P. hirtus adults were collected in March 2013 from White Cave entrance following guidelines defined in National Park Service permit MACA-2015-SCI-0019. Animals were cultured in a light-insulated cave laboratory room. Animals were housed in 60mm polystyrene Petri dishes supplied with a 50 millimeter deep bottom layer of cave soil. The Petri dishes were sealed off at the edges with parafilm. Culture dishes were placed in a Styrofoam box with a layer of moist paper towel at the bottom. The Styrofoam boxes were housed in an incubator at a temperature ranging between 10–12° degrees Celsius. Cultures were fed every two weeks with Fleischmann’s Yeast pellets. All animal handling was conducted under low-intensity red light, given the lack of light stress protective eye pigmentation in the eyelets of P. hirtus (Friedrich et al. 2011). Locomotor activity tests were conducted with offspring animals in 2017.

Circadian activity was monitored using the Trikinetics Activity Monitor (TAM) (Trikinetics Inc., Waltman, MA, USA) placed in a Thermo Scientific Precision Low Temperature Incubator set at 11° degrees Celsius. Moist paper towel was secured to the bottom of the box to maintain humidity. Holes were poked in the plastic caps of the capillaries with a thumbtack for air circulation. Activity data were binned in 30-min intervals. For trials in constant darkness (DD), single beetles were placed into individual monitor capillaries and monitored for up to 14 days without the provision of food. For 12 hour light/12 hour dark regimen trials (LD), beetles were given a 5-hour acclimatization period in darkness, followed by the onset of 12h:12h light/dark cycles. RL5-W4575 White 75 Degree 4500 mcd LED lights (Super Bright LEDs Inc.) were used as light source powered at 2.5V and fixed 3 inches above the TAM, resulting in exposure of test animals to an approximate light intensity of 5×10¹⁶ photons/cm2/second during the 12 hour light phases.

Google spreadsheets and Actogram (Schmid et al. 2011) were used for the preparation of periodograms and actograms. The online implementation BoxPlotR (Spitzer et al. 2014) was used to generate summary box plots of activity intensities, activity periods, and rest periods. Nonparametric analyses of variance were conducted using the Kruskal Wallis Test Calculator hosted by Statistics Kingdom (2017): https://www.statskingdom.com/kruskal-wallis-calculator.html.

In addition to serving as the main tool in Drosophila melanogaster biological clock studies, the TAM setup has found increasing application for studying a broader range of insects (Bahrndorff et al. 2012; Giannoni-Guzmán et al. 2014; Pavan et al. 2016; Wang et al. 2021). Given the similar body sizes of P. hirtus and D. melanogaster, we explored the suitability of the original Drosophila-optimized TAM setup to monitor locomotor activity of adult P. hirtus, starting experiments in the light condition of deep cave zones, i.e. DD. Of a total of 28 individuals tested under these conditions in three separate trials, most perished within 2–5 days, after displaying high levels of locomotor activity (Suppl. material 1). Two individuals, however, completed an observation time span of 14 days alive (Fig. 1a, b and Suppl. material 2). Actogram inspection revealed a decline of locomotor activity during the first three days of the trials. On day 1, individual locomotor activities averaged 29.4 and 18.2 crossings per hour, compared to an average of 14.5 and 11.2 crossings per hour on day 2, and followed by 5.2 and 3.3 crossings per hour on day 3. As the latter numbers were close to the averages of 4.3 and 8.3 crossings per hour observed on day 4, we concluded that the two animals had completed an initial phase of adjustment to the TAM capillary environment by the third day of monitoring. Activity patterns were therefore analyzed starting from day 4 of the observation period (Fig. 1a, b).

Figure 1. 

Initial 7-day activity profiles of P. hirtus adults under DD and LD. Graphs represent the first seven days of locomotor activity monitoring. Light-off phases indicated by grey overlays a, b adjusting animals in DD. Top right arrows indicate start of analyzed time span after the 3-day time window arbitrarily defined as adjustment phase c, d examples of adjusting animals in LD.

For the two adjusted animals monitored at DD, individual average locomotor activity levels amounted to 2.9 and 2.8 crossings per hour over 11.5 days of observation time following the adjustment phase. Further inspection of the corresponding temporal activity plots revealed that these low overall locomotor activities were the result of long resting phases interspersed by short activity bouts (Fig. 1a, b). The latter measured on average 1.6 (+/-0.8) and 1.3 (+/-0.7) hours for each animal in contrast to average resting phase durations of 9.8 (+/-6.5) and 13.8 (+/-10.4) hours. Peak resting phases exceeded 30 hours (Suppl. material 2).

Actogram analyses suggested a random distribution of activity bouts between resting periods (Suppl. material 3). Consistent with this, tests for periodicity in activity distribution failed to detect significant rhythmicity.

To test whether locomotor activity was affected by light in P. hirtus, we monitored animals exposed to 12:12 LD. Overall, the adjustment rate to the TAM environment was slightly higher compared to the tests conducted in DD, with six out of 28 animals tolerating the test tube environment for long-term, producing data for more than 20 days (Suppl. material 4). The activity profiles of adjusted LD animals differed conspicuously from that of the DD animals. Locomotor activity was concentrated in the artificial night (D) phases with an average of 15.6 tube crossings per hour (+/-3.3), which compared to only 1.3 (+/-0.7) in the L phases (Fig. 1c, d) Suppl. material 4). 80% of the L-phases were characterized by non-detectable locomotor activity (Suppl. material 4).

Given the activity pattern differences between DD- and LD-monitored animals, we asked whether the long-term LD-adjusted animals had expended more locomotor activity than the two animals monitored in constant darkness. Comparing the daily numbers of tube crossings revealed that the six LD-adjusted animals had performed an average of 191.3 (+/-36.3) tube crossings per day compared to 66.8 (+/-2.0) crossings by the two DD-adjusted animals (Suppl. materials 2, 4). These numbers revealed a close to threefold higher energy expenditure of the LD animals compared to the two DD-monitored animals.

The difference in daily activity amounts between the LD- and DD-adjusted animals could have been due to differences in activity intensities, activity phase durations, rest phase durations, or a combination of the three variables. To evaluate their relative impacts, we compared their ranges and averages between the LD- and DD-adjusted animals (Fig. 2). Given the preponderance of locomotor activity displayed by the LD-adjusted animals in the presumptive night phases, we focused these comparisons on the D-phase activity profiles of the LD-adjusted animals.

Figure 2. 

Analyses of rest-activity in long-term DD- and LD-adjusted animals a–c box plots, visualizing the relative numbers of observations as widths of boxes median values as back bars inside boxes. Altman whiskers extending from 5^th to 95^th percentile. Each comparison shows the results for the two long-term DD-adjusted animals (DD1 and DD2) followed by the results for the six long-term DD-adjusted animals (LD1, LD4, LD6, LD7, LD8, and LD2-4 in the D-phase a comparison of activity intensities measured as the number of tube crossings per discrete activity bout. Time intervals with less than 2 tube crossings were excluded. Numbers of bouts per animal is given in parentheses. Kruskal-Wallis ANOVA detected no statistically significant differences after adjustment for multiple comparisons b comparison of activity bout durations with numbers of bouts per animal given in parentheses c comparison of resting period durations with numbers of resting periods per animal given in parentheses.

The mean activity intensities of the two DD-adjusted animals were 11 and 16 tube crossings per half hour (Fig. 2a). These values fell within the range of mean D-phase activity intensities in the LD-adjusted animals, which spanned from 11 to 20 tube crossings per half hour (Fig. 2a and Suppl. material 5). None of the activity intensities were detected as significantly different from other tested animals.

The mean activity bout durations of the DD-adjusted animals were 94 (+/-50) min and 80 (+/-42) min (Fig. 2b). These numbers were closely below the lowest mean of 103 (+/-42) min in the LD-adjusted animals and close to twofold lower than the maximum mean of 164 (+/-110) min in the LD-adjusted animals (Fig. 2b and Suppl. material 5). There was also a conspicuous difference in the spread of maximum activity phase durations in the LD-adjusted animals, reaching up to 690 min in contrast to the maximum activity durations of 180 min and 210 min of the two DD-phase-adjusted animals (Fig. 2b, Suppl. material 5). These findings did indicate a possible contribution of activity bout duration to the net activity differences between LD- and DD-adjusted animals. However, only one DD-adjusted vs LD-adjusted animal comparison was detected as marginally significant (Fig. 2b).

The average resting period durations differed most prominently between the DD- and LD-adjusted animals. While the former were characterized by mean resting phase durations of 586 min and 828 min, mean resting phase lengths among the LD-adjusted animals ranged between 107 min to 231 min (Fig. 2c and Suppl. material 5). The maximum resting phase durations of 1,440 min and 2,070 min of the two DD-adjusted animals compared to 720 in the LD-adjusted animals (Fig. 2c, Suppl. material 5). Both DD-adjusted animals differed from four of the six LD-adjusted animals with high statistical significance (Fig. 2b). Taken together, these findings identified the variation in resting phase durations as the key variable underlying the net activity differences between LD- and DD-adjusted animals.

To probe whether the nocturnal locomotor activity rhythm of LD-cultured animals was entrainable in P. hirtus, we transitioned the LD-initiated animals into DD after 17 days. Without exception, all six animals discontinued their circadian activity rhythms, displaying stochastically occurring activity bouts (Fig. 3). Comparing the periodograms covering the DD period of seven days did not produce evidence of shared rhythms. The lack of evidence of free-running locomotor rhythms led us to conclude that the circadian locomotor clock has completely regressed in P. hirtus.

Figure 3. 

Lack of free-running circadian rhythm in LD-entrained animals. Actograms for LD animals 4, 6, 7, and 8 generated with Actogram (Schmid et al. 2011) at default settings. Each row represents two days, shifted by one day forward in relation to the previous row. Grey overlays indicate light-off periods.

While very limited in trial number and likely impacted by artificial stressors, our observations gained from monitoring two animals in DD and six animals in LD yield new compelling insights into the role of light in the biology of P. hirtus and provide preliminary evidence of possibly adaptive deep cave- vs twilight zone rest-activity (RA) modes in this microphthalmic cave beetle species.

The first strongly supported finding of our study is that locomotor activity is affected by exposure to light in P. hirtus, given the highly nocturnal activity patterns of the six monitor-adjusted LD animals. This is further supported by the contrasting activity patterns of the two DD-adjusted animals. It is also consistent with the previously reported light avoidance of P. hirtus as well as the transcriptomic and structural evidence of a reduced but functional visual system (Friedrich et al. 2011). Given the likely lack of extraretinal light perception in P. hirtus based on the failure to detect homologs of extraretinal opsins in the available transcriptome data (Friedrich et al. 2011), it is most likely that the locomotor activity response to light is mediated through the P. hirtus eyelets.

As the second strongly supported finding of our study, our results leave little doubt that P. hirtus lacks the capacity to maintain a free-running circadian locomotor rhythm instructed by light as the zeitgeber. The previously reported expression of the biological clock gene network in the adult head may thus represent the activity of central pacemaker neurons responding to different zeitgeber sources. Alternatively, core clock gene expression may be associated with different outputs. In Drosophila, for instance, the expression of Clk and cyc in specific pacemaker neurons controls non-circadian rest-activity (RA) patterns and their maintenance throughout life history (Keene et al. 2010; Vaccaro et al. 2017).

Transcriptome-based studies in samples of independently evolved cave populations of the Mexican cavefish Astyanax mexicanus revealed the parallel regression of both core clock gene regulation and target gene regulation (Mack et al. 2020), despite the conservation and non-retinal expression of a considerable number of opsin genes (Simon et al. 2019). Unlike in P. hirtus, however, locomotor rhythms are short-term entrainable (Espinasa and Jeffery 2006; Carlson and Gross 2018).

Interestingly, cave populations of Asian loach (Family Balitoridae) species were found to display higher frequencies of conserved circadian clock penetrance and expressivity (Pati 2001; Duboué and Borowsky 2012). Together with the conserved food provision-entrainable clock of P. andruzzi (Cavallari et al. 2011), these findings indicate different trajectories of central clock conservation in fish and beetles. This may be explained by the shorter time frame of cave colonization and the continued interbreeding with surface fish in the case of A. mexicanus. Further light on the generality of this difference could be gained by comparative studies of the considerable number of cave-dwelling congenerics of P. hirtus in the central and southeastern United States of America (Leray et al. 2019) and the large numbers of subterranean beetle species in the Palearctic and Australia (Leys et al. 2003; Ribera et al. 2010).

Another strongly supported finding of our study is the D-phase activity preference of P. hirtus, when cultured in LD. This finding suggests that P. hirtus is a nocturnally active occupant of twilight zones in the open cave system of Mammoth Cave National Park. Our observations during collections confirmed the presence of P. hirtus in twilight areas, but we did not explore RA patterns in the field. It is interesting to note, however, that P. hirtus displayed a similarly strong level of photophobicity as its close relative P. cavernicola in light-dark choice tests (Friedrich et al. 2011). P. cavernicola is a facultative cave species, equipped with fully developed compound eyes (macrophthalmic) and flight-facilitating hind wings, and occupies a wide dispersal area in the Southeast of the United States (Peck 1984). While P. cavernicola has not yet been explicitly tested for nocturnality, this seems very likely given its strong avoidance of light. It seems reasonable, therefore, to speculate that nocturnality is an ancestral shared trait of P. hirtus and P. cavernicola. Ultimate evidence to probe this idea, however, will require comparative studies in both, closely-related cave- and surface species.

As the twilight zones of open caves constitute continuous links between surface and completely light-secluded deep cave space areas, it is further tempting to speculate that nocturnality originated in response to predation pressure at daylight. In further support of this, similar “nocturnal” activity profiles have been reported in all tested microphthalmic cave beetle species so far (Lamprecht and Weber 1977; Rusdea 1990; Friedrich 2013b). Also the phylogenetically more distant anophthalmic cave-dwelling amphipod Niphargus puteanus has been found to display a nocturnal activity profile when exposed to LD cycles (Gunzler 1965). Taken together, these data support the idea that nocturnality or crepuscular behavior are pre-adaptive traits for cave colonization (Romero 2011).

An obvious task towards gaining a better understanding of the significance of nocturnality in P. hirtus would be to elucidate the identity and biology of its assumed predators. Besides P. hirtus, Mammoth Cave is populated by six ground beetle species of the genera Neaphaenops and Pseudoanophthalmus, which are predators of small invertebrates, which might include larval or adult P. hirtus (Barr 1966, 1967; Culver and Hobbs 2017). Indeed, the most abundant of these, Neaphaenops tellkampfii, which is completely devoid of a visual system (Ghaffar et al. 1984) and highly specialized to prey on cave cricket eggs (Kane and Norton 1975), was found to consume P. hirtus larvae in laboratory culture (Griffith 1990). In addition, it is possible that the massively abundant nymphal stages of the cave cricket Hadenoecus subterraneus, a likely omnivore (Levy 1976; Studier et al. 1986; Helf 2003), might exert daytime predation pressure on P. hirtus in twilight zones. H. subterraneus adults possess well-developed compound eyes, populate the cave ceilings, conduct nocturnal foraging dispersal into the surface environment, and have been reported to maintain a long-term free-running nocturnal activity rhythm (Reichle et al. 1965). The activity patterns of juvenile H. subterraneus, however, which populate cave floors reaching into the twilight zone at high frequencies, have not been explored yet. Last but not least, it is furthermore likely that surface species disperse into peripheral sections of the twilight zones, maintaining the ancestral predation pressure on nocturnality.

The final notable finding of our study is that LD-monitored P. hirtus displayed about three-fold higher locomotor activity than the two monitor-adjusted DD animals, mostly due to longer resting periods as opposed to differences in activity levels or activity bout durations. While preliminary, given the small sample size of our study, these findings raise the possibility that P. hirtus engages in light-contingent rest phase modes that may be adaptively optimized to cope with the contrasting nutrient provisions of the deep cave and twilight zone environments.

A rich body of studies in Drosophila confirmed that prolonged resting periods are generally reflective of physiologically and neurologically conserved sleep-like states in insects (Anafi et al. 2019; Beckwith and French 2019; Shafer and Keene 2021). Moreover, studies in vertebrate and invertebrate species converge on supporting the model that the amount of sleep and its circadian distribution are subject to regulatory mechanisms that are independent of the central clock (Borbély 1982; Borbély and Tobler 1996; Duboué et al. 2011; Duboué and Borowsky 2012). Of further potential relevance is that core biological clock genes regulate sleep frequency in Drosophila, as mentioned above (Keene et al. 2010). In light of these data from other species, it is reasonable to speculate that the RA patterns of P. hirtus are the outcome of conserved clock gene expression and yet independent of a central clock.

As a scavenger species, P. hirtus likely has to invest energy in foraging. The nutrient-poor environment of deep cave zones may enforce a more conservative energy investment strategy in the form of a lower food search frequency compared to the more nutrient-rich twilight zones. In the latter, P. hirtus likely enjoys higher probabilities of successful resource discovery, translating into a higher energy budget for sustaining foraging mobility and reproduction. It is therefore conceivable that P. hirtus may be characterized by cave zone ecology-adjusted sleep duration states.

Light regimen contingent sleep duration states have been reported in a number of cave-adapted fish species. In contrast to P. hirtus, however, cavefish species so far analyzed are characterized by higher amounts of energy expenditure in constant darkness due to the shortening of rest phases. This is true for the diverse cave populations of A. mexicanus as well as cave-adapted loach species (Duboué et al. 2011; Duboué and Borowsky 2012; Carlson and Gross 2018). This difference between P. hirtus and cave-adapted fish may be due to different trophic structures of the aquatic subterranean environments of the cavefish populations and the terestrial deep cave zone populated by P. hirtus.

Drosophila has been found to respond to feeding scarcity with activity increase (Lee and Park 2004; Meunier et al. 2007; Keene et al. 2010; Yang et al. 2015). However, fruit flies can also be subjected to selection for starvation resistance, which results in activity reduction due to sleep phase extension (Masek et al. 2014; Slocumb et al. 2015; Miura and Takahashi 2019). It is, therefore, tempting to speculate that the proposed sleep phase extension state of P. hirtus in constant darkness may represent the result of long-term selection to tolerate food scarcity. Equally intriguing, suppressing flight capacity has been found to extend sleep duration in Drosophila (Melnattur et al. 2020). This raises the fascinating possibility that the adaptive sleep-extended state in the flightless P. hirtus may have originated as an integrated outcome of the complete regression of flight capacity. Taken together, our laboratory experiments with P. hirtus uncovered compelling but preliminary evidence of two new paradigms of cave adaptation, which invite further study in this and other cave species systems.