Toward Rigorous Telecoupling Causal Attribution: A Systematic Review and Typology

Round 1

Reviewer 1 Report

The authors of this manuscript review the literature on causal attribution in the telecoupling literature to identify those approaches and methods that allow for a sufficiently rigorous analysis. They find, among others, that the bulk of the reviewed literature focusses mainly on descriptive research methods, not sufficiently making use of the broad range of more rigorous methods. To facilitate the identification of such, in particular technical, approaches and the process of rigorous causal attribution, they suggest employing a standardized terminology and typology.

The general claim and the objective of the article seems reasonable, and the proposed terminology and typology developed, and the best practice examples presented, by the authors certainly helpful to support rigorous analysis of causal attribution in telecoupled systems. They acknowledge that challenge that more researchers in the field of telecoupled systems need to be made aware and/or trained in these rigorous methods. However, in my opinion, they highlight no strong enough – if at all – that descriptive, qualitative methods are similarly important in that regard since they provide valuable contextual information without which some of those ‘robust’ methods would be rather meaningless. Thus, finding the appropriate mix of methods complementing each other should be part of the recommendations, too.

-       Line 98: INUS not yet introduced in text.

Author Response

Reviewer 1

Point 1: The general claim and the objective of the article seems reasonable, and the proposed terminology and typology developed, and the best practice examples presented, by the authors certainly helpful to support rigorous analysis of causal attribution in telecoupled systems. They acknowledge that challenge that more researchers in the field of telecoupled systems need to be made aware and/or trained in these rigorous methods. However, in my opinion, they highlight no strong enough – if at all – that descriptive, qualitative methods are similarly important in that regard since they provide valuable contextual information without which some of those ‘robust’ methods would be rather meaningless. Thus, finding the appropriate mix of methods complementing each other should be part of the recommendations, too.

Response 1: The reviewer raises a good point regarding the need to describe benefits of qualitative methods in addition to quantitative methods. As such, we revised the text so that “rigorous causal attribution” is defined as the combination of qualitative and quantitative methods to triangulate, broaden, and deepen evidence for causes and their associated mechanisms and effects (lines 90-91). We also improved many other areas of the text (where quantitative methods were formerly emphasized; e.g., lines 33-34, 94, 488, 494) by adding details about qualitative methods and the importance of combining them with quantitative methods to facilitate rigorous causal assessments. Ultimately, we recommend that future researchers use qualitative methods in combination with quantitative methods to provide important contextual information regarding causality in telecoupled systems (as mentioned by the Reviewer).

Point 2: Line 98: INUS not yet introduced in text.

Response 2: We clarified the meaning of “INUS” by adding a parenthetical definition (Insufficient, Necessary, Unnecessary, Sufficient) immediately after the first use of “INUS” in line 75.

Reviewer 2 Report

This paper analyzed the recent multi-disciplinary literature using the telecoupling framework. The paper focused on the challenges and future tasks by summarizing the literature and providing detailed examples, and how those examples can be developed in the future.

The paper is generally well written, and is extensively focused on the case study. I believe the paper is suitable for the journal, but some attention to details throughout the paper would help move the paper to next level, and help make a contribution to literature for those not familiar in the telecoupling framework. I added more substantial comments as below:

1)      The logical flow

·         Since authors focused/mentioned the importance of causal attribution in the latter part of the introduction, the section “what is causality” would be better to fit it in the introduction; rather than a separate section.

·         Line 115: the first sentence can be applied to any research design. Please rephrase it to fit “
telecoupling” research.

·         Line 202: while the authors argued that the RCA should be conducted, there is no clear reason and evidence based on the literature

·         Section 5—improving causal assessment— does not flow well with other section. While I understand the importance of terminology, there is no clear connection between section 4 and section 5. Please provide evidence of missing causality inferred by misuse (or inconsistency) of terminology in section 4.

2)      Use of literature reviews

·         Even if the authors reviewed 89 papers, the analysis of papers is not fully delivered. While the authors pointed out that mistreatment and lack of causal attributions on the previous research, that statement was already given in the introduction. If the authors’ finding is stemmed from misuse (or inconsistent use) of terminology, please provide evidence.

·         To address challenges and future tasks, section 6 should be revisited. Instead of, using self-cited papers, I would rather recommend providing how some of the previous papers treat causal attributions and causality as some suggestions. Otherwise, the entire literature review does not tie together with this paper.

Author Response

Reviewer 2

Point 1: The paper is generally well written, and is extensively focused on the case study. I believe the paper is suitable for the journal, but some attention to details throughout the paper would help move the paper to next level, and help make a contribution to literature for those not familiar in the telecoupling framework. I added more substantial comments as below ... The logical flow: Since authors focused/mentioned the importance of causal attribution in the latter part of the introduction, the section “what is causality” would be better to fit it in the introduction; rather than a separate section.

Response 1: As suggested, we moved the “What is causality?” section to the Introduction so that our description of causality, causal effects, causal mechanisms, etc., logically precedes our discussion of the importance of rigorous causal attribution (qualitative and quantitative).

Point 2: Line 115: the first sentence can be applied to any research design. Please rephrase it to fit “telecoupling” research.

Response 2: We rephrased this sentence so that it explicitly references telecoupling research and telecoupled systems. The revised sentence (along with a new follow-up sentence) describes the importance of using existing information about telecouplings (e.g., descriptive, correlational, quasi-experimental) to establish qualitative and quantitative linkages between telecoupled systems and research goals and analyses. These revisions narrow the focus of the original sentence (and the entire paragraph) to telecoupling research, as the Reviewer suggested.

Point 3: Line 202: while the authors argued that the RCA should be conducted, there is no clear reason and evidence based on the literature.

Response 3: Although we recognize that the original text suggested otherwise, our purpose was not to advocate for using Root-Cause Analysis (RCA) in particular. Instead, we intended to highlight the importance of combining qualitative and quantitative methods to evaluate both the proximate and underlying cause(s) of telecouplings, for which RCA is useful. Hence, we mentioned RCA to provide an example method, rather than to suggest it is the only (or most) effective approach. We rewrote and reorganized the text so that our purpose is clear. The paragraph now emphasizes the importance of conducting mixed qualitative-quantitative analyses (as suggested by Reviewer 1), mentioning RCA only as an example method (rather than the focus of the paragraph).

Point 4: Section 5—improving causal assessment— does not flow well with other section. While I understand the importance of terminology, there is no clear connection between section 4 and section 5. Please provide evidence of missing causality inferred by misuse (or inconsistency) of terminology in section 4.

Response 4: We agree that the reasoning for including a terminology section needs to be clarified. However, “evidence of missing causality inferred by misuse (or inconsistency) of terminology” (Reviewer’s comment) was not the reason for our terminology section. In fact, many telecoupling studies to date accurately and consistently discuss causality in terms of “cause” and “effect” (both explicitly defined and used in the telecoupling framework). The problem is that many previous studies do not consider the broader array of cause-related concepts discussed in our terminology section (e.g., causal effect vs. causal mechanism, proximate vs. underlying cause, causal chain). Hence, it is not a matter of previous studies misusing or inconsistently using cause-related terms. Instead, the need for a terminology section reflects: 1) the overall complexity of telecoupling causality that is generally not acknowledged in existing “cause and effect” telecoupling studies; and 2) an overall lack of rigorous qualitative-quantitative causality assessments (and an abundance of descriptive studies) in previous literature, for which a consistent conceptual foundation (i.e., our terminology) is needed. We revised the text (start of section 4, formerly section 5) so that these reasons are clearly described. Consistent terminology will also promote effective communication regarding telecoupling causality within and beyond the scientific community, as we discuss in the revised section 4.

Point 5: Use of literature reviews: Even if the authors reviewed 89 papers, the analysis of papers is not fully delivered. While the authors pointed out that mistreatment and lack of causal attributions on [sic] the previous research, that statement was already given in the introduction. If the authors’ finding is stemmed from misuse (or inconsistent use) of terminology, please provide evidence.

Response 5: We appreciate this comment (and made changes to address it), but we find it somewhat difficult to follow. First, it is unclear what “the analysis of papers is not fully delivered” means, as our text and figures encompass all 89 papers, each of which is described in detail in our supplementary materials. Second, the Reviewer states that we discuss “mistreatment [of causality] and lack of causal attributions” in previous telecoupling research in our Introduction. However, we thoroughly reviewed this section of the paper and found no such discussion. Rather, our Introduction emphasizes the need to fill knowledge gaps regarding: 1) if and how previous researchers have used telecoupling causality concepts beyond simple “cause” and “effect” (lines 100-101); and 2) the specific qualitative/quantitative methods previous researchers have used to identify causes (lines 102-105). Our description of knowledge gaps and their importance for management/governance of telecoupled systems is very different from making the judgment that there is “mistreatment [of causality] and lack of causal attributions” in previous literature. We agree with the Reviewer that the latter statement (reflecting results/conclusions) would be inappropriate for the first section of the manuscript, which explains why we did not include it in the Introduction.  

Overall, we found this comment helpful for refining certain aspects of our manuscript. For instance, our revised objectives now clarify the purpose of our literature review (i.e., to survey the qualitative and quantitative methods used by previous telecoupling researchers, rather than causal terminology per se) and help address the Reviewer’s comment regarding “full delivery.” Please also see our supplementary materials, which provide detailed information about each paper and thereby illustrate the content of (and reasons for) our literature review. We also clarified the relationship between our literature review and terminology section (lines 322-329) and specified reasons for including the latter (e.g., complexity of telecoupling causality, scarcity of rigorous qualitative-quantitative causality research, need for consistent communication; see response to Reviewer 2, Point 4 above).

Point 6: To address challenges and future tasks, section 6 should be revisited. Instead of, using self-cited papers, I would rather recommend providing how [sic] some of the previous papers treat causal attributions and causality as some suggestions. Otherwise, the entire literature review does not tie together with this paper.

Response 6: We appreciate this comment and changed section 5 (formerly section 6) to address it. The objective of our typology is to illustrate different types of telecoupling causes and the utility of classifying them according to a typology. This differs from the Reviewer’s suggestion to describe “how previous papers treat causal attributions and causality,” which we already did in our literature review and associated results (see section 3). Achieving our objective for the typology section is most effective and clearest (from a writing perspective, and thus from the reader’s viewpoint) for systems we are familiar with (i.e., have published on). Hence, our original self-citations (most of them are now replaced with other papers, see below) reflected our desire to clearly describe telecoupling causes for the benefit of readers, rather than a desire to gain more self-citations. However, we changed our typology section so that examples of causes are derived from a broader array of papers. For instance, in the “5.1 Sector-based causes” and “5.2 Origin-based causes” sections, we replaced Peruvian anchoveta examples with those derived from the introduction of Chinook salmon and coho salmon into the Laurentian Great Lakes. Salmon examples were not derived from papers published by authors of the present study, yet they illuminated the utility of classifying causes into multiple typological categories, the ultimate purpose of our typology section. After making revisions such as those described above, the majority of papers we now cite in our causal typology are written by people who did not author the present study, as requested by the Reviewer. 

Reviewer 3 Report

 This review is interesting which focuses on a new concept--telecoupling dealing with complex human-natural interactions. It would be of interest to some interdisciplinary researchers. However, the authors will need to address some issues below.

What are differences between “casual attribution” in telecoupling study and casual-chain dynamics in complex system studies? Some thoughts of this are needed.

The review of literature is a bit shallow in its current state. This is especially the case in Section 4. The authors will need to critically comment on previous studies before making any conclusions. For example, in lines 276-279, it is not clear to me that the previous studies are mostly descriptive.

In the paper, the authors claim that meta-analysis was used for the review. However, it reads to me that the paper just statistically analysed what the existing studies are and what methods have been used, etc (section 4). It is not exactly meta-analysis which focuses on digging into the results of the studies and their relations to the purpose of analysis. The authors may want to address this issue in the revision, e.g. what are causes that have been identified in those studies and what may be the linkage among different various and/or causes?

Author Response

Reviewer 3

Point 1: This review is interesting which focuses on a new concept--telecoupling dealing with complex human-natural interactions. It would be of interest to some interdisciplinary researchers. However, the authors will need to address some issues below. What are differences between “casual attribution” in telecoupling study and casual-chain dynamics in complex system studies? Some thoughts of this [sic] are needed.

Response 1: One notable difference is that “causal attribution” (as defined in our manuscript) is the process of combining qualitative and quantitative information to assign causes, whereas “casual-chain dynamics in complex system studies” refers to patterns (i.e., “dynamics”) rather than an attribution process. However, we believe that evaluating causal chains (which we define in lines 341-342) is a promising method for causal attribution in telecoupling research, particularly for discerning between proximate and ultimate causes (i.e., “links” in causal chains) using qualitatively and quantitatively rigorous methods, as described in lines 342-344. To further address this comment, we revisited results from our literature review and did not encounter casual-chain dynamics in complex system studies, or definitions of “causal attribution” therein. This likely minimized any potential overlap between our definition of “causal attribution” and the definition used in complex systems studies. Hence, we believe that our discussion of causal attribution, and our emphasis on qualitative-quantitative rigor in causal analysis, are unique contributions to telecoupling research that will promote informed, causally accurate management and governance of telecoupled systems.

Point 2: The review of literature is a bit shallow in its current state. This is especially the case in Section 4. The authors will need to critically comment on previous studies before making any conclusions. For example, in lines 276-279, it is not clear to me that the previous studies are mostly descriptive.

Response 2: We thank the Reviewer for this comment but have trouble understanding how this conclusion was reached. Well before lines 276-279 (specified by the Reviewer), we illustrate how the telecoupling literature is mostly descriptive. For instance, please see lines 246-247 (“62.9% of the papers (n=56) made descriptive statements regarding causal effects, causal mechanisms ...”) and lines 248-250 (“Only two of these papers included rigorous [i.e., qualitative and quantitative] causal analysis methods ... in addition to descriptive statements”). To address this comment, we thoroughly reviewed section 3 (formerly section 4) and carefully checked Figure 2 and Tables 1 and 2 (which demonstrate that previous telecoupling literature is largely descriptive) to ensure that the paper accurately and clearly describes the descriptive nature of telecoupling research to date. Furthermore, please see our supplementary materials, which provide detailed paper-specific information from our literature review that collectively illustrates a preponderance of descriptive telecoupling studies and a scarcity of rigorous qualitative-quantitative research.

Point 3: In the paper, the authors claim that meta-analysis was used for the review. However, it reads to me that the paper just statistically analysed what the existing studies are and what methods have been used, etc (section 4). It is not exactly meta-analysis which focuses on digging into the results of the studies and their relations to the purpose of analysis. The authors may want to address this issue in the revision, e.g. what are causes that have been identified in those studies and what may be the linkage among different various and/or causes?

Response 3: We conducted a systematic review, not a meta-analysis, as stated in the manuscript title and at the beginning of section 3 (i.e., line 232: “A systematic review of telecoupling research was performed ...”). Although spelling out “PRISMA” (“Preferred Reporting Items for Systematic reviews and Meta-Analyses”) in lines 232-233 could suggest that we conducted a meta-analysis, the full sentence indicates that we performed a systematic review. It is common practice to cite the PRISMA procedure, so we feel that it is necessary to retain the full definition (“Preferred Reporting Items for Systematic reviews and Meta-Analyses”). However, we feel that the meta-analysis the Reviewer described would be an interesting area for future research to advance the telecoupling framework and its sustainability applications.

Reviewer 4 Report

This paper successfully addressed the importance of causality research in telecoupling studies. It clearly explained the concept of causality and pinpointed the feasible causality methodologies in telecoupling work. The authors comprehensively reviewed the existing literature and pointed out missing causality research in current telecoupling studies. Most importantly, the authors did a great job in proposing ways of improving causality research and identified sources of causes. I believe this paper will attract more attention to the causality research in the telecoupling field. 

Author Response

Reviewer 4

Point 1: This paper successfully addressed the importance of causality research in telecoupling studies. It clearly explained the concept of causality and pinpointed the feasible causality methodologies in telecoupling work. The authors comprehensively reviewed the existing literature and pointed out missing causality research in current telecoupling studies. Most importantly, the authors did a great job in proposing ways of improving causality research and identified sources of causes. I believe this paper will attract more attention to the causality research in the telecoupling field.

Response 1: We thank the Reviewer for these helpful comments and share the hope that this paper will draw attention to the importance of rigorous qualitative-quantitative causality assessments in telecoupling research.

Round 2

Reviewer 3 Report

Thanks for the materials. I am fine with the revisions.