How seasons, weather, and part of day influence baseline affective valence in laboratory research participants?

Many people believe that weather influences their emotional state. Along similar lines, some researchers in affective science are concerned whether testing individuals at a different time of year, a different part of the day, or in different weather conditions (e.g., in a cold and rainy morning vs. a hot evening) influences how research participants feel upon entering a study; thus inflating the measurement error. Few studies have investigated the link between baseline affective levels and the research context, such as seasonal and daily weather fluctuation in temperature, air pressure, and sunshine duration. We examined whether individuals felt more positive or negative upon entering a study by clustering data across seven laboratory experiments (total N = 1108), three seasons, and daily times ranging from 9 AM to 7 PM. We accounted for ambient temperature, air pressure, humidity, cloud cover, precipitation, wind speed, and sunshine duration. We found that only ambient temperature was a significant predictor of valence. Individuals felt more positive valence on days when it was cooler outside. However, the effect was psychologically negligible with differences between participants above c.a. 30 degrees Celsius in ambient temperature needed to generate a difference in affective valence surpassing one standard deviation. Our findings have methodological implications for studying emotions by suggesting that seasons and part of the day do not matter for baseline affective valence reported by participants, and the effects of ambient temperature are unlikely to influence most research.

Introduction

Humans engage in daily activities that elicit positive and negative affect. For instance, people perceive favorable activities—going on a trip with friends and playing in the park with the child, lying in a hammock overlooking the beach—as eliciting positive affect [1]. On the other hand, people evaluate unfavorable activities—spending the holidays alone, having a home destroyed by a tornado, being struck by lightning—as eliciting negative affect [1]. These activities are often determined by contextual factors such as weather conditions and time cycles, including seasons and days. Although the activities themselves mostly determine the emotional experience, the contextual factors also impact how positive or negative people feel in the present moment [2]. Thus, in this study, we focused on contextual factors that are believed to influence emotional affect, namely weather, seasons, and parts of the day.

As for seasons, people feel the worst in winter and feel the best in summer [3, 4]. For instance, one seasonality’s negative effect on affect is conceptualized as winter seasonal affective disorder [5]. The seasonal affective disorder is described in DSM as a variant of categorical mood disorder. However, it has been clear that the tendency of people to experience seasonal changes in mood and behavior is not limited to those severely affected but has an impact on the normal population as well [6-8]. The review of epidemiological research on seasonal affective disorder found that almost in all studies, seasonal variations in mood were found with the depressive symptoms usually peaking in winter [7]. Although most studies used self-report measures in which individuals reported when they felt best or worst during the past year, others measured the mood successively across seasons and supported the mood drops in winter [7].

Furthermore, as almost all processes with a physiological component [9], affect–especially positive affect—also has diurnal cycle components [10-13]. Individuals have an endogenous circadian system that operates with solar time in the day-night cycle [14, 15]. However, there is no consensus related to the peak of affective experience during the day, with research showing conflicting information for when people feel best throughout the day [11, 13, 16]. Studies have shown that people feel the best in the morning [17, 18] around 11:00 [19], around the middle of the day [12, 20], in the middle of the afternoon [11], and in the evenings [13, 16].

Affect is also influenced by weather conditions [21, 22]. For instance, individuals tend to feel better when the day is less cloudy [23-26], barometric pressure is higher [27], precipitation lower [24, 28], and wind power stronger [23]. However, more recent large-scale studies have not replicated many previous findings regarding the link between weather and emotions [29, 30]. For instance, contradictory effects have primarily been focused on temperature. Some researchers have found that temperature increases are associated with increased positive affect and reduced negative affect [24, 27, 28, 31], whereas others have found the opposite [23, 32–34].

The contradictory results may stem from not accounting for seasons and parts of the day when studying weather and affect associations. For instance, the negative association between the daily variation of affect and temperature or sunshine may be driven by the fact that people are more positive at night simply because they are not working, without any relation to lower temperature and sunshine [34]. Furthermore, affect fluctuations related to weather conditions within a single season may not translate into the between-seasons scale. Studies have shown that affect is positively associated with air temperature and barometric pressure in spring [27] but negatively associated in summer [32]. Thus, in our study, we investigated whether people feel differently across time cycles (e.g., seasons) and whether the weather conditions might explain these effects.

Furthermore, the differences between the findings may also stem from differences in used methodologies. For instance, to measure participants’ mood studies used single-item questions [24, 25, 29, 30], standardized questionnaires [23, 27], and automatic mood detection from social media posts [16, 31]. Some studies used a single report from participants [25, 27, 30], whereas other employed repeated measure design with daily diaries [2, 20, 23, 24, 29, 35], and experience sampling methods [36]. Furthermore, the repeated measures lasted from reports over 11 days [24]; 14 days [36], 25 days [37], 30 day [35]; 90 days [2, 20], to two-year period [23]. Thus, the variety of methods may be related to the inconsistencies of the results.

Examination of how individuals feel depending on weather or point in the daily or yearly cycle is also essential for the perspective of advancing laboratory research methodology. Among concerns of experimental researchers is whether testing participants on different occasions significantly increases the noise in the measurement or produces a systematic bias [38]. This is a special case of a more general methodological problem of whether research results are influenced by contextual (and seemingly trivial) factors, e.g., a time point in the semester among students [39-41]. Thus, we focused on baseline affect among laboratory research participants. Namely, we aimed to examine whether individuals transfer any affective results of weather and time cycle into the research. If this was the case, testing individuals in more restricted conditions (e.g., at a similar part of a day or halting data collection in case of significant weather change) might improve the data quality and–possibly—reduce the type II error. If this was not the case, ignoring weather effects or time cycles might decrease the planning burden on the researcher, improve the data collection flow, and extend the time available for data collection in a project.

There is a meaningful difference between the practical methodological aim of our approach and studies that asked about the general link between weather and emotions. Most studies have been concerned with individuals’ diverse daily settings. In contrast, we focused on how individuals feel in a well-controlled laboratory environment that isolates the room environment from the outdoor environment. For instance, most researchers follow the recommendations and aim to maintain stable room temperature within the range of thermal comfort, set the constant light intensity and color temperature by covering windows and using artificial light, and reduce environmental noises (e.g., rain or wind) via the room’s sound attenuation [38, 42–44]. Consequently, this might reduce the impact of weather on affect in this specific group. Moreover, laboratory research often focuses individuals on upcoming tasks. Thus participants might be temporarily less aware of daily factors such as weather, part of day, or seasonal activities. This might even further reduce the impact of weather or time cycles on participants’ affect.

Present study

We used the novel approach to examine the time-rhythmic and weather characteristics of affective experience among laboratory research participants. We focused on affective valence, which is the most fundamental aspect of humans’ emotional response [45]. The reported affect was recorded as a part of the psychophysiological baseline and reflected the resting state before the beginning of the main experiment. The data for this investigation were collected over four years (from November 2016 to March 2019) from seven different laboratory experiments in a mild continental climate in Central Europe.

The uniqueness of our approach is threefold. First, all participants were tested in the same laboratory conditions, without the possible confounding influences of affect-associated behaviors that might bias reported affect in studies using diary methodology [23, 24], experience sampling design [36], or data collected in population studies [30]. Second, we used precise weather conditions that occurred during the experiments rather than using weather variables for the day of self-reports [30, 32]. Third, unlike other studies that examined the association between weather and affect during a single season [29, 32, 36], we investigated data collected during three seasons–winter, spring, and autumn.

Materials and methods

The data for this study were derived from seven laboratory experiments that examined the psychophysiology of emotions [46-50]. Details about the studies are presented in (S1 Table).

Participants

We collected data from 1108 individuals (47% female) that were tested in the same laboratory in Poznan, Poland. Participants were in the age between 18 and 38 (M = 21.86; SD = 2.65). All participants were Caucasian. A power analysis using G*Power 3.1 [51] indicated that examining 954 participants would allow us to detect small effect sizes of f 2 = 0.02, with the power of 0.80, for the regression coefficient. Before participating in each study, we asked volunteers to reschedule if they experienced illness or a major negative life event to eliminate factors that might influence the emotional experience. Each participant provided written informed consent and received vouchers for a cinema ticket for participation in the study. The Institutional Ethics Committee at the Faculty of Psychology and Cognitive Science, Adam Mickiewicz University, approved all seven studies.

Measures

Emotional valence

Participants continuously reported how they felt using a Response Meter (ADInstruments, New Zealand) with a scale ranging from 1 ("extremely negative") to 10 ("extremely positive"). Above the numeric scale, we provided a negative-positive valence graphical scale modeled after the self-assessment manikin [52]. A similar approach was employed in previous studies of the impact of time rhythms and weather of affect [13, 53]. The data was recorded with Powerlab and processed with LabChart 8.19 software (ADInstruments, New Zealand). Participants continuously reported their affect, while waiting for the five minutes without doing any unnecessary actions (resting baseline). We calculated the mean affective valence from the last two minutes of baseline to account for the part of the baseline that was the most proximal to the study and to limit the influence of interaction with the experimenter on affect. Electronic rating scales collect reliable and valid emotion ratings [54-56].

Weather data

We used weather data from the weather station in Poznan, collected by the Polish Institute of Meteorology and Water Management. The weather variables were matched to the experimental data of the participants by date and hour. We examined the impact of the following indicators: ambient temperature, air pressure, humidity, cloud cover, precipitation, wind speed, sunshine duration. Table 1 presents means and standard deviations for weather conditions per season, day of the week, and part of the day.

Table 1

Descriptive characteristics of affect and weather conditions.

	N	Affect		Temperature		Air Pressure		Humidity		Cloud cover		Precipitation		Sunshine duration		Wind speed
		M	SD	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD
Season
Winter	520	5.03	0.91	3.21	5.16	1007.86	11.53	57.65	40.05	5.37	2.69	0.13	0.55	0.28	0.40	1.24	1.10
Spring	252	5.50	1.26	19.53	4.69	1005.77	6.56	49.10	21.18	3.99	2.65	0.18	0.77	0.70	0.42	1.53	1.43
Autumn	336	5.13	1.04	5.15	3.95	1006.41	9.41	49.18	43.10	6.79	4.37	0.24	0.75	0.13	0.29	1.45	1.35
Part of the day
9:00–11:00	68	5.32	1.26	9.62	9.10	1007.21	8.90	76.79	25.07	4.96	2.78	0.08	0.27	0.50	0.48	1.44	1.09
11:01–13:00	216	5.19	1.01	8.98	8.69	1007.64	10.01	48.17	37.39	5.41	2.60	0.14	0.68	0.50	0.46	1.56	1.40
13:01–15:00	260	5.18	1.10	8.71	8.29	1006.41	9.37	51.71	35.44	5.84	4.93	0.20	0.66	0.45	0.45	1.50	1.35
15:01–17:00	282	5.23	1.11	8.45	8.74	1007.00	9.86	50.44	36.71	5.23	2.77	0.21	0.83	0.35	0.43	1.24	1.11
17:01–19:00	282	5.14	1.01	7.10	8.05	1006.39	10.02	53.83	38.31	5.18	2.89	0.18	0.62	0.12	0.30	1.22	1.22

Notes. N = number of participants. Units: Affect = 0–10 Likert scale points, Temperature = Celsius degrees, Air Pressure = Millibar, Humidity = percentage of saturated air at 0 degrees Celsius, Cloud Cover = 0–8 Oktas, Precipitation = millimeters, Sunshine duration = percentage of sunshine during given hour, Wind Speed = meters per second.

Time rhythms

We examined the impact of time rhythms on affect in two ways, i.e., including seasonal and daily variations. We clustered the data based collection moment: season (winter, spring, autumn), and part of the day (early mornings, 9:00–11.00; late morning, 11:01–13:00, early afternoon, 13:01–15:00, late afternoon, 15:01–17:00, and early evening, 17:01–19:00). The number of laboratory visits per season, and part of the day are presented in Table 1.

Analysis

Preliminary analysis

We examined whether the participants tested across seasons and parts of the day differed in age, sex, and BMI, using univariate analysis of variance. To examine differences between the seasons, we used post hoc tests with Bonferroni correction. To address multicollinearity between predictors of affect in our analysis we calculated variance inflation factor (VIF), with values < 5.00, and Tolerance with values > .20 indicating acceptable level of multicollinearity between variables [57, 58].

Main analysis

We examined the rhythmic characteristics of affective valence, including the impact of weather using two-level path analysis with maximum likelihood estimation with robust standard errors (MLR) in mPlus 8.0 [59, 60]. We regressed the affective valence on the mediators (weather conditions) and independent variables (seasons, part of the day). We controlled for age and sex by introducing it as a covariate for affective valence (Fig 1). We dummy-coded seasons and parts of the day, such that significant differences in the model accounted for differences relative to the winter and early mornings, respectively. In the two level-model, we nested individuals data within the studies. We calculated RMSEA, the recommended fit index for the MLR. RMSEA estimator with values < .08, along with the CFI with values above .90, indicates acceptable fit [61]. To interpret the strength of regression coefficients, we used standardized β as an indicator of 0.10 small, 0.30 medium, and 0.50 large effect sizes [62, 63].

Fig 1

Model for role of seasons, time of day, and weather conditions in affective valence.

For presentation simplicity, we grouped weather conditions and control variables in this figure. We examined paths for each variable separately.

Results

Preliminary analysis

We found that the samples examined across the seasons differed in participants age, F (2, 1105) = 10.32, p < .001, sex, F = (2, 1105) = 6.53, p = .002, but not BMI, F (2, 1005) = 0.88, p = .42. The post-hoc tests showed that we tested more women in winter and autumn than in spring, ps < .001. Furthermore, we tested younger participants in spring than in winter and autumn, ps < .05. We found that the samples examined across the parts of the day differed in participants sex, F = (4, 1103) = 2.75, p = .03, but not age F (4, 1103) = 0.17, p = .95, nor BMI, F (4, 1003) = 1.70, p = .15. The post hoc tests did not show any difference between the groups. Based on these preliminary results we controlled for participants age and sex in our main analysis. We found that multicollinearity indices for all predictors of affects were in the recommended range VIFs < 4.09 and Tolerances > .24.

Main analysis

Descriptive statistics and correlation between study variables are presented in Tables 1 and 2. The path model fit the data well, RMSEA = .04, 90% CI [.03, .05], CFI = .94 (Fig 2). For clarity, Fig 2 presents only significant paths. Table 3 presents detailed results.

Table 2

Correlations among study variables.

	M	SD	1.	2.	3.	4.	5.	6.	7.	8.
1. Affect	5.19	1.08
2. Temperature	7.48	8.18	.09**
3. Air Pressure	1006.91	10.06	0.01	-.27**
4. Humidity	53.94	37.99	0.01	-.22**	-.08**
5. Cloud Cover	5.53	3.38	-.07*	-.24**	-.11**	.12**
6. Precipitation	0.18	0.71	-0.01	.06*	-.20**	.15**	.14**
7. Sunshine	0.33	0.43	.06*	.49**	0.05	-.21**	-.49**	-.11**
8. Wind speed	1.37	1.26	0.04	.10**	-0.06	-.10**	0.06	-0.02	.11**
9. Late morinigns	0.20	0.40	0.01	0.04	0.05	-0.07*	0.02	-0.02	0.17**	0.07*
10. Early afternoon	0.24	0.42	-0.03	0.02	-0.02	-0.02	0.07*	0.02	0.14**	0.05
11. Late afternoon	0.25	0.43	0.02	-0.01	0.01	-0.03	-0.02	0.04	-0.01	-0.06
12. Early evening	0.25	0.44	-0.02	-0.07*	-0.02	0.02	-0.05	-0.01	-0.33**	-0.06*
13. Spring	0.26	0.42	0.20**	0.81**	-0.08**	-0.04	-0.22**	0.04	0.48**	0.06*
14. Autumn	0.30	0.46	-0.03	-0.19**	-0.03	-0.10**	0.25**	0.05	-0.31**	0.05
15. Sex	0.47	0.50	-.09**	-.10**	0.01	-.01	.03	-.06	.01	-.02
16. Age	21.86	2.65	0.01	.11**	-0.01	-0.04	-0.04	-0.04	.09**	-.09**

Notes. Sex coded as men = 0, and women = 1; Sunshine = Sunshine duration; Seasons dummy-coded relative to winter; Parts of day dummy-coded relative to early morning.

* p < .05

**p < .01.

Fig 2

Path model for role of seasons and weather conditions in affective valence.

Note. The figure presents only significant paths for the tested path model. For clarity, all non-significant paths from the model were omitted. Thicker lines represent stronger effects. Sex coded as 0 = men, 1 = women. *p < .05, **p < .01, ***p < .001.

Table 3

Path analysis details.

Outcome	Predictors	Std. Estimate	SE	p
Affect
	Temperature	-0.23	0.10	0.03
	Sex	-0.06	0.05	0.23
	Affect	-0.03	0.02	0.20
	Cloudiness	-0.07	0.04	0.09
	Sunshine	-0.07	0.05	0.13
	Wind speed	0.02	0.04	0.61
	Precipitation	0.00	0.04	0.93
	Air pressure	-0.02	0.03	0.34
	Humidity	-0.02	0.03	0.51
	Spring	0.42	0.21	0.04
	Autumn	0.07	0.05	0.17
	Late Morning	0.01	0.06	0.92
	Early Afternoon	-0.03	0.07	0.71
	Late Afternoon	0.00	0.08	0.96
	Early Evening	-0.06	0.08	0.44
Temperature
	Spring	0.85	0.12	0.00
	Autumn	0.12	0.16	0.47
	Late Morning	0.08	0.06	0.18
	Early Afternoon	0.08	0.05	0.10
	Late Afternoon	0.08	0.05	0.11
	Early Evening	0.03	0.06	0.60
Air pressure
	Spring	-0.11	0.08	0.16
	Autumn	-0.07	0.07	0.27
	Late Morning	0.03	0.02	0.13
	Early Afternoon	-0.03	0.02	0.26
	Late Afternoon	-0.01	0.02	0.70
	Early Evening	-0.03	0.03	0.35
Humidity
	Spring	-0.10	0.25	0.69
	Autumn	-0.13	0.37	0.73
	Late Morning	-0.33	0.11	0.00
	Early Afternoon	-0.31	0.09	0.00
	Late Afternoon	-0.33	0.09	0.00
	Early Evening	-0.30	0.11	0.01
Cloudiness
	Spring	-0.15	0.06	0.02
	Autumn	0.19	0.05	0.00
	Late Morning	0.04	0.04	0.28
	Early Afternoon	0.09	0.05	0.09
	Late Afternoon	0.02	0.03	0.43
	Early Evening	0.00	0.04	0.96
Precipitation
	Spring	0.07	0.07	0.33
	Autumn	0.07	0.09	0.42
	Late Morning	0.06	0.04	0.14
	Early Afternoon	0.09	0.03	0.01
	Late Afternoon	0.11	0.04	0.00
	Early Evening	0.07	0.02	0.00
Sunshine
	Spring	0.41	0.13	0.00
	Autumn	-0.18	0.05	0.00
	Late Morning	0.05	0.05	0.33
	Early Afternoon	0.01	0.09	0.88
	Late Afternoon	-0.10	0.08	0.22
	Early Evening	-0.35	0.08	0.00
Wind speed
	Spring	0.09	0.03	0.01
	Autumn	0.07	0.03	0.02
	Late Morning	0.04	0.04	0.27
	Early Afternoon	0.02	0.04	0.59
	Late Afternoon	-0.06	0.04	0.20
	Early Evening	-0.06	0.06	0.33

Note. Sex coded as men = 0, and women = 1; Sunshine = Sunshine duration.

Role of seasons, part of the day, and weather conditions in affective valence

We found a positive direct effect of spring on valence (Table 3) and a negative indirect effect of spring on valence via ambient temperature β = -.20, 95% CI [-.31, -.094]. These two opposing effects canceled each other out, producing a non-significant total effect of spring on valence, β = .21, 95% CI [-.12, .54]. This decomposition of the total effect suggests that participants would feel generally better in spring than in winter if not adverse effects of higher temperatures in spring. Yet, given their joint influence, the effect of spring on valence was non-significant. We found no difference in affective valence between parts of the day.

Of the weather conditions, only the ambient temperature predicted the participants’ affective valence (Table 3). Participants felt better when it was cooler outside. The unstandardized estimate was b = -.03, showing that a decrease of one degree Celsius predicted an increase in an individual’s affect by 0.03 points on the scale from one to ten, an equivalent of a 3.32% valence SD. To further support our findings, we run an exploratory analysis, in which we tested the model for each season separately. We found that people felt better when it was colder outside in spring β = -.27, 95% CI [-.40, -.14]. and in autumn β = -.12, 95% CI [-.24, -.01]. The relationship in winter was not significant β = -.005, 95% CI [-.09, .08].

Seasons and weather conditions

Relative to winter, the ambient temperature was higher in spring but not in autumn (Table 3). The wind speed was higher in spring and in autumn than in winter. The cloudiness was higher, and the sunshine duration was lower in autumn when compared to winter. In contrast, the cloudiness was lower, and the sunshine duration was higher in spring when compared to winter. We found no differences between the seasons in precipitation, air pressure, and humidity (Table 3).

Part of the day and weather conditions

The early mornings were more humid than late mornings, early afternoons, late afternoons, and early evenings (Table 3). The precipitation was higher in early afternoons, late afternoons, and early evenings when compared to early mornings. The sunshine duration was shorter in the early evenings than in the early mornings. We found no differences in temperature, air pressure, cloudiness, and wind speed between the parts of the day (Table 3).

Discussion

We examined whether individuals who start a laboratory experiment report different levels of affect depending on contextual factors such as season, part of a day, and weather conditions. We found that research participants felt better when it was colder outside. However, this effect had negligible practical meaning. Any differences in reported affect would vary within one standard deviation as long as the differences in the temperature between participants on different occasions were below 30 Celsius degrees. We found that participant’s baseline affect did not depend on any other conditions. Thus, we conclude that differences in season, weather, and time of day have little impact on baseline affect among participants for most laboratory research schedules. As our study had a reasonable sample size resulting in high statistical power, we believe that the null results are robust. Our work corresponds well with other large-scale integrative projects that indicated the non-significance of occasion-specific factors in their effect on research participants’ characteristics [39].

Our findings support other research indicating that high ambient temperature is associated with lower positive affect [23, 32, 34]. Some studies suggested the opposite [27, 31], yet they did not consider the seasonal variations. To address the fact that weather is often nested in seasons, we built a multilevel model that accounted for more variance. If we did not include seasons in our analysis, we found a positive correlation between temperature and affect, which might suggest that individuals feel better on warmer days or in warmer seasons. If we we examined the association between temperature and affect within each season, we found that people felt better when it was cooler outside. This finding suggests that it is important to control for seasons when examining the association between weather and affect. However, the effect should be interpreted as small. Thus, temperature differences of as much as 30 Celsius degrees would not be likely to cause deviations from the affect among research participants of more than one standard deviation of the mean valence.

We found that the relationship between seasonal variation and affect was complex. First, individuals felt more positive affect in spring than in winter. Yet, at the same time, springs were much warmer than winters, and participants felt worse on days that were warmer. Consequently, these two effects operated together in opposing directions canceling each other out. This effect is puzzling because simple correlations indicated that the ambient temperature and spring (vs. winter) were positively related to affect. We suggest that the outcomes are best interpreted as avoidance of thermal discomfort related to high temperatures in spring and autumn in our region and low temperatures in winter. Individuals might feel worse during spring and autumn heat, but they also might feel somewhat worse during the winter cold. Furthermore, our findings may suggest that other factors differentiate between the seasons that were not included in the analysis but might have influenced participants’ affect.

Unexpectedly, we did not find influences of daily cycles on affect, when controlled for the weather conditions. Previous studies indicated that the circadian rhythm of affect was consistent with the standard work-rest pattern [11, 13]. In our study, participants could schedule the lab visit before, in between, after the work, due to their own preferences, which may indicate a non-standard work-rest pattern. Future studies could replicate our result with a more homogenous participants pool to account for the work-rest cycles.

Limitations and futures directions

This study has several limitations. First, although we precisely match the weather conditions with the lab visit, there could be a difference between observed objective weather and the experienced weather, indicating the measurement error that could bias our findings [27]. For instance, individuals could differ in their exposure to current weather due to the chosen transportation, e.g., biking vs. arriving by car. Future studies could include time spent outside to control for exposure to the weather conditions.

Second, our data are restricted to observations from one country with a predominately continental climate. It is critical to repeat this analysis in countries with more extreme climates. For instance, individuals could experience the same temperature differently depending on what they are used to.

Third, we did not collect data in the summer, making it difficult to generalize our findings for the whole year. For humans, the most comfortable temperature is around 22°C [64]. Thus the relation between weather and affect may be more explicit when the temperatures go beyond the comfort level.

Fourth, we controlled for only two individual characteristics, namely age, and sex. No other potentially useful moderating variables were assessed. Future studies could include individual differences that would moderate the association between weather and affect [23, 35] or time of the day and affect [13].

Fifth, as in previous studies, we found relatively small effect sizes, which were detectable due to our analyses’ high power [23, 30, 35, 36]. However, as we pointed out, their statistical significance does not warrant predicting meaningful psychological differences.

Sixth, we accounted for several potential predictors in our model, with some of them intercorrelated. Thus, some interpretations of the parameters in our mediation model might be challenging or might be interpreted in different ways. For instance, it is not straightforward to conclude what is the meaning of season or time of day after controlling for the weather. We examined and ruled out the risk of multicollinearity, yet we cannot exclude the likelihood that some associations might have been spurious. This warrants further conceptual work and empirical studies that dissect several causal pathways initiated by one causal factor, i.e., different effects due to seasonal activities (e.g., duties, holidays, more time spent for outdoor leisure) or due to biological effects on the human body (e.g., thermal stress during spring or summer heats).

Finally, our findings have limited generalizability. Aiming to advance laboratory research practice, we focused on baseline affect measures among resting research participants in a well-controlled room environment. For instance, we aimed to keep 23 degrees Celsius temperature in the room, constant dim light, and external sound attenuation. Thus, these findings generalize to individuals under specific conditions that isolate the room environment from the outdoor environment. The results might be different for other scenarios. For instance, individuals might be more prone to weather or time of day if they were less isolated from the outdoor environment, e.g., if the room temperature followed outdoor temperature or the intensity of ambient light (light intensity and light temperature) was influenced by outdoor light, or if participants might observe wind or rain through the window. Our null findings might not hold for laboratories that do not meet some of these standardization criteria, e.g., have poor air conditioning. Moreover, we tested affect among individuals who might have been focused on the upcoming research tasks, and consequently, defocused from other daily factors such as time, weather, or other seasonal activities. Thus, our findings are not generalized to other scenarios where individuals have less restricted focus and might be more attentive to external factors. Our findings also generalize to the effects of weather and time cycles on affect. Our approach does not rule out the possibility that weather and time cycles might affect other processes that are of interest to affective scientists, e.g., cardiovascular circadian rhythms [14, 65].

Conclusions

This study provided novel evidence of how several external contextual factors influence baseline measurements of affect in laboratory studies. Despite an extensive scope of potential factors, we found that resting individuals, anticipating upcoming tasks, well-isolated from the outdoor, presented marginal affective propensity to weather and time-rhythm variation. This seems to suggest that as long as standardized room settings are kept constant, experimental research in affective science is robust to occasion-specific factors offering comparable levels of baseline affect among individuals who participate at a different time of year, time of day, or in different weather conditions.

Overview of studies characteristics.

(DOCX)

Click here for additional data file.

Data for the study.

(XLSX)

Click here for additional data file.

17 May 2021

PONE-D-20-38859

How seasons and weather conditions influence baseline affective valence in laboratory research participants ?

PLOS ONE

Dear Dr. Behnke,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Both Dr. Richard Lucas (who chose to reveal his identity as a reviewer) and I have carefully read your manuscript. On the whole, we both find the topic interesting, but see a number of significant concerns with the paper as written. As Dr. Lucas calls out, the sample, in and of itself, isn’t a problem, but there is insufficient information provided to allow us to evaluate if the different groups are matched on dimensions that may be relevant to this research. Put bluntly, could the results obtained be a result of the groups differing on some, currently, unreported dimension? The bigger problems, however, are analytical in nature. As Dr. Lucas points out, there are questions regarding the model selection, variable inclusion (multicollinearity), and causal inferences. I won’t repeat what Dr. Lucas called out, but I think his comments are completely on point and need to be addressed if this paper is to be published at PLOS ONE. Having said that, if the issues called out by Dr. Lucas can be rectified, I do believe this paper could add value to the scientific literature and I hope you take this opportunity to revise and improve your work.

Please submit your revised manuscript by Jul 01 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Jeff Galak, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1) Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2)  We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

3) Please ensure that you include a title page within your main document. You should list all authors and all affiliations as per our author instructions and clearly indicate the corresponding author.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper uses data from approximately 1,000 participants to assess the associations between mood and weather, after adjusting for season and time of day. The question itself is interesting, and the data have some desirable features. They are not longitudinal, and the sample size might not be quite large enough to detect the types of weather effects that have been found in past studies; but the sample is okay. Below I describe some concerns and suggestions.

The overall sample comes from participants in seven different laboratory studies. The authors ignore this group structure, suggesting that participants from different studies do not differ from one another. However, no evidence for this is provided. I think it would be useful to provide more information about the different samples, including how many participated in different seasons, and whether there are any common variables that should not vary across seasons that could be used to show that these samples are indeed very similar. It might also be helpful to account for this group structure in the analysis itself, perhaps with a multilevel model.

It is not entirely clear exactly what model was tested in Figure 1. The text description implies that every path was included, in which case model fit would not be relevant, because the model would be saturated. Yet the authors do emphasize model fit (which happens to be just okay), which suggests that some paths were, in fact, omitted. So more detail about this model is needed.

I am very concerned about the authors' interpretations of the parameters from their mediation model. First, as I understand recent guidance from methodologists who focus on causal modeling, the association between predictors and outcome after controlling for mediators is difficult to interpret. This is true both for the interpretation of season effects after controlling for weather, but also for the interpretation of time of day effects controlling for weather (though these look consistent with the zero-order correlations).

More importantly, I worry that some of these associations may be spurious, perhaps because of multicolinearity. For instance, temperature is correlated with "spring" .81, and both are included in the model. Notably, the zero-order correlation between temperature and affect is positive, but very weak, which probably aligns well with theory and intuition. However, after controlling for the very strongly correlated "spring" variable, temperature now correlates negatively (and moderately) with affect, which doesn't make much sense theoretically or intuitively. So I'm very concerned that this is an artifact, and I think the authors need to do much more to ensure that that is not the case. For instance, if they simply looked at the association between temperature and affect in each season (perhaps after controlling for time of day), what do the associations look like? I appreciate the authors' goals of making sure that these contextual factors are addressed, but this does introduce some challenging analytic issues that need careful consideration and discussion. I do not have confidence from these analyses that temperature is associated with lower mood.

Related, the authors emphasize that "temperature partly mediated the effects of season on emotions" (p. 11), but they do not discuss the direction of this indirect effect, which is consistently negative. Spring and autumn are warmer than the winter and people are happier in the spring and autumn than the winter; but the indirect effect is actually negative, meaning that this indirect effect doesn't really "explain" the total effect in the way people expect. Rather the indirect effect "explains" why the total effect is not much higher. This should strange finding should not be glossed over, as I believe it would lead readers to (appropriately) question this result.

Minor:

In the very first paragraph, the authors state that "positive affect is elicited by favorable activities such as going on a trip with friends and playing in the park with a child, lying in a hammock overlooking the beach," citing a paper by Cohen et al. (2018). However, this sentence implies that actually being in these situations has been shown to be associated with increased positive affect, which the cited study does not show. Instead, the cited study provided participants with a list of situations and asked them to rate their *hypothetical* reactions to these events. This should be made clear, as—as currently written—the sentence implies that research shows that these affective reactions actually occur.

Similarly, when describing evidence for seasonality, the authors omit important features of the evidence that they review. Notably, at least some of the evidence they cite in support of the idea that those who do not suffer from seasonal affective disorder still experience lower moods in the winter do not actually study changes or even differences in moods across seasons. For instance, the Hardin et al. paper cited as evidence uses a retrospective questionnaire that asks participants to report whether their mood changes in the winter. Because of retrospection problems, this is not strong evidence, and these methodological features should be noted explicitly. Overall, the authors should provide more detail about the studies they review, because as currently written, the lack of detail can sometimes be misleading.

I understand that a figure with all paths would include lots of information. At the same time, readers sometimes skip to figures and tables, glossing over details about those tables and figures and text (which is why many style guides encourage authors to ensure that tables and figures stand alone). Because of this, I think that the figure, which excludes nonsignificant paths is somewhat misleading, as it doesn't reflect what model was actually tested. Given these concerns, combined with the fact that even nonsignificant paths can be important for interpreting overall results, I'd encourage the authors to consider some other way of describing these results; a full table is probably necessary, and not just in the supplemental material.

The authors use stars to indicate significance in Table 2, but they do not appear to be correct. For instance, there are no significant correlations in rows 9 through 14, even though many correlations in these rows exceed other significant correlations in other rows (including the .81 correlation between "spring" and "temperature." These should be checked and corrected.

----

I always sign my reviews: Rich Lucas

I also believe that the role of the reviewer is to identify strengths and weaknesses of a paper, not to provide a recommendation about acceptance versus rejection. Because editorial management systems require a response to questions about recommendations, I almost always select "revise and resubmit." This selection should not be interpreted as a recommendation, but rather as "I choose not to provide a recommendation."

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Richard Lucas

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

1 Jul 2021

Reviewer: 1

This paper uses data from approximately 1,000 participants to assess the associations between mood and weather, after adjusting for season and time of day. The question itself is interesting, and the data have some desirable features. They are not longitudinal, and the sample size might not be quite large enough to detect the types of weather effects that have been found in past studies; but the sample is okay. Below I describe some concerns and suggestions.

#1. The overall sample comes from participants in seven different laboratory studies. The authors ignore this group structure, suggesting that participants from different studies do not differ from one another. However, no evidence for this is provided. I think it would be useful to provide more information about the different samples, including how many participated in different seasons,

and whether there are any common variables that should not vary across seasons that could be used to show that these samples are indeed very similar. It might also be helpful to account for this group structure in the analysis itself, perhaps with a multilevel model.

Reply: As suggested, we tested the two-level model, in which we nested individuals' data within experiments. We also added requested information about the number of participants tested each season and parts of the day within a specific study. Moreover, we also checked whether seasons differed in participants' age, sex, or BMI and found differences in age, sex, but not BMI. Post hoc tests showed that we tested more women in winter and autumn than in spring. Furthermore, we tested younger participants in spring than in winter and autumn. To account for these differences, we controlled for participants' age and sex in the analyses.

Changes in the manuscript:

See Methods section, line number 184-186; 191-198.

See Results section, line number 208-216.

#2. It is not entirely clear exactly what model was tested in Figure 1. The text description implies that every path was included, in which case model fit would not be relevant, because the model would be saturated. Yet the authors do emphasize model fit (which happens to be just okay), which suggests that some paths were, in fact, omitted. So more detail about this model is needed.

Reply: We presented our conceptual model in the revised manuscript (Figure 1). In our analysis, we regressed the affective valence on the mediators (weather conditions) and independent variables (seasons, part of the day). Age and sex were introduced as covariates for valence. Thus, some paths were omitted. In our hypotheses testing, we focus on the significance of specific direct paths and indirect paths.

Changes in the manuscript: See Methods section, line number 203, and Fig 1.

#3. I am very concerned about the authors' interpretations of the parameters from their mediation model. First, as I understand recent guidance from methodologists who focus on causal modeling, the association between predictors and outcome after controlling for mediators is difficult to interpret. This is true both for the interpretation of season effects after controlling for weather, but also for the interpretation of time of day effects controlling for weather (though these look consistent with the zero-order correlations) More importantly, I worry that some of these associations may be spurious, perhaps because of multicolinearity. For instance, temperature is correlated with "spring" .81, and both are included in the model. Notably, the zero-order correlation between temperature and affect is positive, but very weak, which probably aligns well with theory and intuition. However, after controlling for the very strongly correlated "spring" variable, temperature now correlates negatively (and moderately) with affect, which doesn't make much sense theoretically or intuitively. So I'm very concerned that this is an artifact, and I think the authors need to do much more to ensure that that is not the case. For instance, if they simply looked at the association between temperature and affect in each season (perhaps after controlling for time of day), what do the associations look like? I appreciate the authors' goals of making sure that these contextual factors are addressed, but this does introduce some challenging analytic issues that need careful consideration and discussion. I do not have confidence from these analyses that temperature is associated with lower mood.

Reply: We agree that interpretation of the model is challenging. We believe that this reflects the challenge to understand the phenomenon where multiple contextual factors (often intercorrelated) determine how people finally feel while approaching an experiment. We also believe that presenting the full model helps in observing and addressing this complexity and boils it down to essential components. For instance, we show that of the several weather components, temperature holds as predictive of affect while other factors are best fixed as neutral. As suggested, we addressed multicollinearity and found that the predictors were within the recommended range of Tolerance> .02 and VIF < 5. (Dodge, 2008; Everitt &Skrondal, 2010; James et al., 2013; Vittinghoff et al., 2011). We addressed this problem complex models interpretation in the limitations section.

We also agree that our paper might benefit from more emphasis on interpretation challenges, including more analytical work. Thus, we took additional efforts to elucidate our findings further. First, we run analyses in which we tested the model for each season separately. We found that people felt better when it was cooler outside in spring β = -.27, 95% CI [-.40, -.14]. and in autumn β = -.12, 95% CI [-.24, -.01]. The relationship in winter was not significant β = -.005, 95% CI [-.09, .08]. This might suggest that the outcomes are best interpreted as avoidance of thermal discomfort related to high temperatures that are frequent in spring and autumn in our region. Furthermore, the overall positive relationship between affect and temperature might be explained by differences between seasons rather than cumulative effects of differences within seasons. This shows the advantage of accounting for two factors (seasons and ambient temperature) as these two seem to have opposing effects, i.e., individuals feel worse in response to winters' low temperatures, but they also feel worse in spring (and autumn) once it is hot outside.

References:

Dodge, Y. (2008). The Concise Encyclopedia of Statistics. New York: Springer.

Everitt, B. S.; Skrondal, A. (2010), The Cambridge Dictionary of Statistics, Cambridge University Press.

James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An introduction to statistical learning (Vol. 112, p. 18). New York: Springer.

Vittinghoff, E., Glidden, D. V., Shiboski, S. C., & McCulloch, C. E. (2011). Regression methods in biostatistics: linear, logistic, survival, and repeated measures models. New York: Springer.

Changes in the manuscript:

See Methods section, line number 186-189.

See Results section, line number 216-217; 223-238.

See Discussion section, line number 326-335.

#4. Related, the authors emphasize that "temperature partly mediated the effects of season on emotions" (p. 11), but they do not discuss the direction of this indirect effect, which is consistently negative. Spring and autumn are warmer than the winter and people are happier in the spring and autumn than the winter; but the indirect effect is actually negative, meaning that this indirect effect doesn't really "explain" the total effect in the way people expect. Rather the indirect effect "explains" why the total effect is not much higher. This strange finding should not be glossed over, as I believe it would lead readers to (appropriately) question this result.

Reply: As explained in our response to the previous point, we think that this finding indicates that increasing the complexity of the model better describes the complexity of the phenomenon but at the same time complicates the model interpretation. We suggest that the outcomes are best interpreted as avoidance of thermal discomfort related to high temperatures that are frequent in spring and autumn in our region and low temperatures that occur in winter. Individuals might feel worse during spring and autumn heat, but they also might feel somewhat worse during the winter cold. We clarified this issue in the discussion. We presented the relationship between seasonal variation and affect as a complex phenomenon requirng investigations that account for seasons and temperatures.

Changes in the manuscript: See Discussion section, line number 274-297.

Minor issues

#5. In the very first paragraph, the authors state that "positive affect is elicited by favorable activities such as going on a trip with friends and playing in the park with a child, lying in a hammock overlooking the beach," citing a paper by Cohen et al. (2018). However, this sentence implies that actually being in these situations has been shown to be associated with increased positive affect, which the cited study does not show. Instead, the cited study provided participants with a list of situations and asked them to rate their *hypothetical* reactions to these events. This should be made clear, as—as currently written—the sentence implies that research shows that these affective reactions actually occur.

Reply: We clarified this issue, and we made it clear that people evaluate situations as being related to positive or negative affect.

Changes in the manuscript: See Introduction section, line number 39-43.

#6. Similarly, when describing evidence for seasonality, the authors omit important features of the evidence that they review. Notably, at least some of the evidence they cite in support of the idea that those who do not suffer from seasonal affective disorder still experience lower moods in the winter do not actually study changes or even differences in moods across seasons. For instance, the Hardin et al. paper cited as evidence uses a retrospective questionnaire that asks participants to report whether their mood changes in the winter.

Because of retrospection problems, this is not strong evidence, and these methodological features should be noted explicitly. Overall, the authors should provide more detail about the studies they review, because as currently written, the lack of detail can sometimes be misleading.

Reply: We provided additional details about the methods used in the studies mentioned in the introduction.

Changes in the manuscript: See Introduction section, line number 54-59; 86-94.

#7. I understand that a figure with all paths would include lots of information. At the same time, readers sometimes skip to figures and tables, glossing over details about those tables and figures and text (which is why many style guides encourage authors to ensure that tables and figures stand alone). Because of this, I think that the figure, which excludes nonsignificant paths is somewhat misleading, as it doesn't reflect what model was actually tested. Given these concerns, combined with the fact that even nonsignificant paths can be important for interpreting overall results, I'd encourage the authors to consider some other way of describing these results; a full table is probably necessary, and not just in the supplemental material.

Reply: We moved the table with the detailed results of the tested model to the manuscript. We considered the figure with all paths. As presented below, the figure might be unclear to some readers. Thus we only included an additional description of Fig. 2 in the manuscript.

We are ready to add a figure with the full model if still requested.

Changes in the manuscript: See Table 3, line number 257, and Fig 2, line number 253.

#8. The authors use stars to indicate significance in Table 2, but they do not appear to be correct. For instance, there are no significant correlations in rows 9 through 14, even though many correlations in these rows exceed other significant correlations in other rows (including the .81 correlation between "spring" and "temperature." These should be checked and corrected.

Reply: We corrected the Table. Thank you!

Changes in the manuscript: See Table 2, line number 251.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

9 Aug 2021

How seasons, weather, and part of day influence baseline affective valence in laboratory research participants ?

PONE-D-20-38859R1

Dear Dr. Behnke,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Of note, the one reviewer asked to review this revision had technical difficulties with the Plos ONE system, but emailed me his decision privately. In short, he believed that the revision successfully responded to his concerns.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jeff Galak, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

11 Aug 2021

PONE-D-20-38859R1

How seasons, weather, and part of day influence baseline affective valence in laboratory research participants?

Dear Dr. Behnke:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jeff Galak

Academic Editor

PLOS ONE