Introduction
Environments reaching subzero temperatures limit faunal diversity in species able to avoid or tolerate freezing. Whereas this phenomenon may be deleterious to various extents as determined by body size and physiology, the ability to survive partial freezing has evolved independently in various vertebrate lineages (Marchand Reference Marchand1996; Storey and Storey Reference Storey and Storey2017, Reference Storey and Storey1996). Most reported cases scrutinise animal species at high latitudes and altitudes that display seasonal freezing (Cabrera 1996; Sandercock et al. Reference Sandercock, Martin and Hannon2005); yet, the ecology of freezing exposure is highly diverse (Davenport Reference Davenport1992; Costanzo and Lee Reference Costanzo and Lee2013). Depending on the environment, exposure to freezing spells may differ in spatial and temporal thermal variance, predictability, seasonality, duration and absolute temperatures. For example, the Nearctic climate involves seasonal exposure to freezing, is typically associated with regulated and periodic inactivity and concomitant metabolic depression, as occur since hundreds of species of terrestrial insects (especially among Hymenoptera, Diptera, Coleoptera and Lepidoptera), until famous freeze tolerance vertebrates as the wood frog Lithobates sylvaticus (formerly Rana sylvatica) or the garter snake Thamnophis sirtalis, among others (Storey and Storey Reference Storey and Storey1996). In contrast, tropical high elevations such as the Paramos (tropical Andean ecosystems above the tree line) encompass episodic freezing conditions lasting from a few minutes to hours (Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011), a pattern that is sustained all year round, from a few nights to every night, according to local nuances. In addition, such environments display enhanced temporal variance in temperature and extensive diel thermal ranges that may involve freezing spells and warm temperatures over short periods of time (Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013). Due to the phenomenon termed “night sky radiation cooling” (Dobson Reference Dobson2005), such spells are more frequent on unclouded nights, when several species are most active. Tropical high-elevation amphibian communities have not been studied from the standpoint of freezing biology, but a number of species of Nearctic anurans have been investigated regarding patterns and subjacent processes (Layne and Lee Reference Layne and Lee1995). The best-studied species is probably L. sylvaticus, but data are also available on Hyla chrysoscelis, H. versicolor, Pseudacris crucifer and P. triseriata (Schmid Reference Schmid1982; Storey and Storey 2017, Reference Storey and Storey1992; Layne and Jones Reference Layne and Jones2001; Costanzo and Lee Reference Costanzo and Lee2013).
One antifreeze mechanism common to anuran species studied so far is the accumulation of carbohydrates with low molecular weight, such as glucose, which enhances cryoprotection, perhaps as a reactive response to internal ice formation (Voituron et al. Reference Voituron, Barré, Ramløv and Douady2009a). Such carbohydrates, either alone or interacting with other molecules, elevate the osmolarity of body fluids, thus decreasing cell shrinkage during extracellular ice formation (Storey and Storey Reference Storey and Storey2017). In addition, elevated levels of glucose augment tolerance to ischaemia by maintaining ATP turnover under hypoxic conditions and favour metabolic arrest (Storey and Storey Reference Storey and Storey1996; Voituron et al. Reference Voituron, Paaschburg, Holmstrup, Barré and Ramlov2009b). The frequency of freezing and thawing events can also enhance cryoprotectant production, given that rapid sequences of cooling and warming rates promote survival in L. sylvaticus exposed to freezing (Larson and Barnes Reference Larson and Barnes2016). In parallel, Ice-Nucleating Proteins (INPs) control the formation of ice in extracellular fluids (Storey and Storey Reference Storey and Storey1992a), so that the scope of body ice formation relates to both the density of cryoprotectants and the duration of the exposure to freezing temperature (Clausen and Costanzo Reference Clausen and Costanzo1990). It is probable that, in terms of active molecules, the mechanisms just described characterise all freeze-tolerant amphibians, including those in the Paramos. Yet, the nature of responses (e.g., reactive, cyclic or permanent) and the basal line of cryoprotection may vary with the ecology of freezing (Rexer-Huber et al. Reference Rexer-Huber, Bishop and Wharton2015), particularly given the daily exposure to freezing in the Paramo species.
The cryoprotective process of temperate anurans involves differential gene expression in pre-wintering frogs as a preparation to freezing conditions to come (Kiss et al. Reference Kiss, Muir, Lee and Costanzo2011; Storey and Storey Reference Storey and Storey2017). However, such seasonal patterns of expression would be at best secondary for species exposed to freezing on a daily basis, such as Paramo amphibians. Because this habitat is home to at least 40 species representing several genera and families (Bernal and Lynch Reference Bernal and Lynch2008), it seems reasonable to infer that all of them experience, at least occasionally, nocturnal freezing spells followed by higher temperatures during the day (Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011). Accordingly, we postulate that Paramo anurans display ecological, behavioural and physiological traits compatible with freezing. However, previous observations suggest that avoidance mechanisms may not typify Paramo species, at least not as a valid generalisation. For example, frogs in the genus Pristimantis (which are well represented in Paramos) tend to be nocturnal and terrestrial and call from exposed sites, thus experiencing the coldest possible conditions during activity time (Navas Reference Navas1996a). Although Paramos offer thermal diversity across the landscape and some microhabitats offer some thermal protection, avoiding frozen risk on the coldest nights may not be possible (Navas Reference Navas1996a; Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011). Inactive individuals may be exposed to very low temperatures, and even fully active individuals have been reported at temperatures as low as −5 °C (Navas Reference Navas1996b; Lüddecke and Sanchez Reference Lüddecke and Sanchez2002; Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011). Consequently, activity temperatures in some Paramo species may match those that simulate hibernation in L. sylvaticus (Costanzo et al. Reference Costanzo, Lee and Wright1991).
This study is a first attempt to explore physiological freeze tolerance in high-elevation tropical anurans, using as a model Paramo species in the genus Pristimantis. These Paramo frogs experience freezing conditions resembling those suffered by some Nearctic amphibians at the beginning of spring, when they engage in breeding, despite the freezing risk (Storey and Storey Reference Storey and Storey1987). Because Pristimantis frogs from Paramos live permanently under such conditions, our hypothesis is that these high mountain amphibians, when exposed to risk events of freezing, present an increase in glucose (as a cryoprotectant) and internal freezing of fluids, characteristics associated with freeze tolerance strategy (Schmid Reference Schmid1982; Storey and Storey 1996; Reference Storey and Storey2017). Additionally, because this genus originated before the major uplift in the tropical Andes (Heinicke et al. Reference Heinicke, Duellman and Hedges2007), freezing tolerance is unlikely to be an ancestral trait. Therefore, we also expected freeze tolerance to be an exclusive (or more evident) characteristic of Pristimantis montane species that live above the freezing line. To test our hypotheses, we conducted a study focusing on the species P. bogotensis, P. elegans and P. nervicus, with partial comparison with lower elevation counterparts. This comparison is not only relevant to the enhancement of knowledge of freeze tolerance in animals but may also contribute to explaining climate-related changes in the structure of tropical amphibian communities near the tree line (Navas Reference Navas2003; Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013).
Materials and methods
Localities and animal care
We conducted part of this study at the Buitrago area of the Chingaza National Park, located in the highlands of the eastern branch of the Colombian Andean Cordilleras. This Park ranges between 3400 m and 3700 m and has its central point located at 4º 44′ 38″ N and 73º 50′ 11″ W. The Paramo is the ecosystem present in this locality, characterised by a cold and humid climate, with sudden changes in the atmospheric state, and although the annual temperature fluctuation is small (2 to 10 °C), the thermal amplitude during the day varies from below the freezing point (environmental night temperatures below −15 °C) to 40 °C (Sarmiento Reference Sarmiento1986; Madriñan Reference Madriñan2004; Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011). Detailed information on daily temperature shifts in different microhabitats at the locality of work has been published elsewhere (Navas Reference Navas1997; Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013; Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011). It is typical that the coldest night temperatures and the lowest precipitation levels occur simultaneously (between 9 and 17% of annual rainfall, see Vargas and Pedraza Reference Vargas and Pedraza2004).
The other location of the study was the upper basin of the “Gaira” River, on the north-western slope of “Sierra Nevada de Santa Marta”, an area known as “La Cascada” in the province of “Magdalena” in Colombia. This area is located at 1560m and has its central point placed at 11° 10” N, 74° 10´ 45”. The locality is characterised by a tropical montane cloud forest with average annual rainfall of 2446 mm and steep slopes of primary and secondary forest in advanced state of regeneration, with temperatures from 14 to 24 °C (Hernández-Camacho and Sánchez-Páez Reference Hernández-Camacho and Sánchez-Páez1992; Rueda-Solano et al. Reference Rueda-Solano, Navas, Carvajalino-Fernández and Amézquita2016). There is a high diversity of microhabitats in the sector, including rocky areas, sand and leaf litter on the banks of the stream. Inside the riparian forest, undergrowth is found within about 10 m distance from the stream basin. The soil in the forest is covered mostly by leaf litter and dead wood, and there are fern patches where the tree cover is sparse (Carvajalino-Fernández et al Reference Carvajalino-Fernández, Saboya-Acosta, Padilla, Escárraga Fajardo and Porras2012).
Regarding animal collection for data gathering, we divided the samples into two groups: one for field and the other for laboratory experiments:
Animals used in field work
From the Paramo locality, we obtained samples from P. bogotensis (n = 4; weight = 1.2 ± 0.22 g), P. elegans (n = 10; weight = 4.1 ± 1.83 g) and P. nervicus (n = 7; weight = 1.15 ± 0.25 g). From the tropical montane cloud forest locality, we collected the frogs species P. insignitus (n = 11; weight = 2.64 ± 0.91 g), P. megalops (n = 12; weight = 1.9 ± 0.73 g) and P. sanctaemartae (n = 10; weight = 3.62 ± 1.15 g). For P. insignitus, P. megalops and P. sanctaemartae, all the specimens found were juveniles, whereas all other cases were adult males. The “n” values above do not indicate the total number of animals that could be collected, but the number of individuals from which adequate blood samples could be collected. Unfortunately, we were constrained by low effective “n” values because of the extreme difficulty involved in obtaining valid samples for certain species, and the previously established protocol preventing re-use of animals that could not be sampled in a first attempt, all this in the context of permits limiting the total number of animals that could be collected. As a result, the final N values are more related to data collection issues, not to abundance in the field. For example, despite being a common species, field samples derived from cardiac puncture were particularly challenging in P. bogotensis given their small size, and the rapid clotting (and even freezing) of blood samples. We collected the paramo species from December 2009 to January 2010 and tropical montane cloud forest in October 2009.
Animals used in laboratory work
For laboratory experiments, we collected just animals from the Paramo locality; these were P. bogotensis (n = 36; weight = 1.11 ±0.32 g), P. elegans (n = 20; weight = 5.33 ± 1.9 g) and P. nervicus (n = 25; weight = 1.4 ± 0.25 g). We maintained specimens in a laboratory in Bogotá, at 2600 m, where terrarium temperatures ranged from 8 to 17 °C, mirroring trends outdoors. The light cycle was artificially set to 12:12 h daytime/nighttime, and frogs had access to water and arthropods for food ad libitum. All frogs appeared normally active during captivity. The experiments were performed up to three days after capture, and frogs were not used in experiments after 10 days in the laboratory.
Controls
The experiments reported here lead to unambiguous interpretations only if adequate controls are performed. Ideally one should control for a “first test effect” by randomising order when comparing day and night patterns or alternative time-of-day related periods. In addition, controls should be performed to isolate the effects of puncturing, manipulation, time in captivity and several other variables. Whereas we are familiar with such formal aspects of experimental science, our study animals were wild species collected in small numbers (details in Localities and animal care). Accordingly, we prioritised a control for manipulation effects applied only to laboratory tests (described below under Glucose concentration change by freeze exposure). This was considered the most general and informative procedure, and it was applied to five individuals of P. bogotensis. To these specimens, we applied a protocol that preserved all the manipulation and dual puncturing of the freezing protocol but did not subject animals to cold exposure (freezer off, same time in). It turned out that no differences in the glucose level could be detected in animals before nor after this manipulation-only protocol (Wilcoxon paired test: Z: 2.02; p: 0.053; maximum value of glucose: 1.83 mmol/L). Other species may display similar patterns, but we acknowledge this as a supposition to be confirmed. We also controlled for experimental order (randomised) when such procedures did not require an increase in the number of animals.
Body temperature and glucose concentration: night-day contrast
To verify if frogs in the field display enhanced glucose concentration at night when temperatures are lower, we compared paired samples of glucose concentration obtained from blood extracted by cardiac puncture. We applied this protocol to the same individual at 12:00 and at 04:00 hours. We randomised the extraction order, which was applied first at 12:00 hours in some individuals, and at 04:00 hours in others, randomly. Frogs were fasted in a field terrarium during 24 hours prior to sampling, a protocol designed to reduce variance in glucose concentration induced by feeding history. The time of day selected for sampling was that maximising thermal differences according to our field temperature records. Before starting glucose sampling, body temperatures were measured with an infrared thermometer IR201 Extech 6:1. Field blood samples (0.7 µl) were collected by cardiac puncture, and the glucose level was measured with a glucometer (Alphatrak, Abbott Laboratories). The glucometer function was calibrated every five measures with the AlphaTRAK Control Solution (Alphatrak, Abbott Laboratories).
The influence of body mass on glucose modulation between treatments was evaluated, for each species, using ANCOVA tests, performing them with the packages tidyverse (Wickham et al. Reference Wickham, Averick, Bryan, Chang, McGowan, François, Grolemund, Hayes, Henry, Hester, Kuhn, Pedersen, Miller, Bache, Müller, Ooms, Robinson, Seidel, Spinu, Takahashi, Vaughan, Wilke, Woo and Yutani2019), ggplot2 (Wickham Reference Wickham2016), ggpubr (Kassambara Reference Kassambara2020), rstatix (Kassambara Reference Kassambara2021) and broom (Robinson et al. Reference Robinson, Hayes and Couch2021) of the statistical environment R (version 3.6). Body temperature and glucose data were compared intraspecific using Wilcoxon paired test. In addition, we explore differences interspecific of the body temperatures and glucose concentration, recorded during dawn, using a Kruskal–Wallis tests and multiple comparisons with Bonferroni correction tests. The analyses above mentioned were performed by applying the package ggplot2 (Wickham Reference Wickham2016) of the statistical environment R (version 3.6).
Glucose concentration change by freeze exposure
To analyse physiological responses to freezing, we followed blood extraction and glucose analysis protocols similar to those reported in 2.3, although in the context of experiments beginning at 20:00 (night period). From each specimen, we 1) took a glucose sample, 2) moved it to a refrigerator (5 to 8 °C) for one hour (to simulate a Paramo temperature drop), 3) transferred it to an airtight 50 ml top jar and, in turn, submerged it in a salted ice bath at −3 °C for two hours. At this final stage, the expected drop in body temperature was recorded using the NTC thermistor referred to in section (2.5). After procedure 3, we obtained a second glucose sample. The time lapse between sample collections was about three hours. After the second sample, frogs were moved to individual containers kept at 12 – 17 °C for two days. Time recovery was set when the frogs regained the regular breathing, normal posture and righting response, after removing from the container used for freezing treatment. Only after two days were frogs considered recovered from the treatment; we compare species recovery time using ANOVA test and Tukey post hoc test of the statistical environment R (version 3.6). The influence of body mass on the changes of glucose levels was explorer using the same methodology reference in point 2.3 of this manuscript. We performed Wilcoxon pair tests for group comparison using the statistical environment R (version 3.6) as in the others data analysis.
Ice quantification protocol
We quantified body ice formation inside the frog after the freezing protocol described above. We used a glass covered container (10 ml) fitted with glass capillary tubing in the middle of a calorimetric cell whose calorimetric constant could be determined by Joule heat calibration using a 10Ω resistor. All the calorimetric system was insulated with rigid foam (polystyrene). We calculated the exothermic heat produced, during the freezing experiment in the frogs, according to the enthalpy of melting ice by generating a temperature curve with a previously calibrated NTC thermistor of 10 KΩ. This thermistor was connected to a Multimeter Multimaster 570 True RMS (Extech Inc.) able to register data each second with the BS85X software programme (Version 5.1.0.4). Calculations required information on the water content of each species, so we euthanised five frogs of each species with an overdose of 2% lidocaine gel and dried them to constant weight at 70 °C (Schmid Reference Schmid1982). Body water content was expressed as a percentage of wet mass and as the mass of water per dry body mass (g H2O/g dry wet; adapted from Layne & Jones Reference Layne and Jones2001). The quantity of frozen water was calculated in individual specimens from the enthalpy of the melt curve (Schmid Reference Schmid1982) using the Origin7 programme (Originlab, version 7).
Results
Body temperatures under restrained field conditions
In terraria placed in the field and exposed to natural variation in temperature, body temperature of all the species decreased at night (Wilcoxon paired test: P. insignitus, W: 110, p<0.01; P. megalops, W: 168, p<0.01; P. sanctaemartae: W: 132, p<0.01; P. bogotensis, W: 47, p<0.01; P. elegans, W: 210, p<0.01; P. nervicus W: 42, p<0.01; Figure 1). Besides, body temperatures of paramo frogs were substantially lower at dawn (04:00 am) that those of their montane rain-forest counterparts at analogous conditions (KW: 47.83; p<0.01; Figure 1).
Field body temperature of Pristimantis species in terrarium during glucose measures at noon (˜12:00 pm, red boxplot) and dawn (˜4:00 am, blue boxplot). “n” of registers under the name of the species. A, B, C, D, differences of temperatures at dawn between species with letters in each boxplot (using multiple comparisons with Bonferroni correction; significant P≤0.05). +, significant difference between body temperatures at noon and dawn for each species.
Glucose modulation in field and in laboratory
Outcomes found in the field
Body mass did not influence the glucose modulation in experiments performed in the field with both Paramo and tropical montane cloud forest species (ANCOVA: P. insignitus, F: 4.12, p: 0.06; P. megalops, F: 1.06, p: 0.32; P. sanctaemartae, F: 1.33, p: 0.28; P. bogotensis, F: 1.48, p: 0.31; P. elegans, F: 0.31, p: 0.58; P. nervicus, F: 0.149, p: 0.713). Glucose levels of Paramo frogs were substantially higher at dawn (04:00 am) that those of their tropical montane cloud forest counterparts at analogous conditions (KW: 47.83; p<0.01; Figure 2). In terms of diel cycles, P. bogotensis and P. nervicus displayed a substantial increment in glucose levels when compared measures taken at noon and during the dawn (Wilcoxon paired test: P. bogotensis, W: 21, p< 0.03, an increase of 64%; P. nervicus, W: 27; p: 0.02; an increase of 104%; Figure 2).
Differences of glucose levels of Pristimantis species at noon (˜12:00 pm, red boxplot) and dawn (˜4:00 am, blue boxplot). “n” of registers under the name of the species. A, B, C, D, differences of glucose concentration at dawn between species with letters in each boxplot (using multiple comparisons with Bonferroni correction; significant P≤0.05). +, significant difference between glucose concentration at noon and dawn for each species.
Outcomes found in the laboratory
There was also no association between body mass and glucose modulation in the data registered in Paramo species in the laboratory (ANCOVA test: P. bogotensis, F: 1.64, p: 0.25; P. elegans, F: 2.04, p: 0.17; P. nervicus, F: 1.24, p: 0.302). Freezing induced in the laboratory caused an increase in blood glucose concentration in all paramo species (Wilcoxon paired test for all the species: P. bogotensis, W: 121, p< 0.01, an increase of 100%; P. elegans, W: 127.5, p< 0.01, increase of 55%; P. nervicus, W: 104, p: 0<0.01, increase of 126%; Figure 3).
Glucose concentrations changes by experimental freeze exposure in Paramo Pristimantis species. For each species, the glucose levels after a freeze exposure of 2 horas of −3 °C were represented by the blue boxplot and the control groups were represented by the red boxplot (for more details see section 2.4). “n” of registers under the name of the species. +, significant difference between glucose measures of freezing treatment and control.
Recovery patterns after experimental freezing in Paramo frogs
All Paramo frogs recovered from freezing, in a maximum of two days after experimental freezing, with locomotion and feeding patterns seemingly comparable to those of untreated animals. However, for quantification purposes, we considered “time to recover” as the time lapse between the end of the experiment (i.e., the time when frogs were removed from the calorimeter) and the time at which animals displayed regular breathing, normal righting response and ability to jump after light prodding. Species showed differences between times of recover after the freezing treatment (ANOVA test: 28.86; p <0.05). Individuals of P. elegans required the longest time to recover (56 ± 25 minutes), compared with P. bogotensis (Tukey test: 9.26; p<0.05; 4+3 minutes) and with P. nervicus: (Tukey test: 9.34; p<0.05; 3+4 minutes). No differences were found in the recovery times between P. bogotensis and P. nervicus (Tukey test: 0.08; p: 0.99).
Body ice grown during experimental freezing in Paramo frogs
The water content of Pristimantis from Paramos was comparable (P. bogotensis: 77.04 ± 0.49%; P. elegans: 76.62 ± 1.4%; P. nervicus: 79.02 ± 2.37%; n = 5 for each species). Whereas P. bogotensis and P. nervicus presented an exothermic peak when exposed to the experimental freezing protocol, the pattern of body ice formation during the protocol differed between species (Table 1).
Reference value of freezing parameters for Pristimantis elegans, P. bogotensis and P. nervicus.
Discussion
Cryoprotectant synthesis is the result of adaptation to freezing in temperate frogs (Costanzo and Lee Reference Costanzo and Lee1993; Costanzo et al. Reference Costanzo, Lee and Lortz1993), so it is logical to assume a role for species in other ecological settings involving risk of body freezing. Accordingly, anurans exposed to freezing spells in high-tropical elevations must display such responses or equivalent ones (Navas Reference Navas1997; Carvajalino-Fernández et al. Reference Carvajalino-Fernández, Bonilla Gomez and Navas2011; Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013). Our study assumed that similar physiological mechanisms exist in such anuran community, and for simplicity focuses on glucose as a first approach to the problem. However, it is likely that molecules like glycerol (Layne Reference Layne1999) play additional roles, particularly in the context of permanent freeze protection (Moalem et al. Reference Moalem, Storey, Percy, Peros and Perl2005; Rexer-Huber et al. Reference Rexer-Huber, Bishop and Wharton2011), particularly in the most exposed species. Independently of this, our data corroborate the view of glucose as an important and highly flexible molecule associated with freeze tolerance in anurans, including high-elevation tropical lineages. The study also highlights physiological diversity among Paramo anurans and favours the notion of differential impact of freezing spells within the community. Both P. nervicus and P. bogotensis can be formally defined as freeze tolerant due to glucose spikes and body ice grow during freeze events (Storey and Storey Reference Storey and Storey1992), but not P. elegans because of the absence of body ice formation.
The processes leading to protection may be associated with diel shifts enhancing nocturnal cryoprotection or reactive responses to low temperature, but our data were not designed to tell these two possibilities apart. What we know is that a type of daily glucose modulation is possible and that such modulation is compatible with the highest activity period for the Pristimantis in the Paramos during the night, at which time there is a risk of freezing in the Paramo (Navas Reference Navas1996c; Lüddecke and Sanchez Reference Lüddecke and Sanchez2002; Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013). Although from one perspective this may seem comparable to freezing in hibernacula, a factor requiring antifreeze adjustment in Alaskan L. sylvaticus (Sinclair et al. Reference Sinclair, Stinziano, Williams, Macmillan, Marshall and Storey2013; Larson et al. Reference Larson, Middle, Vu, Zhang, Serianni, Duman and Barnes2014; Larson and Barnes, Reference Larson and Barnes2016), the issue requires further research because freezing spells occur more often at the time of day of maximum activity. Given that L. sylvaticus enters a metabolic depression under conditions interpretable as similar, Paramo and temperate species would converge ecologically only when the latter emerge during the spring.
Despite the above-described parallel between Paramo daily conditions and a temperate spring, it is possible that the freezing ecology of Paramo frogs is closer to that of other lineages which tolerate episodic daily freezing, for example, intertidal invertebrates in temperate zones (for example Farke et al. Reference Farke, Riemann and Schrage1984; Hawes et al. Reference Hawes, Worland and Bale2010; Theede et al. Reference Theede, Scheneppenheim and Beress1976; Waller et al. Reference Waller, Worland, Convey and Barnes2006). These animals survive freezing temperatures during aerial exposure at low tides, although their osmoconformity prevents accumulation of low-mass molecules as cryoprotectants. Instead, they accumulate INPs, perhaps the principal physiological mechanisms for freezing tolerance in marine invertebrates (Hawes et al. Reference Hawes, Worland and Bale2010; Storey and Storey, Reference Storey and Storey1996). Furthermore, certain insects from cold environments experience daily freezing (Baust and Nishino Reference Baust, Nishino, Denlinger and Lee1991; Block et al. Reference Block, Wharton and Sinclair1998; Sinclair et al. Reference Sinclair, Worland and Wharton1999; Sinclair and Chown Reference Sinclair and Chown2005), and as Hyla versicolor and Hynobius keyserlingi do (Layne and Jones Reference Layne and Jones2001; Storey and Storey Reference Storey and Storey1992), they use polyols like glycerol or sorbitol as cryoprotectants; whereas sugars like glucose, maltose and trehalose seem secondary in this context (Storey Reference Storey1997).
Our conclusions raise several general questions. Given that significant variation in the physiology of freezing occurs among related species within a single genus, even if those species share certain key ecological traits, we anticipate that the overall physiological diversity in the Paramo amphibian community is substantial. This diversity likely involves diel cycles of protection, constant protective baselines and even the scope of reactive protection. The last mentioned would necessarily involve fast mobilisation of cryoprotectants in short order and may possibly be active after a freezing event (Sinclair et al. Reference Sinclair, Stinziano, Williams, Macmillan, Marshall and Storey2013; Larson et al. Reference Larson, Middle, Vu, Zhang, Serianni, Duman and Barnes2014). This scenario allows for inferring that climatic shifts leading to changes in the frequency or scope of freezing spells may lead to substantial shift in the structure of high-elevation tropical amphibian communities, mainly around the 3500 m contour. Exposure to freezing spells may last minutes or hours (Navas et al. Reference Navas, Carvajalino-Fernández, Saboyá-Acosta, Rueda-Solano and Carvajalino-Fernández2013), so that shifts in this context would make some species less prone to tolerate freeze spells and therefore be limited to lower elevations or less exposed habitats. For example, P. elegans may be constrained to thermally buffered habitats such as the vicinity of streams. On a more evolutionary note, our work supports the idea of a physiologically diverse anuran community in which stock from different lineages has adapted to Paramos acquiring physiological characteristics not present in lower elevation counterparts.
Acknowledgements
J.M. Carvajalino was supported by a grant from the Fundación para el Fomento de la Investigación - Banco de la República - Colombia, fellowship “Jovenes investigadores e innovadores Virginia Gutierrez de Pineda” from Colciencias - Colombia, fellowship “Programa de movilidad en el posgrado” from the Red de Macrouniversidades Publicas de América Latina y el Caribe, and CNPq – Brazil, PhD Scholarship (Process 142096/2012-1). Permission to study and collect animals in the Chingaza Natural Park was granted by Parques Nacionales de Colombia. This project received the support of FAPESP, the State of São Paulo Science Foundation, research grant 2014/16320-7 to C.A. Navas et al.
