• Open Access

Biochar and its effects on plant productivity and nutrient cycling: a meta-analysis

Authors


Correspondence: Lori A. Biederman, tel. 515-509-6346, fax (515)-204-1337, e-mail: lbied@iastate.edu

Abstract

Biochar is a carbon-rich coproduct resulting from pyrolyzing biomass. When applied to the soil it resists decomposition, effectively sequestering the applied carbon and mitigating anthropogenic CO2 emissions. Other promoted benefits of biochar application to soil include increased plant productivity and reduced nutrient leaching. However, the effects of biochar are variable and it remains unclear if recent enthusiasm can be justified. We evaluate ecosystem responses to biochar application with a meta-analysis of 371 independent studies culled from 114 published manuscripts. We find that despite variability introduced by soil and climate, the addition of biochar to soils resulted, on average, in increased aboveground productivity, crop yield, soil microbial biomass, rhizobia nodulation, plant K tissue concentration, soil phosphorus (P), soil potassium (K), total soil nitrogen (N), and total soil carbon (C) compared with control conditions. Soil pH also tended to increase, becoming less acidic, following the addition of biochar. Variables that showed no significant mean response to biochar included belowground productivity, the ratio of aboveground : belowground biomass, mycorrhizal colonization of roots, plant tissue N, and soil P concentration, and soil inorganic N. Additional analyses found no detectable relationship between the amount of biochar added and aboveground productivity. Our results provide the first quantitative review of the effects of biochar on multiple ecosystem functions and the central tendencies suggest that biochar holds promise in being a win-win-win solution to energy, carbon storage, and ecosystem function. However, biochar's impacts on a fourth component, the downstream nontarget environments, remain unknown and present a critical research gap.

Introduction

Atmospheric CO2 concentration is nearly 400 ppm and continues to rise due to human activities, with consequences to global climate, human health (Lafferty, 2009; Anderson et al., 2012), agriculture (Auffhammer et al., 2012), and biodiversity (Bakkenes et al., 2002; Bellard et al., 2012). Among the many proposed solutions to mitigate this increase in atmospheric carbon is to enhance long-term sequestration of carbon through the application of biochar to the soil (Marris, 2006; Lehmann, 2007b). Biochar is a carbon-rich coproduct resulting from pyrolyzing biomass under high-temperature, low oxygen conditions for biofuel production (Lehmann, 2007a; Laird et al., 2009) and although it is similar to other charcoals, biochar is defined by its intentional application to the soil for environmental applications (Lehmann & Joseph, 2009). It contains highly condensed aromatic structures that resist decomposition in soil and thus can effectively sequester a portion of the applied carbon for decades to centuries (Lehmann et al., 2006; Novak et al., 2010; although see Wardle et al., 2008). One analysis estimates that widespread use of biochar could mitigate up to 12% of current anthropogenic CO2 emissions (Woolf et al., 2010).

In addition to biochar's potential to sequester recently fixed atmospheric carbon, its porous structure, high surface area, and affinity for charged particles (Keech et al., 2005) interact with physical and biological components of the soil (Glaser et al., 2002; Steiner et al., 2008a) and can have cascading effects throughout the ecosystem (Hammes & Schmidt, 2009). Recent reviews have highlighted the benefits of adding biochar to agricultural soils (Glaser et al., 2002; Marris, 2006; Lehmann, 2007b; Warnock et al., 2007). These benefits include the promotion of plant growth (Chan et al., 2008; Asai et al., 2009; Major et al., 2009; Graber et al., 2010; Hossain et al., 2010), the improvement of soil water-holding capacity (Laird et al., 2010b), diminishing disease incidence in crops (Matsubara et al., 2002; Elad et al., 2010; Elmer & Pignatello, 2011), limiting the bioavailability of heavy metals (Park et al., 2011), reducing soil N2O emission (Kammann et al., 2011; Taghizadeh-Toosi et al., 2011), and reducing of nutrient leaching loss, which in turn can reduce fertilizer needs (Liang et al., 2006; Laird et al., 2010a). Because biochar is a coproduct of bioenergy production and can contribute to carbon sequestration goals, while also simultaneously increasing yield and reducing fertilizer use, biochar has been touted as a ‘win-win-win’ solution to meeting global environmental challenges (Laird, 2008).

There is, however, considerable variation in plant and soil responses to biochar that cannot be evaluated in a single study and may be lost in the overall message of a literature review. Source material and pyrolysis conditions introduce significant variation in the structure, nutrient content, pH, and phenolic content of the biochar products (Novak et al., 2009a). Interactions with climate, soil type (texture, chemistry, hydrology) (Tryon, 1948; van Zwieten et al., 2010a), and fertilization status (van Zwieten et al., 2010b; Haefele et al., 2011) can also contribute to uncertainty in how biochar interacts with organisms.

There are also many concerns about the production of biochar and the release of this novel material into the environment. In addition to apprehensions about food prices and potential land-use changes due to its manufacture and transport (Hill et al., 2006; Stoms et al., 2012), it remains unclear if there will be negative externalities associated with the widespread application of biochar. Specifically, there has been limited research on the impacts to nonagricultural species, such as soil organisms and perennial plants that inhabit field margins or other nontarget ecosystems, especially aquatic systems. Effective implementation of biochar as a climate-mitigating tool would require the application of vast quantities of biochar into the environment; exposure of nontarget terrestrial and aquatic systems to biochar is likely as wind and water can translocate up to 53% of applied biochar material during application (Major et al., 2009) and biochar materials preferentially erode from the soil (Rumpel et al., 2006). Because many reports on the benefits of biochar draw on the results of relatively few studies, a quantitative understanding of its potential impacts to ecosystems is needed prior to its adoption as a major climate mitigation tool.

We performed a comprehensive meta-analysis of published studies that tested the effects of biochar on one or more ecosystem functions including plant productivity, nutrient uptake, soil properties, and on ecosystem services, such as crop yield. Our primary question was to ask whether the central tendencies in the published empirical literature supported the often-enthusiastic claims of previous reports (Marris, 2006; Lehmann, 2007a; Kleiner, 2009). Furthermore, because the effects of biochar have been described as analogous to effects of fertilization, we compared the effect of biochar vs. fertilizer and evaluated synergistic interactions (Glaser et al., 2002; Chan et al., 2008; Lau et al., 2008). The potential effects of accidental biochar exposure on nonagricultural species were assessed by comparing the responses of annual and perennial plant species and evaluating the effect of biochar on soil organisms. Finally, because biochar as a product can be chemically highly variable, we examined the ability of particular biochar characteristics arising from feedstock and production methodologies to influence plant growth to inform best use practices.

Materials and methods

Literature search

We conducted an exhaustive literature search in Web of Science (thomsonreuters.com) and Google Scholar (scholar.google.com) databases using the keywords ‘biochar’, ‘char’, ‘black carbon’, ‘charcoal’, and ‘agchar’ (most recent search, June 25, 2012). The literature cited sections of published literature reviews (e.g., Glaser et al., 2002; Atkinson et al., 2010; Spokas et al., 2011) were consulted for additional sources. For each identified article, we evaluated the title and abstract to determine if it contained original data and measured the responses of interest (plant growth, soil nutrients, and soil organisms). Those articles that met these criteria were examined in detail.

Although biochar specifically refers to pyrolyzed biomass that is intentionally applied to the soil for environmental applications (Lehmann & Joseph, 2009), we included charcoal derived from hydrothermal carbonization (Rilling et al., 2010). We excluded studies that examined the response to wildfire deposition and historical charcoal applications, such as those found in ‘Terra Preta’ soils, because the exact nature and quantity of amendment material cannot be known. Those studies that included soil contaminants, tested alleopathic interference, averaged responses among different trials, or did not contain appropriate controls were also excluded. This process identified 114 published articles.

Data extraction

Data from ecosystem variables that were measured in more than five studies were extracted from the identified articles, including plant production and nutrient content, soil organism composition, and nutrient availability. Data concerning gas fluxes and process rates were outside the scope of this analysis. Within each article, data were divided into ‘experiments’ based on differences in soil, biochar material, and/or plant species, resulting in 371 independent experiments (Supplemental table 1). Different application rates of biochar and supplemental nutrient addition (inorganic and organic) were considered variables within an experiment.

During data extraction, we selected treatments and experimental conditions that were most representative of normal field conditions. For example, if soil organisms were manipulated within an experiment only data from nonsterilized soils were used. Other decision rules included extracting only the final data of repeated measurements and nutrient data from the uppermost soil layer. We also had to manipulate data presented to allow for comparisons. In 43% the experiments, the biochar application rate was presented as mass per area, these data were converted to percent volume, assuming a soil bulk density of 1.5 g cm−3 unless otherwise provided. pH values measured with CaCl2 were made comparable with pH measured with distilled water using the formula pH-H2O = 1.65 + (0.86*pH-[CaCl2]) (Augusto et al., 2008). Productivity data from plants grown in a mixed-species community were summed. The ratio of aboveground-to-belowground tissue and total biomass were also calculated in those cases where both above- and belowground data were reported. We also obtained auxiliary information whenever possible, including experimental setting, study length, fertilizer type, soil nutrients and pH, biochar composition, feedstock source, pyrolysis conditions, and activation status. The target organisms’ functional group was noted and if it is plant, we recorded its life span.

We used the natural log-transformed response ratio as a measure of effect size (Hedges et al., 1999; Lajeunesse & Forbes, 2003): RRX = ln (T/C), where T is the measured value of the response variable to treatment X [biochar (B), fertilizer (F), or both (BF)] and C is the value in the untreated soils – the control. For those studies where fertilizer was added to both the control and biochar treatments, the response ratio was modified RRX = ln (BF/F). We used RRBFC and RRBFF to distinguish between the true factorial conditions and those where all treatments receive fertilizer, respectively. With the exception of the analysis of biochar rate (described below), we calculated the mean response to material application prior to calculating RRX when studies included multiple biochar (n = 20) or fertilizer application rates. To ensure that our results were not affected by this decision, we calculated a response ratio using the maximum response and compared it with the mean response scenario and they were not significantly different.

The effect of biochar application rate on aboveground productivity was calculated in a manner similar to RRB, but in this case the response to biochar was calculated for each reported biochar application rate, rather than averaged over all application amounts. The slope of each of the response surfaces was determined.

Analysis

All statistical analyses and graphical presentation were performed in R 2.12.2 and the ggplot2 package (Wickham, 1999; R Development Core Team, 2011). Because the data distributions tended to be slightly skewed, we used nonparametric tests. To determine if biochar significantly affected ecosystem variables, we used Wilcoxon signed rank tests to compare the mean RRB with zero. We used paired Wilcoxon signed rank tests to compare the effect size of biochar with that of fertilizer (RRB vs. RRF). To determine if there was a superadditive effect of applying both biochar and fertilizer we used two tests to determine if: First, for factorial application studies we calculated a test statistic (θ) from the original data [θ = ln ((biochar/control) + (fertilizer/control))] that represents the potential additive effect of adding both materials. We compared θ with RRBFC, or the observed effect of adding both biochar and fertilizer, using a one-sided Wilcoxon rank sum test. Second, for those cases where fertilizer was applied to both control and biochar conditions, we used a Wilcoxon signed rank test to determine if RRBFF was different from zero. We compared biochar's effect on annual and perennial plants by comparing productivity responses (RRB) using Wilcoxon rank sum tests.

Our ability to assess how site conditions interact with biochar was limited by the lack of consistent reporting of biochar chemistry and soil characteristics among studies and it was only possible to analyze those variables with more than 100 studies: aboveground productivity and pH. The mean RRB for biochars of different source materials were compared with anova, and the effects of latitude, pyrolysis temperature, and C : N ratio were determined using regression. We used robust standard errors for regressions using pH to correct for heteroskedastity.

Many literature reviews of biochar effects, such as Lehman et al. (2011), Atkinson et al. (2010), and Spokas et al. (2011), include activated charcoal within their broader definition of biochar. We compared the response ratios (RRB) of our biochar data with response ratios from studies that used activated charcoals (Ridenour & Callaway, 2001; Kulmatiski & Beard, 2006; Chan et al., 2008; Lau et al., 2008; Weißhuhn et al., 2009; Wurst et al., 2010; Hale et al., 2011; Hass et al., 2012; Rajkovich et al., 2012), and used Wilcoxon rank sum tests to determine if the responses of these two products are similar (Supplemental table 2).

To evaluate the potential of bias introduced from unpublished data we estimated the hypothetical number of unreported zero effect studies needed to produce a nonsignificant overall effect (Rosenthal, 1979; Harpole et al., 2007). This ‘fail-safe’ number is calculated as X = (k/2.706) [k(math formula)2 – 2.706]. Where k is the number of studies that measured the variable of interest, and math formula is the mean of the normalized standard deviations of the k studies. We excluded studies that did not report the significance of their data. A conservative estimate of math formula = 1.645 can be used for studies with significant effects (P < 0.05) and math formula = 0 for studies reporting nonsignificant responses to biochar (Supplementary table 3).

Results

The addition of biochar to soils resulted in increased aboveground productivity (P < 0.01), crop yield (P <0.01), SMB (P < 0.01), rhizobia nodulation (P < 0.05), plant K tissue concentration (P < 0.05), soil P (P < 0.001), soil K (P < 0.001), total soil N (P < 0.001), and total soil C (P < 0.001), on average, compared with control conditions (Fig. 1; RRB > 0). Also, on average, soil pH increased, or became less acidic, with biochar addition (P < 0.001). Those variables that showed no significant mean effect of biochar included belowground productivity, aboveground : belowground biomass ratio, percent mycorrhizal colonization of roots, plant tissue N and P concentration, and soil inorganic N.

Figure 1.

The relative effect size (mean ± CI) of biochar treatments (RRB) on a range of ecosystem variables. Significance of Wilcoxon signed rank tests: *P < 0.05, **P < 0.01, ***P < 0.001.

The effect size of aboveground productivity did not change as biochar application rate increased (Fig. 2). However, the variation around the mean response at each application rate, as measured by standard error, increased with application rates greater than 0.5%. In the 20 studies that reported multiple biochar application rates the overall mean (±CI) response slope was 0.002 ± 0.181, indicating no clear relationship between productivity and biochar application rate. Of those 20 studies, 8 had negative slopes (<−0.02), meaning the addition of more biochar reduced productivity, 10 had positive slopes (>0.02) meaning increasing biochar application increased productivity, and 2 studies had flat slopes (−0.02 to 0.02) suggesting no effect of increasing biochar application rate.

Figure 2.

The relative effect size (mean ± CI) of biochar treatments (RRB) for aboveground productivity at different log-transformed biochar application rates (equivalent percent application rates on the inside of x axis). Colors represent different experiments. Experiments with multiple application rates (n = 20) are connected with solid lines. Inset: the standard errors for the mean rates for studies with multiple application rates.

The average effect size of fertilization was significantly greater than that of biochar addition for aboveground productivity (P < 0.01), yield (P < 0.05), and soil P (P < 0.01) when compared with control conditions (Table 1 and Fig. 1). Biochar addition, however, was more effective than fertilization at increasing plant tissue P (P < 0.05) and K (P < 0.05) concentration. For all variables except soil total C (P < 0.001), the addition of both biochar and fertilizer (RRBFC) was not significantly different than the additive expectation (image). In those studies where fertilizer was added to all treatments (RRBFF), aboveground productivity (P < 0.01), crop yield (P < 0.001), plant tissue K concentration (P < 0.01), pH (P < 0.005), and soil total C (P < 0.01) increased significantly with biochar.

Table 1. The mean (±CI) effect size for the fertilizer (RRF) and fertilizer plus biochar treatments (RRBFC and RRBFF). Bold values of RRF were significantly different than the effect size of RRB. Bold values of RRBFC indicate a significant difference than θ, a test statistic that represents the potential additive effect of biochar and fertilizer. For those cases where fertilizer was applied to both the control and biochar treatment (RRBFF), bold values indicate a significant difference from zero
 RRFRRBFCθRRBFF
Aboveground 0.65 ± 0.23 0.96 ± 0.29 1.22 0.18 ± 0.1
Belowground0.64 ± 0.741.38 ± 1.41.480.23 ± 0.18
Above : below ratio0.41 ± 0.70.56 ± 0.690.950.03 ± 0.18
Yield 0.69 ± 0.35 0.83 ± 0.37 1.18 0.35 ± 0.16
SMB0.31 ± 0.450.48 ± 0.620.920.05 ± 0.52
Rhizobia nodules0.21 ± 0.10.55 ± 0.330.89Insufficient data
% Mycorrhizae col.−0.05 ± 0.25 0.21 ± 0.14 0.780.22 ± 0.18
Tissue N conc.0.25 ± 0.17 0.23 ± 0.17 0.840.06 ± 0.16
Tissue P conc.0.05 ± 0.07 0.08 ± 0.15 0.760.09 ± 0.12
Tissue K conc.0.04 ± 0.08 0.39 ± 0.25 0.76 0.08 ± 0.08
Soil inorganic N3.83 ± 1.03 0.55 ± 1.16 1.060.00 ± 0.19
Soil P 1.43 ± 0.73 1.66 ± 0.73 2.160.21 ± 0.41
Soil K0.23 ± 0.2 0.66 ± 0.37 1.15 0.51 ± 0.33
pH0.002 ± 0.03 0.11 ± 0.04 0.76 0.08 ± 0.03
Soil total N0.04 ± 0.08 0.33 ± 0.12 0.890.13 ± 0.16
Soil total C0.11 ± 0.08 1.27 ± 0.27 0.93 0.94 ± 0.51

Belowground biomass in perennial plants did not change in biochar treatments, which was significantly (P < 0.01) different than the belowground response by annual plants (Fig. 3). This difference contributed to a significantly smaller total productivity for perennial plants in biochar-treated soils (P < 0.001). Perennial and annual plants did not differ in aboveground productivity and the ratio of aboveground : belowground tissue in response to biochar.

Figure 3.

The relative effect size (mean ± CI) of biochar treatments (RRB) for annual and perennial plant productivity. Significance of Wilcoxon signed rank tests: *P < 0.05, **P < 0.01, ***P < 0.001.

Aboveground productivity RRB varied significantly with latitude (adj. R2 = 0.095, P < 0.01); the effect size of biochar was more positive in the tropical regions than in temperate zones (Fig. 4). There was also a significant effect (P < 0.01) of feedstock source of the biochar material on aboveground productivity: grass- and manure-origin biochars increased productivity compared with control conditions (Fig. 5a) and forb-origin biochar reduced productivity, although there were limited data for this class of biochar. The C : N ratio of the biochar source had no predictable effect on productivity (adj. R2 = 0, P = 0.96) (Fig. 5b). Higher pyrolysis temperature biochars produced greater effects (adj. R2 = 0.076, P < 0.05) (Fig. 5c) and the pH of the biochar product significantly influenced aboveground productivity (adj. R2 = 0.172, P < 0.01), with alkaline biochars having a more pronounced positive effect than acidic biochars (Fig. 5d). The change in soil pH following the application of biochar is a function of both the initial soil pH (adj. R2 = 0.069, P < 0.01) and the pH of the biochar (adj. R2 = 0.139, P < 0.001). Acidic soils showed a greater positive response to biochar than alkaline soils. The effect of biochar on soil pH was greater for alkaline biochars. Both biochar and activated charcoal affected the ecosystem variables similarly. The average length of the studies analyzed was 113.4 days and the longest study was 3 years.

Figure 4.

The relative effect size of biochar treatments (RRB) for aboveground productivity as a function of latitude. (Adj. R2 = 0.079, P < 0.01).

Figure 5.

The relative effect size of biochar treatments (RRB) for aboveground productivity as a function of (a) source material (P < 0.01), (b) pyrolysis temperature (Adj. R2 = 0, P = 0.9), (c) biochar pH (Adj. R2 = 0.059, P < 0.05), and (d) C : N ratio (Adj. R2 = 0.172, P < 0.01).

For most variables, the number of hypothetical studies necessary to overturn the results of the meta-analysis was sufficiently large that our results are unlikely to be biased by unreported null responses (Supplemental table 3). For example, at least 1172 unreported null-response studies would be needed to cause the overall mean effect of biochar on aboveground productivity to be nonsignificant. For SMB, rhizobia nodules, and tissue nutrient concentration, however, the number of hypothetical unreported nonsignificant studies that would cause an overall nonsignificant mean effect would be less than 25.

Discussion

In our meta-analysis of 371 independent experiments we find that despite variability introduced by soil type, climate, and production methods, the average effect of biochar on various ecosystem properties was neutral to positive. This is consistent with previous reviews reporting the beneficial aspects of this product (Lehmann et al., 2006; Atkinson et al., 2010). Most importantly for agricultural systems, aboveground production and yield were increased in biochar-treated soils.

Biochar has been shown to promote plant productivity and yield though several mechanisms. Physical conditions change with biochar; its dark color alters thermal dynamics and facilitates rapid germination, allowing more time for growth compared with controls (Genesio et al., 2012). Biochar can also improve soil water-holding capacity (Laird et al., 2010b), facilitating biomass gain (Kammann et al., 2011). Plant growth can also be affected by biochar-induced changes in soil nutrient conditions, particularly the cycling of P and K (Dempster et al., 2012a,b; Taghizadeh-Toosi et al., 2012). In this analysis, we found that plant tissue K concentration and soil P and K increased following biochar application (Fig. 1). These nutrients may be directly introduced to the soil through labile organic compounds associated with biochar and become available as these compounds weather (Topoliantz & Ponge, 2005; Yamato et al., 2006; Rajkovich et al., 2012). This effect, however, depends on the production variables of the biochar (Hass et al., 2012), and is short lived, as the nutrients are used by plants and/or are leached from the soil (Steiner et al., 2007; Major et al., 2010). Long-term effects of biochar on nutrients occur through complex physiochemical reactions with soil particles (Spokas et al., 2011). One such reaction that affects P is biochar-induced increases in soil alkalinity (liming). In acidic soils phosphorus can be adsorbed onto iron oxides, which makes it unavailable to plants. Liming agents reduce the concentration of iron and aluminum in the soil solution and the previously bound phosphorus then becomes available (Cui et al., 2011). Liming also reduces the mobility of toxic elements, such as Al and Cd (Ritchey & Snuffer, 2002; Hass et al., 2012). Changes in plant productivity with biochar pH support this mechanism; alkaline biochars are more effective at increasing biomass than acidic biochars (Fig. 5d). Biochar was also most effective at changing soil pH in acidic soils, which would be particularly beneficial in low latitudes (Fig. 4) where soils are acidic and agriculture is limited by P availability (Steiner et al., 2008b).

Another way biochar may affect soil nutrients is through the reduction in leaching losses (Laird et al., 2010a). Biochar's porous structure, large surface area, and negative surface charge (Bird et al., 2008; Cheng et al., 2008; Downie et al., 2009; Novak et al., 2009a) increase the soil's cation exchange capacity and allow for the retention of nutrients, such as K (Liang et al., 2006; Major et al., 2011). Biochar can also slow cation loss by inducing a shift in soil water nutrient transport from bypass to matrix flow (Laird et al., 2010a). Nutrients, such as P, can be adsorbed to biochar's surface, which slows leaching (Laird et al., 2010a; Beck et al., 2011).

The analysis of application rate provided little insight into how biochar maybe best applied; there was no obvious threshold or trend with increasing application rates (Fig. 2). Variability in plant response, however, increased at with application rates (Fig. 2 inset). Other studies have found application thresholds, beyond which growth was reduced: Rajkovich et al. (2012) found a general application threshold of 2% (26 t ha−1); Kammann et al. (2011) found that quinoa growth was retarded at 100–200 t ha−1 (6.7–13.3%); and Baronti et al. (2010) observed a threshold of 10 t ha−1 (0.7%) for durum wheat.

Biochar–fertilizer interactions

Although biochar altered the soil nutrient environment and promoted plant growth, it is effects are not equivalent to that of fertilizer, as fertilizer alone was more effective for improving plant productivity and soil P (Table 1). Furthermore, unlike in the fertilizer treatments, plant allocation patterns did not change in biochar treatments. Plants typically respond to improved nutrient conditions by reducing the allocation of tissue belowground (Brouwer, 1962; Poorter & Nagel, 2000), and a lack of change following biochar exposure suggests that it is not similarly alleviating belowground competition for nutrients. In contrast, biochar amendments were better than fertilizer at increasing plant P and K tissue concentrations. As explained above, biochar can improve the availability of these nutrients through soil liming and by reducing leaching losses. Despite the differential effects of these materials, however, there was limited evidence of a superadditive or synergistic effect when both biochar and fertilizer are applied.

Our analysis finds that biochar's effects on N are limited, as both soil available N and the concentration of N in plant tissues were unaffected by its application. Furthermore, aboveground productivity did not change with the nitrogen content of the biochar, as measured by C : N ratio. This contradicts studies that find complex interactions between fertilizer, biochar, and the N cycle (DeLuca et al., 2006; Laird et al., 2010a; Taghizadeh-Toosi et al., 2012). Soil type and existing soil N status are strong controls on soil N cycling and because this meta-analysis generalizes across a wide range of soils these effects may be masked. There were, however, increases in total soil N, presumably because the N within the structure of the biochar material contributed to this pool, but is unavailable to plants and microbes.

Effects on perennial species and soil organisms

Belowground, annual plants responded positively to biochar, whereas perennial species (including native and naturalized grasses and forbs, forage crops, and sugar cane) had no response and the difference between these two life forms was significant (Fig. 3). Rather than producing differences in tissue allocation, biochar affects overall plant productivity; increasing it for annual plants and with limited effect on perennials. There was also considerable variability in plant response, especially for annual plants belowground. Numerous volatile and biologically active compounds, such as ethylene, butyric acids, benzoic acid, quinones, and 2-phenoxyethanol, are introduced into the soil with biochar amendment (Graber et al., 2010; Spokas et al., 2010) and depending on concentration, these compounds may promote growth, or produce toxic effects (Keely & Pizzorno, 1986; Elad et al., 2010; Meller Harel et al., 2012). Differences in allocation and phenology, such as higher growth rates in annual species (Pitelka, 1977), may contribute to functional group sensitivity to these compounds. Documented changes following historical charcoal applications (not included in the analysis) underscore biochar's potential effect on plant community composition (Chidumayo, 1988). For example, annual weed cover and legume density was increased, and perennial sprouting reduced, on Terra Preta soils compared with adjacent nonaffected areas (Major et al., 2005). Plant species composition is also affected by the alkaline soils and altered calcium concentrations found in former charcoal production areas (Mikan & Abrams, 1995; Young et al., 1996).

Biochar has variable effects on plant-associated soil microbes. Root nodulation by rhizobia generally increased (Fig. 1), presumably because conditions associated with efficient N-fixation, such as slightly alkaline soil and access to P, have improved (Graham, 1981; Rondon et al., 2007; Lehman et al., 2011). In contrast, biochar did not significantly alter root colonization by mycorrhizal fungi, although the wide confidence intervals suggest that there is considerable variability. Biochar changes the plant's nutrient environment, increasing P availability for example, and this may reduce plant dependence on mycorrhizae (Raznikiewicz et al., 1994). Biochar could also affect the adsorption and desorption of signaling compounds that would otherwise promote root–fungi connections (Akiyama et al., 2005; Warnock et al., 2007).

Biochar application increased total soil C, thus it contributes to the sequestration of carbon at least in the short term (3 years) (Fig. 1). Although much of this carbon sequestration is due to the inert portions of the biochar material, there was also an increase in SMB, another soil carbon pool. Biochar can contribute to SMB through various mechanisms. It augments the availability of micropore habitat providing refugia for soil microbes from larger fauna, thus increasing microbe population size (Zackrisson et al., 1996; Pietikäinen et al., 2000). Labile organic compounds, a by-product of the production process, are introduced to the soil with biochar and decompose readily. Microbial food resources are also enhanced by the retention of native dissolved organic matter on the charged surface of biochar (Steiner et al., 2008a; Deenik et al., 2010) and through biochar-induced increases in plant productivity (Graber et al., 2010; Jones et al., 2012). It is also possible that biochar addition could increase microbe populations by ‘priming’ the decomposition of native soil carbon, leading to a net loss of carbon to the atmosphere (Wardle et al., 2008; Rogovska et al., 2011). Recent studies, however, have found priming to be negligible in soils that are unaffected by phenolic compounds (Bell & Worrall, 2011; Jones et al., 2011; Zimmerman et al., 2011).

Biochar characteristics

The net ability of biochar to enhance ecosystem services depends, in part, on the specific qualities of the biochar and site conditions (Haefele et al., 2011). Within the literature surveyed there was little uniformity on the production methods of biochar; materials were created in a range of pyrolyzers from highly sophisticated industrial equipment to primitive earthen mounds. Temperature conditions also varied, and there was no precise definition of terminology, such as the time and temperature conditions of ‘fast’ or ‘slow’ pyrolysis processes (Mohan et al., 2006). Furthermore, the reporting of production (feedstock source, time and temperature of pyrolysis, kiln type) variables and the resulting qualities (pH, C : N ratio, and nutrient content) of the biochars was inconsistent and there were few data available to correlate with the variables of interest. Because pyrolysis conditions affect the behavior of biochar in the soil (Bruun et al., 2012), the lack of data limits our ability to understand biochar's interaction with soil and its organisms (Spokas et al., 2011).

There was considerable variation in how feedstock source influences aboveground productivity (Fig. 5); biochars from grass and manure/sewage increased aboveground productivity. This was unexpected given the reputation of grass- and manure-sourced biochars for producing unpredictable effects due to high concentrations of silicates (Lehman et al., 2011). We also found that biochars produced at higher temperatures were more effective at promoting aboveground productivity. High-temperature biochars tend to be alkaline (Bagreev et al., 2001; Novak et al., 2009b) and contain less biologically active volatile compounds (Gundale & DeLuca, 2006; Hale et al., 2012) that can otherwise limit plant growth. High-temperature biochars are also more resistant to decomposition and would, therefore, be better candidates to fulfill the C sequestration function (Novak et al., 2010; Harvey et al., 2012).

Future research needs and environmental concerns

We found that the addition of biochar generally improves, or at least does not harm, many aspects of the ecosystem and its functioning, including plant productivity and soil nutrient content. This is consistent with the findings of other nonquantitative reviews (Glaser et al., 2002; Marris, 2006; Lehmann, 2007b; Warnock et al., 2007). However, to achieve meaningful goals for carbon sequestration, such as 12% of current anthropogenic CO2 emissions (Woolf et al., 2010), large quantities of biochar would have to be applied to a significant portion of the earth's arable land. Because of the ability of applied biochar to be transported by wind and water, nontarget organisms will be affected by this activity, but there are limited data available for biochar's effects on nonagricultural species, including native plant communities, aquatic systems, and soil organisms. For example, only 16 studies have tested biochar effects on perennial plants and relatively few studies are needed to overturn the positive findings for rhizobia and SMB. Changes in community composition following biochar application are also very limited and the different responses by annual and perennial plants demonstrate that we still do not understand the mechanism by which it interacts with organisms. Perennial plants and soil creatures perform many critical ecosystem services, such as erosion prevention and pest predation, in agricultural systems. It is imperative that we understand how biochar interacts with all aspects of the environment prior to its widespread application.

Acknowledgments

We wish to thank the Leopold Center for Sustainable Agriculture for funding this project, Joseph Garrison, Erich Sneller, and Emily Zimmerman for assistance, and Dean Adams and David Laird for technical advice.

Ancillary