The Continuum of Eco-Anxiety Responses: A Preliminary Investigation of Its Nomological Network

Insights from a wide range of disciplines, though highly valuable, have seemed to contribute to difficulties with conceptual clarity and notable heterogeneity in how eco-anxiety is defined (e.g., Coffey et al., 2021; Martin et al., 2022; Pihkala, 2020a). Although eco-anxiety can be viewed in many ways, we believe Passmore et al.’s (2022) definition provides a useful summary of previously proposed accounts. These authors define the construct as “persistent feelings of worry, anxiety, dread, or doom regarding environmental degradation and the impacts and implications of climate change on our planet as a whole” (p. 3). Despite lingering ambiguities in the literature, a commonality between this definition and other views of the phenomenon is that eco-anxiety tends to capture a relatively strong form of distress (e.g., Clayton & Karazsia, 2020; Hogg et al., 2021; Pihkala, 2020b; Verplanken et al., 2020). At the same time, however, it is also often acknowledged that symptoms can fall along a continuum ranging from milder forms of worry to clinically significant responses in which one experiences emotional, cognitive, social, or functional impairments (e.g., Clayton & Karazsia, 2020; Passmore et al., 2022; Pihkala, 2020a; Soutar & Wand, 2022), a view taken of the concept of anxiety more broadly (Endler & Kocovski, 2001; Lazarus, 1991). This indicates that although eco-anxiety represents a rational response to serious issues (e.g., Clayton, 2020; Hogg et al., 2021; Passmore et al., 2022; Pihkala, 2020a), it may sometimes require professional support. As such, it is important to recognize that some forms of eco-anxiety may be more (mal)adaptive than others. The notion that eco-anxiety can encompass a range of responses could be leveraged to help explain this variation; however, there is a relative lack of research on this issue (Pihkala, 2020a).

This continuum notion may also provide useful insights into how eco-anxiety, which is often considered a relatively new construct, is related to environmental concern, a construct that has been a topic of scholarly inquiry for over 50 years (Cruz & Manata, 2020). Like eco-anxiety, environmental concern can be defined in many ways. Herein, we view environmental concern as an affective attitude toward environmental issues (e.g., Cruz & Manata, 2020; Landry et al., 2018; Takàcs-Sànta, 2007). From this definition, it can be inferred that a link between affect and environmental issues is a shared feature of both eco-anxiety and environmental concern. Therefore, it may be unsurprising that climate change worry, which could be considered an indicator of eco-anxiety, is often discussed or operationalized as climate change concern and vice versa (Martin et al., 2022; Stewart, 2021; see also van der Linden, 2017), a tendency resembling Lazarus’ (1991) identification (outside of the domain of eco-anxiety per se) of concern, worry, and anxiety as belonging to the same emotion family. While inquiring about participants’ levels of worry versus concern may seem merely an issue of semantics, some suggest that this distinction is not trivial. For example, van der Linden (2017) notes that it is possible to be concerned about climate change without actively worrying about it. As such, concern is often considered to be less severe than eco-anxiety (Feather & Williams, 2022; Helm et al., 2018; McBride et al., 2021; Pihkala, 2020b; see also van der Linden, 2017). At the same time, asking about participants’ levels of concern may be beneficial, as it seems to share less content overlap with measures of ill-being (McBride et al., 2021). It is therefore possible that some measures of environmental concern could be considered to fall on the less severe end of the continuum of eco-anxiety responses. To explore this idea, we examined how measures of environmental concern relate to measures of eco-anxiety that differ in severity and if these sets of measures exhibit (dis)similar patterns of relationships with mental health and a pro-environmental orientation.

It is often suggested that eco-anxiety may pose a threat to psychological health (e.g., Clayton, 2020; Passmore et al., 2022; Pihkala, 2020a); however, evidence for this has been somewhat mixed (Ojala et al., 2021). The most consistent support for this proposition comes from studies that tend to report positive associations between eco-anxiety and aspects of ill-being such as anxiety, depression, stress, pathological worry, and general psychological distress, though variability in this pattern of findings has emerged (ranging from r = -.05 to r = .59; Clayton & Karazsia, 2020; Feather & Williams, 2022; Helm et al., 2018; Hogg et al., 2021; Mouguiama-Daouda et al., 2022; Reyes et al., 2021; Schwartz et al., 2022; Stanley et al., 2021; Stewart, 2021; Verplanken et al., 2020; Verplanken & Roy, 2013; Wullenkord et al., 2021). These correlations are unsurprising given that the content of eco-anxiety measures, especially those capturing more severe symptoms (e.g., Clayton & Karazsia, 2020; Hogg et al., 2021), are often based on or overlap with ill-being scales (McBride et al., 2021; Ojala et al., 2021). Measures of environmental concern, which tend to share less of this overlap, seem to exhibit weaker associations with ill-being (ranging from r = -.02 to r = .11; Feather & Williams, 2022; Landry et al., 2018; McBride et al., 2021).

A more complex set of findings emerges when considering the relationship between eco-anxiety and indices of well-being. To begin, people’s endorsement of emotion terms included in eco-anxiety measures (e.g., worry, fear), as well as their overall scores on eco-anxiety measures, have been found to be positively associated with negative (e.g., angry, sad; ranging from r = .24 to r = .70) and positive (e.g., hope, interest; ranging from r = .35 to r = .57) emotions experienced when they think about climate change or global warming (Clayton & Karazsia, 2020; Smith & Leiserowitz, 2014; Verplanken et al., 2020).¹ When considering well-being measures that do not have an environmental framing, Ogunbode et al. (2021) found anxiety-related negative emotions about climate change to be negatively associated with a general measure of well-being in many countries around the world (r_pooled = -.24). By contrast, a more severe measure of climate change anxiety (i.e., Clayton & Karazsia, 2020) was unrelated to a general measure of well-being in a sample of Filipinos (r = -.05; Reyes et al., 2021) and, in a German context, was negatively associated with the satisfaction of autonomy (r = -.14) and relatedness (r = -.10) needs and positively associated with the frustration of autonomy (r = .20), relatedness (r = .27), and competence (r = .29) needs (Wullenkord et al., 2021). An Australian sample yielded a negative correlation between a more severe measure of eco-anxiety and life satisfaction (r = -.12; Hogg et al., 2021). In their review, Ojala et al. (2021) discussed several studies that were not easily accessible to the present authors (e.g., due to being published in a non-English language) that demonstrated mixed associations between measures of climate/environmental worry and various aspects of well-being. Interestingly, measures of environmental concern have also exhibited mixed associations with well-being (ranging from r = -.19 to r = .33; McBride et al., 2021; McConnell & Jacobs, 2020; Verplanken et al., 2020).

Taken together, there seems to be an overall trend in which more severe measures of eco-anxiety are related to poorer psychological functioning. At the same time, it should be noted that not all relationships observed were statistically significant (e.g., Reyes et al., 2021; Verplanken & Roy, 2013) and there tended to be notable variation in effect sizes, perhaps reflecting the particular mental health indicator being assessed. Finally, the above works tended to adopt single measures of eco-anxiety, mental health, or both. We were curious to discover if simultaneously examining how several measures of eco-anxiety are related to multiple indicators of mental health would offer additional insights into these relationships, such as whether more severe forms of eco-anxiety are more associated with poor psychological functioning than less severe forms.

Despite some evidence that eco-anxiety is related with poorer mental health, there is also research highlighting its more adaptive nature, as demonstrated by its relationships with aspects of a pro-environmental orientation. More specifically, eco-anxiety (regardless of severity) tends to be positively associated with predictors and measures of pro-environmental intentions and behaviour, though some variability in this pattern of findings is present (Clayton & Karazsia, 2020; Helm et al., 2018; Hogg et al., 2021; Mouguiama-Daouda et al., 2022; Ojala et al., 2021; Schwartz et al., 2022; Stanley et al., 2021; Verplanken et al., 2020; Verplanken & Roy, 2013; Wullenkord et al., 2021).² These findings suggest that eco-anxiety may be largely viewed as a form of practical anxiety (i.e., a motivating response that facilitates behavioural engagement; Kurth, 2018; Pihkala, 2020a) rather than a form of eco-paralysis (i.e., a demotivating response that hinders action; Albrecht, 2011).

In the current work, we sought to further examine how eco-anxiety relates to a set of individual difference variables indicative of a general pro-environmental orientation—namely belief in the occurrence of climate change, belief in the anthropogenic (human) causation of climate change, collective eco-guilt, empathy with nature, and nature connectedness. Although it can be argued that one could suffer from paralyzing forms of eco-anxiety and still score high on these constructs, it is important to note that they each have been shown to be positively associated with pro-environmental behaviour.

We conducted five studies across three countries to examine how measures that may encompass a continuum of environment-related worry and anxiety are related to each other and to measures of environmental concern. Next, we considered if these various measures exhibited (dis)similar patterns of relationships with various aspects of both mental health (i.e., measures of ill-being; measures of well-being and its correlates, such as meaning in life) and of a pro-environmental orientation (e.g., nature connectedness, empathy with nature). Materials, data, and syntax for all studies can be found on the open science framework (OSF): https://osf.io/krfnv/

In Study 1, we explored the interrelations among two measures of eco-anxiety and a measure of environmental concern. We then examined how these measures were related to multiple aspects of well-being, as well as to nature connectedness, a key correlate of pro-environmental behaviour (e.g., Mackay & Schmitt, 2019). Those who feel a broader sense of connection to nature may also report elevated levels of eco-anxiety, as such individuals may be more aware of environmental issues and the threats they pose (e.g., Clayton & Karazsia, 2020). This study was not preregistered.

Participants were 242 undergraduate students enrolled in psychology courses at MacEwan University located in western Canada. No demographic information was collected. Data collection took place from February 14^th to June 14^th, 2019. Prior to analyses, four participants were removed due to excessive amounts of missing data (i.e., completing < 50% of the survey questions), resulting in a final sample of 238 participants.

Participants completed the following scales listed below (see OSF for full scales and order of administration).

Eco-Anxiety and Environmental Concern. We were unaware of any standard measures of eco-anxiety at the time of data collection. As such, we employed two measures we considered to be proxies of the construct. The first was a 4-item measure of perceived ecological stress initially developed by Homburg et al. (2007) and later adapted by Helm et al. (2018). Participants were asked to indicate how stressed they feel because of the following environmental problems: (1) global environmental problems like the ozone hole and climate change, (2) pollution of the oceans and the environment, (3) extinction of species, and (4) loss of forests and spread of uninhabitable desert areas. Each item was rated on a 5-point scale (1 = not at all stressed, 5 = extremely stressed). The second proxy of eco-anxiety was an item from Verplanken and Roy (2013): “How often do you have thoughts about the environment, which you find worrying, uncomfortable, or upsetting?” rated on a 5-point scale (1 = never, 5 = all of the time). When considering where these two measures fall on the continuum of eco-anxiety responses, we reasoned that the second measure would capture a more severe form of eco-anxiety, as the item content appears to focus on a stronger form of distress.

We assessed environmental concern using one item from Berenguer et al. (2005): “To what extent are you concerned about the situation of the environment in general?” rated on a 7-point scale (1 = not at all, 7 = totally). This item has been used to measure environmental concern in prior research (e.g., Landry et al., 2018).

Measures of Well-Being. The Well-Being Scale (Lui & Fernando, 2018) is a 29-item measure that assesses overall well-being, as well as five distinct domains: hedonic, eudaimonic, social, physical, and financial. Items were rated on a 6-point scale (1 = strongly disagree, 6 = strongly agree). The Multidimensional Existential Meaning Scale (George & Park, 2017) is a 15-item measure that assesses purpose, comprehension, and mattering as aspects of meaning in life. Items were rated on a 7-point scale (1 = very strongly disagree, 7 = very strongly agree).

Nature Connectedness. The Nature Relatedness Scale (NRS; Nisbet et al., 2009) is a 21-item measure that assesses one’s subjective connection to nature. The scale provides a total score, as well as three subscale scores for NR-Self, NR-Perspective, and NR-Experience. Items were rated on a 5-point scale (1 = disagree strongly, 5 = agree strongly).

Descriptive statistics and correlations are reported in Table 1 (see Table S1 for confidence intervals for correlations). Mean scores on the two eco-anxiety measures and the environmental concern measure were above the respective scale midpoints. Correlational analyses demonstrated positive and large correlations among these three measures. Interestingly, these measures were all generally unrelated to well-being with the exception that all three measures correlated positively with eudaimonic well-being. Moreover, the measure of environmental concern evidenced a significant positive association with social well-being. All three variables exhibited moderate to strong positive associations with the various aspects of nature connectedness. Taken together, these results do not point to a sharp differentiation among the two measures of eco-anxiety and the measure of environmental concern.

In Study 2, we introduced a measure of climate change worry as an additional measure of eco-anxiety and examined its associations with the eco-anxiety and environmental concern indicators from Study 1. We also considered how these measures were related to aspects of ill-being. Lastly, we evaluated how belief in the occurrence of climate change, belief in the anthropogenic causation of climate change, collective eco-guilt, empathy with nature, and nature connectedness were associated with eco-anxiety and environmental concern. In selecting these variables, we first reasoned that since climate change is a key contributor to various environmental issues, believing in its occurrence would be a potent source of eco-anxiety. Further, we anticipated that people will report feeling more eco-anxious if they consider climate change to be the result of human activity, as doing so would directly implicate humans as being responsible for numerous environmental calamities. A sense of responsibility for harmful acts also features prominently in conceptualizations of guilt. For instance, one may feel a sense of collective (eco-) guilt when one’s nation is seen as responsible for committing environmental wrongdoings (Mallett, 2012). Perceived harm against the environment may also set the stage for empathic responses towards it. Upon considering nature’s plight, some may be more prone to understanding and sharing its distress (Tam, 2013), potentially evoking feelings of anxiety about the natural world. Relatedly, a broader sense of connection to nature may be an important correlate of eco-anxiety, as suggested in Study 1 and in prior work (e.g., Clayton & Karazsia, 2020). We pre-registered Study 2 at https://osf.io/dx7rt.

We aimed to recruit 300 participants (expecting exclusions) based on Schönbrodt and Perugini’s (2013) recommendation that sample sizes should approach 250 in order to obtain stable correlation estimates. We recruited 496 undergraduate students who were enrolled in psychology courses at the University of British Columbia Okanagan located in western Canada. Data collection took place from October 9^th to November 28^th, 2019. Prior to analyses, 53 participants were removed due to either failing an attention check (n = 49), rating their comprehension of written English as 2 or lower on a 5-point scale (n = 3), or completing < 50% of the survey questions (n = 1), resulting in a final sample of 443 participants (77.7% female; M_age = 20.48, SD_age = 3.00; age range 17-42). The first exclusion criterion was preregistered, while the latter two were not.

Participants completed the following scales listed below. Additional qualitative questions were completed by participants as part of another study (see OSF for full scales and order of administration).

Eco-Anxiety and Environmental Concern. We assessed these constructs using the same measures as in Study 1, with the addition of a 1-item measure of eco-anxiety adapted from Ballew et al. (2019): “How worried are you about climate change”, rated on 4-point scale (1 = not at all worried, 4 = very worried). Of note, the original measure used “global warming”; however, we used “climate change”. We again perceived the content of this measure to be less severe than the measure from Verplanken and Roy (2013).

Ill-Being. The Depression, Anxiety, and Stress Scales (Lovibond & Lovibond, 1995) is a 21-item instrument that assesses depression, anxiety, and stress. Participants rated how often they experienced each item over the past week on a 4-point scale (0 = did not apply to me at all, 3 = applied to me very much, or most of the time).

Measures of Pro-Environmental Orientation. We assessed the extent to which one believes that climate change is currently taking place and is caused by human actions with the 14-item Occurrence and Anthropogenic Causation Scale (Brownlee & Verbos, 2015). Items were rated on a 7-point scale (1 = completely disagree, 7 = completely agree). We assessed collective eco-guilt using a 5-item measure from Mallet (2012) that asks participants the extent to which they feel guilty because of negative environmental actions committed by Canadians as a group. Items were rated on a 7-point scale (1 = not at all guilty, 7 = extremely guilty). We employed the 10-item Dispositional Empathy with Nature Scale to assess an individual’s tendency to empathize with the natural world (Tam, 2013). Items were rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree). We assessed nature connectedness as in Study 1.

This study had preregistered hypotheses and analysis plans. Specifically, we first hypothesized that belief in the occurrence of climate change, belief in the anthropogenic causation of climate change, collective eco-guilt, empathy with nature, and nature connectedness total (and subscale scores) would exhibit positive correlations of .20-.30 with the three measures of eco-anxiety and measure of environmental concern.³ Next, we hypothesized that, using multiple regression analysis, belief in the occurrence of climate change, belief in the anthropogenic causation of climate change, collective eco-guilt, empathy with nature, and nature connectedness total would positively predict the three measures of eco-anxiety and the measure of environmental concern. Lastly, we hypothesized that, using hierarchical multiple regression analysis, the aforementioned predictors would positively predict each of three measures of eco-anxiety and the measure of environmental concern over and above ill-being (i.e., depression, anxiety, and stress).

Descriptive statistics and correlations are reported in Table 2 (see Table S2 for confidence intervals for correlations). Mean scores on three eco-anxiety measures and the measure of environmental concern were above the respective scale midpoints. Correlational analyses demonstrated positive and large correlations among these four measures. The three measures of eco-anxiety exhibited notably similar patterns of relations with ill-being. That is, they were each unrelated to depression, and weakly positively related to anxiety and stress. Environmental concern was unrelated to all three measures of ill-being. Table 2 also shows that the three measures of eco-anxiety and the measure of environmental concern evidenced similar positive associations with our five indices of a pro-environmental orientation.

We conducted a series of regression analyses to reflect the stated analytic approach in our preregistration of this study; as regressions were not conducted for the remaining studies, please see the supplemental materials for regression results for Study 2. In brief, similar patterns to those in the bivariate analyses emerged in the regression analyses in which belief in the occurrence of climate change, belief in the anthropogenic causation of climate change, collective eco-guilt, empathy for nature, and nature connectedness total were used to predict the different eco-anxiety/environmental concern criterion variables (see Table S3). Moreover, these patterns persisted when ill-being variables were entered in an initial regression step (see Table S4).

In summary, we found the three eco-anxiety measures and the measure of environmental concern to be strongly related. The three eco-anxiety measures were positively associated with anxiety and stress but unrelated to depression. The measure of environmental concern was unrelated to ill-being. All three eco-anxiety measures and the measure of environmental concern exhibited relatively similar positive associations with aspects of a pro-environmental orientation.

In Study 3, we continued to examine interrelations among the eco-anxiety measures and the measure of environmental concern used in Studies 1 and 2. We then considered how these measures were related to various aspects of ill-being and well-being simultaneously, as well as nature connectedness. Notably, we drew on a sample of undergraduate students from China instead of Canada, as opposed to Studies 1 and 2, heeding calls for more eco-anxiety research to be conducted in non-Western countries (e.g., Coffey et al., 2021). This study was not preregistered.

Participants were 1635 (53.0% female; M_age = 20.04, SD_age = 1.14; age range 17-32) undergraduate students attending Zhejiang Ocean University located in southeast China. Data collection took place between February 7^th and 17^th, 2020. All participants complied with instructions and thus there were no participant exclusions prior to analyses.

Participants completed Chinese versions of the following scales listed below. If a Chinese version was not available, the fifth author (fluent in English and Chinese) translated the scale into Chinese. Additional qualitative questions were completed by participants as part of another study (see OSF for full English versions of scales and order of administration).

Eco-Anxiety and Environmental Concern. We assessed eco-anxiety and environmental concern using the same measures outlined in Study 2.

Ill-Being. We used the same measure of ill-being outlined in Study 2 (Chinese version: see L.-C. Jiang et al., 2020; Wang et al., 2016).

Measures of Well-Being. The Scale of Positive and Negative Experiences (Diener et al., 2010) is a 12-item measure that assesses positive and negative affect. Participants rated how often they generally experienced each item on a 5-point scale (1 = very rarely or never, 5 = very often or always). The Satisfaction with Life Scale (Diener et al., 1985; Chinese version: Xiong & Xu, 2009; see also Yuen, 2002) is a 5-item measure of overall life satisfaction. Items were rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree). The Meaning in Life Questionnaire (Steger et al., 2006; Chinese version see Steger, n.d., and Y. Jiang et al., 2016) is a 10-item scale that assesses the presence of, and search for, meaning in life. Items were rated on a 5-point scale (1 = absolutely untrue, 7 = absolutely true). The Basic Psychological Need Satisfaction and Frustration Scale (Chen et al., 2015) is a 24-item measure that assesses the satisfaction and frustration of the needs for autonomy, relatedness, and competence. Items were rated on a 5-point scale (1 = completely disagree, 5 = completely agree).

Nature Connectedness. The Connectedness to Nature Scale (Mayer & Frantz, 2004; Chinese version: see Geng et al., 2015) is a 14-item measure that assesses one’s subjective connection to nature. Items were rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree).

Descriptive statistics and correlations are reported in Table 3 (see Table S5 for confidence intervals for correlations). Mean scores on the Helm et al. (2018) and Ballew et al. (2019) eco-anxiety measures, as well as the measure of environmental concern were above the respective scale midpoints; however, the mean score on the Verplanken and Roy (2013) eco-anxiety measure was notably lower than in Studies 1 and 2 and was below the midpoint of the scale. Correlations among these four measures were also notably smaller than in Studies 1 and 2, especially those involving the Verplanken and Roy measure. With respect to ill-being, the Helm et al. and Verplanken and Roy eco-anxiety measures evidenced weak positive associations with depression, anxiety, and stress, whereas the Ballew et al. eco-anxiety measure and the environmental concern measure demonstrated very weak negative associations with depression. An interesting pattern of relations emerged upon examining how these measures were related to well-being. The Helm et al. measure was generally associated with greater well-being (e.g., positive affect, need satisfaction). The Ballew et al. measure largely mirrored these associations and demonstrated additional well-being relationships (e.g., correlating negatively with negative affect and need frustration). The associations involving the measure of environmental concern were highly similar to those exhibited by the Helm et al. and Ballew et al. eco-anxiety measures but tended to be notably stronger. By contrast, the Verplanken and Roy measure was generally associated with poorer well-being (e.g., lower positive affect and life satisfaction and higher negative affect). The three eco-anxiety measures and the measure of environmental concern correlated positively with nature connectedness; however, these associations were lower than in Studies 1 and 2, especially for the Verplanken and Roy measure. As a general observation, it is important to note that even though many correlations derived from this large sample reached statistical significance due to its high power, the vast majority were small to moderate in magnitude.

In Study 4, we continued to examine how the eco-anxiety measures and the measure of environmental concern were related to each other, well-being, and a pro-environmental orientation. We drew on a sample of adults from the United Kingdom, as opposed to undergraduates from Canada or China as in Studies 1-3. This study was not preregistered.

We recruited 316 adults living in the UK via Prolific Academic. Each participant was compensated £1.50. Data collection took place on April 17^th, 2020. Prior to analyses, one participant was removed due to completing < 50% of the survey questions, resulting in a final sample of 315 participants (73.0% female; 87.3% White; M_age = 32.30, SD_age = 10.54; age range 18-72).

Participants completed the following scales listed below. Additional measures were completed by participants as part of another study, including alternative measures of nature connectedness. We focused on the Nature Relatedness Scale-6 to maximize consistency across studies (see OSF for full scales and order of administration).

Eco-Anxiety and Environmental Concern. We used the measures outlined in Study 2; however, the Helm et al. (2018) eco-anxiety measure was slightly altered to ask participants how stressed/anxious they feel on the same 5-point scale but removing “stressed” from the response options (i.e., 1 = not at all, 5 = extremely).

Measures of Well-Being. We assessed positive affect and negative affect using the measure adopted in Study 3. Satisfaction of the needs for autonomy, relatedness, and competence were assessed using the measure adopted in Study 3; however, we did not employ the subscales that assessed the frustration of these needs. Additionally, we altered the instructions of the need satisfaction measure to ask participants to base their responses on how they generally felt.

Measures of Pro-Environmental Orientation. We assessed belief in the occurrence of climate change and collective eco-guilt using the measures employed in Study 2. We altered the instructions for the collective eco-guilt measure to ask participants the extent to which they felt guilty because of negative environmental actions committed by the citizens of the country they live in as a group. We assessed nature connectedness with the Nature Relatedness Scale-6 (Nisbet & Zelenski, 2013), a 6-item version of the scale used in Study 1.

Descriptive statistics and correlations are reported in Table 4 (see Table S6 for confidence intervals for correlations). Mean scores on three eco-anxiety measures and the measure of environmental concern were above the respective scale midpoints. Correlational analyses demonstrated positive and large correlations among these four measures. The Helm et al. (2018) eco-anxiety measure was associated with both greater positive and negative affect, as well greater autonomy and relatedness need satisfaction. The Verplanken and Roy (2013) and Ballew et al. (2019) eco-anxiety measures were both only associated with greater autonomy satisfaction. Lastly, the measure of environmental concern was associated with greater postive affect, as well as greater autonomy and relatedness need satisfaction. Table 4 also shows that the three measures of eco-anxiety and the measure of environmental concern evidenced similar positive associations with the three indices of a pro-environmental orientation.

To summarize, the three eco-anxiety measures and the measure of environmental concern emerged as strongly related and, regarding their associations with well-being, they most consistently evidenced a positive association with autonomy satisfaction. They also exhibited relatively similar associations with aspects of a pro-environmental orientation.

One notable limitation of Studies 1-4 is that our primary research questions were often examined in a piecemeal fashion. In Study 5, we attempted to be more comprehensive in our approach while also incorporating some important additions. First, while planning for the study, we became aware of a formal measure of eco-anxiety, Clayton and Karazsia’s (2020) Climate Change Anxiety Scale that assesses cognitive-emotional and functional impairments in response to climate change. We included this measure, along with a widely used scale of environmental concern from Schultz (2001). Second, we assessed multiple aspects of ill-being and well-being simultaneously. Lastly, we included environmental cognitive alternatives (Wright et al., 2020) as an additional marker of a general pro-environmental orientation, as well as more direct indices (i.e., pro-environmental consumption behaviour and willingness to engage in environmental activist behaviour). This study was not preregistered.

Participants were 319 undergraduate students enrolled in psychology courses at Carleton University located in eastern Canada. Data collection took place from January 26^th to March 5^th, 2021. Prior to analyses, 11 participants were removed for failing two attention checks, resulting in a final sample of 308 participants (78.6% female; 63.0% White/Caucasian; M_age = 20.15, SD_age = 3.95; age range 17-47).

Participants completed the following scales listed below. Additional measures were completed by participants as part of other studies, including an alternative measure of nature connectedness. We focused on the Nature Relatedness Scale-6 to maximize consistency across studies (see OSF for full scales and order of administration).

Eco-Anxiety and Environmental Concern. We assessed eco-anxiety and environmental concern using the measures outlined in Study 4, with the addition of two scales. The first was the Climate Change Anxiety Scale (CCAS; Clayton & Karazsia, 2020), a 13-item measure that provides a total score and two subscale scores that capture cognitive-emotional impairment and functional impairment. Items were rated on a 5-point scale (1 = never, 5 = almost always). It is worth noting that during the initial scale development process, the items of this measure were a part of a larger item pool that, when factor analyzed, produced a four-factor solution that consisted of the two subscales noted above, as well as two factors reflecting the experience of climate change and pro-environmental behavioural engagement. Clayton and Karazsia noted that only the former two factors comprised the CCAS, and thus, they were the only factors assessed here. The second was the Environmental Concerns Scale (Schultz, 2001), a 12-item measure that assesses biospheric, altruistic, and egoistic concerns. Items were rated on a 7-point scale (1 = not important, 7 = supreme importance). We focus on the biospheric subscale, as it has been shown to more clearly tap environmental concern than the other two subscales (Cruz & Manata, 2020).

Ill-Being. We used the same measure of ill-being outlined in Study 2.

Measures of Well-Being. We assessed positive affect, negative affect, and life satisfaction using the measures adopted for Study 3. We also employed the 16-item Multidimensional Meaning in Life Scale (Costin & Vignoles, 2020) to assess meaning in life judgments, as well as the facets of coherence, purpose, and mattering. Items were rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree). Lastly, we used the Adult Hope Scale (Snyder et al., 1991), which provided a total score and two subscale scores for agency and pathways. Items were rated on a 8-point scale (1 = definitely false, 8 = definitely true)

Measures of Pro-Environmental Orientation. We assessed belief in the occurrence of climate change, collective eco-guilt, and empathy with nature using the same measures outlined in Study 2 and we assessed nature connectedness using the same measure employed in Study 4. We also assessed the extent to which one is able to imagine what a sustainable relationship between humans and nature may look like using the 10-item Environmental Cognitive Alternatives Scale (Wright et al., 2020). Items were rated on a 7-point scale (1 = strongly disagree, 7 = strongly agree). Lastly, we assessed pro-environmental consumption behaviour and willingness to engage in environmental activist behaviour each with 10 items from Schmitt et al. (2019). The consumption items were rated on a 5-point scale (1 = never, 5 = always) and the activist items were rated on a 7-point scale (1 = extremely unwilling, 7 = extremely willing).

Descriptive statistics and correlations are reported in Table 5 (see Table S7 for confidence intervals for correlations). Mean scores on the eco-anxiety and environmental concern measures were above the respective scale midpoints except for total and subscale scores on the CCAS, which were below the scale midpoint. Correlational analyses demonstrated positive and large correlations among the three eco-anxiety measures and measure of environmental concern collected in Study 4. These measures correlated positively but less strongly with the CCAS and the biospheric concern measure. The Verplanken and Roy (2013) eco-anxiety measure correlated most strongly with the CCAS. The biospheric concern measure was unassociated with the CCAS. All eco-anxiety and environmental concern measures correlated positively with altruistic concern except the CCAS. The Helm et al. (2018) eco-anxiety measure and the environmental concern measures correlated positively with egoistic concern.

All eco-anxiety measures correlated positively with ill-being, but correlations involving the CCAS tended to be stronger. The Berenguer et al. (2005) environmental concern measure evidenced weaker positive associations with ill-being, and no significant correlations emerged between biospheric concern and ill-being.

With respect to well-being, the Helm et al. (2018) eco-anxiety measure was positively associated with hope total and hope pathways. The Verplanken and Roy (2013) eco-anxiety measure was negatively associated with positive affect and meaning in life judgments, and positively associated with negative affect. No associations emerged between the Ballew et al. (2019) eco-anxiety measure and the well-being indices. The CCAS was associated with greater negative affect. The Berenguer et al. (2005) measure of environmental concern was positively associated with hope total and pathways. Biospheric concern mirrored these relationships but was also associated with greater coherence and purpose.

All measures of eco-anxiety and environmental concern tended to be positively associated with the various indices of a pro-environmental orientation. The exceptions were that the CCAS tended to be unrelated to the belief in the occurrence of climate change and collective eco-guilt, though the cognitive-emotional impairment subscale was positively related to collective eco-guilt. Additionally, functional impairment was unrelated to pro-environmental consumption behaviour. Of note, the CCAS tended to evidence smaller associations with pro-environmental consumption behaviour and environmental activist willingness than the other measures.

To summarize, the three eco-anxiety measures and the measure of environmental concern collected in Study 4 again emerged as strongly related. These measures correlated less strongly with the CCAS and biospheric concern. The Verplanken and Roy (2013) eco-anxiety measure and the CCAS were associated with some aspects of poorer well-being (e.g., negative affect), whereas the remaining measures tended to be associated with some aspects of greater well-being (e.g., hope). All eco-anxiety and environmental concern measures tended to be positively associated with various aspects of a pro-environmental orientation in similar ways, except for the CCAS, which exhibited some weaker relationships.

Eco-anxiety is a relatively new construct that holds considerable promise in helping to understand the psychological and behavioural consequences arising from people’s in(direct) exposure to environmental issues. The aim of the present research was to gain additional insight into this concept by examining 1) the notion that eco-anxiety can encompass a continuum of responses (one that is possibly inclusive of the older concept of “environmental concern”), and whether these various responses are differentially linked to aspects of 2) mental health and 3) a pro-environmental orientation. We conducted five studies across three countries to help examine these issues.

To probe the notion that eco-anxiety can encompass a continuum of responses, we employed a series of measures that differed in severity of the eco-anxiety symptoms they assessed, and we examined how scores on those measures related to each other and to measures of environmental concern, an older construct that could potentially be considered to capture the less severe end of eco-anxiety responses. In first comparing measures of eco-anxiety, we observed that, across all studies, mean scores were above the respective scale midpoints, except on the Verplanken and Roy (2013) measure in Study 3 and Clayton and Karazsia’s (2020) CCAS collected in Study 5 in which mean scores were below the scale midpoint. These findings are consistent with previous research using the CCAS (Clayton & Karazsia, 2020; Feather & Williams, 2022; Reyes et al., 2021; Schwartz et al., 2022; Wullenkord et al., 2021).

We also observed, across studies, that the Helm et al. (2018), Verplanken and Roy (2013), and Ballew et al. (2019) eco-anxiety measures were strongly positively correlated, except in Study 3 (China), in which these correlations were notably weaker, especially those involving the Verplanken and Roy measure. In addition to possible culture differences, translating the items to Chinese in Study 3 may have contributed to these lower correlations and the notably lower mean on the Verplanken and Roy measure in this sample (see limitations section). In Study 5, these three measures correlated less strongly with the CCAS. Of note, the Verplanken and Roy measure correlated most strongly with the CCAS, providing some support for our contention that these measures capture more severe forms of eco-anxiety. At the same time, the fact that the Helm et al., Verplanken and Roy, and Ballew et al. measures tended to correlate notably more strongly with each other than with the CCAS may seem troubling, especially when viewed in the context of past work questioning whether the CCAS can be considered a measure of climate/eco-anxiety. Indeed, although eco-anxiety is often described as an emotional response to environmental issues (e.g., Clayton, 2020; Clayton & Karazsia, 2020; Ojala et al., 2021; Passmore et al., 2022), Wullenkord et al. (2021) note that they consider the CCAS to assesses various impairments stemming from the climate crisis rather than the emotional experience of climate change anxiety. These authors also noted that the CCAS “needs further development to capture gradations and degrees of severity of climate anxiety” (p. 16; see also Feather & Williams, 2022, p. 142). Some of these concerns may be addressed by the recently proposed Hogg Eco-Anxiety Scale (HEAS; Hogg et al., 2021), which includes an affective symptoms subscale. However, both the CCAS and the HEAS may have difficulties capturing degrees of severity of eco-anxiety, as they both seem to assess a more severe response (e.g., extant work indicates that mean scores tend to be notably below the respective scale midpoints, which may suggest that a considerable number of items from both measures are difficult to endorse unless one is more strongly eco-anxious).

We agree that the CCAS does not seem to capture the emotional core of eco-anxiety; however, its assessment of various impairments is theoretically consistent with the continuum notion of eco-anxiety (and anxiety more broadly) wherein eco-anxiety can become so severe that it may impede functioning (e.g., Clayton & Karazsia, 2020; Endler & Kocovski, 2001; Lazarus, 1991). As such, it could be argued that the CCAS is a measure of eco-anxiety, though one that captures more severe and perhaps less emotion-focused forms than the other measures we employed. Moreover, since eco-anxiety is commonly viewed as a more severe form of distress, one could argue that scales consisting of more severe items are assessing the part of the continuum in which researchers and clinicians seem to be most interested.

With respect to the measures of environmental concern, we observed that mean scores were above the scale midpoints. These results, coupled with those above, are consistent with the idea that it is possible to report significant concern/worry about the environment but not necessarily be impaired by these concerns/worries (Clayton & Karazsia, 2020; Feather & Williams, 2022; Schwartz et al., 2022). Correlational analyses further revealed that the Berenguer et al. (2005) environmental concern measure tended to be strongly positively correlated with the eco-anxiety measures, with the exception of the CCAS. Interestingly, the correlations between Schultz’s (2001) measure of biospheric concern and the above measures were notably weaker, and biospheric concern did not correlate with the CCAS. Still, it seems unlikely that eco-anxiety is a more self-focused version of concern, as egoistic concern was even less associated with eco-anxiety indicators.

In Studies 2, 3, and 5 we examined eco-anxiety’s associations with self-reported depression, anxiety, and stress. Overall, we found small to moderate positive associations between these variables, with some exceptions. In particular, none of the eco-anxiety measures were associated with depression in Study 2 and the Ballew et al. (2019) eco-anxiety measure evidenced a very weak negative association with depression and no associations with anxiety and stress in Study 3. Although we considered the Verplanken and Roy (2013) measure to assess a more severe form of eco-anxiety than the Helm et al. (2018) and Ballew et al. (2019) eco-anxiety measures, it did not tend to exhibit notably stronger associations with ill-being. However, in Study 5, the CCAS did have notably stronger associations with ill-being than the other eco-anxiety measures. These findings, coupled with those from prior work, point to a somewhat consistent positive association between eco-anxiety and ill-being, and also provide (albeit more limited) evidence for a stronger association between eco-anxiety and ill-being involving measures that assess more severe instances of eco-anxiety. Additional research is needed to draw a firmer inference about this trend.

In line with previous work (Feather & Williams, 2022; Landry et al., 2018; McBride et al., 2021), the measures of environmental concern tended to exhibit null or weak positive associations with ill-being. This may be because environmental concern measures tend to share less content overlap with ill-being scales than measures of eco-anxiety (McBride et al., 2021; see also Ojala et al., 2021) and/or that environmental concern indexes a less intense form of eco-anxiety. Another possibility is that aspects of general negativity (e.g., trait neuroticism) may strongly shape how eco-anxiety measures are related to measures of ill-being. Controlling for such personality features in future investigations may offer useful insights into the continuum of eco-anxiety responses (e.g., the dispositional tendency to experience negative affects may be a particularly strong driver of the association between more severe measures of eco-anxiety and ill-being).

In Studies 1, 3, 4, and 5 we examined eco-anxiety’s associations with various indices of well-being. Overall, we found a mixed set of findings across studies, wherein our eco-anxiety indices exhibited positive, negative, and null well-being relationships, broadly mirroring the current state of the literature. Of note, a greater number of significant relationships regarding well-being emerged in Study 3 than in the other studies, which may be due both to the high statistical power achieved in Study 3 and to the possibility that some items functioned differently due to cultural differences or translations (see limitations section). In studies where we did observe well-being relationships, we found that the Helm et al. (2018) measure tended to be positively associated with aspects of well-being (e.g., positive affect and autonomy need satisfaction in Studies 3 and 4), as did the Ballew et al. (2019) measure (e.g., autonomy need satisfaction in Studies 3 and 4). By contrast, the Verplanken and Roy (2013) measure was generally associated with poorer well-being (e.g., lower positive affect and greater negative affect in Studies 3 and 5), but not always (e.g., autonomy satisfaction in Study 4). In Study 5, the only well-being relationship we observed involving the CCAS was with greater negative affect. These results provide some evidence of more severe eco-anxiety measures being associated with poorer well-being. However, it is crucial to highlight that many of the associations were rather weak.

The measures of environmental concern tended to exhibit positive or null relationships with various aspects of well-being, which generally aligns with findings from past research (McBride et al., 2021; McConnell & Jacobs, 2020; Verplanken et al., 2020). Of note, the Berenguer et al. (2005) environmental concern measure had some notably stronger associations with aspects of well-being than the eco-anxiety measures, particularly in Studies 1 and 3.

In summary, although there were indications that eco-anxiety may be associated with poorer mental health, particularly when more severe instances are considered, not all relationships were significant and effect sizes were small to moderate overall. As such, it seems eco-anxiety may not represent as strong of a threat to mental health as has been previously proposed (e.g., Clayton, 2020; Passmore et al., 2022), at least as suggested by our data. It is also possible that some of the negative mental health impacts associated with eco-anxiety may be mitigated by well-being benefits derived from pro-environmentalism (see below). However, it is crucial to note that the seemingly smaller number of people experiencing more severe eco-anxiety is not inconsequential and that many environmental issues are projected to worsen, which may lead to more pronounced symptoms of eco-anxiety and accompanying mental health challenges (e.g., Clayton, 2020; Passmore et al., 2022). Continuing to monitor this relationship is a research priority, in our view.

Across studies, we tended to observe consistent positive medium-large associations between our measures of eco-anxiety and pro-environmental orientation (e.g., collective eco-guilt, empathy with nature, nature connectedness), suggesting eco-anxiety may have motivating and adaptive potential. At the same time, there were some notable caveats, particularly in Study 5. More specifically, CCAS total and subscale scores were unrelated to belief in the occurrence of climate change and collective eco-guilt, with the exception of the cognitive-emotional impairment subscale being positively related with collective eco-guilt. The findings concerning the belief in the occurrence of climate change were especially surprising given past work linking the CCAS with lower climate change denial (Wullenkord et al., 2021). It possible that a combination of potential floor (stemming from low means on the CCAS) and ceiling (stemming from the high mean on the present climate belief measure) effects resulted in notable attenuations in our observed relationships. CCAS total and subscale scores also tended to evidence notably smaller associations with pro-environmental consumption behaviour and environmental activist willingness than the other eco-anxiety measures, while the functional impairment subscale was unrelated to pro-environmental consumption behaviour. This is fairly consistent with previous research using the CCAS, which has found the scale to correlate positively with measures of pro-environmental intentions or behaviour, though not always (Clayton & Karazsia, 2020; Mouguiama-Daouda et al., 2022; Schwartz et al., 2022; Wullenkord et al., 2021). Given that the CCAS is intended to capture forms of impairment, this pattern of results may seem unsurprising. It is nonetheless noteworthy that despite the scale’s severe nature, it was still positively associated with the various indices of a pro-environmental orientation. Finally, mirroring past work (Ballew et al., 2019; Landry et al., 2018; McConnell & Jacobs, 2020; Verplanken et al., 2020), the measures of environmental concern also tended to exhibit consistent positive medium-large associations with indices of a pro-environmental orientation.

This set of findings are also encouraging from a mental health perspective, as some of these pro-environmental variables such as nature connectedness and pro-environmental behaviour have been associated with increased well-being (for meta-analyses, see Capaldi et al., 2014; Pritchard et al., 2020; Zawadzki et al., 2020). Therefore, although eco-anxiety may sometimes have a direct negative relationship with well-being, it may also have a positive indirect relationship through some of these pro-environmental indices.

At this point, we feel it is worthwhile to discuss how environmental concern may be related to eco-anxiety more extensively. To begin, if we were to rank our focal measures on a conceptual continuum of severity from low to high based on our results, we would place the CCAS on the high end followed by the Verplanken and Roy (2013) measure, the Helm et al. (2018) and Ballew et al. (2019) measures in no particular order, and lastly the Berenguer et al. (2005) measure on the low end. Past work indicates that concern, worry, and anxiety can be considered to belong to the same emotion family (Lazarus, 1991), that a link between affect and environmental issues is common to both eco-anxiety and environmental concern (e.g., Cruz & Manata, 2020; Passmore et al., 2022), and that asking about concern seems to share less content overlap with ill-being measures (McBride et al., 2021). These propositions are generally supported by our results in which the Berenguer et al. (2005) environmental concern measure tended to evidence strong positive associations with our more emotion-focused eco-anxiety measures and had more positive mental health relationships (e.g., sometimes stronger associations with well-being and lower associations with ill-being). As such, it could be argued that some measures of environmental concern, like the one from Berenguer and colleagues, could be considered to tap the low (less severe) end of the eco-anxiety continuum. Given our findings, we are unsure whether this may also apply to the Schultz (2001) measure of biospheric concern we collected. In outlining these points, we want to stress that we are not suggesting that the last 50 years of research on environmental concern should now be considered research on eco-anxiety. On their own, measures of environmental concern seem to be missing the assessment of more severe anxiety/worry related symptoms and phenomena, which are increasingly recognized as defining features of eco-anxiety. Concern could potentially be worked into new or existing eco-anxiety scales in an attempt to better capture the breadth of eco-anxiety responses. Future work should continue to explore how environmental concern is theoretically and empirically related to eco-anxiety. In the meantime, it is likely best to avoid using measures of environmental concern as measures of eco-anxiety until more work on this issue is done. Instead, we encourage researchers to employ formal eco-anxiety indices in their studies (e.g., Clayton & Karazsia, 2020; Hogg et al., 2021; Stewart, 2021).

The current research has notable limitations that should be considered when interpreting our results and conclusions. First, the reported studies were cross-sectional in nature, leaving ambiguous the causal nature of the relationships examined herein. Second, six measures employed in Study 3 (i.e., the three eco-anxiety measures, environmental concern, positive and negative affect, need satisfaction and frustration) were translated to Chinese by the fifth author. More rigorous translation procedures (e.g., back-translations) were not used and, even though the fifth author is fluent in English and Chinese, our translations may have been inexact. This, in addition to possible cultural differences, may help to explain some of the results of Study 3, such as the lower correlations among our focal measures. Still, general patterns of relationships observed in Study 3 were also observed in other studies (e.g., positive correlations among our focal measures across studies), though, on the whole, it is worth noting that the shifting power across studies may contribute to imprecision in our reported estimates.

Third, some of our focal measures differed in the environmental issue(s) participants were asked to respond about (e.g., climate change, pollution, or the “environment” more generally). Although it has been shown that people’s stress/anxiety about various environmental conditions load well onto a single factor (Helm et al., 2018; Hogg et al., 2021), participants still may have perceived or reacted to certain questions differently based on the issue(s) they were presented. Fourth, our focal measures not only differed in the environmental issues being assessed, but also in gradation, or “the psychological content that varies in magnitude along the continuum” (Tay & Jebb, 2018, p. 381). More specifically, the Helm et al. (2018), Ballew et al. (2019), Berenguer et al. (2005), and Schultz (2001) measures seem to capture variations in intensity, whereas the Verplanken and Roy (2013) measure and CCAS seem to capture a mix of behavioural extremity⁴ and frequency. Therefore, our measures are not technically on a single continuum. We considered the various measures we collected to fall on a conceptual eco-anxiety continuum that varies in severity, though we did not take a strong stance on the nature of the gradation and instead used the broader idea as an interpretive lens to make sense of the measures and their correlates. We recognize that our approach was rather crude and that there could be many differently graded eco-anxiety continua. As such, we consider the present research a preliminary investigation of the eco-anxiety continuum notion. We recommend researchers consider taking steps to integrate the concept of continua when revising existing definitions, models, and scales of eco-anxiety or creating new ones (see Tay & Jebb, 2018). Adopting common gradations and accompanying response scales for a given investigation, as well as using large samples, multi-item measures, factor analyses, and item response theory will assist such efforts.

Fourth, we would like to recognize that there are multiple factors besides those described above that may constrain the generalizability of our findings. Setting, cultural context, and age are particularly important to consider. With respect to setting, it is possible that measures of eco-anxiety could lead to more pronounced results among samples of people that are more vulnerable to the direct impacts of climate change than those studied here (e.g., Gibson et al., 2020; Hickman et al., 2021). Relatedly, there are also likely important cultural differences (e.g., in beliefs and attitudes concerning the environment) that shape the experience and consequences of eco-anxiety. These differences may have contributed to some of the divergent results we observed between our western samples (Canada, UK) and our sample from China. Our samples were also mainly comprised of relatively young adults, and past research has shown that younger people may be particularly likely to experience eco-anxiety (e.g., Clayton & Karazsia, 2020; Hickman et al., 2021). Therefore, our findings may not generalize to samples that differ notably from those studied here. At the same time, we propose that the notion of a continuum of eco-anxiety responses may be widely applicable. It seems likely that the experience of eco-anxiety is not limited to the severe end of the continuum for all people. As an example, even individuals from parts of the world that seem especially vulnerable to the direct impacts of climate change (e.g., Tuvalu, the Philippines) exhibit variation in reported symptoms, even if more clustered around the severe end (e.g., Gibson et al., 2020; Hickman et al., 2021). Taken together, even though results may differ across samples (e.g., eco-anxiety may be experienced as more severe and thus potentially more maladaptive in samples especially affected by environmental issues), there will likely still be (at least some) variability in responses across the continuum and attempting to capture and study a fuller range of these responses will contribute to a richer holistic understanding of this important construct.

Limitations notwithstanding, the present research may provide important preliminary insights into the continuum of eco-anxiety responses and its relationships with environmental concern, mental health, and a pro-environmental orientation. Environmental crises, on a global level, will only continue to rise over the coming years, and with it a range of eco-anxiety responses. As such, deeper insights into these various responses and their implications will be vital as we individually, and collectively, deal with the challenges ahead.

Contributed to conception and design: PKL, H-AP, AJH, JMZ

Contributed to acquisition of data: PKL, H-AP, AJH, JMZ, YY, MR

Contributed to analysis and interpretation of data: PKL, H-AP, AJH, JMZ

Drafted and/or revised the article: PKL, H-AP, AJH, JMZ, YY, MR

Approved the submitted version for publication: PKL, H-AP, AJH, JMZ, YY, MR

The research was supported by a Social Sciences and Humanities Research Council (SSHRC) of Canada Insight Grant to John M. Zelenski [435-2020-0852]. Paul K. Lutz was supported by a SSHRC Graduate Scholarship – Master’s (CGS-M). SSHRC did not have a role in determining the aims and outcomes of the current research or in the decision to submit the article for publication.

The authors have no competing interests to declare.

Materials, data, and syntax for all studies can be found on OSF: https://osf.io/krfnv/

We thank the editor and two anonymous reviewers for their important contributions to this work.