Introduction

Insect conservation is increasingly gaining recognition globally as a top priority since insects “are ruling the world”; being interlinked with biodiversity components and regulating several ecosystem processes (Losey and Vaughan 2006; Weisser and Siemann 2008; Schowalter 2013; Noriega et al. 2018; Samways et al. 2020). Insects are one of the most diverse group of animals exhibiting rapid population declines and high species extinction risk (Stork 2018; Eisenhauer et al. 2019; Cardoso et al. 2020). It has been predicted that half of the one million species that face extinction in the future decades are insects (IPBES 2019). The key drivers for their population declines are habitat loss, fragmentation and degradation, pesticide use, invasive species spread, climate change, and overexploitation (Díaz et al. 2019; IPBES 2019; Cardoso et al. 2020).

Particularly for insect populations that are highly specialized, even minor disturbances on their microhabitat can have considerable effects on their populations (Didham et al. 1996; Hunter 2002; Steck et al. 2007). Habitat fragmentation reduces biodiversity and impairs key ecosystem functions; the main process being that smaller and more isolated habitat fragments can lead to reduced inter-fragment movements thus hampering potential fragment recolonization after local extinctions (Haddad et al. 2015). It is well documented that the disruption of continuous habitat negatively affects insect species richness and population densities (Didham et al. 1998; Wettstein and Schmid 1999; Bazelet et al. 2014), constrains dispersal movements and limits genetic flow (Jha and Kremen 2013; Ortego et al. 2015), and finally compromises populations’ viability, resulting in local extinctions (Frankham 2005; DiBattista 2008; Mendez et al. 2014). Additionally, distribution range contractions can increase the effects of random genetic processes which can in turn lead to the reduction of the effective population size (Ne), the loss of genetic variability and the overall decrease of a species’ evolutionary potential (Hedrick and Kalinowski 2000; Pertoldi et al. 2007). The use of genetics in conservation biology has been repeatedly used to identify such processes. Furthermore, it has already contributed to the understanding of the effects of habitat fragmentation and genetic diversity loss on species’ extinction risk (Frankham et al. 2010).

Insects are poorly studied and underrepresented in Red List assessments and conservation agendas, hence a coordinated conservation action is urgently needed, covering knowledge gaps, monitoring all taxonomic, genetic and functional facets of insect diversity, and accelerating insect inclusion in relevant legislation frameworks (Cardoso 2012; Hochkirch 2016; Eisenhauer et al. 2019; Samways et al. 2020) is needed. One of the many understudied insect groups that play a key role in ecosystem functions (e.g., food webs and nutrient cycling) are grasshoppers (Orthoptera: Acrididae) (Latchininsky et al. 2011), and especially the genus Chorthippus. Chorthippus is one of the most diverse and species-rich genera of the family Acrididae, showing substantial incomplete lineage sorting coupled with little genetic differentiation between the species (Vedenina and Mugue 2011; Nolen et al. 2020; Hawlitschek et al. 2022). The family counts over 1,000 species in Europe, out of which 68% are endemic, and faces a great extinction risk with about one-fourth of the European species being threatened (Hochkirch et al. 2016). Yet, there is limited focus on their conservation, with only 1% of them listed in the annexes of the Habitats’ Directive (92/43/EC), due to the lack of basic taxonomic information and scarce knowledge on genetic data (Kim 1993; Hochkirch et al. 2016).

Steno-endemic species are defined by very restricted distribution and are particularly prone to extinction, since human-driven habitat loss and fragmentation decrease species distribution range and increase population extirpation (Fonseca 2009; Dirnböck et al. 2011; Salles et al. 2019; Mayani-Parás et al. 2021). One such species is the Epirus dancing grasshopper, Chorthippus lacustris. The species strongly depends on the seasonally flooded grasslands of the Epirus region in NW Greece. During the last 50 years, the species has lost 85–99% of its habitat due to degradation and fragmentation, land use changes and unsustainable grazing (Kati et al. 2006, 2012; Willemse et al. 2016), thus the species is characterized as Critically Endangered (CR) at a global scale under the IUCN criteria (Willemse et al. 2016). Preliminary results of an ongoing conservation project indicate eight breeding populations of the species, while two known populations being probably extinct the last 20 years (Noutsou et al. 2022). Water drainage processes have been repeatedly taking place in the Epirus prefecture, mainly for agricultural encroachment, further compromising the viability of the species’ populations (Kati et al. 2012; Chiotelli 2015).

The present study aims to use genetic data to explore the genetic diversity and divergence of the Epirus dancing grasshopper populations. Specifically, we employed both mitochondrial and nuclear genetic markers to (a) assess the levels of genetic diversity of the species, (b) describe the population genetic structure and (c) investigate the spatial patterns of genetic differentiation among its populations. Our results provide an assessment of the genetic status of the species populations and, ultimately, contribute significantly to the conservation and management of the Critically Endangered Epirus grasshopper and its habitat.

Methods

Sample collection

We collected material (hind leg) from 81 adult specimens of C. lacustris (n = 26 in 2016; n = 55 in 2022) across eight known population localities (4–13 samples per locality) for the species, in Epirus region in Northern-Western Greece (Fig. 1). Sampling took place from mid-August onwards, at the end of species breeding cycle, to avoid impacting the populations. We stored tissue samples at -20 °C and extracted total genomic DNA using the NucleoSpin Insect kit (Macherey-Nagel), following the manufacturer’s protocol.

Fig. 1
figure 1

Map showing the sampling populations of C. lacustris in Epirus region. Pie charts represent haplotype frequencies of the two mitochondrial genes (left: COI and right: CytB) for each analyzed population

Full size image

Laboratory procedures

Partial sequences from two mitochondrial genes, the COI (Cytochrome Oxidase subunit 1) and the CytB (Cytochrome B) were successfully amplified for 60 samples. We used primer the combinations LCO1490/ HCO2198 (Folmer et al. 1994) and CB-L/ TS-I-N (Jermiin and Crozier 1994), respectively (Table S1). PCR amplifications were carried out in a final volume of 25ul containing 0.2 units of FastGene Optima DNA Polymerase blend, 4mM MgCl2, 0.4mM for each dNTP, 0.5uΜ for each primer, 1X FastGene Optima Buffer and approximately 50ng DNA. COI protocol included an initial denaturation at 95 °C for 3 min, 35 cycles of 95 °C for 30 s, primer annealing at 52 °C for 1 min, primer extension at 72 °C for 90 s and a final extension of 72 °C for 10 min. Cytb protocol included an initial denaturation step at 95 °C for 2 min, followed by 30 cycles of denaturation at 95 °C for 45 s, annealing at 47 °C for 30 s, primer extension at 72 °C for 60 s and a final extension for 4 min at 72 °C. PCR products were purified using the NucleoSpin ExtractII (Macherey-Nagel) cleanup kit and single strand sequencing was conducted by CEMIA-Cellular & Molecular Immunological Applications, Larisa, Greece.

The AFLP genotyping was performed according to a modified protocol from Vos et al. (1995), for 57 samples collected from five chosen sampling sites around Lake Pamvotis. Total genomic DNA was digested using two restriction enzymes TaqI and EcoRI (Takara Bio Inc.) and the genetic profile was generated using a combination of 10 different primers according to Whitlock et al. (2006, Table S1). Digestion was carried out in a final volume of 20ul containing 1mM TaqI buffer, 0.1mM BSA, 0.5U of EcoRI and 200ng of genomic DNA for 2 h at 37 °C. After the addition of 0.5U TaqI, samples were incubated for two more hours at 65 °C. Ligation was carried out in a final volume of 30ul containing 1X ligase buffer (Takara Bio Inc.), 1.8 μm of each adaptor, 1U T4 DNA ligase and the digested DNA, and the reaction was incubated overnight at 16 °C. The digested-ligated DNA fragments were diluted 25-fold to be used as templates for the pre-amplification reaction in a total volume of 50ul. Pre-selective reactions contained 1X Taq buffer, 1.5mM MgCl2, 0.2mM of each dNTP, 0.3µΜ of each primer (T01P2, T02P2), 0.3 µΜ of Presel ECO primer and 10 µl of the diluted ligation product. The PCR amplifications were carried out using the following profile: an initial denaturation step at 94 °C for 120 s followed by 20 cycles of 30 s at 94 °C, 60 s at 56 °C, and 60 s at 72 °C with a final extension step at 72 °C for 300 s. Pre-amplification products were diluted 25-fold to be used as template for the selective amplification. The selective amplifications were performed in a total volume 25ul containing 10 mM Taq buffer (Takara Bio Inc.), 3mM MgCl2, 0.3mM of each dNTP, 0.2 μm of EcoRI primers, 0.5 μm of 5 selective primers (T101P2, T105P2, T106P2, T204P2 & T205P2) and 5ul of diluted pre-amplified DNA. Selective amplification was carried out using a touchdown protocol with an initial denaturation step at 94 °C for 120 s, 30 s at 94 °C, 60 s at 65 °C followed by 11 cycles where the annealing temperature was gradually reduced 0.7 °C per cycle followed by 23 cycles of 30 s at 94 °C, 30 s at 56 °C, and 60 s at 72 °C. Primers used in each pre-selective and selective PCR reactions are presented in Table S2. Selective products were separated in Fragment Analyzer (Advanced Analytical Technologies Inc.) using the dsDNA 910 Reagent Kit and AFLP patterns were visualized and processed using the PROsize 2.0 Software (Advanced Analytical Technologies). AFLP profiles were scored according to the presence/absence of peaks, simultaneously by two persons to reduce scoring bias. Individuals with more than 10% missing data were discarded from further genetic analysis, resulting in a total of 53 individuals.

Genetic analyses

Mitochondrial sequences were edited by eye in MEGA v.11 (Tamura et al. 2021) and aligned with CLUSTAL W (Thompson et al. 1994). Haplotype and nucleotide diversity of the sequenced individuals were calculated in DnaSP v.5 software (Librado and Rojas 2009). Sequence divergence (p-distance) was estimated using MEGA v.11 through 1,000 bootstrap replications. To visualize the relationships among the detected haplotypes, a Median Joining (MJ) network (Bandelt et al. 1999) was constructed with the software PopArt (Leigh et al. 2015). A relationship among genetic and geographic distance measures was investigated using Mantel tests performed with the “mantel.randtest” function in R package “ade4” (Dray and Dufour 2007). Genetic and geographic distance matrices were created using the average number of base pair differences and latitude and longitude coordinates, respectively.

Inferences of the demographic history of the population in the study area were based on mismatch distribution analyses (Rogers and Harpending 1992) using DnaSP v.5.10.01 (Librado and Rozas 2009) and Arlequin v. 3.5.2.2 (Excoffier and Lischer 2010), while Tajima’s D (Tajima 1989) and Fu’s Fs (Fu 1997) tests for departures from neutrality were also calculated.

AFLP data were used to estimate Nei’s (1973) gene diversity and levels of genetic differentiation between the sampling populations. We employed a Bayesian approach with non-uniform prior distribution; 1,000 bootstrap replicates were perfomed in the software AFLP-SURV (Vekemans 2002), under the methodology of Lynch and Milligan (1994). In addition, population differentiation was also estimated using a Bayesian approach as implemented in the software Hickory v. 1.1 (Holsinger 2003). All models were analyzed with default parameters (burn-in = 5000, sample = 25,000, thin = 5), while the best model was based on the Deviation Information Criterion (DIC). We performed a clustering analysis with Structure 2.3.3 (Pritchard et al. 2000) using 200,000 burn-in iterations and 500,000 MCMC under the admixture model using the population origin of each individual. Ten replicates for K = 1 to 6 were run and the best K was evaluated following the method of Evanno et al. (2005) in Structure Harvester (Earl and vonHoldt 2012). Secondly, in order to infer spatial population genetic structure, we incorporated a Bayesian Markov Chain Monte Carlo (MCMC) method using Geneland (Guillot et al. 2005) in R 3.4.1 (2022). Geneland uses genotypic data and geographical coordinates to determine the most likely number of gene pools and the probabilities of each individual belonging to each of them. Five independent MCMC runs were performed using 200,000 iterations with 100 thinning intervals with K free to vary, to infer the optimal value for this parameter as suggested in Guillot et al. (2005). All replicates indicated an estimate of K = 3. Five independent MCMC runs were performed with the same parameter settings, using the correlated allele frequencies model and setting Kmax = 3. The run with the highest probability was chosen for post-process analysis. The modal population of individuals and the probability of membership in each cluster were computed.

Results

Genetic diversity

The analysis of 60 sequences for the COI gene resulted in eight haplotypes [H1-H8; Accession numbers: OR198168 (H1), OR198169 (H2), OR198170 (H3), OR198171 (H4), OR198172 (H5), OR198173 (H6), OR198174 (H7), OR198175 (H8)], with haplotype diversity ranging from 0 to 0.833 (HD index) and nucleotide diversity ranging from 0 to 0.002 (Pi index) (Table 1). Haplotype H1 was the most frequent and was found in all eight populations (Fig. 1). Five private haplotypes were recorded (Table 1), distributed in four populations (1, 2, 3 and 8), while two private haplotypes were observed in population 3 (Table 1). The analysis of 31 sequences for the CytB gene resulted in four haplotypes [H1-H4; Accession numbers: OR209173 (H1), OR209174 (H2), OR209175 (H3), OR209176 (H4)], with haplotype diversity ranging from 0 to 0.500 (HD index) and nucleotide diversity ranging from 0 to 0.004 (Pi index) (Table 1). Three private haplotypes were observed, in three different populations each (3, 5, and 7; Fig. 1).

Table 1 Haplotypes and levels of genetic diversity for each population. Ν: number of samples sequenced. n: number of individuals exhibiting each haplotype (HX), HD: haplotypic diversity, Pi: nucleotide diversity, SD: Standard Deviation. Private haplotypes are shown in bold
Full size table

The Median Joining networks for each gene revealed a shallow star-like structure, with haplotypes differing by one or two nucleotide bases (Fig. 2). Mismatch distribution analysis and neutrality tests support a sudden expansion of the species in the study area (Table S2, Fig. S1).

Fig. 2
figure 2

Median Joining haplotype networks for the two mitochondrial genes. Vertical lines correspond to the number of differences (mutations) among the observed haplotypes. The spatial distribution of haplotypes is presented in Fig. 1

Full size image

Regarding the AFLP genotyping, the primer pairs resulted in 74 AFLP fragments with an average of 84.8 ± 4.7% of the fragments being polymorphic at the level of 5% within the populations (Table 2). Levels of expected heterozygosity under Hardy-Weinberg genotypic proportions, Hj (Nei 1987), ranged from 0.26 to 0.32, with a mean of 0.29 (± 0.01) (Table 2). Similar results were obtained from the Hickory analyses as shown in Table 2, regarding the levels of genetic diversity within each population of C. lacustris.

Table 2 Levels of genetic diversity of C. lacustris based on AFLPs. N: number of individuals, Hj: Nei’s genetic diversity, SD: Standard Deviation, Hs: genetic diversity within population, P %: percentage of polymorphic loci
Full size table

Spatial patterns of genetic differentiation

Genetic p-distance between the sample localities ranged between 0.02 and 0.16% and 0.25-1% for the genes COI (Table S3) and CytΒ (Table S4), respectively. No significant association between genetic and geographic distances were found for either of the mitochondrial genes (COI: slope = 0.52, p = 0.1, CytB: slope = 0.14 p = 0.5).

Levels of genetic differentiation based on the AFLP genotyping were found to be low to moderate between populations (Fst ranging between 0.007 and 0.025; Table S5). We observed low but statistically significant genetic differentiation for the total of the five populations in Ioannina basin (Fst = 0.01, p = 0.008 and Hickory Gst = 0.032, p = 0.008). Structure analysis based on a Bayesian approach, revealed two genetic clusters based on the five chosen populations around Lake Pamvotis (Table 2).Individuals from populations 6, 7 and 8 were included in the first group, while individuals from populations 3 and 4 clustered together in the second group (Fig S2). Geneland analysis suggested the presence of three clusters in the study area (Fig. 3, S3). Probability of membership of each individual (values > 0.5) revealed three clusters separating the populations 3 and 4, while the populations 6, 7 and 8 are clustering together in space (Fig. 3).

Fig. 3
figure 3

Spatial output from Geneland for K = 3. a) Map of estimated posterior probability of population membership. Numbers correspond to populations in Table 1. b-d) Plots representing the assignment of pixels and posterior probalities to belong to each of the three genetic clusters. The highest membership values are in white and light yellow and the contour lines indicate the spatial position of genetic discontinuities between populations

Full size image

Discussion

Genetic diversity

Genetic diversity is closely linked to the evolutionary and adaptive potential of species (Allendorf and Luikart 2009), and its evaluation is essential for their conservation, especially for species that are characterized as endangered. In the present study, we investigated levels of genetic diversity of the steno-endemic and Critically Endangered (CR) Chorthippus lacustris, by employing both mitochondrial and nuclear genetic markers, and revealed moderate to high levels of genetic diversity within populations. The analysis based on the two mitochondrial genes, showed increased haplotypic diversity coupled with low nucleotide diversity indicating small differences between the observed haplotypes. This is also evident from the haplotypic network, which shows a star-like structure with only one or two differences between the haplotypes, indicative of a population expansion arising from a small effective population size (Rogers and Harpending 1992; Rogers 1995; Grant and Bowen 1998).

Sudden population expansion affects levels of genetic diversity, and the relationships among haplotypes, resulting in closely related haplotypes as we observe in the present study. The overall negative values resulting from the neutrality tests further support the population expansion model, probably during the interglacial periods in Pleistocene in the southern refugia of Balkan peninsula (Hewitt 1996; Vedenina and Helversen 2009). However, small populations with small generation lengths might lose genetic diversity at an accelerated rate, compromising their overall fitness and viability (Frankham et al. 2010). The localization of different haplotypes in several focal populations of C. lacustris is likely due to previous bottlenecks and fragmentation of a formerly robust population, with subsequent local isolations in favorable habitats.

Spatial patterns of genetic differentiation

There is an apparent subdivision of populations around Lake Pamvotis into three different genetic clusters, however the genetic distances between them were generally low. The genetic discontinuity found between populations 3 and 4 is congruent with the location of Lake Pamvotis that acts as a probable barrier between them (Figs. 1 and 4). Populations 6, 7 and 8 are differentiated from populations 3 and 4, which could reflect their spatial arrangement, since these populations are located on the opposite sites of a main highway (Figs. 1 and 4). Transport infrastructures represent barriers to the movements of several insect species, and further increase isolation of populations in the adjacent habitats (Vandergast et al. 2009; Muñoz et al. 2015).

The observed genetic structure of the Epirus grasshopper might indicate a recent restriction of connectivity due to habitat fragmentation and significant changes in land use, leading to ongoing inhibition of gene flow (Kati et al. 2006). Sufficient gene flow may act as a buffer against the effects of geographical subdivision and isolation, especially in flying insects that can disperse or migrate (Brower and Boyce 1991; Estoup et al. 1996; Freeland et al. 2003).

Although little is known about the dispersal capabilities of the Epirus dancing grasshopper, grasshoppers of the genus Chorthippus can disperse from 30 m per generation (Bailey et al. 2003; Tim and Julian 2004) up to 1,344 m through suitable habitat patches (Bridle and Butlin 2002). Even though, dispersal capabilities in insects are generally correlated with increased gene flow, contradicting results have also been observed where mainly sedentary insect species show weaker isolation by distance and decreased genetic differentiation between populations than more mobile species (Gill 1981; Peterson and Denno 1998; Blanchet et al. 2012). For example, surprisingly little genetic structure was found during a landscape study on a flightless cricket, the Pholidoptera griseoaptera (Diekötter et al. 2010). Indeed, dispersal ability is species-specific, depending on wing-morphology, body size, the degree of sexual size dimorphism and even habitat requirements (generalist vs. specialists) (Reinhardt et al. 2005; Wiesner et al. 2011).

Furthermore, patterns of genetic variation and structure are also affected by the historical and present landscape configuration, the variability of landscape features, the presence of habitat patches between population localities that might act as suitable settlements for dispersing insects and/or life history trade-offs (Hanski et al. 2006; Holzhauer et al. 2006; Ortego et al. 2012; Keller et al. 2013). Here, the spatial genetic analysis indicates fragmentation due to landscape features (Lake Pamvotis and highway roads), but low genetic differentiation between the focal populations. For many Orthoptera species inhabiting floodplains, environmental disturbances might either lead to high active dispersal and emigration to suitable breeding sites or high reproduction rates coupled with passive dispersal of many offsprings (Hopper 1999; Simmons and Thomas 2004; Dziock et al. 2011). However, increased habitat fragmentation could lead to high mortality of mobile species searching for suitable areas (Reinhardt et al. 2005).

Conservation implications

The present study marks the first attempt to describe the genetic diversity and population structure of the Critically Endangered Chorthippus lacustris. Despite the changes in land use and urbanization around the Lake Pamvotis in the last decades, the species maintains moderate to high levels of genetic diversity within populations, although this may be a transient state from the recent disturbances. However, some signs of genetic differentiation between the sampled populations could be identified, pointing to some kind of restriction in population functional connectivity.

The limited distribution and specific habitat requirements of the Epirus dancing grasshopper might render its survival sensitive to habitat loss and fragmentation. Hence, the conservation of local populations focusing on retaining current levels of genetic diversity is fundamental for the species. Management efforts should focus on the preservation of favorable habitats and breeding sites and minimize the effects of land use changes in order to ensure the survival of the species.

Further field research for possible other remnant populations in the area and assessment of dispersal movements might shed light on the current distribution and the levels of connectivity among suitable habitats for the species.