivity.
The 11% increase in richness of AMF taxa in response to alternative tillage suggests that lower intensity soil disturbance creates more niche space in the rhizosphere and root zone that accommodates tillage-sensitive taxa, for example those that rely more on intact root fragments or extraradical mycelia vs. spores for AM formation. Other studies showing that changes in diversity and community composition of AMF are possible (e.g. Ramos-Zapata et al. 2012; Higo et al. 2013; Säle et al. 2015), but not consistent (e.g. Hu et al. 2015; Njeru et al. 2015), with alternative tillage or cover cropping suggest that determining how to manage AMF community composition will be somewhat site-specific, and tailored to farming goals for productivity and environmental quality.
Impacts of Cover Cropping and Alternative Tillage on AMF Colonization
The 28·5% increase in AMF colonization of cash crop roots following a winter cover crop may be a result of increased AMF spore abundance in soil (Lehman et al. 2012; Njeru et al. 2015). Since AMF are obligate biotrophs, they require C resources from roots to grow and reproduce (Smith & Read 2008), which are not available during a fallow period. Reduced AMF colonization in crops grown after long plant-free periods has been associated with poor crop growth and P and zinc deficiencies (Thompson, Clewett & Fiske 2013). The larger effect of cover cropping on colonization rates reported in a previous meta-analysis (90% increase; Lekberg & Koide 2005) may be due to the inclusion of greenhouse experiments in that meta-analysis, which showed a greater positive response than field experiments (Lekberg & Koide 2005). The stronger response of AMF colonization to legume cover crops compared to graminoids or non-legume dicots likely reflects the high mycorrhizal dependency of typical legume cover crops (e.g. V. villosa and Trifolium spp.), which could lead to greater spore production and higher levels of colonization in the cash crop (Galvez et al. 1995; Njeru et al. 2014).
Whereas reductions in AM formation could be expected following a non-AM cover crop species, either as a result of a reduction in AMF populations (similar to a bare fallow) or production of fungal inhibitory compounds like isothiocyanates by Brassicaceae, experimental results have been inconsistent (Gavito & Miller 1998; Pellerin et al. 2007; White & Weil 2010; Koide & Peoples 2012). In this study, the change in AMF colonization in cash crop roots was indeed greater following cover crops that were AMF hosts compared to non-AMF hosts, but there was still a significant increase in colonization following a non-AM cover crop. This may in part be related to the presence of weeds that are AMF hosts in the non-AM cover crop treatment (Njeru et al. 2014) or differences in soil moisture and temperature patterns in cover cropped vs. fallow soils that impact spore viability, for instance. It is also possible that additional organic matter from non-AM cover crops make soil physical properties more conducive to hyphal growth and colonization of subsequent crops (Drew, Murray & Smith 2006).
The 27·0% increase in AMF colonization in alternative tillage regimes compared to conventional tillage reflects the detrimental effects of soil disturbance on AMF hyphal networks (Evans & Miller 1990) and resulting reductions in root colonization (Lekberg & Koide 2005). The stronger response of no-till compared to other forms of alternative tillage was expected since it eliminates below-ground disturbance and thus leaves mycelial networks intact, which form an important component of inoculum potential (Evans & Miller 1990; Kabir 2005). In this meta-analysis, the positive interaction between tillage and cover cropping may have been accentuated because all cover crops included in trials evaluating alternative tillage were legumes.
Soil P availability did not strongly affect the response of AMF colonization of cash crop roots to cover cropping or alternative tillage. Using just the most commonly reported test for available P (Olsen; 34% of studies) also showed no relationship with the change in colonization. Nor was a relationship found between soil available P levels in control treatments and the level of AMF colonization, which may be more affected by soil available P than the response ratio (Bolan, Robson & Barrow 1984). High soil P may reduce the plant growth response from AM more than the rate of colonization (Sorensen, Larsen & Jakobsen 2005), but not necessarily (Köhl, Lukasiewicz & van der Heijden 2016).
Impacts of Cover Cropping and Alternative Tillage on AMF Communities
The slight (11%) increase in AMF species richness in response to alternative tillage, but lack of changes in a diversity index (measured by the Shannon Index), suggests that alternative tillage has relatively small effects on AMF habitats, or arrival of additional taxa is slow after a change in soil disturbance. Taxonomic changes can occur in the absence of changes in AMF diversity or richness (e.g. Jansa et al. 2002, 2003) in response to types of tillage. For instance, Jansa et al. (2003) observed that Scutellospora sp. were absent in maize roots from ploughed or chisel tilled plots but present in no-till plots, while several species in the genus formerly known as Glomus (Krüger et al. 2012) were more prevalent in tilled soils, and Gigaspora sp. were present in all treatments, suggesting differing dependences of these genera on an intact hyphal network for survival and root colonization.
The few field-based studies that examined the effect of fall/winter cover cropping on AMF communities show limited changes in response to cover cropping (Table 2). For instance, Ramos-Zapata et al. (2012) showed on average approximately four more AMF taxa (identified from spores in trap cultures) following a velvetbean Mucuna deeringiana [Bort] Merr. cover crop compared to a non-weeded fallow (10·7 vs. 6·3 species) at the end of a 13-year experiment. Specifically, spores from Acaulospora and Rhizophagus sp. were found only in cover cropped soils. But several other studies (Higo et al. 2014, 2015; Njeru et al. 2015) did not show any changes in AMF richness or diversity in soil or roots following cover crops in multi-year trials. Fallow treatments tended towards AMF genera with larger spores, perhaps indicating greater viability during long fallows, whereas cover crops tended to support greater abundance of some species in the former genus Glomus (Ramos-Zapata et al. 2012; Higo et al. 2013). While cover crops and associated weeds offer a different host environment and more resources compared to fallows, other management practices (e.g. continued tillage) may constrain changes in AMF community composition.
| Cover crops | Cash crop | Effect on AMF species richness? | Effect on AMF species diversity? | Effect on AMF community composition? | Study |
|---|---|---|---|---|---|
| Vicia villosa Roth, Brassica juncea (L.) Coss, Mix | Solanum lycopersicum L. | No | No | No | Njeru et al. (2015) |
| Ttriticum aestivum (L.), Trifolium pretense L., Brassica napus L. | Glycine max (L.) Merr | No | No | No | Higo et al. (2014) |
| T. aestivum | G. max | Yes, T. aestivum > fallow | Yes, T. aestivum > fallow | Yes | Higo et al. (2013) |
| T. aestivum, B. napus | G. max | No | No | Yes | Higo et al. (2015) |
| Mucuna deeringiana (Bort) Merr. | Zea mays L. | Yes, M. deeringiana > fallow | NA | Yes | Ramos-Zapata et al. (2012) |
Management Implications and Conclusions
This meta-analysis shows that cover cropping and reducing soil disturbance are strategies that farmers can use to increase AM formation and potentially alter the AMF community across a wide range of soil types and cash crops. Specifically, combining no-till and legume cover cropping would best increase AMF colonization of cash crop roots, highlighting positive interactions across management practices. But cover cropping even appears to counteract some of the negative impacts of soil disturbance on AM formation. System approaches that combine cover cropping and reduced tillage with other AM-promoting practices like crop diversification and organic management (Oehl et al. 2004; Verbruggen et al. 2010) may offer the most promise for enhancing AM communities, while also increasing soil C storage and nutrient cycling, and reducing nutrient losses and soil erosion (Quemada et al. 2013; McDaniel, Tiemann & Grandy 2014; Schipanski et al. 2014). Fostering indigenous AMF communities through plant choices and soil management could become an essential component of ecological intensification, which relies on such ‘service providing organisms’ to support crop productivity while reducing environmental impacts and external inputs (Bender, Wagg & van der Heijden 2016). Future work that links changes in AMF root colonization and functional diversity with specific ecosystem functions would help optimize agricultural systems for both food production and environmental quality.
Acknowledgements
We sincerely thank the authors of all the field studies that contributed to this meta-analysis. Funding to T.M.B. was provided by a U.S. NSF GROW Fellowship and the University of Adelaide. Funding to T.R.C. was supported by the Australian Research Council (FT120100463) and the University of Adelaide. NIFA support through the California Agricultural Experiment Station (CA-D-LAW-2085-H) is acknowledged.
Data accessibility
Data used for the meta-analysis are archived in Dryad Digital Repository http://doi.org/10.5061/dryad.1756f (Bowles et al. 2016).
