Compassion Towards Nature and Well-Being: The Role of Climate Change Anxiety and Pro-Environmental Behaviors

Abstract

The current study examined the structure and psychometric characteristics of the Portuguese version of the Climate Change Anxiety Scale (CAS), exploring the climate change anxiety and environmental action mediator effects on the relationship between connection to nature and well-being, and whether self-compassion and receiving compassion from others would moderate these associations. The sample was composed of 522 participants from the general population who completed a set of self-report measures through an online survey. Exploratory factor analysis results of the CAS extracted four factors, and a confirmatory factor analysis revealed an acceptable fit of this structure to the data (χ2 (164) = 328.67; p < 0.001, CMIN/DF = 2.004; CFI = 0.94; TLI = 0.93; RMSEA = 0.06 [90% CI 0.05–0.07; p < 0.001]). The CAS demonstrated good internal consistency (α = 0.92) and convergent, concurrent, and divergent validities. Two moderated mediation models showed a significant direct effect of connection to nature on well-being, and self-compassion and compassion from others significantly moderating this relationship. Furthermore, in both models, climate change anxiety and environmental action significantly mediate the association between connection to nature and well-being. Overall, connection with nature seems to enhance well-being in lower and medium levels of self-compassion and from others but may increase climate change anxiety in all compassion levels, reinforcing the importance of promoting nature-based interventions combined with compassion-focused programs to foster well-being.

1. Introduction

Currently, climate change issues represent one of the foremost challenges confronting humanity. Media outlets frequently report on severe weather events such as intensified storms, hurricanes, extreme heat, and sea-level rise, attributing them to anthropogenic impacts on the climate [1]. Reports from the Intergovernmental Panel on Climate Change underscore the role of human activities in driving atmospheric, oceanic, and terrestrial warming. Hence, there has been a notable increase in weather and climate extremes, including heatwaves, heavy precipitation, droughts, and tropical cyclones, resulting in widespread damage to ecosystems and posing threats to food and water security [2]. Moreover, there is an elevated risk of epidemics and pandemics due to these climate-related changes [3]. Climate change has also had significant impacts on human physical and mental health. Specifically, extreme climate events have been linked to adverse mental health outcomes, including poor mental health, increased psychiatric hospitalizations, elevated mortality rates among individuals with mental illness, and heightened suicide rates [4]. This risk is particularly heightened among populations with pre-existing mental illness, low- and middle-income individuals, and those residing in tropical climate zones. Research has demonstrated a positive association between mortality and organic mental disorders, suicides, and self-harm, as well as between morbidity and mood disorders, organic mental disorders, schizophrenia, and anxiety disorders [5]. Furthermore, studies have identified challenges in coping with emotions such as fear, anxiety, worry, guilt, and sadness in the face of climate-related stressors [6]. Thus, direct exposure to extreme events and environmental disasters often leads to profound psychological trauma, characterized by feelings of terror, anger, and shock. This exposure is also linked to elevated rates of various psychopathological conditions, including post-traumatic stress disorder, generalized anxiety disorder, depression, phobias, as well as substance abuse disorders involving alcohol and drugs [7].

1.1. Connection to Nature, Mental Health, and Well-Being

From an evolutionary standpoint, humans have been intimately intertwined with the natural world since time immemorial, evolving within this relationship to adapt and survive. Through the utilization of senses, emotions, and cognition, humanity has adapted to the overarching goals of survival and reproduction. Mental patterns have evolved in ancestral minds, facilitating the achievement of these goals [8] (pp. 49–66). One such example is the intrinsic need for belonging, where ancestral human species recognized that staying together increased chances of survival, particularly during times of scarcity [9]. Attachment theory elucidates this process, highlighting the intrinsic motivation of young children to seek comfort and security from their attachment figures [10]. Additionally, the concept of place attachment underscores the search for places that provide comfort and security [11] (pp. 3–37). Despite nature offering sustenance and shelter, it also presented numerous dangers, threatening human lives. The evolution of higher mental processes, such as planning, anticipation, and imagination, enabled humanity to develop better defenses, distancing itself from the perils of the natural world. Nevertheless, physical and mental health, as well as overall well-being, remain reliant on nature, given that 99% of human existence transpired within this environment [8]. Moreover, the biophilic hypothesis [12] posits the existence of an intrinsic affinity towards life within human identity. This hypothesis encompasses various aspects, including attraction, rationality, aversion, exploitation, affection, control, self-identity, and symbolism, all of which contribute to a deep-seated connection with nature [13]. Summarily, the biophilic hypothesis asserts a biological predisposition towards being attracted to, dependent on, and affiliative with nature [14].

Three classical theories have been proposed to elucidate the benefits of connecting with nature. Firstly, the stress-reduction theory [15] posits that exposure to non-threatening natural environments reduces stress levels and negative emotions while improving sustained attention. Secondly, the attention-restoration theory [16] suggests that interacting with nature aids in restoring attention by shifting from top-down attentional processes (e.g., thoughts, rumination) to bottom-up processes (e.g., sensory engagement). Lastly, the prospect-refuge theory [17] argues that landscapes offering both expansive views (prospect) and places for concealment and protection (refuge) contribute to enhanced cognitive restoration [18]. Presently, research is beginning to demonstrate that increased engagement in nature-based activities, such as forest bathing (FB), can improve mental health and well-being [19,20]. Furthermore, combining FB with compassion-focused interventions (e.g., compassion mind training) has been shown to improve nature connection scores. Consequently, the development of a reliable measure of connection to nature becomes crucial, prompting environmental psychologists to conceptualize this construct. Various measures, such as the Connectedness to Nature Scale [21], emphasize the presence of emotions associated with belonging to the natural world, while others, like the Inclusion of Nature in the Self, focus on cognitive beliefs [22]. Additionally, the Commitment to Nature scale [23] proposes an interpersonal relationship between humans and nature, highlighting their deep interdependence on well-being. Multidimensional scales, such as the Environmental Identity Scale [24], assess interaction with nature, the importance of nature, membership in nature, and positive emotions towards nature. Overall, there exists a robust convergence among these measures, except for the Inclusion of Nature in the Self [25]. Consequently, a broader construct termed “Connection to Nature” has been delineated, emphasizing its strong association with well-being and pro-environmental behaviors [25], grounded in the innate connection between human beings and nature. In addition, a meta-analysis has shown that more connection to nature predicts more pro-environmental behaviors [26].

In this study, nature is defined expansively, encompassing all spaces where plants grow, whether by human design or not, referring to areas typically characterized by greenery and the presence of water [27].

1.2. Connection to Nature and Climate Change Anxiety

In the latest report, the Directorate-General for Climate Action of the European Commission [28] highlights that Portuguese residents over 15 years old perceive climate change as the third most serious global issue, with 89% considering it an extremely severe problem. Furthermore, 68% of respondents have taken action to combat climate change in the past six months, and 28% of them believe they are personally responsible for these efforts [28]. Climate change poses a psychologically distant risk to humans, as it is anticipated to occur in the future and affects other people and places [29]. Therefore, understanding climate change risk perception necessitates consideration of four psychological dimensions: cognitive, experiential, socio-cultural, and sociodemographic characteristics. Moreover, an anxiety cycle emerges only when a perception of threat is present [30], with research indicating varied risk perceptions across these dimensions [31].

In the cognitive dimension, knowledge about climate change contributes to increasing risk perception, although distinctions between subjective and objective knowledge remain unclear. The experiential process involves affect and emotion guiding information processing [32], with individual experiences and affective evaluations from media influencing risk perception, particularly when directly experiencing climate change events [33]. While indirect experiences of climate change exist, their impact on risk perception is lower. Socio-cultural influences, encompassing culture, values, worldviews, and social norms, contribute to variations in climate change risk perception. This includes differences between environmental domains with egoistic values, which prioritize individual outcomes, socio-altruistic values, oriented towards caring for others, and biospheric values, focused on caring for non-human nature and the biosphere itself [34,35]. Overall, evidence suggests that younger individuals, females, those with higher education levels, politically liberal individuals, and racial minorities exhibit greater concerns about climate change [36].

Henceforth, individuals exhibiting higher scores in connection with nature also demonstrate elevated levels of climate change risk perception and heightened anxiety. This emotional response serves an adaptive purpose, as it is oriented towards the future, involving the anticipation of potential threats and the formulation of strategies to mitigate them and has a survival function by priming both the physical body and mind for rapid action [37]. However, excessive anxiety can have maladaptive consequences, leading to emotional dysregulation [38]. Within the realm of climate psychology, this specific manifestation of anxiety is conceptualized as climate change anxiety. Consequently, individuals who feel a stronger connection to nature and exhibit greater environmental concern are more likely to be vigilant towards climate change threats and thus experience heightened anxiety [39]. Nevertheless, several factors influence the escalation of climate change anxiety. For instance, individuals who lack a sense of connection to nature but have experienced climate change-related events directly or indirectly, such as storms, droughts, or loss of habitat, are prone to heightened anxiety characterized by worry and trauma [36]. Conversely, individuals who harbor a stronger sense of connection to nature, accompanied by heightened feelings of care and concern, tend to exhibit greater vigilance towards environmental threats. Consequently, they experience elevated levels of climate change anxiety, along with increased symptoms of stress and depression, and lower levels of overall well-being [39].

1.3. Climate Change Anxiety, Mental Health, and Well-Being

The threat of climate change is intricately linked to mental health outcomes, though the nature of this association varies in its relation to mortality and morbidity. Certain populations, particularly those in tropical and subtropical regions, as well as individuals aged over 65 years, exhibit greater vulnerability to these mental health impacts [5]. Additionally, a relationship has been identified between latitude, temperature, and mortality, with lower minimum mortality rates observed in higher latitudes. Moreover, numerous studies have linked high ambient temperatures with an increased risk of suicide [4,40], as well as an elevated risk of mental-health-related emergency department visits [40,41], and higher mortality rates among individuals with pre-existing mental health conditions [4]. One study even reported that increased suicide rates in the United States and Mexico were influenced by the use of depressive language on social media [40].

Furthermore, the threat of climate change is directly associated with trauma and indirectly affects community well-being [42]. This impact is particularly pronounced among family farmers, leading to a compromised sense of self-identity and chronic forms of place-based distress, heightening perceived risks of depression and suicide [43]. Conversely, individuals in the United States and New Zealand may choose to go childfree for both biospheric and altruistic reasons [23].

1.4. Climate Change Anxiety and Pro-Environmental Behavior

The challenges posed by anxiety often prompt individuals to engage in a range of coping mechanisms, especially those associated with defensive responses to (physical, social, or psychological) threats, such as fight, flight, or freeze behaviors. Flight behavior typically involves denial of climate change, freezing behavior may include attempts at distraction or pathological worry [44], and fight behavior often results in the adoption of pro-environmental behaviors [45]. Furthermore, research indicates that individuals with a stronger connection to nature are more inclined to engage in the latter [26]. Pro-environmental behaviors are conscious actions aimed at reducing the negative impact of human activity on the natural or built environment [46]. These behaviors are influenced by internal factors like personality traits and values, as well as external factors such as political, social, cultural, and economic circumstances.

Moreover, environmental consciousness encompasses knowledge (e.g., understanding climate change), emotions (e.g., fear of climate change), and values/attitudes (e.g., concern for the natural world) [46]. Values that drive pro-environmental behaviors include childhood experiences in nature, witnessing environmental degradation, familial and organizational values, influence from role models such as friends or teachers, and educational background [47].

Additionally, individuals with higher levels of biospheric environmental concerns are more inclined to perceive ecological stress, employ ecological coping strategies, and demonstrate pro-environmental behaviors [48]. Conversely, those with social-altruistic concerns may engage in ecological coping without necessarily perceiving ecological stress, while individuals with egoistic concerns neither perceive ecological stress nor engage in ecological coping [48].

In another study, evidence suggests that risk perception indirectly influences anxiety and worry, while response efficacy (belief in one’s effectiveness and desired consequences) indirectly and directly influences behavior, and psychological adaptation (affective, cognitive, and motivational readiness to address climate change issues) directly influences behavior [29]. Conversely, individuals with a strong green identity (someone who identifies as friendly to the natural environment) are more likely to exhibit pro-environmental behaviors [49].

Moreover, research highlights various motives associated with different pro-environmental behaviors, without discernible patterns [50]. These motives are often linked to emotions, with empathic feelings and attitudes, stemming from perspective-taking, fostering helping behaviors [34]. Different studies have identified a range of emotions that drive pro-environmental behaviors, including anxiety [39], worry [44], fear, guilt, and hope [51], as well as eco-anger, eco-anxiety, and eco-depression [52] and feelings of hopelessness [53].

Interestingly, these behaviors can manifest as either individualistic actions (e.g., recycling, conserving energy) or collective endeavors (e.g., participating in political protests), and individuals with the same motives may exhibit either individual or collective behaviors [50].

1.5. Pro-Environmental Behavior, Mental Health, and Well-Being

Although ecological coping has been shown to decrease depressive symptoms [48] and sustainable behaviors have been associated with increased happiness [54], psychopathology linked to anxiety often arises when coping mechanisms, such as pro-environmental behaviors, prove ineffective in addressing and resolving the underlying issue, thereby perpetuating a sense of threat perception. For example, in the context of the climate change threat, the sixth report from the Intergovernmental Panel on Climate Change underscores the ongoing escalation of the threat [2]. Consequently, coping behaviors like denying the problem (flight), engaging in distraction or pathological worry (freeze), or actively participating in pro-environmental actions (fight) may not yield the desired outcomes, thereby exacerbating anxiety and diminishing overall mental health and well-being [55].

Furthermore, pro-environmental behaviors could exacerbate distress and anxiety because they may fail as a coping strategy [56]. Different emotions predict varying coping outcomes, with one study identifying that eco-anger predicts better mental health compared to eco-depression and eco-anxiety, which are less adaptive and lead to lower well-being [52].

Overall, there is more evidence suggesting that environmental activism protects against mental distress and improves well-being compared to individual pro-environmental behavior [57]. It acts as a buffer, with a study concluding that collective activism significantly attenuated the association between climate change anxiety and major depressive disorder more than individual pro-environmental behavior [45]. It is asserted that activism protects against feelings of hopelessness [53], lowers symptoms of general anxiety disorder, and fosters feelings of empowerment and social support [58], contributing to long-term mental health [59].

1.6. The Role of Compassion

Compassion is a concept deeply rooted in various spiritual and moral philosophical traditions, originating from the Latin term “compassio”, which denotes the act of sharing suffering with others. It encompasses understanding another person’s pain, empathizing with their perspective, and desiring to alleviate their suffering [60]. Compassion plays an evolutionary role intertwined with attachment processes, fostering affiliative and cooperative behaviors through physiological mechanisms such as the activation of the vagus nerve [61]. As an evolved motivational system, compassion emerges from a blend of innate caregiving motives shared with other mammals and higher-order cognitive processes associated with social and emotional intelligence unique to humans [62]. Compassion entails various competencies and motivations, including care for well-being, sensitivity (attentiveness to suffering), sympathy (emotions arising from experiencing suffering), distress tolerance (ability to tolerate feelings arising when engaging with suffering), empathy (seeing the world through others’ eyes), and non-judgment (open acceptance that aids distress tolerance). These competencies facilitate wise actions such as attentive behavior (recalling knowledge and wisdom), imagery (envisioning compassionate practices), reasoning (sound judgment), behaviors (compassionate actions), sensory awareness (being aware of bodily states and cultivating compassionate bodily states), and feelings (emotions of compassion) [62]. Within a compassionate mindset, three flows are recognized: self-compassion, compassion towards others, and others’ compassion towards oneself. Furthermore, compassion is depicted as an intra- and interpersonal dynamic process involving these three flows [63]. Viewing compassion as motivation prompts consideration of its role in the relationship between connection to nature, climate change threat perception, pro-environmental behaviors, and their consequent impacts on well-being. There is evidence in the literature indicating an association between connection to nature and compassion. On one hand, prosocial behaviors are known to increase with a stronger connection to nature [64,65]. A common hypothesis explaining this association involves the experience of self-transcendent emotions, which include wonder, awe, gratitude, and compassion [66]. On the other hand, heart rate variability (HRV), a marker of parasympathetic nervous system activity, has been found to be associated with both compassion [67] and connection to nature [68]. Thus, enhancing one’s connection to nature might elevate levels of compassion, and both factors can increase HRV, thereby fostering prosocial behavior [69] and pro-environmental behaviors [65]. Furthermore, Cassidy and Elliot [70] hypothesized a mediation of self-compassion [71] in the association of pro-environmental behaviors and mental well-being, linking self-compassion with hope. They concluded that higher levels of self-compassion are associated with increased pro-environmental behaviors and higher levels of well-being but did not clarify the direction of causality between self-compassion, pro-environmental behaviors, and well-being.

Interestingly, some research has established a link between connection to nature and the theoretical construct of nondual compassion [72]. Nondual compassion is described as a state of unified compassion and unconditional love, coupled with a comprehensive understanding of the nature of the mind [73]. This construct is characterized as an altered state of consciousness and is viewed as a trait that functions along a spectrum of compassion for self, others, and both self and others [72]. It is associated with nondual awareness, which is characterized by human experience devoid of the dualities of subject and object, self and other, and mind versus body [74,75]. Changes have been observed in delta, alpha, and theta brain activity in response to both exposure to nature and nondual awareness [76] leading to an integration of internal and external experiences [72], simultaneously activating the default mode network and goal-oriented networks [74], and evidencing a state of oneness, uniting extrinsic and intrinsic experiences. Additionally, feeling a part of a broader system, such as nature, rather than separate from it, could increase nondual compassion and understanding interconnection and interdependence of all life forms, promoting a harmonious and compassionate relationship with the natural world.

1.7. The Current Study

In the Anthropocene era, a term coined by Paul Crutzen and Eugene Stoermer [77] to describe the current geological epoch marked by significant human impact on Earth and climate change, it is crucial to understand and improve the relationship between humans and nature to create a balanced, win-win interaction. Numerous studies indicate that a better relationship with nature enhances well-being and physical and mental health [25]. However, the nature of this relationship can vary significantly based on how individuals relate to nature [34]. For instance, a person who has greater concerns with environmental issues will have decreased well-being and mental health when closer to nature. Therefore, it is essential to understand the mechanisms that underline the connection between nature and well-being. By doing so, we can enhance our social, economic, and political systems to foster human and environmental well-being as a symbiotic entity. Exposure to nature, which enhances well-being, could also increase awareness of climate change, potentially leading to direct or indirect suffering. Emotions elicited by climate change (e.g., anxiety) can drive coping mechanisms, such as pro-environmental behaviors, decreasing anxiety, and thereby enhancing well-being [54]. Interestingly, literature differentiates between individual and collective actions, showing varied outcomes based on the type of behavior [57]. Consistent with existing research, we hypothesize that environmental action positively influences well-being by improving physical and mental health, fostering social connections, promoting economic benefits, and ensuring long-term sustainability and equity [58]. Given the presence of suffering when humans interact with nature through increased awareness of climate change, which is associated with difficult emotions such as fear, anxiety, guilt, and anger [78], and considering that compassion is a motivational construct with strong evidence linking it to improved well-being and mental health [79], we hypothesized that compassion may play a role in the relationship between humans and nature. Moreover, it is important to distinguish between receiving compassion (from the self and others), which entails engaging and acting towards the prevention/alleviation of one’s own suffering, and compassion for others, which entails engaging with and preventing/alleviating the suffering of others [63].

From a psychological standpoint, recognizing and understanding suffering is the initial step towards alleviating it. We hypothesize that individuals who are more aware of their own suffering may experience lower levels of climate change anxiety. Although awareness of others’ suffering can lead to prosocial behaviors and pro-environmental behaviors [48], these actions are responses to human suffering rather than environmental suffering. Individuals with higher compassion for others may not necessarily demonstrate increased compassion for nature.

Therefore, the main aim of the current study was to investigate what mechanisms/factors could explain the relationship between connection to nature, climate change anxiety and well-being, and whether, among individuals with higher levels of self-compassion and compassion from others, a stronger connection to nature is associated with lower climate change anxiety and higher well-being. The specific aims were to: (1) Adapt the Climate Change Anxiety Scale [39] to the Portuguese language and examine its factor structure and psychometric properties in the Portuguese population; (2) Analyze a moderated mediation model, with connection to nature as a predictor of well-being, mediated by climate change anxiety and pro-environmental behaviors; (3) Examine compassion’s flow towards oneself (from self and others) as moderator, investigating whether higher levels of compassion’s flow towards oneself, are associated with a weaker relationship between connection to nature and climate change anxiety and a stronger relationship between connection to nature and well-being.

Deriving from Specific Aims 2 and 3, the following hypotheses were formulated: (1) Connection to nature would be associated with well-being and with climate change anxiety; (2) Climate change anxiety would mediate association between connection to nature and pro-environmental behaviors, and between connection to nature and well-being; pro-environmental behaviors would mediate the relationship between connection to nature and well-being, and between climate change anxiety and well-being; (3) Compassion flow towards oneself (from self and others), would moderate the associations between connection to nature and climate change anxiety, and between connection to nature and well-being.

2. Materials and Methods

2.1. Ethics

Ethics approval for the present study was granted by the Ethics and Deontology of Research Commission at the Faculty of Psychology and Education Sciences in the University of Coimbra (UC; CEDI:FPCEUC:81/4). All research was performed in accordance with regulations specified in the ethics approval and in accordance with the Declaration of Helsinki.

2.2. Procedure

Participants were recruited online from the Portuguese general population between February and March 2024, informed about the study’s aims and procedures, provided informed consent prior to participation, and guaranteed complete anonymity. The study was conducted using an online survey, self-paced and about 20 min long, combining different measures, hosted at the University of Coimbra institutional account in the online platform https://www.limesurvey.org/pt/ (accessed on 20 May 2024), and was disseminated through social media platforms, institutional emailing lists, and using snowball sampling, and there was no payment for completing the survey. Inclusion criteria were being between 18 and 65 years of age and living in Portugal.

2.3. Participants

A sample of 522 participants was recruited. The average age of participants was 44.49 years (SD = 11.68), with ages ranging from 18 to 65 years. The sample consisted of 361 identified as cisgender women (69.16%), 158 as cisgender males (30.27%), 1 non-binary (0.19%), and 2 not answered (0.38%). Marital status distribution included 320 individuals (61.3%) who were married or living with a partner, and 202 individuals (38.7%) who were single, divorced, or widowed. The majority of participants were Caucasian (n = 445, 85.2%). Participants resided in urban areas (n = 234, 44.8%), rural areas (n = 209, 40%), and suburban areas (n = 79, 15.1%). Educational attainment varied, with 333 participants (63.8%) having completed high school and 189 participants (36.2%) having completed basic or secondary education. Most participants had children (n = 350, 67%), while 172 participants (33%) did not have children. Regarding socioeconomic status, 243 participants (46.6%) were at a medium level, 122 participants (23.4%) at a low level, 106 participants (20.3%) at a high level, and 51 participants (9.8%) did not provide income information. Most participants reported good physical health (n = 265, 50.8%) and good mental health (n = 266, 51%). A minority of participants were undergoing psychiatric or psychological treatment (n = 67, 12.8%), while 453 participants (86.8%) were not. Additionally, the majority (77.8%) did not engage in daily contemplative practice, whereas 109 participants (20.9%) reported having contemplative practice.

For this study, the sample was randomly subdivided into two subsets (n1 = 261 and n2 = 261). These two subsets didn’t present statistically significant differences regarding distribution of sociodemographic information (age, gender, area of residence, ethnicity, education level, marital status, having children, socioeconomic level, physical and mental health level, psychiatric or psychological counseling (yes/no), and contemplative practices (

Table A1. Sociodemographic differences between samples.

Variables	Sample 1 (n = 261)	Sample 2 (n = 261)	t/χ2	d/V
Age M(SD)	44.29 (11.82)	44.65 (11.71)	−0.34 (520)	0.03
Gender n (%) Female	187 (71.6)	168 (67.47)	3.97	0.09
Residence n (%) Urban	115 (44.06)	113 (45.38)	1.16	0.05
Ethnia n (%) Caucasian	214 (81.99)	220 (88.35)	10.64	0.14
Education level n (%) Higher education	165 (63.22)	160 (64.26)	0.06	0.01
Marital status n (%) Married	163 (62.45)	148 (59.44)	0.49	0.03
Children n (%) Yes	183 (70.11)	159 (63.86)	2.26	0.07
Socioeconomic level n (%) Medium	123 (47.13)	117 (40.21)	0.01	0.00
Physical health n (%) Good	145 (55.56)	115 (46.18)	8.02	0.13
Mental health n (%) Good	131 (50.19)	130 (52.21)	2.25	0.07
Psi/psi counseling n (%) No	227 (86.97)	216 (86.75)	0.01	0.00
Contemplative practices n (%) No	207 (79.30)	190 (76.31)	2.38	0.07

of Appendix A).

2.4. Measures

2.4.1. Climate Anxiety Scale (CAS) [39]

Climate Change Anxiety Scale is a self-reported questionnaire, designed to measure anxiety to climate changes, with two subscales, Cognitive–Emotional Impairments with 8 items (e.g., “I have nightmares about climate change”) describing difficult thoughts and feelings related to climate changes, and Functional Impairments with 5 items (e.g., “My friends say I think about climate change too much”), describing difficulties in daily routines and relationships related to climate changes. Higher scores indicate higher levels of anxiety about climate change. Clayton and Karazsia also identified as two factors related to climate changes anxiety, Experience of Climate Change with 3 items (e.g., “I was directly affected by climate change”), describing direct experience with climate change phenomenon, and Behavioral Engagement with 6 items (e.g., “I recycled”), describing pro-environmental behaviors. Those 22 items were included in the exploratory (EFA) and confirmatory (CFA) factor analysis. Participants indicated how often they experienced each statement over the last two weeks with a 5-point Likert scale (1 = never, 2 = rarely, 3 = sometimes, 4 = often, 5 = almost always). The Cronbach alphas indicated good internal consistencies (Cognitive–Emotional = 0.97, Functional = 0.79, Experience = 0.86, and Behavioral = 0.94). These scales were then translated and adapted to the Portuguese population, and the back translations were examined by a bilingual English teacher to examine the accuracy and fidelity of the original scales.

2.4.2. Hogg Eco-Anxiety Scale (HEAS) [80,81]

The Hogg Eco-Anxiety Scale is a self-reported questionnaire designed to measure eco-anxiety with four subscales: Affective Symptoms with four items (e.g., “Feeling afraid”), Rumination with three items (e.g., “Unable to stop thinking about losses to the environment”), Behavioral Symptoms with three items (e.g., “Difficulty sleeping”), and Personal Impact Anxiety with three items (e.g., “Feeling anxious about the impact of your personal behaviors on the earth”). Higher scores indicate higher levels of eco-anxiety. Participants indicated how often they experienced each factor of eco-anxiety over the last two weeks when thinking about climate change and other global environmental conditions with a 4-point Likert scale (0 = not at all, 1 = several of the days, 2 = over half the days, 3 = nearly every day). The Cronbach alphas indicated good internal consistencies, with Affective Symptoms = 0.92 (Portuguese version = 0.85), Rumination = 0.90 (Portuguese version = 0.89), Behavioral Symptoms = 0.86 (Portuguese version = 0.86), and Personal Impact Anxiety = 0.88 (Portuguese version = 0.92). In this study, the Cronbach alpha was 0.85, 0.89, 0.85, and 0.91, respectively.

2.4.3. Environmental Action Scale (EAS) [57,82]

The Environmental Action Scale is a self-reported questionnaire, aimed at evaluating the degree of involvement of people in collective actions in favor of the environment, with two subscales, Participatory Actions with 10 items (e.g., “Participated in an educational event”) and Leadership Actions with 8 items (e.g., “Organized an educational event”). Higher scores indicate higher involvement in collective actions. Participants indicated, in the last six months, how often, if at all, they have engaged in the related environmental activities and actions with a 4-point Likert scale (0 = never, 2 = sometimes, 4 = frequently). The Cronbach alphas indicated good internal consistencies in Participatory Actions = 0.92 (Portuguese version = 0.90) and Leadership Actions = 0.92 (Portuguese version = 0.96). In this study, the Cronbach alphas were 0.90 and 0.83, respectively.

2.4.4. Brief Version Connectedness to Nature Scale (CNS-Brief Version) [83,84]

Brief Version Connectedness to Nature Scale is a self-reported questionnaire, designed to measure connection to nature with 7 items (e.g., “I often feel a kinship with animals and plants”). Higher scores indicate higher connection to nature. Participants were invited to answer, on a 5-point Likert scale ranging from 1 (completely disagree) to 5 (completely agree). The Cronbach alpha indicated good internal consistency, α = 0.87 (Portuguese version = 0.83). In this study the Cronbach alpha was 0.89.

2.4.5. Compassionate Engagement and Action Scales (CEAS) [85,86]

This self-report questionnaire encompasses three subscales measuring the three orientations/flows of compassion: Self-compassion, Compassion to Others, and Compassion from Others [85]. Each subscale has 13 items and is divided into two sections. The first (engagement) is related to the motivations and capacities to deal with suffering. It includes the six compassion engagement elements: motivation to care for well-being, attention/sensitivity to suffering, sympathy, distress tolerance, empathy, and being accepting and non-judgmental [85]. The second section (compassionate action) is related to the ability to pay attention to learn about and act on what is helpful—developing the wisdom and commitment to do something about it. It includes compassion action elements: directing attention to what is helpful, thinking and reasoning about what is likely to be helpful, taking helpful actions, and creating feelings of support, kindness, helpfulness, and encouragement to deal with distress. The items are rated on a 10-point Likert scale (1–10). Higher scores indicate higher levels of compassion. The CEAS was found to have robust psychometric properties. In the scale Compassion to Others—engagement (e.g., “I tolerate the various feelings that are part of other people’s distress”), the Cronbach alpha was α = 0.90 (Portuguese version α = 0.82), and in this study, 0.88. The Cronbach alpha for “Compassion to others—Actions” (e.g., “I am able to take the actions and do the things that will be helpful to others.”) was α = 0.94 (Portuguese version α = 0.90), and in this study, 0.95. In Compassion for Self—engagement the Cronbach alpha for the 2 items emotional sensitivity scale (e.g., “I am emotionally moved by my distressed feelings or situations.”) was α = 0.77 (Portuguese version α = 0.72; in this study 0.68) and α = 0.72 (Portuguese version α = 0.63; in this study 0.82) for the 4 items’ engagement with suffering scale (e.g., “I am motivated to engage and work with my distress when it arises”). In the scale Compassion for Self—actions (e.g., “I direct my attention to what is likely to be helpful to me.”), the Cronbach alpha was α = 0.90 (in this study was 0.94), like in the Portuguese version. In Compassion from Others—engagement (e.g., “Others are emotionally moved by my distressed feelings.”), the Cronbach alpha was α = 0.89, like in the Portuguese version (in this study was 0.93). Finally, in the scale Compassion from Others-Actions (e.g., “Others direct their attention to what is likely to be helpful to me.”), the Cronbach alpha was α = 0.91 (Portuguese Version α = 0.92; in this study 0.95). For the purposes of this study, participants completed the Compassion for Self and Compassion from Others subscales.

2.4.6. Depression, Anxiety, and Stress Scale (DASS 21) [87,88]

The Depression, Anxiety, and Stress Scale is a self-reported questionnaire, designed with three subscales, Depression (e.g., “I felt that life was meaningless”), Anxiety (e.g., “I was aware of dryness of my mouth”) and Stress (e.g., “I found myself getting agitated”), with 7 items each one. Higher scores indicate higher levels of depression, anxiety, and stress. Participants were invited to answer, on a 4-point Likert scale ranging from 0 (did not apply to me at all) to 3 (applied to me very much, or most of the time). The Cronbach alphas indicated good internal consistencies in Depression, α = 0.91 (Portuguese version = 0.85; in this study 0.90), in Anxiety, α = 0.84 (Portuguese version = 0.74; in this study 0.88), and in Stress, α = 0.90 (Portuguese version = 0.81; in this study 0.91).

2.4.7. PERMA-Profiler (PERMA) [89,90]

PERMA-Profiler is a self-reported questionnaire, designed with seven subscales: Positive Emotion (P) with 3 items (e.g., “In general, to what extent do you feel contented?”); Engagement (E), 3 items (e.g., “In general, to what extent do you feel excited and interested in things?”); Relationship (R), 3 items (e.g., “To what extent do you feel loved?”); Meaning (M), 3 items (e.g., “In general, to what extent do you lead a purposeful and meaningful life?”); Accomplishment (A), 3 items (e.g., “How often are you able to handle your responsibilities?”); Negative Emotion, 3 items (e.g., “In general, how often do you feel anxious?”); Physical Health, 3 items (e.g., “In general, how would you say your health is?”); Loneliness, 1 item (“How lonely do you feel in your daily life?”); and Happiness, 1 item (“Taking all things together, how happy would you say you are?”). The average of Happiness together with the 15 PERMA items shapes the overall well-being subscale. The answers range on a Likert scale from an 11-point response format. The results of the PERMA-Profiler are calculated by averaging the items that make up each factor. The scores of the PERMA-Profiler indicate the flourishing psychological profile, the level of overall well-being, negative emotion, physical health, and loneliness. The Cronbach alpha of scale Positive Emotion was α = 0.88 (Portuguese version = 0.87; in this study 0.94), of Engagement was α = 0.72 (Portuguese version = 0.56; in this study 0.73), of Relationship was α = 0.82 (Portuguese version = 0.77; in this study 0.79), of Meaning was α = 0.90 (Portuguese version = 0.84; in this study 0.88), of Accomplishment was α = 0.79 (Portuguese version = 0.71; in this study 0.82), of Negative Emotion was α = 0.71 (Portuguese version = 0.71; in this study 0.76), and of Physical Health was α = 0.92 (Portuguese version = 0.86; in this study 0.90).

2.5. Data Analysis

Data were analyzed using the Statistical Package for Social Sciences (IBM SPSS, version 27.0) and AMOS 27 software [91]. Preliminary analyses of all variables in the study assessed the normality of the data distribution, following Kline’s recommendations [92]. Normality was examined based on skewness (SK) and kurtosis (Ku) values, considering values above |3| and |10|, respectively, as indicative of non-normal distribution (see

Table A2. Normality analysis of variables in study.

Variables	M	SD	Skewness	Kurtosis
CNS	25.53	6.59	−0.52	−0.29
Cognitive impairment	11.80	4.45	1.78	3.94
Functional impairment	7.12	2.91	1.72	2.83
CAS	18.92	6.89	1.82	4.02
Participatory action	11.11	8.73	1.12	1.00
Leadership action	2.85	4.63	2.26	5.06
Environmental action	13.96	12.66	1.43	1.84
Wellbeing	7.08	1.69	−1.10	2.04
Self-compassion engagement	37.58	9.99	−0.52	0.59
Self-compassion action	28.24	8.31	−0.60	−0.13
Self-compassion_total	65.82	16.62	−0.67	0.60
Compassion from others engagement	34.11	10.84	−0.11	−0.26
Compassion from others action	24.07	8.01	−0.27	−0.23
Compassion from others_total	58.17	18.22	−0.19	−0.25

Note. CNS = Brief Version Connectedness to Nature Scale; CAS = Climate Change Anxiety Scale.

of Appendix A). To address the first research objective, which involves examining the factor structure and psychometric properties of the CAS, the sample was randomly divided into two subsets. In one subset, an EFA was conducted using the principal component analysis approach with direct oblimin (oblique) rotation. Criteria for factor extraction included examination of the low loading items (r < 0.3) [93], Kaiser-Meyer-Olkin Measure (KMO > 0.6) [93], and Bartlett’s Test of Sphericity (χ2, p ≤ 0.050) [93]. The Kaiser criterion (i.e., eigenvalues greater than 1.0) was used, and the scree plot was inspected to determine the optimal number of factors [93]. To confirm whether the structure of the CAS derived from the EFA presents an adequate fit to the data in the Portuguese population, a confirmatory factor analysis (CFA) using maximum likelihood estimation was conducted in AMOS 27 software [91] in the other randomly generated sample subset. The structure was specified to have a second-order factor, the Climate Anxiety Scale (CAS), which contains two subscales: Cognitive Impairment and Functional Impairment. The specified model includes two other intercorrelated factors, Experience of Climate Change and Behavioral Engagement, correlated with the CAS (Figure 1).

The normality of the items was evaluated by the Sk and Ku values (

Table A3. Normality Analysis Of Items Of CAS.

Items	M	SD	Skewness	Kurtosis
CAS1	1.99	0.95	0.59	−0.47
CAS2	1.72	0.82	0.90	0.26
CAS3	1.40	0.72	2.01	4.23
CAS4	1.27	0.60	2.43	6.03
CAS5	1.51	0.80	1.48	1.28
CAS6	1.25	0.59	2.66	7.70
CAS7	1.25	0.61	2.60	6.37
CAS8	1.41	0.73	1.88	3.22
CAS9	1.35	0.67	1.96	3.36
CAS10	1.77	0.94	0.97	−0.01
CAS11	1.29	0.58	2.29	6.36
CAS12	1.32	0.63	2.17	5.13
CAS13	1.39	0.78	2.20	4.43
CAS14	1.68	0.92	1.24	0.81
CAS15	1.79	1.02	1.09	0.30
CAS16	2.32	1.19	0.48	−0.74
CAS17	2.88	1.10	−0.01	−0.56
CAS18	4.28	0.96	−1.30	1.11
CAS19	4.45	0.76	−1.57	3.25
CAS20	4.03	0.94	−1.00	1.05
CAS21	3.20	1.26	−0.20	−0.95
CAS22	3.65	1.09	−0.51	−0.31

of Appendix A). No item had indicators of severe violations to the normal distribution (Sk < |3| and Ku < |10|) [92]. Multivariate outliers were screened using the Mahalanobis squared distance (D²) method [92], and 12 outliers were removed to improve the model fit. However, removing the thirteenth and subsequent outliers resulted in a decrease in model fit, so these outliers were retained. The sociodemographic differences between the subsets (see Table A1 of Appendix A) were observed through Student’s t-test, to continuous variables and chi-square to categorial variables. It also analyzed the magnitude of effect, through Cohen’s d (continuous variables; <0.02, small; 0.021–0.079, moderate; >0.08, large) [94], and through V of Cramér (categorial variables; 0.10–0.29, small; 0.30–0.49, moderate; >0.50, large) [94,95]. Model fit was ascertained using Normed chi-square (χ2/df), with 2 to 5 indicating a good fit. Because the chi-square index (χ2) is very sensitive to sample size and to minor or potentially nonconsequential violations of model fit, we assessed the fit of the model with three additional indicators: the comparative fit index (CFI), the Tucker–Lewin Index (TLI), and the root-mean-square error of approximation (RMSEA). Criteria for adequate and good model fit were CFI and TLI values above 0.90 and RMSEA values between ≤ 0.05 and ≤ 0.08 [92,93,96]. The mean (M) and standard deviation (DP) of each factor were calculated, and reliability was assessed by estimating Cronbach’s alpha coefficients for each subscale, classified as acceptable (≥0.70) and good (≥0.80) [97].

To examine the validity of the CAS, Pearson product-moment correlation coefficients were computed. Cohen’s [94] guidelines were used to describe the effect sizes of Pearson’s correlations (i.e., weak for correlations between 0.10 and 0.29, medium for those between 0.30 and 0.49, and strong for correlations at 0.50 or higher). Convergent validity was evaluated using Pearson correlation coefficients between the CAS and subscales and the Depression, Anxiety, and Stress Scale [87] and with the Experience of Climate Change factor [39]. Concurrent Validity was assessed by evaluating the Pearson correlation coefficients between the CAS and the Hogg Eco-Anxiety Scale [80]. Divergent validity was assessed by evaluating the correlation between the CAS and the Behavioral Engagement factor [39].

Finally, to accomplish the second and third objectives of our study, two sequential moderated mediation models were evaluated. Although some differences were found in the study variables when comparing levels/categories within the sociodemographic variables (see

Table A4. Comparing categories within sociodemographic variables in study variables.

F Test	Gender				Area					NSE					Etnia
	F	M	F Test	η2	R	SU	U	F Test	η2	L	M	H	F Test	η2	F Test	DF	p-Value	η2
Variable	M (SD)	M (SD)			M(SD)	M(SD)	M(SD)			M(SD)	M(SD)	M(SD)
CNS	26.17 (6.40)	24.02 (6.82)	4.37 **	0.03	25.23 (6.45)	27.88 (5.46)	25.04 (6.90)	5.74 **	0.02	25.27 (6.89	25.26 (6.70)	26.56 (5.35)	1.06	0.01	0.97	10.499	0.47	0.02
CAS	19.33 (7.32)	17.99 (5.73)	1.70	0.01	19.43 (7.66)	19.25 (7.03)	18.34 (6.05)	1.48	0.01	20.46 (7.72)	18.40 (6.58)	18.39 (5.94)	2.65 *	0.02	2.08	10.499	0.02	0.04
EAS	14.10 (12.32)	13.57 (13.51)	0.24	0.00	14.93 (12.88)	15.03 (14.39)	12.73 (11.77)	1.96	0.01	14.45 (14.11)	14.91 (13.57)	11.37 (8.33)	1.97	0.01	0.41	10.499	0.94	0.01
WB	7.09 (1.67)	7.05 (1.71)	1.53	0.01	7.20 (1.64)	7.32 (1.59)	6.89 (1.75)	2.76	0.01	6.56 (1.78)	7.20 (1.69)	7.42 (1.32)	5.60 ***	0.03	0.76	10.499	0.67	0.02
SC	67.59 (15.92)	61.79 (17.26)	7.12 ***	0.04	66.29 (16.38)	67.09 (17.16)	64.96 (16.69)	0.60	0.00	64.42 (16.90)	65.63 (17.49)	68.27 (12.33)	1.09	0.01	1.79	10.499	0.06	0.04
CFO	59.20 (18.52)	55.46 (17.27)	2.40	0.01	57.54 (17.35)	58.71 (18.44)	58.57 (18.96)	0.21	0.00	57.81 (18.29)	57.48 (18.83)	60.33 (15.44)	0.62	0.00	0.39	10.499	0.95	0.01

* p < 0.050 ** p < 0.010 *** p < 0.001. CNS = Brief Version Connectedness to Nature Scale; CAS = Climate Change Anxiety Scale; EAS = Environmental Action Scale; WB = Well-Being (PERMA); SC = Compassionate Engagement and Action–Self-Compassion scale; CFO = Compassionate Engagement and Action-Compassion from Others Subscale; F = female; M = male; R = rural; SU = semi-urban; U = urban; L = low; M = middle; H = high.

of Appendix A), an inspection of the potential covariance of sociodemographic variables revealed small effect sizes (η2 = 0.01, small; 0.6, moderate; 0.14, large) [94] and weak-to-moderate correlations (see

Table A5. Correlation table between sociodemographic variables and variables in study.

Correlation (r/rs)	CNS	CAS	EAS	WB	SC	CFO
Education level a (r)	0.08	−0.10 *	0.14 **	0.06	0.11 *	0.11 *
Marital_Status b (r)	0.01	0.02	−0.03	0.09 *	−0.09 *	−0.09 *
Children c (r)	0.03	0.05	−0.01	0.01	−0.03	−0.15 **
Psi/psi counseling d (r)	−0.09	−0.06	−0.07	0.09	0.03	0.04
Contemplative practices e (r)	0.21 **	0.08	0.14 **	0.06	0.10 *	0.02
Age (rs)	0.17 **	0.10 *	0.05	0.10 *	0.01	−0.12 **
Physical health (rs)	−0.01	−0.14 *	−0.01	0.27 **	0.15 **	0.11 *
Mental health (rs)	0.01	−0.19 **	−0.07	0.39 **	0.24 **	0.15 **

* p < 0.050. ** p < 0.010. Note. (r) Pearson’s r; (r_s) Spearman’s rho; ^a Education: 0 = secondary school or below, 1 = Higher education. ^b Current marital status: 0 = single, 1 = married. ^c Children: 0 = no children; 1 = children. ^d Psychological or psychiatric counseling: 0 = no; 1 = yes. ^e Contemplative practices: 0 = no; 1 = yes. CNS = Brief Version Connectedness to Nature Scale; CAS = Climate Change Anxiety Scale; EAS = Environmental Action Scale; WB = Well-Being (PERMA); SC = Compassionate Engagement and Action–Self-Compassion scale; CFO = Compassionate Engagement and Action-Compassion from Others Subscale.

of Appendix A) with the variables included in the model, and therefore these were not included as covariates. In the moderated mediation models, we examined the direct and indirect effects, considering climate change anxiety (CAS) and pro-environmental behaviors (EAS), as sequential mediators, in the association between connection to nature (CNS) and well-being (PERMA; Figure 2 and Figure 3), using PROCESS macro for SPSS [98]. To maintain a parsimonious model, compassion (self-compassion—Figure 2 and compassion from others—Figure 3) was hypothesized as a moderator in the more theoretically relevant but minimum number of paths to reduce the possibility of Type I error. Therefore, in this study, compassion for self and from others were hypothesized as moderators of the associations between connection to nature and well-being, and between connection to nature and climate change anxiety. To illustrate the moderation effect, a graphic was plotted considering one line for each of the three levels of the moderator (M − SD, M, M + SD) [99]. For tests of mediation, the significance of direct, indirect, and total effects was assessed using χ2 tests [92]. The bootstrapping resampling method was further used to test the significance of the mediational paths, using 5000 bootstrap samples and 95% confidence intervals (CI) [92]. For all the analyses performed in this study, results were considered statistically significant at a p-value lower than 0.05.

3. Results

3.1. Climate Change Anxiety Scale

3.1.1. Exploratory Factor Analysis

In one of the subsets (n = 261), an EFA was conducted, using the principal component analysis with direct oblimin (oblique) rotation. Although a solution emerged with five factors with eigenvalue above one, after analyzing the scree plot, and in line with the original study [39], one factor, with an eigenvalue of 1.002, was dismissed. Moreover, one of the factors, which in the original study included Items 17, 18, 19, 20, 21, and 22, emerged in this study formed by Items 18, 19, 20, and 22 (Cronbach’s α of 0.72). The fifth factor, with Items 17 and 21, was not retained (Cronbach’s α of 0.53). The model, with four factors, accounted for 60.98% of the variance among the items. Factor loadings are presented in

Table 1. Exploratory factor analysis using principal component analysis (n = 261).

Items	FACTOR
	1	2	3	4	5
CAS1	0.69
CAS2	0.76
CAS3	0.75
CAS4	0.79
CAS5	0.80
CAS6	0.78
CAS7	0.67
CAS8	0.78
CAS9					0.79
CAS10					0.78
CAS11					0.80
CAS12					0.78
CAS13					0.56
CAS14			−0.85
CAS15			−0.87
CAS16			−0.80
CAS17 a				−0.69
CAS18		0.79
CAS19		0.70
CAS20		0.84
CAS21 a				−0.65
CAS22		0.60
Cronbach’s alpha	0.89	0.72	0.80	0.53	0.83

^a not retained.

.

All the presented factor loadings range between 0.56 and 0.87. Factor 1 measures Cognitive and Emotional Impairment in response to climate change, Factor 2 is related to Behavioral Engagement, Factor 3 represents Personal Experience of Climate Change, and Factor 5 measures Functional Impairment influenced by concern about climate change.

3.1.2. Confirmatory Factor Analysis

In the other sample subset (n = 261), a CFA was conducted using the four factors that resulted from the EFA (Figure 1), similar to the original study [39]. Results showed that this model has a poor fit to the data: χ2 (165) = 455.54; p < 0.001, CMIN/DF = 2.76; CFI = 0.89; TLI = 0.87; RMSEA = 0.08 [90% CI 0.07–0.09; p < 0.001]. Furthermore, the intercorrelation between the CAS and Behavioral Engagement was removed because it was not statistically significant. Outliers were screened using the Mahalanobis squared distance and 12 of them were removed to improve the model fit: χ2 (166) = 413.12; p < 0.001, CMIN/DF = 2.49; CFI = 0.91; TLI = 0.90; RMSEA = 0.08 [90% CI 0.07–0.09; p < 0.001]. An inspection of the modification indices indicated that correlating the measurement errors of two pairs of items of the Cognitive Impairment factor (Items 1 and 2; Items 5 and 8) would significantly increase the model fit. After correlating these two pairs of items, results revealed an improvement in the model fit (Figure 4), with this adjusted model presenting an acceptable fit to the data χ2 (164) = 328.67; p < 0.001, CMIN/DF = 2.004; CFI = 0.94; TLI = 0.93; RMSEA = 0.06 [90% CI 0.05–0.07; p < 0.001]. The first-order factors, Cognitive Impairment and Functional Impairment, were significantly loaded on the second-order factor of Climate Change Anxiety (0.93 and 0.91, respectively). Regarding local fit, in the Cognitive Impairment subscale, items revealed standardized regression weights (SRW) ranging from 0.56 (Item 1) to 0.92 (Item 6), and in the Functional Impairment subscale from 0.63 (Item 10) to 0.92 (Item 12; Figure 3). Squared multiple correlations (SMC) results indicated that in the Cognitive Impairment factor, values ranged from 0.31 (Item 1) to 0.85 (Item 6), and from 0.40 (Item 10) to 0.85 (Item 12) in the Functional Impairment subscale.

3.1.3. Descriptive Statistics and Reliability

Table 2. Descriptive statistics—variables under study.

Scale	M (SD)	Range	Cronbach’s α
Climate Anxiety Scale	18.92 (6.89)	13–55	0.92
Cognitive impairment	11.80 (4.45)	8–35	0.89
Functional Impairment	7.12 (2.91)	5–20	0.86
Experience with climate change	5.78 (2.64)	3–15	0.79
Engagement behaviour	16.41 (2.74)	4–20	0.70
HEAS	5.33 (6.18)	0–34	0.94
EAS	13.96 (12.66)	0–62	0.93
CNS	25.59 (6.56)	7–35	0.89
CEAS-SC	65.98 (16.55)	10–100	0.90
CEAS-CFO	58.20 (18.13)	10–100	0.96
Well-Being (PERMA)	7.08 (1.69)	0–10	0.96

Note. N = 510. CAS = Climate Change Anxiety Scale; HEAS = Hogg Eco-Anxiety Scale; EAS = Environmental Action Scale; CNS = Brief Version Connectedness to Nature Scale; CEAS–SC = Compassionate Engagement and Action–Self-Compassion Scale; CEAS–CFO = Compassionate Engagement and Action–Compassion from Others Subscale.

shows the means, standard deviation, and Cronbach’s alpha of each factor, and other scales (N = 510). The Climate Anxiety Scale had a Cronbach’s alpha of 0.92, with 0.89 in the Cognitive Impairment subscale, and 0.85 in the Functional Impairment subscale, indicating a good internal consistency. Experience with climate change had α = 0.79, and Engagement behavior α = 0.70.

Table 3. Correlation matrix—variables under study.

Variables	1	2	3	4	5	6	7	8	9	10	11
CAS	1
Cognitive impairment	0.96 **	1
Functional Impairment	0.90 **	0.74 **	1
Experience with Climate Change	0.46 **	0.43 **	0.44 **	1
Behavioral Engagement	0.08	0.06	0.10 *	0.18 **	1
HEAS	0.74 **	0.68 **	0.70 **	0.38 **	0.09	1
EAS	0.32 **	0.26 **	0.36 **	0.38 **	0.33 **	0.36 **	1
CNS	0.15 **	0.13 **	0.15 **	0.23 **	0.40 **	0.16 **	0.35 **	1
CEAS-SC	−0.09 *	−0.08	−0.09	0.04	0.19 **	−0.11 *	0.10 **	0.29 **	1
CEAS-CFO	−0.04	−0.03	−0.05	0.12 **	0.19 **	−0.03	0.08	0.20 **	0.46 **	1
Well-Being (PERMA)	−0.19 **	−0.17 **	−0.19 **	0.03	0.18 **	−0.19 **	0.14 **	0.25 **	0.56 **	0.45 **	1
DASS-21	0.35 **	0.32 **	0.35 **	0.18 **	−0.05	0.46 **	0.08	0.07	−0.17 **	−0.09 *	−0.37 **

Note. N = 510. CAS = Climate Change Anxiety Scale; HEAS = Hogg Eco-Anxiety Scale; EAS = Environmental Action Scale; CNS = Brief Version Connectedness to Nature Scale; CEAS–SC = Compassionate Engagement and Action–Self-Compassion Scale; CEAS–CFO = Compassionate Engagement and Action-Compassion from Others Subscale; DASS-21 = Depression, Anxiety, and Stress Scale. * p < 0.050. ** p < 0.010.

shows the correlation between these subscales and the other scales. Behavioral engagement was not correlated with cognitive or functional impairment. Moreover, Factors 1 and 5 were strongly correlated.

Additionally, the analysis identified a moderate multicollinearity risk between the variables CAS, Cognitive Impairment, and Functional Impairment. The variance inflation factor (VIF) was 2.24, falling within the range indicative of moderate multicollinearity [100,101], while the tolerance value of 0.45 exceeded the critical threshold of 0.1, suggesting no substantial multicollinearity risk [100].

3.1.4. Convergent, Concurrent, and Divergent Validity

For the convergent validity, correlation analyses indicated a moderate association between the Depressive Anxiety and Stress Scale with the Climate Change Anxiety Scale (r = 0.35; p < 0.010), Cognitive Impairment (r = 0.32; p < 0.010), and Functional Impairment (r = 0.35, p < 0.010). Experience of Climate Change was also moderately correlated with the CAS (r = 0.46; p < 0.010), Cognitive Impairment (r = 0.43; p < 0.010), and Functional Impairment (r = 0.44, p < 0.10). Moreover, Hogg Eco-Anxiety Scale [80] was strongly correlated with the CAS (r = 0.74; p < 0.010), Cognitive Impairment (r = 0.68; p < 0.010), and Functional Impairment (r = 0.70; p < 0.010), thus supporting concurrent validity. Conversely, for discriminant validity, behavioral engagement showed weak correlations with the CAS (r = 0.08), Cognitive Impairment (r = 0.06), and Functional Impairment (r = 0.10; p < 0.050).

3.2. Moderated Mediation Model Analysis

Two moderated mediation models were analyzed (Figure 2 and Figure 3).

Table 4. Direct effects on different levels of self-compassion and indirect effects.

Effect	Path	Unstandardized Coefficient	95% Confidence Interval
			Lower	Upper
Direct	CNS -> WB	0.100 ***	0.042	0.162
Direct	CNS -> WB|Self-compassion = −1 SD	0.040 ***	0.019	0.066
Direct	CNS -> WB|Self-compassion = 0 SD	0.020 *	0.002	0.042
Direct	CNS -> WB|Self-compassion = +1 SD	0.002	−0.023	0.028
Indirect	CNS -> CAS -> WB	−0.009	−0.016	−0.004
Indirect	CNS -> EAS -> WB	0.010	0.003	0.016
Indirect	CNS -> CAS -> EAS -> WB	0.002	0.001	0.003

Note. N = 510. CAS = Climate Change Anxiety Scale; EAS = Environmental Action Scale; CNS = Brief Version Connectedness to Nature Scale; WB = Well-Being (PERMA). * p < 0.050. *** p < 0.001.

and

Table 5. Direct effects on different levels of compassion from others and indirect effects.

Effect	Path	Unstandardized Coefficient	95% Confidence Interval
			Lower	Upper
Direct	CNS -> WB	0.123 ***	0.067	0.178
Direct	CNS -> WB |Compassion from others = −1 SD	0.065 ***	0.040	0.090
Direct	CNS -> WB |Compassion from others = 0 SD	0.039 ***	0.018	0.059
Direct	CNS -> WB |Compassion from others = +1 SD	0.012	−0.016	0.040
Indirect	CNS -> CAS -> WB	−0.009	−0.016	−0.004
Indirect	CNS -> EAS -> WB	0.010	0.004	0.016
Indirect	CNS -> CAS -> EAS -> WB	0.001	0.000	0.003

Note. N = 510. CAS = Climate Change Anxiety Scale; EAS = Environmental Action Scale; CNS = Brief Version Connectedness to Nature Scale; WB = Well-Being (PERMA). *** p < 0.001.

present the results of the direct and indirect effects of each model. Correlations between the study variables and the sociodemographic variables revealed weak to moderate effect sizes (between 0.01 and 0.37; see Table A5 of Appendix A). Consequently, these sociodemographic variables were not included in the models as covariates. Results of the direct effects of connection to nature on well-being are presented at low, medium, and high levels of the significant moderators, self-compassion, and compassion from others.

Regarding the first model with self-compassion as moderator presented in Figure 2, results showed a significant direct effect of connection to nature on well-being (b = 0.10, SE = 0.031, p < 0.001, 95% CI = [0.042, 0.162]), with a significant moderation effect of self-compassion (b = −0.0012, SE = 0.0004, p < 0.010, 95% CI = [−0.0021, −0.0003]; see Table 4).

As depicted in Figure 5, increased connection to nature was significantly associated with enhanced levels of well-being in participants with low (b = 0.040, SE = 0.012, p < 0.001, 95% CI = [0.019, 0.066]) and medium (b = 0.020, SE = 0.010, p < 0.050, 95% CI = [0.002, 0.042]) levels of self-compassion, but not in participants scoring high self-compassion (b = 0.002, SE = 0.013, p = 0.87, 95% CI = [−0.023, 0.028]).

There was also a significantly indirect effect of connection to nature on well-being, significantly mediated by climate change anxiety (b = −0.009, SE = 0.003, 95% CI = [−0.016, −0.004]), an indirect effect of connection to nature on well-being significantly mediated by environmental action (b = 0.010, SE = 0.003, 95% CI = [0.003, 0.016]), and an indirect effect of connection to nature on well-being significantly mediated by climate change anxiety/environmental action (b = 0.002, SE = 0.001, 95% CI = [0.001, 0.003]).

The moderation effect of self-compassion on the association between connection to nature and climate change anxiety was not significant (b = −0.002, SE = 0.002, p = 0.435, 95% CI = [−0.006, 0.003]). Therefore, the index of moderated mediation in the overall model was not significant, b = 0.000, SE = 0.000, 95% CI = [−0.0001, 0.000], meaning that self-compassion did not moderate the mediation path (between connection to nature and climate change anxiety), only the direct effect of connection to nature on well-being. This mediation model with a moderation of direct effect explained 36.89% of the variance of well-being.

In regard to the model with compassion from others as a moderator and presented in Figure 3, results showed a significant direct effect of connection to nature on well-being (b = 0.123, SE = 0.028, p < 0.001, 95% CI = [−0.067, 0.178]), with a significant moderator effect of compassion from others (b = −0.001, SE = 0.001, p < 0.010, 95% CI = [−0.002, −0.001]; see Table 5).

The inspection of the graphic of moderation indicates that higher connection to nature was significantly associated with enhanced levels of well-being, in participants with low (b = 0.065, SE = 0.013, p < 0.001, 95% CI = [0.040, 0.090]) and medium levels (b = 0.039, SE = 0.011, p < 0.001, 95% CI = [0.018, 0.059]) of compassion from others, but not in participants with increased levels of compassion from others (b = 0.012, SE = 0.014, p = 0.39, 95% CI = [−0.016, 0.040]; Figure 6).

There was also a significantly indirect effect of connection to nature on well-being, significantly mediated by climate change anxiety (b = −0.009, SE = 0.003, 95% CI = [−0.016, −0.004]), an indirect effect of connection to nature on well-being, significantly mediated by environmental action (b = 0.010, SE = 0.003, 95% CI = [0.004, 0.016]) and an indirect effect of connection to nature on well-being, significantly mediated by climate change anxiety/environmental action (b = 0.001, SE = 0.001, 95% CI = [0.000, 0.003]). The moderation effect of compassion from others on the association between connection to nature and climate change anxiety was not significant (b = −0.002, SE = 0.002, p = 0.435, 95% CI = [−0.007, 0.002]). Therefore, the index of moderated mediation in the overall model was not significant, b = 0.000, SE = 0.000, 95% CI = [−0.0001, 0.000], meaning that compassion from others did not moderate the mediation path (between connection to nature and climate change anxiety), only the direct effect of connection to nature on well-being. Thus, the mediation model with a moderation of this direct effect explained 29.79% of the variance of well-being.

4. Discussion

In a world facing increasing challenges related to climate change caused by human activities, known as the Anthropocene Era [77], understanding how humans relate to nature could help change behaviors and promote a balanced life. Climate change anxiety is an emotional reaction that can stem from various causes, particularly socio-altruistic values (caring for others) and biospheric values (caring for non-human nature and the biosphere itself), but not from egoistic values [34]. Given the intrinsic motivation for interdependence with nature [14], people who feel a stronger connection to nature and have greater environmental concern are more likely to be vigilant towards climate change threats, experiencing heightened anxiety [39], poorer mental health, and lower levels of well-being [1]. Pro-environmental behaviors are deliberate actions aimed at reducing the negative impact of human activity on the natural or built environment and function as a coping mechanism [46]. Individuals with higher levels of biospheric values are more inclined to perceive ecological stress and engage in pro-environmental behaviors. Those with higher levels of socio-altruistic values also engage in pro-environmental behaviors but without necessarily perceiving ecological stress [48]. Additionally, individuals with higher levels of self-compassion and compassion from others exhibit more pro-environmental behaviors and higher levels of well-being [70]. Therefore, the current study aimed to better understand the relationship between connection to nature and well-being by examining the mediator role of climate change anxiety and environmental action and the moderator effect of self-compassion and compassion from others on this association.

Following the adaptation of the CAS to Portuguese, we studied its factor structure and psychometric properties within the Portuguese population. The original study by Clayton and Karazsia [39], an exploratory factor analysis (EFA), identified four factors, named Cognitive Impairment, Functional Impairment, Personal Experience with Climate Change, and Behavioral Engagement. Our EFA revealed five factors. Three of these matched the original study, namely Cognitive Impairment (factor loadings between 0.67 and 0.80), Functional Impairment (factor loadings between 0.56 and 0.80), and Personal Experience with Climate Change (factor loadings between −0.80 and −0.87). However, the Behavioral Engagement factor, comprising Items 17 to 22, was subdivided into two factors in our study: one included Items 17 and 21 and exhibited unacceptable internal consistency, and the other comprised Items 18, 19, 20, and 22 with acceptable internal consistency. Due to the differing grammatical construction and focus on consequences of not acting pro-environmentally, Items 17 (“I wish I behaved more sustainably”) and 21 (“I feel guilty if I waste energy”) were hence excluded from the final scale composition in this Portuguese adaptation. Consequently, in our study, Behavioral Engagement is represented by Items 18, 19, 20, and 22.

In line with the original study [39], a confirmatory factor analysis (CFA) using a four-factor structure was then conducted and tested for a second-order factor (Climate Change Anxiety Scale) comprising the cognitive engagement and functional impairment factors. Initially, this second-order factor was estimated to be correlated with the personal experience climate change factor and the behavioral engagement factor. The CFA results indicated a poor model fit and revealed a weak correlation between the Climate Change Anxiety Scale and the Behavioral Engagement factor. Upon inspecting the Mahalanobis distance and measurement error covariances, we identified significant outliers and covariation between the measurement errors of Item 1 with Item 2 and Item 5 with Item 8, which differ from the CAS’s original study [39]. Items 1 and 2 that start with “Tenho dificuldades em…” in the Portuguese language differed grammatically from the other items in the Cognitive Impairment subscale. Similarly, Items 5 and 8, starting with “Penso: Porque é que…” in the Portuguese language, showed covariation due to their distinct grammatical structure.

The model was thus respecified by removing extreme outliers and covarying the measurement errors of Items 1 with 2 and 5 with 8. The CFA of the adjusted model indicated an adequate model fit, corroborating the factor structure with four factors: Cognitive Impairment, Functional Impairment, Personal Experience Climate Change, and Behavioral Engagement. The CAS demonstrated good internal consistency and reliability, evidencing internal validity similar to the original study [39]. It was also important to compare the CAS with other scales that measure similar and different constructs. The CAS evidenced good convergent validity with a scale that measures anxiety (DASS-21) and with a scale that measures personal experience with climate change [39]. Interestingly, results from this research differed from the original study. In our study, the Cognitive and Functional Impairment subscales showed a low-moderate correlation with the DASS-21 and a higher-moderate correlation with Personal Experience with Climate Change compared to the original study [39]. The Hogg Eco-Anxiety Scale [80] was translated and validated in Portugal [81] and measures eco-anxiety, which includes anxiety related to climate change issues and various environmental calamities. Unlike the CAS, which focuses solely on climate change-related cognitive and functional impairment, the Hogg scale addresses a broader range of environmental concerns. Therefore, extending the original CAS study [39], we included this scale in our study and verified its concurrent validity through a strong correlation with the CAS. Nevertheless, our study was cross-sectional, indicating that future research in Portugal should seek to explore the temporal consistency of the CAS.

Although the Experience Climate Change factor was moderately correlated with Cognitive Impairment and Functional Impairment, Clayton and Karazsia proposed a second-order factor (Climate Anxiety Scale) comprising Cognitive and Functional Impairment [39]. The correlation analysis revealed a moderate multicollinearity risk among CAS, Cognitive Impairment, and Functional Impairment. VIF indicated a moderate but non-severe multicollinearity, and the tolerance value suggests no substantial multicollinearity risk. Therefore, it aligned with previous validation studies of this scale across diverse cultural contexts [102,103,104,105] that have retained the two-factor structure despite similar multicollinearity levels. Given the absence of severe multicollinearity in our data and alignment with established methodological precedents, we elected to maintain the original two-factor structure for consistency and comparability. Studies have found an association between natural disasters and poorer mental health [7], suggesting that future research could examine the impact of direct and indirect climate change experiences on climate change anxiety. This could help explain the observed correlations and enable the design of more effective psychological programs for individuals affected by climate change events. Overall, the CAS proved to be a reliable measure for assessing climate change anxiety. The increasing number of people experiencing direct or indirect effects of climate change underscores the need for validated measures in Portugal to assess anxiety related to this issue. Although the Portuguese validation of the CAS was conducted with the general population, resulting in possible floor effects (i.e., low mean scores for each item), it is crucial to study the CAS in a clinical population and in specific groups, such as climate change activists [57] or in younger or female populations [36], to identify cut-off points that distinguish adaptive anxiety from clinical anxiety.

Robust evidence linking nature connection to well-being raises questions about potential mediator and moderator mechanisms on this relationship. Our second and third aims are an attempt to answer those questions. A moderated mediation model was evaluated by examining whether climate change anxiety and environmental action would function as mediators on the association between connection to nature and well-being and testing whether self-compassion and receiving compassion from others would moderate the associations between connection to nature and climate change anxiety and connection to nature and well-being.

4.1. Connection to Nature and Climate Change Anxiety

Results from our model indicated that a greater connection to nature was associated with increased climate change anxiety. Participants with a stronger connection to nature may exhibit heightened environmental concern and are more likely to be vigilant to climate change threats, leading to an emotional response such as climate change anxiety [39].

4.2. Connection to Nature and Environmental Action

Our findings indicated that a stronger connection to nature was associated with increased environmental action. Participants who identify closely with nature and possess a sense of self-care may perceive the protection and care of nature as analogous to taking care of themselves. This aligns with previous research, which found that empathic feelings and attitudes, derived from perspective-taking, serve as a motivation for engaging in pro-environmental behaviors [34].

4.3. Connection to Nature and Well-Being

Our findings indicate that a stronger connection to nature enhances well-being. Previous research has described that feeling connected to nature can evoke emotions such as wonder, awe, gratitude, and compassion [66]. Additionally, psychological theories suggest that psychopathology is often associated with mental inflexibility [62,106]. Therefore, a connection to nature may facilitate a shift from a narrow self-perspective to a broader perspective, thereby promoting psychological flexibility.

4.4. Climate Change Anxiety as a Mediator

Our results demonstrated that climate change anxiety mediated the association between connection to nature and well-being, as well as between connection to nature and environmental action. Numerous studies have correlated climate change with poor mental health and lower levels of well-being [4,40,41]. Therefore, participants with a higher connection to nature exhibited worse well-being, which could be attributed to an anxiety response related to environmental concern and heightened vigilance to climate change threats [39].

Additionally, a stronger connection to nature was linked to increased environmental actions, an association also mediated by climate change anxiety [39]. Previous studies have identified this mediation, suggesting that environmental action serves as a coping mechanism to alleviate the distress caused by climate change anxiety [59]. Furthermore, it has been found that collective environmental activism significantly reduces the impact of climate change anxiety on major depressive disorder more effectively than individual pro-environmental behavior [45]. Our results are in line with these previous studies, but further research is needed to explore and compare the effects of collective environmental activism and individual pro-environmental behaviors as responses to climate change anxiety.

4.5. Environmental Action as a Mediator

Our study confirmed that environmental action acts as a mediator in the relationship between climate change anxiety and well-being, as well as between connection to nature and well-being. Consistent with other research, our findings suggest that environmental activism enhances well-being and mitigates climate change anxiety [45,57]. Furthermore, we found that a stronger connection to nature predicts higher well-being. This can be explained by the similarity between feelings, attitudes, and behaviors towards nature and the concept of self-care and consequently by environmental action [34].

4.6. Self-Compassion and Receiving Compassion from Others as Moderators

In our models, self-compassion and receiving compassion from others emerged as moderators in the direct effect of connection to nature on well-being. The results indicated that, for participants with lower to medium levels of self-compassion (both self-directed and received from others), a higher connection to nature significantly enhanced well-being. We initially predicted that greater self-compassion would strengthen the positive impact of connection to nature on well-being. However, the findings revealed that among participants with higher levels of self-compassion, a greater connection to nature did not further enhance well-being. This suggests that individuals with higher self-compassion might have already achieved their maximum potential well-being, thus, a greater connection to nature does not seem to enhance well-being in them.

It is important to note that the construct of connection to nature encompasses various aspects such as heightened sensitivity (attentiveness to the suffering of nature), sympathy (emotional responses to the suffering of nature), empathy (seeing the world from nature’s perspective), care for nature’s well-being, distress tolerance (ability to endure feelings arising from nature’s suffering), and non-judgement (open acceptance that aids distress tolerance). These attributes seem to closely resemble the competencies and motivations associated with compassion. Future research could seek to further examine the potential correspondence and associations between the competencies/components of both constructs.

Additionally, physiological markers like heart rate variability, an indicator of parasympathetic nervous system activity, are associated with both connection to nature [68] and compassion [67]. Numerous studies have found a strong correlation between compassion and well-being [79,107]. This raises the question: Could interventions that combine nature immersion with compassion training accelerate the development of compassion’s competencies and motivations and have beneficial psychophysiological effects? Aligned with studies that integrate connection to nature with compassion [19], future research could explore this relationship further. Specifically, examining the psychophysiological effects of psychological intervention programs that target both connection to nature and compassion, such as forest bathing [20] and compassionate mind training [61], could provide valuable insights into the feasibility, usefulness, and effectiveness of such interventions to improve well-being.

Conversely, self-compassion and receiving compassion from others did not moderate the association between connection to nature and climate change anxiety. Across varying levels of self-compassion and compassion from others, a higher connection to nature was consistently linked to increased climate change anxiety. This sample, drawn from the general population, exhibited a potential floor effect in the CAS results. Consequently, the levels of climate change anxiety observed can be considered normative and adaptive. Therefore, higher levels of compassion did not mitigate the increase in climate change anxiety associated with a greater connection to nature. Future research could investigate this model within a clinical sample and in specific groups (e.g., climate change activists, youth, or female populations) where non-adaptive climate change anxiety could be more prevalent.

5. Limitations

This study has limitations that should be considered when interpreting our findings and that future research should seek to address. First, as a cross-sectional study, the CAS was not evaluated for temporal stability. The observed multicollinearity between factors may suggest limited discriminant validity in the bifactor structure of the scale. However, variance inflation factor (VIF) values fell below critical thresholds (VIF < 5), indicating moderate rather than severe multicollinearity. Tolerance metrics supported this interpretation, with values exceeding 0.1, which aligns with established thresholds for acceptable collinearity. Furthermore, in our moderated mediation model, the directionality of the effects was hypothesized based on existing literature; however, the results should be regarded as preliminary. We encourage the scientific community to replicate this cross-sectional study. Nevertheless, the causality of the relationship between variables cannot be determined from our findings, without longitudinal or experimental research, causal inferences regarding these relationships cannot be drawn, as the associations may be bidirectional. In addition, this model explained 37% of the variance in levels of well-being, raising the question of whether there are other variables that could better explain variance in levels of well-being when associated with a higher connection to nature. Furthermore, our study did not assess compassion towards nature. The CEAS self-compassion scale and receiving compassion from others scale do not have items addressing questions related to compassion engagement and action towards nature suffering. Interestingly, some literature links connection to nature to a different form of compassion, known as unified compassion. This type of compassion is characterized by a human experience that transcends the dualities of subject and object, self and other, and mind versus body [74,75]. Due to the absence of a measure for this construct, this study was unable to assess its existence.

The sample was collected exclusively online, and although routine strategies were used to mitigate the risks associated with the online modality [108], the sample might not be representative of the population due to self-selection bias. Finally, participants using self-response instruments are influenced by social desirability factors, which can bias their answers due to these factors.

6. Conclusions

Adapting the CAS assessment tool to Portuguese and evaluating its psychometric properties is a crucial step towards validating its use in clinical psychology and advancing climate change research. The study underscores the complex relationship between humans and nature, showing that a stronger connection to nature can both enhance well-being and increase climate change anxiety. This dual effect highlights the evolutionary roots of human interactions with nature and the challenges posed by modern environmental issues. Importantly, the research found that individuals with lower levels of self-compassion and compassion from others experience greater improvements in well-being when they feel more connected to nature, whereas this benefit is not as pronounced in those with higher compassion. These findings suggest that nature-based interventions, when combined with compassion-focused approaches such as compassionate mind training, could be particularly effective in promoting well-being and helping individuals cope with climate change anxiety. Overall, fostering both a connection to nature and compassion may be key to supporting mental health and encouraging sustainable behaviors in the face of environmental challenges.