Interactions among conspecifics play a crucial role in the social structure of many species, contributing to the maintenance of their health, reproductive output, and survival (Neumann, 2009; Buwalda et al., 2013). Furthermore, the positive effects of social interaction attributed to social buffering serve to promote and/or maintain social cohesion (Hennessy et al., 2009). In this context, maternal presence constitutes the primary partner with which an animal forms a strong affiliative relationship, contributing to the formation and maintenance of filial attachment or pair-bonds throughout its lifespan (Shionoya et al., 2007; Hennessy et al., 2009). Social interactions are not only essential for many life processes, such as communal rearing of young and cooperation within a group (Neumann, 2009; Blumstein et al., 2010; Ebensperger et al., 2012), but they also play a protective role against stress-related events such as depredation, resource acquisition or defense, and social thermoregulation (Ebensperger, 2001; Gilbert et al., 2009).

Recently, it has been emphasized that stress is not only a “fight-or-flight” response to environmental threats, but also an ongoing process by which the body and brain adapt to daily experiences, including social environment, physical activity, and hazardous environments (McEwen and Akil, 2020). The physiological process of achieving stability or homeostasis through the production of mediators, such as hormones of the hypothalamus-pituitary–adrenal (HPA) axis, in response to challenges of daily life is known in stress biology as “allostasis” (McEwen and Wingfield, 2003). As life stressors persist, the adaptation process can be costly for organisms, triggering neurological and psychiatric disorders, including anxiety, emotional disturbances, and aggressive states, as well as atrophy of brain structures, cardiovascular disease, and other systemic disorders (McEwen and Wingfield, 2003; McEwen and Akil, 2020). “Allostatic load” is the price the body pays for adaptation to adverse situations (McEwen and Wingfield, 2003).

Disruptions to the social environment (e.g., social isolation or social instability) are associated with a range of physiological, neuroendocrine, and behavioral dysfunctions in both humans and nonhuman species that can lead to increased mortality and/or reduced lifespan (Beery and Kaufer, 2015; Mumtaz et al., 2018). These effects vary depending on species, sex, age, life stage, duration of exposure, and individual perceptions of the stressors (Lapiz et al., 2003; Neumann, 2009; Mumtaz et al., 2018; Rivera et al., 2020). Interestingly, among highly social mammals such as humans and primates, positive social relationships lead to substantial mental and physical health benefits in stress situations, reducing the risk of mortality (Taylor, 2006; Beery and Kaufer, 2015). Animal models, such as rodents, demonstrated that a lack of social interaction has behavioral effects, including impaired recognition memory, reversal learning, and an inability to make decisions (Zhao et al., 2009; Mumtaz et al., 2018), as well as physiological effects, such as an increase in chronic sympathetic tone, oxidative stress, and activation of the HPA axis, and a reduction in inflammatory control (McEwen and Wingfield, 2003; Cacioppo et al., 2011). Neurobiological consequences were also observed, including changes in the neuronal structure and brain activity (Lapiz et al., 2003; Westenbroek et al., 2004; Schoenfeld and Gould, 2012). Given these pervasive effects, more sophisticated assessment techniques using multidimensional measures are needed to test the impact of social isolation stress on animal performance.

Biological systems exhibit great complexity. This complexity results from the interactions of multiple structural units and regulatory feedback loops that enable living organisms to effectively cope with the diverse stresses and challenges of daily life (Lipsitz, 2004). The stated variables of healthy organisms exhibit complex fluctuations characterized by long-range power-law correlations that arise through multifractal cascades operating over a wide range of timescales (Goldberger et al., 2002). The breakdown of this multi-scale complexity has been recognized as a hallmark of pathologies and aging (Hausdorff et al., 2001; Costa et al., 2008). In this context, multifractal analysis of biosignals has become a useful tool in biology and biomedicine. It allows us to understand the abnormalities that lead to many pathologies and to provide simple numerical descriptions of complex biological systems and their diseases, including COVID-19 (Li, 2021), Parkinson’s disease (Marmelat and Meidinger, 2019), obstructive sleep apnea (Zhang et al., 2009), and cardiovascular disease (Peskin and McQueen, 1994). Thus, this approach is emerging as a powerful diagnostic tool for determining physiological health, and it holds great promise for improving our understanding of the physiological consequences of stress.

Animal studies on the effects of social isolation stress are a valuable counterpart to human studies. Due to the potential risk of damaging effects, randomized and experimental manipulations of social isolation in humans are limited in intensity and duration. In addition, the last few years have seen increased interest in the effects of long-term social isolation, mostly associated with the quarantines imposed during the COVID-19 pandemic. Thus, longitudinal studies in animals are key to elucidating anxiety-related effects on individuals’ behavior and physiology due to isolation-derived stress. Here, we selected the degu (Octodon degus) as a suitable model for studying long-term social-affective biological behavior since this rodent shares many physiological and behavioral characteristics with humans, including the ability to form social bonds and extended social group families (Silk, 2007; Colonnello et al., 2011; Rivera et al., 2016). The main advantage of using the degu as an animal model is the degus’ relatively long lifespan under captive conditions [7–8 years, Lee (2004)], which allows researchers to assess the effects of natural aging and age-related diseases. Aging degu brains naturally develop amyloid plaques and tau phosphorylation, two hallmarks of Alzheimer’s disease (Inestrosa et al., 2005; Tarragon et al., 2013; Hurley et al., 2018). Furthermore, during the aging process, degus exhibit cognitive decline and memory impairment similar to that seen in humans with dementia (Ardiles et al., 2012; Rivera et al., 2021b). Recently, we reported that long-term social isolation not only affects stress physiology by impairing the negative feedback loop of the HPA axis, resulting in prolonged elevated blood cortisol levels following a stressor, but also has effects on behavioral, functional, and molecular traits related to spatial and social memory-related tasks in degus (Rivera et al., 2020, 2021a). This study aimed to investigate cognitive and physiological performance and multifractal properties of metabolic time series r(VO₂) in degus exposed to prolonged social isolation stress. To the best of our knowledge, this is the first study to examine the fractal properties of time series of r(VO₂) and cognitive performance in a long-lived animal model exposed to prolonged social isolation stress, providing novel insights into the impact of such a stressor on physiological and cognitive functions in a social mammal.

Laboratory pregnant female degus obtained from our colony at the Faculty of Biological Sciences, Pontificia Universidad Catόlica de Chile were kept in pairs and housed in clear acrylic terrariums (length × height × depth: 50 × 35 × 23 cm) with hardwood chip bedding. Each cage contained one nestbox made of clear acrylic (22 × 12 × 15 cm). We checked for litters daily, and the day of birth was defined as postnatal day (PND) 0. To avoid litter differential parental effects within a litter, the whole litter was randomly assigned to one of the following rearing conditions: (i) unstressed controls (CTRL, N = 12) in which the male litters were left undisturbed with their family. The siblings remained together until PND 90, and thereafter they were raised as male-matched groups of three siblings from PND 91 until the end of the experiment; (ii) Chronically isolated individuals (CI, N = 12) in which the male pups were removed from their mothers and kept individually in small opaque cages for 1 h daily (between 09:00 a.m. and 12:00 p.m.) from PND 1 to PND 35 (day of weaning). During separation, the pups had acoustic and olfactory but no visual and social contact with their siblings or mother. From PND 36 until the end of the experiment, they were individually housed in standard rodent cages with no physical contact with conspecifics (though olfactory, acoustic, and some visual contact could not be fully prevented).

All animals were kept in a ventilated room and exposed to a 12 L:12D setting and ambient temperature (yearly minimum = 13.4 ± 0.2°C; yearly maximum = 24.9 ± 0.2°C) and were fed a standard rabbit commercial pellet diet (Champion, Santiago, Chile) and ad libitum water. All animal protocols followed the guidelines of the National Institutes of Health (NIH, Baltimore, MD, USA). All efforts were made to minimize animal suffering and to reduce the number of animals used.

The assumptions of normality and homoscedasticity were confirmed using Shapiro–Wilk and Levene’s tests, respectively. For univariate analyses, we used t-tests to analyze the effect of stress treatment groups on male degus across each variable of the behavioral test. All behavioral data are presented as the mean ± SEM. In social interaction and NLR-NOR tests, the IR were analyzed. Also, we assess the effects of social interaction on BMR using a linear model with following variables: BMR * Mb. To assess the multifractal properties of the oxygen consumption time series for CTRL and CI groups, we considered six parameters: the Hurt exponent (H) and the variability of h(q) (∆h(q)), which were used to characterize the dependence profile of h(q) on q. For the singularity spectrum, we used the width of spectrum (∆a), amplitude of spectrum (∆f), the asymmetry parameter (B), and dominant exponent (a₀). To estimate the parameters of multifractal properties, we used the package fractal. We used a t-test to compare these parameters between the treatments.

To analyze the effects of social stress on the physiology and behavior of the degus in an integrated manner, we carried out a Principal coordinate analysis (PCoA) and a multivariate non-parametric analysis of dissimilarities (PERMANOVA) using a Manhattan dissimilarity matrix. Pairwise distances were calculated for all variables (behavior-cognitive related traits in Table 1, and physiological traits BMR, Mb, Table 2), and the PCoA coordenates used to test differences between the CTRL and CI groups. The PERMANOVA was implemented in the Adonis2 function within the vegan R package.¹

Animal sociality is adaptive, which means that being social increases survival and improves reproductive performance (Kikusui et al., 2006; Snyder-Mackler et al., 2020). Positive social relationships serve to buffer against allostatic load imposed by ecological adverse experiences, such as predation risk, resource competition, infanticide, and social isolation (Rukstalis and French, 2005; Kikusui et al., 2006; Brent et al., 2014; Siracusa et al., 2022). Here, we investigated the effects of prolonged chronic social isolation stress (from post-natal and post-weaning into adulthood) on behavioral, cognitive, and physiological performance in social degus. Overall, and in agreement with our predictions, prolonged chronic social isolation led to an increase in anxiety-like behaviors, a decrease in social and working memory, and a decrease in the complexity of the time series of VO₂ consumption. Interestingly, our results suggest that using MF-DFA on VO₂ time series could be a useful diagnostic tool for the aging process and neurodegenerative disease (see Zorick et al., 2020 for MF-DFA in EEG signals).

It has always been suggested that more than one test should be used to measure anxiety in order to accurately characterize anxiety-related behavior and to strengthen interpretations (Bailey and Crawley, 2009). These are the main reasons we also used the novel object open field and the light–dark box tests in this study. The classical open field test, which assesses exploration of a novel environment and general locomotor activity, has been employed as an initial screening test for anxiety-related behavior in rodents. No significant differences between treatments were observed in the open field test; however, in the novel object open field test, the CI group behaved differently than the CTRL group. CI degus were less keen to explore the novelty and spent less time in the center of the arena. Similarly, during the light–dark box test, CI males displayed a lower number of transitions (i.e., number of times the animals crossed from the light box to the dark box, and vice versa), and it took them almost three times as long to enter the dark box, suggesting that stressed animals were more insecure. Moreover, CI degus had a longer latency to enter the dark chamber for the first time, suggesting less willing behavior to novelty. Using both tests, we demonstrated high levels of anxiety and hyperreactivity behavior in socially isolated animals. Our results are consistent with previous reports in rats, where anxious, hyperreactive, or hyperaroused rats were less likely to explore novelty, suggesting a deficit in motivation (Rygula et al., 2005; Thorsell et al., 2006). Similar levels of anxiety-like behavior and locomotor hyperactivity have been observed in socially isolated mice and rats (Levine et al., 2007; Medendorp et al., 2018). In addition, anxiety-like behavior using the light–dark box test, specifically the number of light–dark transitions, has been used as an index of activity exploration in mice (Bourin and Hascoët, 2003).

To evaluate how prolonged social isolation stress affects sociability and social novelty preference, we used the three-chamber paradigm. Social recognition memory, which is also necessary for the stability of social groups, reflects the ability to identify and remember conspecifics (Kogan et al., 2000). Using the Recognition Index (RI), we found in session 1 that CI degus had lower values of RI compared to CTRL. These data suggest that degus’ sociability was negatively affected by stress treatment. Both groups showed a clear preference for the compartment containing the unfamiliar conspecific. CTRL and CI animals spent about 35 and 25%, respectively, of their time interacting with the unfamiliar degus, and 5 and 8%, respectively, of their time with the inanimate object. Previous studies demonstrated that chronic stress reduces social motivation and social interaction, particularly in highly anxious animals (Lukkes et al., 2009; Sandi and Haller, 2015; Rivera et al., 2021a). However, the prolonged chronic isolation stress did not diminish social abilities (Sandi and Haller, 2015; Rivera et al., 2021a). As with in humans, stressful situations can promote affiliative behavior or group cohesion to provide a buffer against stress and improve health and well-being (Kikusui et al., 2006; Taylor, 2006; Oppenheimer et al., 2016; Hsiao et al., 2018). In this context, previous studies in stressed rodents have shown a willingness to interact with conspecifics to potentially ameliorate negative emotions for positive neurochemical rewards (Kikusui et al., 2006; Gilles and Polston, 2017). During Session II, the social recognition memory was assessed by the spontaneous approach behavior observed in an individual who was re-exposed to a familiar conspecific. Our results indicated that CI degus had lower RI values compared to CTRL. Moreover, we found that CI males spent almost twice as much time interacting with Partner I compared to CTRL, indicating that prolonged social isolation affects social memory. Our results are in line with the reported long-lasting effects of early life stress in memory performance in other animal models (Ibi et al., 2008; Mumtaz et al., 2018) and even degus (Rivera et al., 2021a).

We used the novel object recognition test to evaluate cognition, particularly working memory, attention, and preference for novelty (Popović et al., 2009). We observed that working episodic memory (measured by RI in the NLR/NOR task) was impaired in male degus under CI treatment. During the NLR session, both groups showed similar time interacting with the moved object; however, CI animals spent almost three times as much time interacting with the familiar object as CTRL degus. Furthermore, during the NOR session, the CI animals spent less than half as much time interacting with the novel object compared to the unstressed animals, and almost twice as much time interacting with the familiar object. These results suggest that CI degus exhibit an impairment of the rodents’ innate preference for novel experiences (such as exploring new objects). Previous works using mice and rats have similarly described deficits of spatial memory resulting from social isolation (Voikar et al., 2005; McLean et al., 2010; Famitafreshi and Karimian, 2018). Moreover, we previously found that chronic isolation affected the cognitive performance in the NLR/NOR task using the same isolation protocol described in this paper (Rivera et al., 2020).

We also investigated the potential effects of social isolation stress on degus’ energy expenditure (e.g., BMR). We predicted that more anxious and hyperactive individuals would have a higher energy expenditure than non-stressed individuals (Mueller and Diamond, 2001). While it is intuitive to expect that organisms that are physically active, bold, and exploratory will acquire and expend energy at higher rates than organisms with opposite behaviors, previous evidence is contradictory (Mueller and Diamond, 2001; Careau et al., 2008). High levels of stress, anxiety, and hyperactivity can put a strain on the sympathetic nervous system, increase cardiac output and respiration, and cause individuals to maintain higher activity levels or muscle tone than calm, relaxed individuals (Gilmour et al., 2005). However, our results did not support the stress and energy expenditure relationship. Social isolation stress had no effect on BMR, which could be attributed to the fact that BMR represents the minimum maintenance cost of organisms and is measured in post-absorptive, non-reproductive, resting. and thermoneutral zone organisms (McNab, 2002). Its estimation took into account a short segment of the metabolic record with the lowest and most stable values (5–10 min after steady state, see Methods). Thus, by focusing only on the steady-state period, BMR estimation excludes fluctuations in metabolic rates that could potentially reflect behavioral syndromes (i.e., hyperactivity, anxiety-like behavior). Body mass also did not vary between treatments, although it has been reported to be a variable strongly influenced by stress levels (Patterson and Abizaid, 2013). While chronic social isolation did not affect BMR, it did affect the multifractality of the VO₂ time series. Importantly, MF-DFA allowed us to detect anomalies in the regularity of the time series. The negative effects of social isolation stress on cognitive performance were accompanied by the loss of complexity VO₂ time series with a dominant pattern of long-range correlation, reflected in a lower ∆h and ∆α in the multifractal spectrum. The decrease in complexity observed in oxygen consumption rate, a reliable indicator of metabolic status, could potentially indicate a decrease in the regulation of homeostatic processes (Modell et al., 2015). Previous studies have reported that a reduction in physiological complexity predisposes organisms to frailty syndrome, a manifestation of physiological decline due to decoupling of integrated multimodal networks and disabling control loops, resulting in functional decline (Lipsitz, 2004; Schubert et al., 2009). In our study, the degradation or loss of structural complexity in stressed animals may be associated with an increase in inflammatory markers, making them more susceptible to premature immune aging. Indeed, exposure to adversity associated with traumatic stress events (e.g., war, abuse) in humans appears to accelerate markers of biological aging (Jergović et al., 2014; Lohr et al., 2015) and even makes them more susceptible to acceleration, progression, and worsening of symptoms of neurodegenerative disorders such as Alzheimer’s, Parkinson’s disease, and Huntington’s disease.

Overall, our results reveal that chronic social stress significantly affects anxiety-like behavior, social and working memory, and the fractality of the VO₂ time series. Likewise, the simultaneous analysis of all variables through PCoA showed a clear segregation of variables into two well-defined clusters, corresponding to healthy and stressed degus. To the best of our knowledge, this is the first study integrating cognitive and behavioral performance with the multifractal dynamics of physiological signals in response to chronic prolonged social stress.

Our results highlight: (1) the importance of incorporating methods, such as MF-DFA, that capture the complex nature of biological systems; and (2) the need to deepen our knowledge of organisms’ response to stress through studies that integrate different levels of organization (from physiology to cognition and behavior). Because stress responses vary among individuals (i.e., sex-dependent manner) and may be determined by social experiences during early development and adulthood (Lukkes et al., 2009; Rivera et al., 2021a), future studies focusing on the cross-generational effects of stress and including other complexity analyses of biological signals will allow us to better predict how social environments affect individuals, populations, and species. This data can then be useful for disease detection and diagnosis (Dutta et al., 2013; Zorick et al., 2020).

The original contributions presented in the study are publicly available in Zenodo https://doi.org/10.5281/zenodo.8397447.

The animal study was approved by Bioethical and Biosafety Committee of the Faculty of Biological Sciences of the Pontificia Universidad Católica de Chile (CBB-121-2013). The study was conducted in accordance with the local legislation and institutional requirements.

DR contributed to behavioral formal analysis, writing, reviewing, editing, original draft preparation, funding acquisition, and visualization. GC contributed to physiological formal analysis, writing, reviewing, editing, original draft preparation, and visualization. JB contributed to multifractality formal analysis, writing, and visualization. FB contributed to conceptualization, supervision, and funding acquisition. All authors contributed to the production of the article.