Reader TTS

Carbon dioxide removal is key to climate change mitigation, yet implications of its deployment remain unclear. Recent exponential growth in literature is rapidly filling this gap but makes

the synthesis of the evidence on carbon dioxide removal side effects increasingly challenging. Here we address this issue by mapping this literature and proposing a taxonomy to synthesize

and compare evidence on carbon dioxide removal side effects. The expansive evidence warrants the use of machine learning to systematically select relevant research and provide an inventory

of nearly 400 co-benefits, challenges, and limits. We find rich evidence in Europe but little information for Africa, South America, and Oceania, where large-scale carbon dioxide removal is

nevertheless projected. There is a predominance of articles discussing negative effects compared to positive ones. Starting from the limitations of our analysis and literature gaps, we

provide entry points for future studies that can build on our literature-based taxonomy.

To comply with the Paris Agreement and to limit global warming to 1.5 °C, rapid and deep reductions in gross CO2 emissions need to be complemented by active carbon dioxide removal (CDR) from

the atmosphere1,2,3,4,5,6,7. CDR may contribute to climate change mitigation by accelerating the realization of net-zero CO2 emissions, by offsetting residual emissions, which are often

claimed to be hard to abate, and by eventually achieving net-negative emissions to reverse a potential temporary overshoot of the carbon budget3,8,9 therewith gradually declining warming

towards lower and safer levels10. Mitigation pathways for 1.5 °C warming—be they with no, limited, or high overshoot—therefore typically imply substantial amounts of CDR, although actual

deployment rates vary as a function of policy choices11,12,13.

CDR is continuously gaining attention and importance, partly due to ongoing delay in deep emission reductions but also as more and more net-zero pledges are being put forward. Most of the

currently discussed CDR options, however, are not yet available at the scale required to substantially contribute to climate change mitigation14,15. While this implementation gap for CDR is

already growing, there are also substantial environmental, socio-political, and economic implications arising from the deployment of CDR, which have not yet been sufficiently understood.

Previous literature reviews have identified both benefits and risks of CDR deployment16,17,18,19,20,21,22. However, the CDR literature has been growing exponentially in recent years, which

makes it increasingly challenging to comprehensively track and synthesize evidence on potential co-benefits, challenges, and limits15. In addition, the absence of a taxonomy to categorize

and analyze evidence severely hampers the synthesis and comparability of knowledge.

We address this issue by systematically mapping the recent literature evidence on co-benefits, challenges, and limits for six land-based CDR options: afforestation and reforestation (AR),

bioenergy with carbon capture and storage (BECCS), biochar, direct air capture with carbon capture and storage (DACCS), enhanced weathering (EW), and soil carbon sequestration (SCS). These

options currently dominate the discussion on land-based CDR and are increasingly incorporated in integrated assessment models (IAMs) that are used to inform long-term mitigation

strategies11,12,15,23. In this study, we consider accompanying or consequential effects of CDR deployment as well as phenomena hampering successful CDR deployment—details can be found in the

Supplementary Note 2. These effects can be co-benefits, challenges, or limits of CDR deployment. In the following, we collectively refer to these effects as positive or negative side

effects of CDR deployment.

We first show how the literature evidence on CDR side effects has evolved over time. Based on the recent literature, we present an initial taxonomy of CDR side effects across multiple

categories and aggregation levels to provide a comprehensive overview of the literature evidence. We compare the literature-based effect profiles of the six considered CDR options and

evaluate the available evidence regarding the desirability of the effects identified, that is, whether effects are associated with societal, environmental, and economic benefits or

disbenefits. Ultimately, we explore geographic differences in the literature coverage and point towards potential literature gaps. More detailed information on the study’s approach is

provided under Methods.

We identified 982 peer-reviewed documents discussing side effects for the here-considered CDR options (Fig. 1a). The publication dates of these studies span across the last three decades,

with steep growth in the number of published documents in recent years. More than 50% of these documents have been published since 2018 (Fig. 1b). We find a large variety in study designs

and methodologies across studies, which can be categorized into five different study types. In recent years, the largest study type group was composed of quantitative analyses and modeling

studies, including IAM studies, life cycle assessments (LCAs), and other quantitative approaches to estimate effect sizes across CDR options. A smaller group of studies is focused on

qualitative analyses and conceptual deliberations, often focused on exploring policy implications and theoretical implementation challenges of CDR deployment.

a gives a stylized overview of the peripheral (at least remotely related), identified, relevant, and eventually synthesized literature evidence and how the different evidence components

refer to each other. A detailed overview of the evidence identification, selection, and synthesis process is provided in the Methods section, as well as the Supplementary Note 1. b shows how

the evidence on side effects has evolved over the course of the last three decades and gives an overview of the different study types in the set of relevant studies (n = 982).

Empirical evidence mostly comes from field experiments—often focused on soil-related implications of AR, biochar, and SCS—and partly from survey and interview studies, which often study the

perception and acceptance of potential future CDR deployment. In addition to these original research studies, there is a group of documents composed of reviews and meta-studies, often

focused on an individual side effect or CDR option.

Our literature inventory of CDR side effects in the full-texts selected for synthesis (n = 233) resulted in nearly 400 individual effects, covering a wide range of environmental,

socio-political, technological, and economic aspects. These side effects were structured in a literature-based taxonomy for the six considered CDR options, grouped in three overarching

effect spheres and spanning three levels of effect aggregation (Fig. 2). Each effect category is assigned a unique identifier to make the taxonomy easily operable and broadly applicable

(Supplementary Fig. 2). The literature-based taxonomy is built on the peer-reviewed evidence for the six land-based CDR options considered in this study, including information on unspecific

CDR as a general mitigation concept. Beyond direct side effects, the taxonomy also contains phenomena that impact successful CDR deployment.

The taxonomy shows three levels of effect categories with higher levels of aggregation towards the center of the circular plot. Overarching effect spheres are shown in color, covering

Economic prosperity & well-being, Environmental & human health, and Implementation challenges. An abbreviation list and unique identifiers for all effect categories are provided

(Supplementary Tables 3, 4). This taxonomy is based on the synthesized literature information for AR, BECCS, Biochar, DACCS, EW, SCS, and option-unspecific CDR to provide an overview of

relevant side effects of these CDR options. Not all categories are applicable to all CDR options. The here presented taxonomy reflects the recent literature evidence and is not conclusive,

meaning that categories might evolve as new evidence arises.

Three overarching effect spheres were identified. The largest sphere covers a wide range of effects related to environmental and human health implications of CDR deployment. A second sphere

includes the potential impacts of CDR deployment on economic prosperity and overall societal well-being. A third sphere entails a variety of potential implementation threats and challenges

that may undermine the successful contribution of CDR to climate change mitigation. Generally, the spheres on environmental and human health, as well as on economic prosperity and societal

well-being, primarily cover the direct side effects of actual or hypothesized future CDR deployment, while the sphere on implementation challenges mostly covers threats and barriers to

successful CDR deployment. The literature-based taxonomy consists of 18 effect categories with several subcategory levels. An overview of these effect categories is given in Table 1.

For each included CDR option, we evaluated the option-specific coverage of side effects in the considered literature (Fig. 3). For none of the considered CDR options, the evaluated

literature body provides information on all 18 effect categories. Between one and three effect categories are not available per CDR option, with many more subcategories either missing or not

applicable. In each case, a careful assessment is needed of whether the effect does not occur for a particular CDR option or whether the missing subcategory points to an identified

literature gap. The individual effect profiles and the level of detail of the available information vary considerably between the considered CDR options. Side effects of AR, biochar, and

BECCS are well-covered by the literature, while information on DACCS and EW is more limited in terms of the number of articles and the spectrum of covered effects. The coverage of SCS ends

up somewhere in the middle between the former and the latter group. The literature on option-unspecific CDR primarily deals with effects within the spheres of implementation challenges and

economic prosperity and well-being, with less information on effects within the sphere of environmental and human health.

The layout position of effect categories and spheres is identical to Fig. 2. Red nodes and lines indicate that an effect category was present in the evaluated literature body for the

respective CDR option; Gray nodes and lines indicate that an effect category was not present in the evaluated literature for the respective CDR option.

In addition to structuring the literature and identifying the coverage of individual side effects per CDR option, the analysis of the number of articles and the spectrum of covered effects

per category is important to understand and synthesize the insights. Figure 4 shows, for each category, the number of evaluated articles focusing on positive and/or negative effects

(desirability)—irrespective of effect sizes, significance, or study contexts.

Where available, information on the desirability of effects was directly taken from the respective articles. Otherwise, effect desirability was assigned manually as long as desirability was

unambiguous and not strongly context-dependent. More information on the study’s approach is provided in the Methods section. a shows the share of articles mentioning positive versus negative

effects per effect category and CDR option. b shows the number of articles per mentioned effect desirability. Effect desirability is not aggregated within or across articles. Thus, double

counting of articles is possible if articles mention multiple different effect desirabilities per effect category and CDR option.

AR: Changes to the water cycle, impacts on the flow of nutrients and minerals, biodiversity implications, soil changes, high land demand, and thermal impacts are among the most widely

discussed side effects of AR. Slightly more evidence on negative (n = 9 articles) than positive (n = 6) implications for the water cycle are found, with some articles (n = 6) describing the

effects of unclear impact. The impact desirability on nutrients in soils such as nitrogen, phosphorus, and potassium is mostly unclear due to context-dependency, with several articles

indicating negative effects due to soil nutrient losses. The literature reports both positive and negative impacts on biodiversity with a small set of effects for which desirability is

unclear. For several articles (n = 8), the desirability of effects on soils is unclear due to context-dependency, with a slight dominance of articles on positive (n = 6) compared to negative

(n = 4) effects. Both positive and negative thermal impacts, mostly in terms of albedo changes, are reported for AR and strongly depend on the baseline conditions of the respective

studies24,25. The land use impact of AR is described as negative in nine out of 11 articles.

BECCS: Impacts on land use, water cycle, and energy clearly dominate the considered literature body on BECCS side effects with still a comprehensive set of articles covering most other 15

effect categories—BECCS is the CDR option with the most available literature on side effects (n = 72). Effects on land use and the water cycle documented in the literature are predominately

undesirable, while for energy, there are more studies (n = 17) indicating net energy production potential than studies (n = 9) indicating net energy demand for BECCS26,27,28,29. Many of the

predominately negative effects on biodiversity, food, and yields, as well as nutrients and minerals, are related to the high land demand for bioenergy plantations for BECCS.

Biochar: Implications for food and yield, nutrients and minerals, and general soil conditions are the primarily discussed side effects of biochar soil amendment. Articles mentioning impacts

on food and yield are predominantly positive (n = 20) compared to negative (n = 5) due to observed biochar-related yield increases for various crops30,31. Described implications for

nutrients and minerals appear to be both positive and negative, with many effects for which desirability is unclear. The literature on soil effects of biochar is more positive (n = 12) than

negative (n = 5), but many articles (n = 10) describe effects for which desirability is ambiguous or context-dependent and, therefore, unclear. The benefits of biochar described in the

considered literature are the most manifold compared to the other CDR options in this study.

DACCS: Impacts on the water cycle, energy, acceptance, land use, and nutrients and minerals are the most discussed effect categories for DACCS. Articles on DACCS predominantly discuss

undesirable side effects (19 out of 24 articles). The literature body also holds some information on desirable effects of DACCS deployment compared to respective baselines or counterfactual

scenarios, e.g., reduced pressure on biodiversity or water and land demand32,33, however, the number of articles is comparatively low (n = 4).

EW: Nutrients and minerals, acceptance, pH change, food and yield, and energy are the most discussed side effect categories for EW, while the evaluated literature evidence base on side

effects of EW is the most limited (n = 17) among the considered CDR options. Positive effects of EW include the provision of essential nutrients and minerals such as phosphorus, potassium,

calcium, and magnesium to soils and plants34,35. EW may also reduce soil acidification. The high energy demand for grinding rocks is described in several articles (n = 4) as a main drawback

of this CDR option35,36.

SCS: For SCS, impacts on nutrients and minerals, as well as on food and yield, are the most widely discussed side effects. Changes to general soil conditions, the water cycle, and

biodiversity are also studied in several articles. Both positive and negative effects on the abundance of nutrients and minerals such as nitrogen, potassium, phosphorus, and related

compounds in soils and their leaching are described, with a substantial set of effects for which the desirability is unclear37,38. More articles report benefits (n = 10) than downsides (n = 

5) for food and yield due to observed yield increases for a variety of agricultural products, including maize, soybeans, and tomatoes39,40. Similar to biochar, the positive effects of SCS

predominate for the majority of the effect categories considered, based on the evaluated literature.

Unspecific CDR: The reviewed articles on side effects of option-unspecific or aggregated CDR strongly focus on implications for policy response followed by information on biodiversity

impacts and acceptance. Overall, the side effect information for unspecific CDR is dominated by undesirable effects. An array of different negative policy developments to the availability of

CDR, including reduced and delayed climate action or issues of burden sharing36,41, is described, as already laid out in detail in the previous section.

The evaluation of effect desirability shows comparatively comprehensive literature evidence for positive effects for biochar and SCS across most effect categories. For BECCS, DACCS, and

option-unspecific CDR, available information for the 18 evaluated effect categories appears to be more negative than positive in most cases. The number of articles covering positive versus

negative side effects of AR appears to be balanced—for EW, the evidence on side effects is relatively scarce.

The general and CDR option-specific information on side effects can be complemented by looking at the geographical coverage of the evaluated literature (Fig. 5). There is substantial

research on CDR side effects in Europe (n = 58) despite the continent’s comparatively small size, while few studies on CDR side effects in Africa (n = 13), Oceania (n = 13), and South

America (n = 11) have been identified. The few available studies for these three regions predominately discuss negative aspects—for at least 50% of the considered effect categories, there is

no literature evidence on desirable effects. A large part of the available literature covers a global or multi-regional scope. For all regions considered, there is a predominance of

articles describing negative effects compared to positive ones. Available information on CDR benefits is especially rich for Asia and Europe. For Africa, there is no evidence for DACCS, EW,

and option-unspecific CDR in the evaluated evidence base. No articles studying SCS implications in South America were found. For all other regions, there is information on potential side

effects for all considered CDR options. Interestingly, the evidence on benefits appears to be more constrained to individual regions than the evidence on disbenefits for DACCS, EW, and

option-unspecific CDR.

a shows the number of articles per CDR option and world region across all effect categories. b shows the number of articles per effect category and world region across all CDR options.

“Multiple regions” refers to geographical study scopes covering more than one of the six listed continents—details are provided in the Supplementary Note 2. Studies without information on

geographical scope were not considered in this figure. Double counting of articles is possible if articles mention multiple different effect desirabilities per region and effect category or

CDR option. The blue column in (a, b) on all effect types also contains studies on effects with unclear or neutral desirability. Crossed cells indicate that no information is available in

the evaluated evidence base.

The presented taxonomy of CDR side effects, the comparison of effect profiles for the considered options, and the evaluation of available evidence on benefits versus disbenefits, including

geographic differences, provide a comprehensive overview and map of the diverse and rapidly growing literature evidence on CDR implications. Our inventory of nearly 400 partly interrelated

side effects underlines the multi-layered and complex nature of CDR as a climate change mitigation option. The diversity in CDR effect profiles and the perceived parallel existence of

benefits and disbenefits across several CDR options indicate the potential to optimize climate change mitigation strategies and portfolios to foster advantages and minimize risks42. Our

literature-based taxonomy of CDR side effects can be an initial but seminal tool for future studies to efficiently and comprehensively enhance knowledge on individual aspects of CDR

implications and thus help close remaining research gaps. For policymaking, our map provides an overview of the various aspects that need to be carefully considered in the context of

national and international CDR legislation and regulation. Below, we discuss current limitations as well as entry points for future extensions and opportunities for impactful analyses.

While taking a comprehensive and systematic approach to scan for evidence, the findings presented here are, by design, constrained by the currently available peer-reviewed literature. This

study synthesizes the evidence on CDR side effects that were found and discussed by recent studies across different geographic contexts, suggesting a structured taxonomy for further analysis

and highlighting where evidence is comparatively comprehensive or scarce. However, this reviewed literature might not yet be a complete representation of all potential effects and their

desirability—categories might evolve as new evidence arises.

The search for recent CDR studies was as broad as possible, yet the selected body exclusively consists of articles where side effects of CDR were explicitly framed as such. While for the

refined literature selection and full-text analysis, both explicit and implicit side effects of CDR were considered, the initial literature identification was restricted to articles written

in English and explicitly articulating side effects or phenomena hampering successful CDR deployment in their title, abstract, or keywords (see Methods and Supplementary Table 1 for details

on the applied side effect identification). Evidence that uses a different vocabulary or language to discuss CDR side effects might thus have been missed. Furthermore, only articles that

consider side effects in an explicit CDR context have been considered. Articles that deal with CDR components or options in non-CDR contexts, such as afforestation as a nature-based solution

or bioenergy without CCS as an energy source, were not considered. This more focused approach is meaningful, as many side effects of the considered CDR options are setup-specific. However,

there may be several side effects that have not been discussed in an explicit CDR context but would also occur when a specific CDR option is deployed.

The focus on original research articles published since 2018 allowed for a more in-depth evaluation of the individual full-texts on up-to-date effects; however, it excluded side effect

literature that was published before. To evaluate potential blind spots, the inventory of side effects of this study was compared to the CDR side effects described in other review articles

and the IPCC AR6 report16,17,18,19,20,21,22,43. Potential implications for non-CO2 emissions are mentioned for AR19 and BECCS17,18 in previous reviews, which were not covered in our

literature-based inventory. For biochar, ref. 17. mention policy implications concerning potential competition over biomass resources, which is also not part of our literature body. Apart

from this, we conclude that despite the restriction to articles published after 2018, the vast majority of effect categories per CDR mentioned in other reviews is comprehensively covered

and, in many cases, further detailed by our study. However, modifying the set of considered research articles would likely have an impact on the evaluation of effect desirability in terms of

the distribution between articles discussing positive versus negative effects—especially for CDR options where the available evidence is comparatively scarce.

There are also limitations induced by the assessment of the identified literature. The evaluated articles are very heterogeneous. They differ in terms of methodological approaches,

considered CDR deployment volumes, effect sizes, effect baselines, and CDR intervention contexts with various temporal and geographical scopes. Furthermore, effect desirability is determined

in comparison to an associated baseline or counterfactual scenarios defined within the context and scope of a respective study, which might make wider comparability challenging. Our

approach allowed for an integration of a wide range of different study designs, including IAM modeling, LCAs, small to medium-scale field experiments, surveys, as well as conceptual works

(Fig. 1b). This resulted in a rich evidence base and provided an overall picture of the diverse CDR side effects and their current academic discussion. However, the current approach does not

yet allow for a comparison of the sizes or significance of identified benefits and disbenefits.

The proposed taxonomy and effect profiles of studies per CDR option allow to identify key gaps in the literature. For example, several theoretically relevant side effect categories are not

extensively covered in the evaluated literature for BECCS. These gaps are apparent in direct comparison to AR, for which there is comprehensive information on albedo changes or impacts on

soil composition and quality, while for the bioenergy plantations that are integral to BECCS, this is not the case despite expected effect similarities. Importantly, for bioenergy production

without CCS, the literature body for the two mentioned effect categories is much larger44,45,46, which partly links to limitations discussed in the previous section but also to challenges

that need to be overcome when synthesizing evidence on complex CDR options. Similarly, impacts on non-CO2 greenhouse gas emissions from soils are discussed in several articles on biochar and

SCS, while there is little information in the evidence base of recent literature for AR, BECCS, and EW even though these three options also actively influence soils, as previous reviews

highlighted17,18,19. These differences in literature coverage, despite similarities in the expected effects, indicate potential research gaps. Generally, the here presented effect profiles

per CDR option may support future investigations to evaluate whether missing information on certain effect categories points to their non-existence in the real world or only in the

literature.

Our map identifies clear gaps in geographical coverage. Comparatively few studies focus explicitly on CDR side effects in Africa, South America, and Oceania, while there is substantial

evidence of side effects in Europe, Asia, and on a global or multi-regional basis. This insight is also supported by recent findings on literature coverage on CDR in general15. This

underrepresentation of Africa and South America is critical as these regions are considered essential for CDR deployment in mitigation pathways in IAMs15,23,47 and, therefore, require urgent

further investigation. This is especially the case since the few available studies for these regions mostly highlight negative aspects. Our analysis does not allow to identify a clear

reason for these regions’ underrepresentation.

Interlinkages and overlaps between side effect categories, observed throughout conducting this study, were highlighted in part in Table 1, e.g., for land use and biodiversity or soil

conditions and the water cycle. While beyond the scope of this study, a more systematic analysis of such effect chains and feedbacks could further enhance the understanding of CDR

implications and potentially feed into further extensions of the here presented taxonomy.

The comparison and evaluation of implications of different CDR options would further benefit from systematic reviews of the evidence on effect sizes and has already been started to some

degree16. Our map and taxonomy can provide guidance for determining effect categories and related evidence as entry points for more comprehensive effect quantifications. Several side

effects, such as pH change or nutrient cycling, may have optimal ranges, where benefits may be turned into disbenefits and vice versa, depending on whether effect sizes are within the

optimal range or not. This is highly context and option-setup-dependent and has implications for sustainable CDR potentials, which warrants further analysis.

Ultimately, we hope this evidence map and taxonomy will facilitate more comprehensive and consistent analyses of CDR side effects to ensure an evidence-based integration in mitigation

strategies and CDR portfolios that minimize disbenefits and maximize benefits.

This systematic map of CDR side effects consists of four main methodological steps, namely, literature identification, literature selection, literature coding, and literature synthesis. The

individual steps are detailed below.

Potentially relevant peer-reviewed literature on CDR for this study was systematically identified via the abstract and citation database Scopus, using one keyword-based search query per

considered CDR option and an additional query for unspecific CDR as a general mitigation concept. The developed search queries were partly informed by the queries used in the review and map

by ref. 17, and ref. 48. The CDR queries were combined with additional subqueries to restrict the selection to articles explicitly discussing positive or negative side effects of CDR, as

well as potential threats to CDR deployment. The queries are presented in Supplementary Table 1. By the time the queries were applied (May 20, 2022), 14,704 individual articles were

identified.

Not all literature resulting from the search queries was indeed relevant to this study. For the literature selection, we used a machine-learning-assisted selection process to separate

relevant from irrelevant studies. To ensure consistency and transparency, first, a set of inclusion and exclusion criteria for selecting relevant studies was developed (Supplementary Table 

2).

Based on the inclusion and exclusion criteria, a random sample (n = 1010) of the potentially relevant literature (n = 14,704) was manually labeled as relevant or irrelevant based on titles,

abstracts, and keywords, making use of the NACSOS platform49. The labeled subset was used to train a machine learning relevance classifier, which is made available. The classifier was used

to predict the relevance of the unseen literature. The prediction scores allowed us to sort the remaining literature in descending order of predicted relevance for the further screening

process. The sorted unseen literature was screened by hand and iteratively tested if a recall target of 95% with p