Exploring the relationship between physical activity, flood experience, eco-anxiety, and youth well-being in a changing climate

Abstract

Climate change threatens human health, but its psychological effects on young people remains relatively understudied. Conversely, ample evidence demonstrates that Physical Activity (PA) positively impacts the psychological well-being in young people. Thus, this cross-sectional study investigated whether PA moderates the impact of a large flood experience on the relationship between eco-anxiety and well-being in young people. One thousand twenty-one (N = 1.021) adolescents aged 9 to 16 years old (Mage = 12.35 ± 1.38) completed measures for PA frequency, flood experience, eco-anxiety, and psychological well-being, six month after Daniel extreme flood event in Greece. Moderated moderation analysis was conducted to test the hypothesized model using PROCESS macro. We examined whether the relationship between eco-anxiety and well-being was moderated by flood experience alone, PA frequency alone, and/or flood experience and PA frequency combined. However the three-way interaction predicted only a very small of unique variance in well-being, ΔR² = 0.007. Further examination of the three-way interaction showed that when PA frequency scores were lower than 2.71 days/week, flood experience potentiated the negative effect of eco-anxiety on well-being. In contrast, the moderating influence of flood experience on the relationship between eco-anxiety and well-being was buffered when PA frequency scores were higher than 6.25 days/week. Findings provide initial support that frequent PA participation (6–7 days /week) may improve the well-being of children and adolescents in times of crisis, potentially by reducing the effect of flood-related experience on the negative relationship between eco-anxiety and well-being.

Introduction

The Intergovernmental Panel on Climate Change (IPCC) defines climate change as “a change in the state of the climate that can be identified by changes in the mean and/or the variability of its properties, and that persists for an extended period, typically decades or longer”¹. Human activities contribute to the increase of the earth’s temperature. By the end of the 21st century, the global mean temperature is projected to rise between 0.3 and 4.8 degrees Celsius based on scenarios related to future greenhouse gas emissions and mitigation efforts¹. The most significant indicators of climate change include global average warming, rising sea levels, and increases in the frequency and intensity of natural disasters, such as heat waves, droughts, and intense precipitation leading to floods. Lately, extreme heat waves, droughts, and severe floods have impacted large areas, including Pakistan, Europe, Asia, and South America². From September 4th to 6th, 2023, Greece’s region of Thessaly experienced the biblical climate change phenomenon of the “Daniel” storm, which brought over 900 mm of rainfall, surpassing the area’s annual average rainfall, marking one of the worst rainfall events in Greece’s history². This flood led to death many people, caused massive infrastructure destruction and left a series of villages submerged causing problems to infrastructure but also to people’s well-being. In this context, the ability to adapt to environmental changes is crucial, thus, effective coping strategies to climate change and involved experiences are needed. Various strategies are recommended for managing psychological challenges related to climate change and mitigating its adverse impacts on mental well-being³. Among the evidence-based options, engaging in regular Physical Activity (PA) emerges as an accessible and effective means of easing anxiety, lowering stress levels, and fostering overall psychological well-being⁴. Their findings indicated that PA interventions in children and adolescents are associated with moderate reductions in depressive symptoms (ES 0.43 to 0.61), with larger effects typically observed in clinically symptomatic populations compared to non-clinical samples. Also interindividual variability in intervention outcomes has been reported, with effect magnitudes influenced by factors such as initial mental health status, type and intensity of activity, and implementation context. Finally, a very recent study⁵² large-scale study in Swedish children suggests that preadolescence, a key period for the onset of psychiatric disorders, may also be a sensitive period in which PA plays a crucial role in preventing conditions such as anxiety, depression, and addiction.

Eco anxiety, climate disaster experience, and youth well being

Climate change-related environmental and socioeconomic outcomes could lead to increased rates of depression, anxiety, post-traumatic stress, other mental health disorders, and the emerging concept of eco-anxiety⁵. Eco-anxiety is an umbrella term referring to experiences of anxiety related to environmental crises^3,6. It includes “climate change anxiety”, which is the state of anxiety specifically related to anthropogenic climate change (i.e., global warming, increased frequency of natural disasters, and extreme weather events). Eco-anxiety may also include anxiety about a variety of environmental calamities possibly caused by climate change, encompassing the extinction of entire ecosystems and plant and animal species, worldwide mass pollution, and deforestation⁶. According to the World Health Organization (WHO), eco-anxiety does not constitute mental illness, but it is a rational and normal response to the environmental crisis, and turns into a real global threat⁷. Eco-anxiety is also a constructive response because it can trigger environmentally sustainable behaviours⁵. However, when eco-anxiety is high, it may cause a decline in mental health via symptoms of subjective distress, rumination, sleep or somatic disturbance, and hopelessness, which are features of anxiety disorders⁷. Eco-anxiety can profoundly affect mental well-being, especially when it becomes intense, hard to regulate, and disruptive to daily life⁷. Earlier measures of climate change anxiety and eco-anxiety have primarily focused on affective symptoms. However, contemporary measures recognize the multidimensional nature of this construct, encompassing four dimensions of eco-anxiety: affective symptoms, rumination, behavioral symptoms, and anxiety about one’s negative impact on the planet⁶.

Nowadays, eco-anxiety is highly prevalent among children and young adults^8,9. Hickman et al.¹⁰ examined climate anxiety in young people across 10 countries with differing climate-related vulnerabilities and found that all the participants were worried about climate change. Particularly, 59% were very or extremely worried and 84% were at least moderately worried. Also, 45% stated that their feelings about climate change negatively affected their daily life and functioning, whereas many expressed increased negative thoughts due to climate change¹⁰.

Studies from high-income countries revealed equally discouraging findings^8,9. Specifically, children and adolescents experience eco-anxiety and negative affective states in response to awareness of climate change. This might include intense emotional reactions like sadness, anger, fear, powerlessness, helplessness, guilt, and hope, which may occur without meeting full diagnostic criteria. However, clinically defined mental disorders such as depression and anxiety might also be a consequence^8,9. An eight-year longitudinal study revealed substantial interindividual variability, with adolescents showing high persistent worry experiencing more depressive symptoms over time, while increased concern about climate change did not uniformly translate into poorer mental health¹¹.

Another study showed that direct experience of climate-related events significantly increased eco-anxiety, climate distress, and the impact of climate change on future planning¹². Finally, a large sample of US youth (aged 16–24) showed that self-reported direct experience of climate-related events, (e.g., floods, wildfires, extreme temperatures), significantly increased eco-anxiety, climate distress, and the impact of climate change on future planning, but also psychological adaptation, meaning-focused coping and climate agency. These findings suggest that self-reported direct experience of climate-related events, such as floods, may act as a moderator in the relationship between eco-anxiety and well-being.

PA, youth well-being, and eco-anxiety

PA as was stated above is a well-established intervention for improving youth mental health, with meta-analyses linking it to reduced depression and anxiety^4,13. Beyond symptom reduction, participation offers positive experiences that support healthy physical, mental, and social development during the childhood years. In a recent study, involving 17.055 children⁵², found that PA at age 11 was protective against anxiety in both sexes, with a trend toward protection against depression, and reduced addiction risk in boys. Participation in organized sports showed broad protective effects, lowering the incidence of anxiety and addiction in both boys and girls. These findings highlighted that increased PA in boys and girls as a preventive mental health strategy and underscore its relevance for policymakers and healthcare professionals.

However, physical inactivity affects nearly 31% of the world’s adult population and 80% of adolescents, and significantly contributes to the development of several Non-Communicable Diseases (NCDs)¹⁴. These diseases cause 41 million deaths annually, accounting for 74% of all deaths globally¹⁵. Given the global physical inactivity phenomenon and the importance of PA participation for better health, the WHO guidelines for better health recommend 150 min of moderate-intensity activity for adults, 75 min of vigorous-intensity activity, or an equivalent combination per week¹⁶. For children and adolescents, the WHO guidelines recommend at least an average of 60 min/day of moderate-to-vigorous intensity PA, mostly aerobic-type PA¹⁶. Numerous epidemiological studies, however, indicate that children and adolescents around the world continue to exhibit low levels of PA, falling short of the WHO’s 2020 guidelines for daily PA^14,17. Moreover, there was an increase in adolescents’ physical inactivity, screen time, and unhealthy eating behaviors during the COVID-19 pandemic^18,19. The global phenomenon of physical inactivity and its association with poor mental health¹⁹ raise serious concerns, because young people may often experience poor mental health. Poor mental health causes human suffering, and affects everyday living including school performance with serious consequences for the overall well-being. While various conventional mental health therapies are employed, they are frequently associated with long waiting lists, ineffectiveness, inaccessibility, costs and/or unpleasant side effects.

Implementing early-targeted educational and coping strategies such as PA during childhood and adolescence⁴ is particularly critical, as this is a developmental period in which many mental health disorders first manifest²⁰. PA may act as a medium through which such strategies are conveyed or strengthened. PA can foster self-regulation, resilience, and adaptive coping skills that transfer to broader life contexts. In this way, PA serves not only to promote mental health directly but also to provide an experiential framework for practicing and reinforcing educational and coping strategies⁵¹ to manage emerging challenges such as eco-anxiety¹⁰. Also lays the foundation for long-term psychological wellbeing with broader implications for reducing future healthcare costs through early mental health promotion²¹.

According to Clayton et al.²² PA can boost mood, confidence, and emotional regulation, even in the aftermath of disasters such as floods. By releasing endorphins, PA reduces stress and enhances well-being, effects particularly relevant to eco-anxiety. Resilience research further highlights PA as a key factor in fostering psychological strength and adaptability²³. Given these benefits, PA should be incorporated into preventive and therapeutic strategies and considered a potential moderator in managing eco-anxiety and promoting overall health.

A moderated moderation framework between PA, floods experience, eco-anxiety and well-being

System-level and individual-level strategies are essential for supporting youth mental health in the face of climate-related threats. While system-level/community-wide adaptations such as policies and infrastructure provide broad resilience against environmental hazards, individual behaviors offer immediate and personal benefits for mental and emotional well-being²⁴. PA as already stated has consistently been shown to reduce anxiety and depressive symptoms, regulate mood and foster resilience and coping capacities^4,22. These effects are especially important for youth exposed to natural disasters such as floods, where direct experience may heighten eco-anxiety and its negative impact on well-being¹⁰. Together, these approaches create a complementary framework: broad structural changes lower overall risk, and personal coping through PA empowers people to stay mentally, emotionally but also physical healthy. Promoting PA should be prioritized in preventive and therapeutic interventions targeting eco-anxiety and climate-related mental health challenges among young people²⁵.

The present study advances the current literature by testing a moderated moderation model. Specifically, we hypothesize that flood experience moderates the association between eco-anxiety and well-being, such that the association is stronger for those with direct flood experience. This hypothesis is grounded in research showing that personal exposure to climate-related disasters can heighten eco-anxiety and its psychological costs^12–14. Children and adolescents who have lived through floods face not only the general fear of climate change but also direct experiences, which can make eco-anxiety stronger and more harmful to their well-being. Further, we propose that this moderation effect is itself conditional on PA frequency, that is, higher levels of PA may buffer the negative impact of eco-anxiety on well-being, especially among youth with direct flood experience. This three-way interaction is depicted in Fig. 1. Our approach integrates an individual-level adaptation framework, highlighting PA as a practical, resilience-enhancing strategy for at-risk youth exposed to climate-related disasters.

Fig. 1.

Hypothesized moderated moderation effect of flood experience and PA frequency on the eco-anxiety and well-being relationship.

Methods

Study design and participants

In this cross-sectional population-based study, the sampling frame consisted of all adolescents studied in publicly funded elementary (fifth and sixth grade) and secondary (first, second, and third grade) schools within central Greece during the academic year 2024–2025. An a priori power analysis using the software G*Power (version 3.1.9.) for a regression based moderated moderation analysis indicated that the minimum sample size to yield a statistical power of at least 0.80 with an alpha of 0.05 and a small effect size (f² = 0.01) is 787. A quota sampling was employed to ensure equal representation across key demographic variables: gender (almost 50% males and 50% females), educational level (almost 50% students from elementary schools and 50% students from high schools), and flood experience (almost 50% students with a flood experience and 50% students without a flood experience). Participants were recruited via school administrators from January to March 2024. Inclusion criteria in the study were the ability to complete the self-reported study questionnaires in Greek. The final sample consisted of 1.021 students (448 males, 565 females, eight missed to provide their gender) aged 9 to 16 years old (Mage = 12.35 ± 1.38), and coming from different schools in central Greece (elementary schools: n = 542; high schools: n = 479). In this study, to minimize potential bias, we used validated and widely used scales to reduce measurement error associated with self-reported data. Participants completed the questionnaires from March to June 2024, six months after the “Daniel” storm.

Measures

Demographic variables

A detailed description of the demographic variables is presented in Table 1.

Table 1.

A list of the demographic variables, with the respective demographic questions, and their response options.

Demographic variables	Question	Response options
Gender	What is your gender?	1. Boy 2. Girl 3. Non-binary 4. Other 5. Do not know
Age	How old are you?	Write your age in years.
City	What is the name of the city where you live now?	Write the name of the city where you live now
Ethnicity	What is your ethnicity?	Write your ethnicity (e.g., Greek)
Educational level	What is your stage of education?	1. Elementary school 2. High school

PA

The brief two-item questionnaire (PACE+) was employed to record the frequency of adolescents’ engagement in moderate-to-vigorous PA²⁶. Under the stem “Over the past 7 days or a typical or usual week, on how many days were you physically active for a total of at least 60 minutes per day?”, participants responded on an 8-point Likert scale from 0 (0 days) to 7 (7 days). In this study, we preferred to use the PACE + questionnaire over accelerometers or other age-specific tools for the following reasons: First, in epidemiology, self-report questionnaires are frequently used due to their low cost, convenience, and ease of administration. Second, this PA measure has already been applied and tested in studies with adolescents globally, in Europe, and in Greece²⁷. A composite score was created, by taking the mean of the responses.

Well-being

The five-item scale of WHO’s, also known as WHO-5 Well-Being Index, was used to assess adolescents’ psychological well-being (e.g., “I feel calm and relaxed” or “I feel active and vigorous”)²⁸. Participants responded on a 6-point Likert scale ranging from 0 (At no time) to 5 (All the time), with higher scores representing better quality of life. This scale is frequently used in epidemiological studies with adolescents worldwide¹⁹. A composite score was created, by taking the average (mean) of the responses.

Flood experience

A single item was employed to capture participants’ previous flood experience “Have you personally experienced a recent flooding event in your area?”. Participants had two options for response (Yes or No). Similar items have been used in previous environmental studies to capture the experience of flooding²⁹.

Eco-anxiety

The 13-item Hogg Eco-Anxiety Scale was employed to assess participants’ anxiety about a variety of ecological crises⁶. Under the stem “Over the last 2 weeks, how often have you been bothered by the following problems, when thinking about climate change and other global environmental conditions (e.g., global warming, ecological degradation, resource depletion, species extinction, ozone hole, pollution of the oceans, deforestation)?”, participants provided their responses (e.g., “Worrying too much” or “Feeling anxious about the impact of my personal behaviors on the earth”) on a 4-point Likert scale ranging from 0 (Not at all) to 3 (Nearly every day). This scale has already been translated, validated, and successfully used in several languages worldwide³⁰. A composite score was created, by taking the average (mean) of the responses.

Procedure

The research was conducted following international ethical guidelines that are consistent with the guidelines of the American Psychological Association and with the Declaration of Helsinki. Also, we received approval for this study from the University Ethics Committee (Ref. Number 09/17.1.2024). Informed consent was secured from both parents/caregivers and pupils before the beginning of the study. All students voluntarily participated in the study and completed a multi-section paper-based questionnaire tapped the studied variables during a PE class. A researcher was always present during the completion of the questionnaire. All students were informed about the purpose of the study, the anonymity and the confidentiality of their responses, and their right to drop out of the survey at any time they want. To further address potential social desirability bias, the participants, besides their reassurance regarding the anonymity and confidentiality of their responses, also received instructions aiming at reducing their socially desirable responses (e.g., that there are no right or wrong answers and to answer the questions as honestly as possible).

Data analyses

Data analyses were performed using IBM SPSS Statistics 24 (IBM Corp., Armonk, NY, USA). The factorial validity of the Greek version of the Hogg Eco-Anxiety Scale⁴ was assessed through confirmatory factor analysis (CFA) using the SPSS statistical program AMOS 22 Two alternative models were tested. First, in line with Hogg et al.⁶ we hypothesized and estimated a correlated four-factor, first-order eco-anxiety model in which items were only allowed to load on their intended factor, factors were allowed to correlate, but error terms were not. We also tested a hierarchical eco-anxiety model in which the four first-order latent factors were represented by one higher-order latent factor. In both models, for purposes of identification and latent variable scaling, one item from each factor was fixed to 1.0. To evaluate the fit of the data to the hypothesized models, we use a selection of goodness-of-fit indices, including Chi-square (χ²), Comparative Fit Index (CFI), Tucker-Lewis index (TLI), and Root Mean Square Error of Approximation (RMSEA). CFI and TLI values > 0.95 and RMSEA values < 0.06 are considered indicators of excellent fit, whereas CFI and TLI values > 0.90 and RMSEA < 0.08 are considered indicators of acceptable fit³¹.

Descriptive statistics, internal consistency coefficients (Cronbach’s alphas), and Pearson’s correlations for all study variables were also calculated. To examine the hypothesized moderated moderation (three-way interaction) model (Fig. 1) we used the PROCESS macro for SPSS-version 4.2, an observed variable path analysis modeling tool that is based on regression³². PROCESS macro is widely used through the social, business, and health sciences to perform mediation, moderation and conditional process analysis. In this study, we performed a moderated moderation analysis using Model 3. In the model, we entered eco-anxiety as the independent variable, well-being as the outcome variable, flood experience disaster as the primary moderator W, and PA frequency as the secondary moderator Z. In all analyses conducted, we treated all participants regarding their age range (9–16) as a single group, that is, adolescents, because adolescence has been defined as the phase of life between childhood and adulthood that begins with the onset of puberty (between the age of 8 and 10 years) and ends in the mid-20s³³. Finally, listwise deletion was used to handle missing data, as this method is relatively “safe” when there are less than 5% missing values across all participants, considered to be missing completely at random, as in our study.

Results

The factorial validity of the Greek version of the Hogg eco-anxiety scale

The original 13-item four-factor correlated model of eco-anxiety demonstrated an excellent fit to the data: χ² (59, Ν = 1,021) = 149.85, p < 0.001, CFI = 0.98, TLI = 0.97, RMSEA = 0.04, RMSEA (90% CI 0.03, 0.05). Likewise, the hypothesized 13-item hierarchical model of eco-anxiety displayed an acceptable fit to the data, slightly worse compared with the original four-factor correlated model of eco-anxiety: χ² (61, Ν = 1,021) = 248.23, p < 0.001, CFI = 0.96, TLI = 0.94, RMSEA = 0.06, RMSEA (90% CI 0.05, 0.06]. For both models, all items loaded on their respective factors significantly and adequately (standardized factor loadings ranging from 0.61 to 0.92).

Descriptive statistics, reliabilities, and correlations

The extent of missing data, descriptive statistics (means, standard deviations), Cronbach’s alpha reliability coefficients, and correlations for study variables are presented in Table 2. The completion rate was above 97.69% for all study variables. Examination of Cronbach’s alpha coefficients (ranging from 0.81 to 0.87) revealed adequate internal consistency (a > 0.70) for all scales. Four hundred fifty-nine (45.9%) reported having experienced a recent flood disaster in the place where they live, whereas 529 (51.8%) reported that they have not, while 23 (2.3%) didn’t provide any answer. In general, young people scored low on the eco-anxiety scale and moderately to highly on the PA frequency, and well-being scales. Eco-anxiety had non-significant correlations with flood experience and PA frequency, and low negative significant correlations with well-being. Flood experience had a low negative significant correlation with well-being and non-significant correlations with PA frequency. Finally, PA frequency had low positive significant correlations with well-being.

Table 2.

The extent of missing data, descriptive statistics, cronbach’s alpha, and correlations for all study variables.

Variables	Missing Data N = 1021		M	SD	a	1	2	3	4
	n	%
1. Eco-anxiety	11	1.10	0.98	0.60	0.87	-
2. Flood experience	23	2.30	0.53	0.50	NA	− 0.01	-
3. Physical activity frequency (days per week)	0	0	3.94	1.73	0.84	0.02	− 0.03	-
4. Well-being	21	2.21	3.09	1.06	0.81	− 0.07*	− 0.10**	0.23***	-

Note: Flood experience was coded as 0 = Yes, 1 = No. The mean for flood experience reflects the percentage of young people not experiencing a flood disaster recently. NA = Not Applicable. Eco-anxiety scores (possible range 0–3); Physical activity frequency (days per week) scores (possible range 0–7); Well-being scores (possible range 0–5). *p < 0.05, **p < 0.01, ***p < 0.001.

Moderated moderation analysis

The PROCESS macro for SPSS, version 4.2 (Model 3) was used to test the moderated moderation (three-way interaction) model (Fig. 1; Table 3). Particularly, we examined whether the relationships between eco-anxiety and well-being were conditional upon flood experience disaster and PA frequency; and whether the two-way interaction between eco-anxiety and flood experience disaster was conditional upon PA frequency. In other words, we examined whether the relationship between eco-anxiety and well-being was moderated by flood experience disaster alone, PA frequency alone, and/or flood experience disaster and PA frequency combined (Fig. 1).

Table 3.

Summary of moderated moderation results by flood experience and PA frequency with unstandardized regression coefficients for the total sample and for the adolescent girls and boys separately.

	Dependent Variable: Well-being
	Total Sample					Adolescent Girls					Adolescent Boys
	B	SE	t	p	95% CI	B	SE	t	p	95% CI	B	SE	t	p	95% CI
Simple effects
Eco-anxiety	−0.68	0.20	−3.36	0.001	−1.08 – −0.28	−0.79	0.31	−2.89	0.004	−1.33– −0.25	−0.62	0.30	−2.05	0.041	−1.22– −0.03
Flood experience	−0.87	0.32	−2.77	0.006	−1.49 – −0.25	−0.76	0.27	−1.61	0.108	−1.68–0.17	−1.35	0.44	−3.07	0.002	−2.21– −0.49
PA frequency (days per week)	0.03	0.05	0.55	0.580	−0.07– 0.13	−0.05	0.08	−0.62	0.533	−0.20– 0.10	0.04	0.07	0.51	0.609	−0.10– 0.18
Two-way interaction
Eco-anxiety x Flood experience	0.74	0.28	2.69	0.007	0.20–1.29	0.76	0.40	1.90	0.058	−0.03–1.55	0.85	0.39	2.15	0.032	0.07–1.62
Eco-anxiety x PA frequency	0.12	0.05	2.83	0.005	0.04–0.21	0.18	0.06	2.88	0.004	0.06–0.30	0.09	0.06	1.44	0.151	−0.03– 0.21
Flood experience x PA frequency	0.16	0.07	2.18	0.029	0.02–0.30	0.23	0.12	1.94	0.053	−0.01– 0.46	0.18	0.09	1.89	0.060	−0.01– 0.36
Three-way interaction
Eco-anxiety x Flood experience PA frequency	−0.17	0.06	−2.79	0.005	−0.30 – −0.05	−0.25	0.10	−2.58	0.010	−0.44– −0.06	−0.13	0.08	−1.61	0.108	−0.30– 0.03

Note. SE = Standard error; CI = Confidence interval.

For the total sample and girls, PROCESS revealed a statistically significant three-way interaction of eco-anxiety x flood experience disaster x PA frequency for well-being (Table 3; Figs. 2 and 3). The moderated moderating model explained 7.82% of the variance in well-being for the total sample, F (7, 989) = 11.98, p < 0.001; and 6.44% of the variance in well-being for girls, F (7, 547) = 5.38, p < 0.001. The statistically significant three-way interaction accounted for only a very small of unique variance in well-being for both the total sample, ΔR² = 0.007, ΔF (1, 989) = 7.81, p = 0.005, and girls, ΔR² = 0.011, ΔF (1, 547) = 6.67, p = 0.010.

Fig. 2.

The conditional effect of eco-anxiety on young people’s well-being as a function of flood experience and PA frequency from the moderated moderation model in Fig. 1.

Fig. 3.

The conditional effect of eco-anxiety on adolescent girls’ well-being as a function of flood experience and PA frequency from the moderated moderation model in Fig. 1.

For both the total sample and girls, to further explore the moderating effect that PA frequency was having on the moderated effect of flood experience disaster on the relationship between eco-anxiety and well-being, we graphically plotted and probed the three-way interaction using the simple slope analysis, and the Johnson-Neyman technique³². Simple slope analysis estimates the conditional effect of the XW interaction at value(s) of Z (e.g., the mean value plus one SD above and below the mean) and conducts an inferential test for this conditional interaction at that value of Z³². Conversely, the Johnson-Neyman technique indicates where the conditional interaction between X and W transitions between statistically significant and not along the distribution of Z³².

For the total sample, simple slope analysis revealed that at low levels of PA frequency (1 SD below the mean; Z = 2.22 days per week), the effect of eco-anxiety on well-being is significantly moderated by flood experience disaster, [b = 0.36, F (1, 989) = 5.02, p = 0.025]. However, at moderate levels of PA frequency [Mean; Z = 3.94 days per week; b = 0.06, F (1, 989) = 0.29, p = 0.591] and at high levels of PA frequency [1 SD above the mean; Z = 5.67 days per week; b = −0.24, F (1, 989) = 2.64, p = 0.105], flood experience disaster did not significantly moderate the effect of eco-anxiety on young people’s well-being. In contrast, for girls simple slope analysis revealed that at low levels of PA frequency [1 SD below the mean; Z = 2.02 days per week; b = 0.26, F (1, 547) = 1.21, p = 0.271] and at moderate levels of PA frequency [Mean; Z = 3.65 days per week; b = −0.15, F (1, 547) = 0.90, p = 0.343]. However, at high levels of PA frequency [1 SD above the mean; Z = 5.28 days per week; b = −0.56, F (1, 547) = 6.88, p = 0.009], flood experience disaster significantly moderate the effect of eco-anxiety on adolescent girls’ well-being.

For the total sample, the Johnson-Neyman technique indicated that the two-way interaction between eco-anxiety and flood experience disaster was statistically significant when PA frequency scores were lower than 2.71 (b = 0.27, 95% CI = 0.00, 0.55, p = 0.050) or higher than 6.25 (b = −0.34, 95% CI = −0.68, 0.00, p = 0.050). These results indicate that at low levels of PA frequency, flood experience disaster increases the negative effect of eco-anxiety on well-being, whereas at very high levels of PA frequency, the negative effect of eco-anxiety on well-being held for young people without flood experience disaster but not for those experiencing a flood disaster for whom the effect of eco-anxiety on well-being was positive (Fig. 2). In other words, the moderating-exacerbating influence of flood experience disaster on the negative relationship between eco-anxiety and well-being was buffered at very high levels of PA frequency. For adolescent girls, the Johnson-Neyman technique indicated that the two-way interaction between eco-anxiety and flood experience disaster was statistically significant when PA frequency scores were higher than 4.35 (b = −0.33, 95% CI = −0.65, 0.00, p = 0.050). These results indicate that the moderating-exacerbating influence of flood experience disaster on the negative relationship between eco-anxiety and well-being was buffered at moderate to high levels of PA frequency (Z > 4.35 days per week), but not at low levels of PA frequency (Fig. 3). Also, importantly, and similarly to our findings above for the total sample, at high levels of PA frequency (Z = 5.28 days per week), the negative effect of eco-anxiety on well-being held for adolescent girls without flood experience disaster but not for girls experiencing a flood disaster, for whom the effect of eco-anxiety on well-being was positive (Fig. 3).

Finally, PROCESS revealed a non statistically significant three-way interaction of eco-anxiety x flood experience disaster x PA frequency for adolescent boys’ well-being (Table 3). The moderated moderating model explained 11.15% of the variance in adolescent boys’ well-being, F (7, 427) = 7.65, p < 0.001. However, PROCESS revealed a statistically significant conditional interaction between eco-anxiety x flood experience disaster. This result means that when PA frequency (Z) is equal to zero (0), flood experience disaster moderates (increases) the negative effect of eco-anxiety on adolescents boys’ well-being.

Discussion

Natural disasters like floods driven by climate change are a major source of eco-anxiety in the sense of a lived experience for young people and pose a serious threat to overall human well-being^7–11. While some previous studies have examined the relationships between eco-anxiety, flood experience disaster, and adolescents’ well-being, their interplay with PA remains unexplored^9–12,24. The benefits of PA on mental health and overall well-being in young people are well documented^4,13, but its relationships with eco-anxiety and flood experience disaster remain unexplored. This study shed some light on this complex synthesis of associations by examining (a) the relationships among eco-anxiety, flood experience disaster, and adolescents’ well-being, and (b) whether PA moderates the moderating role of a large flood disaster experience on the relationship between eco-anxiety and adolescents’ well-being.

We found that adolescents’ well-being had a low negative significant correlation with eco-anxiety and a low positive significant correlation with PA frequency. These results are in line with previous research findings in adolescents indicating that well-being is negatively related to eco-anxiety^8–11 and PA frequency^4,13. However, non-significant correlations between eco-anxiety and both flood experience and PA frequency emerged. These results suggest that in our sample of adolescents, eco-anxiety was not related to flood experience disaster and PA frequency. Potential reasons for these non-significant findings might be the fact that we analyzed only the total score of participants’ eco-anxiety and that we measured participants’ flood experience by only a single-item, limitations that future research could address.

Despite these non-significant findings, moderated moderation analyses in the total sample and girls indicated that PA frequency significantly moderated the moderating role of the flood experience disaster on the eco-anxiety and well-being relationship, supporting the initial hypothesis. Nevertheless, for both the total sample, and adolescent girls, the statistically significant three-way interaction accounted for only a very small of unique variance in well-being. In line, previous cross-sectional^6,34 and longitudinal research³⁴ found that eco-anxiety had a small negative relationship with well-being in undergraduate students. According to Dawson, significant interactions generally have a small effect size³⁶. Consistent with Dawson’s conclusions on effect size³⁶, the statistically significant three-way interactions found in our study accounted for only 1% of unique variance in adolescents’ well-being. Finally similar findings have been reported in previous research, for example Aguinis et al. reported a median f² value of just 0·002 across 30 years of research in applied psychology³⁷.

Following Dawson’s recommendation, we did not focus on the size of the effect per se, but rather the practical relevance of significant interactions³⁶. To aid in interpreting the practical relevance of a significant interaction, we graphically plotted and subsequently probed the interaction³⁸ using the simple slope analysis, and the Johnson-Neyman technique. These techniques help translate significant interactions into actionable insights. They move beyond “there is an interaction” to answer how, for whom, and under what conditions, which is key for policy-making, treatment design, and applied interventions³². In particular, interaction plots visualized the conditional effect of the two-way interaction of eco-anxiety x flood experience on well-being at different values of PA frequency (Z = 2.00, Z = 4.00, Z = 6,50 corresponding to the 16th, 50th, and 84th percentiles respectively). The real-world implication of interaction plots is that helps us to intuitively see patterns, and particularly how the conditional effect of the two-way interaction of eco-anxiety x flood experience on well-being changes at different values of PA frequency. Simple slope analysis estimated the conditional effect of the eco-anxiety x flood experience two-way interaction at specific value(s) of Z (i.e., the mean value plus one SD above and below the mean) and conducted an inferential test for this conditional interaction at that value of Z³². The real-world implication of simple slope analysis is that helped us to determine for whom the effect of eco-anxiety on well-being is significantly moderated by flood experience. In particular, for the total sample, simple slope analysis revealed that the effect of eco-anxiety on well-being is significantly moderated by flood experience disaster in adolescents with low levels of PA frequency. In contrast, flood experience disaster did not significantly moderate the effect of eco-anxiety on well-being in adolescents with moderate or high frequency of PA participation. For adolescent girls, simple slope analysis revealed that at low levels of PA frequency and at moderate levels of PA frequency, flood experience did not significantly moderate the effect of eco-anxiety on adolescent girls’ well-being. However, at high levels of PA frequency, flood experience significantly moderated the effect of eco-anxiety on adolescent girls’ well-being.

Overcoming a serious limitation of simple slope analysis, which picks entirely arbitrary points (e.g., Mean +/- 1 SD)³², the Johnson-Neyman technique helped us to identify the exact values of PA participation at which the conditional interactions between eco-anxiety and flood experience transitioned from a significant to a non-significant predictor of well-being, which is a strength of this study. Particularly, for the total sample, when PA frequency participation was < 2.71 days/week, flood experience disaster exacerbated the negative effect of eco-anxiety on well-being. When PA frequency was > 6.25 days/week, however, the moderating-exacerbating influence of flood experience disaster on the negative relationship between eco-anxiety and well-being was weakened. Similarly, for adolescent girls, when PA frequency scores were higher > 4.35 days/week, the moderating-exacerbating influence of flood experience disaster on the negative relationship between eco-anxiety and well-being was buffered, but not at low levels of PA frequency. These precise thresholds are extremely valuable for designing PA guidelines for adolescents, developing effective intervention, and setting policies.

Overall, our findings indicate that adolescents experiencing flood disasters might face heightened vulnerability to the negative psychological impacts of eco-anxiety, particularly at lower levels of PA. Specifically, lower frequencies of PA could exacerbate eco-anxiety’s negative impact because individuals lacking regular PA may have fewer resources both physiologically³⁹ and psychologically⁴⁰ to cope effectively with traumatic or stress-inducing events. This finding aligns with our hypothesized moderated moderation model (Fig. 1), which predicts that the eco-anxiety and well-being relationship depends jointly on flood experience and PA frequency. As depicted in Figs. 2 and 3, moderate to high frequencies of PA mitigate the negative impact of flood experience on adolescents’ well-being. Several physiological and psychological mechanisms can explain this buffering effect. Physiological mechanisms include reductions in stress hormones such as cortisol and adrenaline, increased release of neurotransmitters like dopamine, serotonin, norepinephrine, and endorphins, which reduce symptoms of anxiety and depression³⁹. Psychological mechanisms through which PA enhances adolescents’ mental health include⁴, better stress management, emotional regulation, and coping skills, increased self-efficacy and resilience, and improved social skills and community integration⁴⁰. Thus, regular PA may significantly bolster adolescents’ ability to effectively manage eco-anxiety and minimize its detrimental effects following climate-related disasters like floods.

In contrast to our findings above for the total and adolescent girls, we found a non statistically significant three-way interaction of eco-anxiety x flood experience disaster x PA frequency for adolescent boys’ well-being. One possible explanation for this finding, compared to the statistically significant three-way interaction that we found for girls in our study is the following. Previous research showed that compared to boys, girls reported higher levels of eco-anxiety⁴¹, lower levels of well-being¹⁹,and lower PA frequency^17,19. Therefore, moderate to high levels of PA frequency are able to help girls to mitigate the negative effect of eco-anxiety on their well-being, and become more resilient to climate-change challenges, compared to boys who already display high levels of PA^17,19, and are more resilient (higher well-being¹⁹, and lower eco-anxiety⁴¹ compared to girls. This explanation was supported also by our findings for the total sample, indicating that only high levels of PA frequency, and specifically > 6.25 days/week are capable of helping adolescents boys and girls to lessen the negative effect of eco-anxiety on their well-being.

Overall, the above findings suggest that girls compared to boys are more sensitive to the combined effects of eco-anxiety and flood experience but also more amenable to the buffering benefits of regular PA. This could be further explained by a mix of socialization processes and hormonal and neurobiological differences. Regarding socialization processes, girls are often socialized to be more emotionally expressive, empathetic, and to internalize their feelings. This means they are more likely to deeply process and dwell on threats like climate change (eco-anxiety) and the direct trauma of a flood experience. Boys, conversely, might be socialized to externalize stress (e.g., through aggression, risk-taking) or to suppress emotional vulnerability⁴². Moreover, from early childhood, boys are often encouraged and provided with more opportunities for sport and PA⁴³. Conversely, academic achievement, social relationships, and domestic responsibilities are often more heavily emphasized for girls, potentially leading to less time and energy for PA⁴³. However, PA is a well-established buffer against the higher internalizing problems (e.g., anxiety, depression) on which girls are more prone⁴. Therefore, for girls, who may have fewer “socially sanctioned” outlets for externalizing stress, PA might become a particularly crucial and effective coping strategy⁴⁴.

Concerning neurobiological differences, PA impacts neurotransmitters (like serotonin and dopamine) and neurogenesis³⁹, which are involved in mood regulation. If girls are more susceptible to disruptions in these systems due to stress (as suggested by higher rates of anxiety/depression), then the positive neurobiological effects of PA might be more impactful and evident in their well-being outcomes, especially when combined with significant stressors. Lastly, the hormonal differences between the two sexes may also amplify the buffering benefit of PA in girls. Especially, the interplay between estrogen and oxytocin significantly impacts well-being, particularly in women⁴⁵. Therefore, oxytocin release from PA⁴⁶ may buffer stress more effectively and enhance well-being in girls than in boys.

Overall, our findings above are consistent with 2020 WHO PA guidelines for children and adolescents, which recommend an average of at least 60 min of moderate-to-vigorous PA per day (mostly aerobic)¹⁶. Thus, our findings support current evidence for the beneficial effect of PA on mental well-being and health^4,13. Available reviews on the relationship between climate change and PA in children have predominantly focused on the effects of climate change on PA^35‚47 and emphasized the association of climate change with reduced participation in PA and in sport engagement during heat waves⁴⁷. However, these reviews have overlooked the potential reciprocal effects of PA on climate change, such as reductions in air pollution though active transport including walking or biking⁴⁸.

Climate change is a threat that requires examination from a multidisciplinary perspective. In this respect, education in young people, who are the future generations of the planet, emerges as an essential tool to respond to the socio-environmental problems of a more sustainable planet⁴⁹. PA may influence environmental impact depending on the level and type of activity or inactivity. For example, a sedentary lifestyle is typically associated with higher carbon emissions due to the increased dependence on motorized transport and activities such as the use of the elevator instead of the stairs or prolonged television viewing instead of playing outside. To this extent, sedentary lifestyle contributes to greater energy consumption, thereby elevating carbon levels. Substituting sedentary behaviours with PA is advantageous not only for physical and psychological health but also contributes positively to environmental sustainability.

Practical and intervention implications

In practical terms, our findings suggest that interventions aimed at reducing eco-anxiety in youth should explicitly integrate PA with environmental sustainability initiatives. For example, school-based “active transport” programs (e.g., walk-and-bike to school days) and routine stair-vs-elevator campaigns at school and at home can simultaneously lower carbon emissions and foster empowerment through PA. Incorporating brief, structured “green exercise” breaks such as “brain-boost” exercises accompanied by eco-information can reinforce the link between personal well-being and planetary health. Initiatives with parents or peers such may enhance motivation and peer support, and parental engagement. Moreover, community events (e.g., sustainable eating) help sustain behavior change beyond the school setting. Finally, embedding strategies within the formal curriculum of PE will be essential to translate individual activity choices into broader climate resilience and mental-health benefits.

Based on our findings that PA could be beneficial for children and adolescents who have directly experienced climate-related crises such as floods, we recommend the development of enjoyable PA interventions tailored for this age group. Providing opportunities for fun and play is essential, as engaging in enjoyable activities can offer moments of relief and happiness even in difficult times. Such programs could include structured, supportive group activities in schools and communities.

Finally, in light of our findings on gender differences, it is important that strategies are gender-sensitive, taking into account variations in eco-anxiety experiences and sociocultural frameworks. Previous research showed that compared to boys, girls reported higher levels of eco-anxiety⁴¹ and lower PA frequency^17,19. Therefore, for girls PA promotion strategies in the context of eco-anxiety should prioritize social connection, emotional expression, and nurturing engagement with nature. Team-based formats that embed PA within adventure or climate resilience narratives may also resonate more strongly with boys’ coping styles and motivational preferences⁵⁰. It is important to acknowledge, however, that these practical implications are not grounded in empirical data and should therefore be regarded as speculative.

Limitations and future research directions

With respect to the study limitations and future research directions, the cross-sectional design employed restricts the ability to infer causal relationships between the variables under examination. This limitation highlights the need for longitudinal and experimental research. A point regarding the effect sizes reported in this study that may inform future research is that although the interactions account for only 1% of the variance in adolescents’ well-being, according to Aguinis et al.³⁷ this small effect may be meaningful in practice. To determine the practical significance of this interaction, Aguinis and colleagues³⁷ recommended that qualitative methods could be adopted to probe the importance of the findings for specific “stakeholders”. In the case of the effect of the three-way interaction of eco-anxiety x flood experience x PA frequency on adolescents’ well-being, stakeholders may include adolescents experiencing eco-anxiety and flood disaster, as well as adolescents experiencing eco-anxiety and without flood disaster, who could be interviewed to understand the implications of different PA frequencies on their well-being.

Moreover, we did not account for potential clustering of participants within schools, which may violate the assumption of independent observation and may bias standard errors and p-values since adding multilevel data into a moderated moderation framework exceeds the capabilities of Hayes’ PROCESS macro. Another study limitation is that the single-item measure for flood experience may have failed to capture the nuance of flood experiences, such as varying degrees of exposure, impact, and proximity. Also, the study did not account for key potential confounders such as socioeconomic status, family support systems, or pre-existing mental health conditions that may simultaneously affect both access to PA and the psychological impact of flood exposure. Finally, this study examined only the participants’ eco-anxiety total score in relation to the study variables and not the four dimensions of eco-anxiety. Thus future research could explore how the four eco-anxiety dimensions might differentially relate to PA frequency and flood experience, to enrich our understanding of eco-anxiety’s impact on well-being, and provide a clearer picture of the underlying mechanisms at play. Finally, future research should examine how the unique cultural, social, and environmental characteristics of Greece shape these findings and assess their generalizability across other countries and regions.

Conclusions

Given the exploratory nature of this study, our results only suggest that a higher frequency of PA is associated with greater psychological well-being in children and adolescents, potentially by reducing the negative impact of flood experiences and eco-anxiety on overall well-being. Also, existing evidence may guide future research on the key role of PA in enhancing physical and psychological health and in providing impact to environmental sustainability by reducing energy consumption, which is linked to mitigating climate change.

Author contributions

All authors contributed to the study design and conceptualisation. NC supervised the research activity.NC and YT acquired the financial support for the project leading to this publication. NC, CK, MH, AK, and YT identified the measures used in the survey. AV, CK, and ED performed the data collection and data entry. NC and NK analyzed and verified the underlying data NC and AK completed the literature search and review, and drafted the manuscript. NC, AK, IM, CK, YD, MH, AV, and AP revised and commented on the manuscript and approved the final version.All authors confirm they had full access to all the data of the study and accept responsibility for the decision to submit for publication.

Funding

This project is carried out within the framework of the National Recovery and Resilience Plan Greece 2.0, funded by the European Union – NextGenerationEU (Implementation body: HFRI) (H.F.R.I. Project Number: 16567, ENLITE Project, ELKE: 7590). https://greece20.gov.gr.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due that this research is part of a funding project that has not finished yet but are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.