Tracking climate change in the spatial distribution pattern and the phylogeographic structure of Hyrcanian wood frog, Rana pseudodalmatina (Anura: Ranidae)

1 INTRODUCTION

Climatic changes through time have important effects on the evolutionary history and spatial distribution of species (Bontrager & Angert, 2019; Foden et al., 2013; Hopley & Byrne, 2019; Omann et al., 2009; Razgour et al., 2019; Velásquez-Tibata et al., 2012; Vieira et al., 2018). Historical changes in regional climate also play an important role in conditioning patterns of gene flow among populations, with species-specific life histories and demographics also contributing to shaping species distributions and their temporal dynamics (Baek et al., 2018; Bell et al., 2010; Hua, 2013; Safner et al., 2011; Schierenbeck, 2017). Numerous phylogeographic studies have shown that during cold phases of the last glacial period the distribution ranges of many species contracted to climatic refugia (Bao et al., 2015; Leipold et al., 2017; Malekoutian et al., 2020; Song et al., 2018). Subsequently, as temperatures increased during the Holocene, distributions expanded again to occupy newly suitable habitats (Chen et al., 2017; Cox et al., 2019; Igea et al., 2013; Liepelt et al., 2009; Zhao et al., 2019). On shorter, more recent time scales, climate change has been linked to distribution shifts in many species, sometimes resulting in local or regional extinction (Devictor et al., 2012; He et al., 2018; McCarty, 2001). According to 131 studies, extinction rates in response to climate change are estimated to range from 0% to 54%, with an average of 7.9% (Urban, 2015). Therefore, understanding how species respond to climate change is of fundamental importance in conservation planning.

Identifying Quaternary climatic refugia due to stable climatic conditions can be critical to the conservation of species for future climate change in terms of predicting the effects on population persistence (Provan & Bennett, 2008), especially in globally threatened groups like amphibians (Alroy, 2015; Duarte et al., 2012; Pounds et al., 2006). Due to their biphasic life-cycles, highly permeable skin, and unshelled eggs, amphibians are particularly vulnerable to alterations in environmental conditions, and climate change has been identified as one of the most important factors threatening their populations (Carey & Alexander, 2003; Corn, 2005; Wake & Vredenburg, 2008). Amphibians are regarded as good bioindicators of environmental changes due to their physiological constraints and relatively low mobility (Carnaval et al., 2009). Genetic studies can provide evidence for historical distribution shifts of amphibian populations in response to climatic oscillations during the Quaternary (Afroosheh et al., 2019; Chiocchio et al., 2017; Fouquet et al., 2012; Provan & Bennett, 2008; Yodthong et al., 2015; Zeisset & Beebee, 2008; Zhang et al., 2016; Zhou et al., 2014). Integrative studies combining phylogeographic analyses with species distribution modeling (SDM) can be used to assess the impact of past climatic changes on the evolutionary history and historical demography of populations and help predict future changes in their geographical ranges (Afroosheh et al., 2019; Ahmadzadeh et al., 2013; Nicolas et al., 2018; Velo-Antón et al., 2013).

The Hyrcanian wood frog, Rana pseudodalmatina Eiselt & Schmidtler, 1971, is a brown frog (Anura, Ranidae) endemic to temperate forests in northern Iran (also known as Hyrcanian forests). It ranges from the Talesh region to the Golestan National Park, on the northern slopes of the Alborz Mountains and the southern edge of the Caspian Sea of Iran (Najibzadeh et al., 2018). The species composition and diversity of the Southwest Asian forests, rich in endemic species, suggests that the Hyrcanian region has likely acted as a glacial refugium for many taxa during the Quaternary glaciations (Röhrig, 1991; Saberi-Pirooz et al., 2018). Phylogeographic studies and climatic models have shown that the southern Caspian Sea has acted as a glacial refugium for a broad range of taxa, including Iranian rock lizards (Darevskia chlorogaster and Darevskia defilippii, Ahmadzadeh, Flecks, et al., 2013), Iranian brown bears (Ursus arctos, Ashrafzadeh et al., 2016), fat dormouse (Glis glis, Ahmadi et al., 2018), Persian mountain salamanders (Paradactylodon persicus, Ahmadzadeh et al., 2020), and Caspian green lizard (Lacerta strigata, Saberi-Pirooz et al., 2021). During the last half-century, the climate became warmer in the Hyrcanian forests, with the average annual temperature increasing between 1.28 and 2.45℃ and annual precipitation declining between 55.6 and 409.4 mm (Tohidifar et al., 2016). Due to the close relationship between temperature and rainfall and amphibian distributions (Ortiz-Yusty et al., 2013), this warming trend is expected to have a negative impact on the distribution of R. pseudodalmatina. This species is categorized as Least Concern (LC) in the IUCN Red List, but few studies have assessed the species’ conservation status and threats (Sharifi et al., 2009).

The main aims of this study are (a) to investigate the evolutionary history and demographic history of R. pseudodalmatina based on two mitochondrial DNA fragments (cytochrome b, cytb, and 16S ribosomal RNA, 16S); (b) to assess the impact of past (Last Glacial Maximum and Mid-Holocene) and more recent (1970–2000) climate fluctuations on the spatial distribution pattern of the species using Species Distribution Models (SDMs); and (c) to investigate whether R. pseudodalmatina will behave as a single distributional unit in response to future climate change (2070).

2 MATERIALS AND METHODS

2.1 Sampling

Twenty specimens of R. pseudodalmatina and five specimens of Rana macrocnemis were collected across their distribution range in the Hyrcanian forests (Figure 1, Appendix 1). Tissue samples consisted of third toe clips; all sampled individuals were subsequently released at the site of capture. Samples were preserved in 96% ethanol and kept at −20℃ prior to DNA extraction. Details of sampling locations, accession numbers for newly generated mtDNA sequences, and code ID are presented in Appendix 1.

3 RESULTS

3.2 Estimating divergence time

Based on the combined mtDNA genes, the MRCA of Hyrcanian/Anatolia lineage and other species of the genus Rana in Europe lived at 13.6 Mya (95% HPD; 6.23–21.11 Mya). At the intra-species level, R. pseudodalmatina haplotypes shared a MRCA at 1.6 Mya (95% HPD; 0.58–2.54 Mya; Figure 2).

3.3 Genetic structure and demographic analysis

Molecular diversity indices based on the cytb and 16S genes including the number of H, h, and π within R. pseudodalmatina and its regional patterns are presented in Table 1. The cytb fragment contained 16 variable sites (12 transitions and four transversions), of which seven were parsimony-informative and nine were singletons. The mean nucleotide composition was A: 22.91%, T: 27.61%, C: 33.76%, and G: 15.73%. The 16S fragment contained 16 variable sites (14 transitions and two transversions), of which one was parsimony-informative and 15 were singletons. Mean nucleotide composition was A: 29.79%, T: 25.27%, C: 24.33%, and G: 20.61%.

TABLE 1. Molecular diversity indices based on cytb and 16S genes within Hyrcanian wood frog, Rana pseudodalmatina and western (W) and eastern (E) regional patterns. Sample size (N), number of haplotypes (H), haplotype diversity (h), and nucleotide diversity (π)

	Gene	N	H	h	π	Tajima's D	Fu's Fs
Rana pseudodalmatina	cytb	39	12	0.78	0.005	−0.52	−2.07
	16S	48	13	0.69	0.002	−2.15*	−9.28*
E	cytb	17	4	0.50	0.0008	−1.04	−1.47*
	16S	19	6	0.46	0.001	−2.04*	−3.86*
W	cytb	22	8	0.60	0.001	−2.07*	−4.42*
	16S	29	7	0.37	0.001	−2.28*	−4.68*

• * p < 0.05.

AMOVA results showed significant genetic differences between two regional patterns (E and W) based on mtDNA genes (cytb: F_st = 0.44, p < 0.05; 16S: F_st = 0.58, p < 0.05). Most values of Tajima's D and Fu's Fs in R. pseudodalmatina and its regional patterns were significant and negative (Table 1). The MMD diagrams for the E and W displayed a unimodal pattern with right-skewed consistent with recent demographic expansion (Figure 3c, d). The BSP based on cytb showed no obvious trends in N_e through time in each regional pattern (Figure 3a, b). The haplotype parsimony network for the cytb gene showed that the western and eastern groups were separated from each other with two-step mutations (Figure 4).

3.4 Biogeographic analysis

Ancestral distributions of R. pseudodalmatina were inferred from S-DIVA and BBM analyses (Figure 5). In both reconstructions, the most recent common ancestor of the species was distributed in the Hyrcanian forests; subsequently, it expanded into the western and eastern areas. According to S-DIVA, a vicariance event took place in the ancestral node of the two areas, and dispersal occurred in the western distribution. BBM reconstruction indicated that vicariance and dispersal events shaped the species’ current distribution (Figure 5).

3.5 Species distribution modeling (past–present–future)

The potential distributions as suggested by our models for past (Last Glacial Maximum, LGM [21 Kya] and Mid-Holocene [6 Kya]), current (1970–2000), and future (2070) climatic conditions are shown in Figure 6, and Figure S2, wherein the average performance was AUC_test = 0.828. On average, Temperature Annual Range (BIO7: 57.3%) had the highest contribution, followed by Precipitation of Wettest Month (BIO13: 25.2%), Mean Temperature of Wettest Quarter (BIO8: 9.0%), and Mean Temperature of Driest Quarter (BIO9: 8.5%).

Based on the SDMs, the central part of the Caspian coast in the southern region was identified as most suitable habitats for R. pseudodalmatina during the LGM (Figure 6c). But in the Mid-Holocene, the potential distribution range of the species expanded to include the eastern, central, and western Hyrcanian forests (Figure 6c). Under current climatic conditions, the potential range of the species increased, with habitat suitability uniformly expanding in the Hyrcanian forests from the Talesh region to the Golestan National Park on the southern edge of the Caspian Sea and in the northern slopes of the Alborz Mountains (Figure 6a).

According to the RCPs 2.6, 4.5, 6, and 8.5 scenarios of future climate projections, the distribution range of R. pseudodalmatina would tend to shift to higher altitudes especially in the western part of the range in the northern slope of Alborz Mountain (Figure 6e-h). Stability through time suggests that the Hyrcanian forests acted as historical refugia for this species (Figure 6d). Based on landscape analyses, connectivity under current climatic conditions is maximum in the central part of the range and decreases in the easternmost and westernmost extremes (Figure 7).

4 DISCUSSION

We examined the genetic diversity patterns of the Hyrcanian wood frog, R. pseudodalmatina, throughout its geographic range in the Hyrcanian forests. Also, we used phylogeography and species distribution modeling approaches to investigate the putative glacial refugia and the effect of climate oscillations on the species.

We analyzed two mtDNA genes (cytb and 16S) and found the Hyrcanian/Anatolian lineage and other species of the genus Rana in Europe had MRCA during the Miocene (13.6 Mya: 95% HPD; 6.23–21.11 Mya; Figure 2). It is assumed that the uplift of the Turkish–Iranian plateau and the physical vicariance such as the salt-water barrier may have led to the speciation among the Anatolian, the Caucasian, and the Iranian brown frogs (Najibzadeh et al., 2017). Rana pseudodalmatina displayed a phylogeographic pattern with two regional groups across its distribution range in the western and eastern Hyrcanian forests (Figure 1, Figure S1). The results of Najibzadeh et al. (2018) corroborate our findings that the species forms two distinct groups in the east and west of the geographic range. However, Najibzadeh et al. (2017) estimate of TMRCA of the two groups is older compared to our findings. Nevertheless, the authors noted that their methodology is prone to misleading results. According to our result, the MRCA of two regional patterns lived 1.6 Mya (95% HPD; 0.58–2.54 Mya), which is much more recent compared to the equivalent estimate of Najibzadeh et al. (2020), which was ca. 4 Mya. This difference is possibly due to the different set of markers, the number of individual samples, prior settings, and calibration points used in the two studies. We used two mtDNA markers (cytb and 16S) for divergence time estimations; however, Najibzadeh et al. (2020) only used the 16S gene. Our estimated divergence time is comparable to the Caspian green lizard (L. strigata) and Fat Dormouse (Gilis gilis) that turned into the eastern and western lineages about 0.9 Mya (95% HPD: 0.55–1.16 Mya) and 1.19 Mya (95% HDP: 0.55–1.91 Mya), respectively, in the Hyrcanian region (Ahmadi et al., 2018; Saberi-Pirooz et al., 2021).

According to the AMOVA results, the two populations are almost isolated from each other because of the F_st value of nearly 0.5. The MRCA of the two regional groups of R. pseudodalmatina was during the Pleistocene (1.6 Mya [95% HPD; 0.58–2.54 Mya]; Figure 2), which suggests that this species may have been affected by the Quaternary climate fluctuations. It seems that the presence of two groups of R. pseudodalmatina in the west and east of the distribution may indicate the long-term residence within two different glacial refugia during the Pleistocene. The existence of cryptic refugia and the idea of “refugia-within-refugia” in the Hyrcanian forests that is proposed here for R. pseudodalmatina has been previously discussed for other species that live in this region (Ahmadi et al., 2018; Ahmadzadeh, Flecks, et al., 2013; Saberi-Pirooz et al., 2021). The assessments of demographic history revealed that the recent expansion occurred in the populations (Table 1 and Figure 3).

Our findings identified a successive vicariance event correspond to the divergence of the regional eastern and western groups from the ancestral node (Figure 5). We suggested that the increase in the Caspian Sea water level during the Pleistocene led to the occurrence of vicariance event for the species (Lateef, 1998; Mamedov, 1997). On the other hand, the results of S-DIVA and the BBM analysis showed that dispersal events also played an important role in creating this disjunct distribution pattern (Figure 5). The parsimony haplotype network based on the cytb gene indicated that the more derived haplotypes were found in the western region (Figure 4). This suggested that the species derived from the west of the Hyrcanian forests and then migrated to the east of the distribution area.

In response to climate change, species shifted their distribution range to reach more favorable conditions (Guedes et al., 2020). Based on SDMs results, R. pseudodalmatina showed a scenario of past range contraction followed by recent expansion (Figure 6 and Figure S2). We assumed during the LGM, the distribution of the R. pseudodalmatina has been concentrated in two different glacial refugia in the east and west of the distribution range (Figure 6c). Given the warmer conditions since the Mid-Holocene compared to the recent climatic conditions (Akhani et al., 2010), its distribution in the more suitable habitat and a longitudinal shift was expanded throughout the Hyrcanian forests (Figure 6a,b). The longitudinal shift of the distribution range is also shown in the study of Duan et al. (2016) on Chinese amphibians.

The optimal temperatures for R. pseudodalmatina range between 11 and 18℃, and species activity decreases widely at temperatures above 23℃ (Pesarakloo et al., 2012). According to our result, Temperature Annual Range (BIO7) had the highest contribution in the model in the past, recent, and future climate change (57.3%). Based on the future climate (2070) projection, the distribution range of the R. pseudodalmatina will shift upward to higher altitudes in the northern slope of Alborz Mountains where the species currently occurs (Figure 6). The results of our study confirm the general trend that amphibian species migrate to higher altitudes under the influence of climate warming. For example, Forero-Medinaet al. (2011) investigated 46 amphibian species of tropical mountains and showed that under future climate changes, their range of distribution will shift to higher altitudes. In contrast to these trends and our findings, Najibzadeh et al. (2020) showed that the suitable habitat for R. pseudodalmatina has remained stable since the LGM and will remain the same in the future.

Based on the results of the Circuitscape analysis, the highest connectivity was observed between the populations located in the central parts (the westernmost part of the eastern regional population) to the westward (the easternmost part of the western regional population) of Mazandaran (Figure 7). These results may indicate secondary (postglacial) contact of these populations (Austin et al., 2004; Zamudio & Savage, 2003), which requires further investigation in future studies. According to the results of the stability map through time in this study, Hyrcanian forests acted as climate refugia for the R. pseudodalmatina during climate change (Figure 6d). Therefore, Hyrcanian forests may be buffered to contemporary climate change over time and provide resistance to valued socio-cultural, ecological, and physical resources.

CONFLICT OF INTEREST

There are no conflicts of interest.