Obtaining estimates of population size and connectivity is a critical step within conservation research, however, for many threatened amphibians this is a difficult task due to the cryptic nature or rarity of such species. In this study of L. littlejohni, where we were unable to obtain sufficient recaptures to estimate population sizes through traditional mark-recapture methods, however, we were able to estimates effective population size in five genetically distinct populations using genetic methods. Additionally, we revealed the restriction of gene flow at the regional and local scale. This additional knowledge would be unlikely obtained through standard capture-recapture methods. Two population clusters are clearly geographically and genetically isolated, but the other three are located on the same plateau. Climate and geography were not strong drivers of genetic isolation. High levels of inbreeding, small population size and reduced genetic diversity were detected in the two geographically isolated populations and one of the populations on the Woronora plateau. These results provide strong evidence that the rarity of this frog is likely due to small population sizes rather than low detection rates and urgent conservation action is needed to ensure the persistence of this threatened frog.
Population Structure:
When comparing the major sampling regions, strong genetic isolation was observed in L. littlejohni. Based on the results of the pRDAs, we suggest that genetic drift due to population bottlenecks/founder effects is the main driver of genetic variation across populations, rather than large climatic difference or geographic distance between populations (Capblancq & Forester 2021; Coleman et al. 2013). While Isolation by Distance was significant, the partial RDA indicated that only a small percentage of genetic variation could be attributed to geography. In the partial RDA models, population genetic structure was the strongest predictor of genetic variation indicating that demographic history rather isolation by adaption or isolation by distance have driven patterns of genetic variation in L. littlejohni (Capblancq & Forester 2021). However, these variables were strongly confounded and over 65% of variation in the full model could not be exclusively attributed to climate, geography or genetic structure alone. This suggests any future work to investigate loci under selection or conduct further landscape genetic analysis in L. littlejohni must consider ways to account for this collinearity (Frichot et al. 2015; Hoban et al 2016; Capblancq & Forester 2021).
Increased genetic structuring of populations has been linked to disease-related population bottlenecks in numerous vertebrate species. In these species, population bottlenecks led to increased population fragmentation and exacerbated genetic drift causing populations to become more genetically differentiated (Phillips et al. 2020; Serieys et al. 2014; Lachish et al. 2011). Both the Central Coast Range and Blue Mountains have effective population sizes below 50, high inbreeding values and lower genetic diversity compared to other L. littlejohni populations, indicating that these populations have indeed suffered from population bottlenecks. Between these two regions are largely continuous natural areas consisting of seemingly suitable habitat where there were historical sightings of L. littlejohni, thus these regions were once connected (Mahony et al. 2020). Due to a lack of pre-chytrid genetic samples, we cannot confirm chytrid as the driver of L. littlejohni declines in this study, although this disease is the most likely driver of sudden decline in this winter active, stream and pond frog (Hero et al. 2005). However, it seems likely that these two populations have become further isolated both geographically and genetically due to disease driven population declines with the resulting small population sizes exacerbating the consequences genetic drift.
Large scale damming and population crashes due to disease may explain the strong genetic isolation between the Blue Mountains and Woronora Plateau populations. Damming of riverine systems alters population connectivity by fragmenting suitable habitat (Heggenes & Roed 2006; Emel & Storfer 2012; Hansen et al. 2014; Werth et al. 2014). In other frog species, river regulation has led to changes in frog species composition, increased genetic isolation, and loss of genetic diversity (Naniwadekar & Vasudevan 2014; Peek et al. 2021, Grummer & Leache 2017). We hypothesize that the construction of the Warragamba Dam in the 1960s may have disrupted gene flow between the Blue Mountains and Woronora populations. The Warragamba dam creates Lake Burragorang, however, L. littlejohni does not occupy large lakes. Rather, this frog is typically associated with first order streams and small fire dams (Mahony et al. 2020, Lemckert 2010). Thus, the lake may act as barrier due to unsuitable habitat.
The degree of genetic structuring observed on the Woronora Plateau was unexpected as the landform is contiguous and there is little urban development beyond major roads. This region has the largest population of L. littlejohni, thus it was anticipated that gene flow would occur between the sampled river catchments. Instead, we detected three genetic clusters with low to moderate genetic differentiation (O’Hares, Cataract and Cordeaux; Fig. 2). Distinct genetic structure at a local scale is typically seen in species that have high site fidelity and inherently low dispersal abilities (Zickovich & Bohonak 2007; Richardson et al. 2020), however, we do not think this is the case for L. littlejohni. The geographic size of the Cordeaux cluster (> 84km2) indicates that L. littlejohni is not limited by its ability to disperse across substantial distances. We propose three factors that may explain the fine-scale structuring of L. littlejohni populations: 1) roads that prevent movement between catchments; 2) dispersal is restricted to creek lines; and 3) habitat disturbance that has isolated populations locally. As discussed with the other populations, the genetic structuring may also be exacerbated by general distribution-wide declines due to chytrid.
Roads (1) restrict movement of a variety of vertebrate species, leads to genetic isolation of populations (Hale et al. 2013; Holderegger & Di Giulio 2010; Serieys et al. 2014; Youngquist et al. 2017). For example, major urban infrastructure (housing and dual carriage roads) has been linked to the genetic sub-division of southern bell frog (L. raniformis) (Hale et al. 2013), and the presence of highways is predictive of genetic distance in Blanchard’s cricket frogs (Acris blanchardi) (Youngquist et al. 2017). The Woronora Plateau has several large dual carriageway roads which traverse it (Picton Road and Appin Road) (Fig. 1), which are possible barriers to dispersal in L. littlejohni.
Dispersal primarily along creek lines (2) rather than through terrestrial landscapes may contribute to isolation of populations. For some aquatic and semi aquatic species, patterns of gene flow reflect the hierarchical nature of river systems and/or the isolation of populations to independent river basins or drainages (Brauer et al. 2016; Coa et al. 2020; Fonseca et al. 2021). Whilst the river catchments on the Woronora Plateau are geographically close to each other, they are parts of different major river drainages (Fig. 1). The Cordeaux Cluster and Cataract Catchment flow into the Hawkesbury-Nepean River system, while the O’Hares Catchment flows into the Georges River/Tucoerah River. O’Hares and the Cataract site is essentially the border of these two major river catchments. Although we did detect movement between Cataract and O’Hares sites, the moderate genetic differentiation of these two clusters indicates movement is not common despite the short distance between them (less than 5 km between Cataract and closest O’Hares site). As L. littlejohni tadpoles are aquatic, larval dispersal along creeks (particularly during high rain periods) is highly probable. Additionally, during our surveys we often encountered adult frogs utilizing the sandstone bedrock of creek lines to move between water bodies. We only detected a frog more than thirty metres from water once, despite accessing sites by walking. There have been reports of L. littlejohni found under rocks away from streams during other fauna surveys (G. Daly unpubl. data), thus some terrestrial movement does occur, but it is suspected to be limited.
In addition to roads, the main habitat disturbance (3) occurring in the Woronora is longwall coal mining. Longwall mining can lead to the diversion of water underground due to subsidence causing cracks in the surface rock (Booth 2006). The Cataract River Catchment has been undermined extensively, however, to what degree this mining has impacted frog species is unclear due to a lack of research (WaterNSW 2016). In the last ten years, sightings of L. littlejohni within Cataract have been limited in number and are restricted to small pockets within the catchment. It is plausible that mining activity has fragmented populations locally as drying of creeks may result in limited successful dispersal events. However, longwall mining also occurs within the Cordeaux Cluster, which has no population sub-structuring, thus there is no clear trend as to how creek drying due to mining impacts frog movement.
Small populations and Inbreeding:
The small effective population size across all L. littlejohni populations indicate that the rarity of this species is indeed due to historical declines and not just strict habitat preference or cryptic behaviour. These estimates are also of conservation concern as the fate of fragmented populations is determined by the effective population size of the isolated population, rather than the species as a whole (Frankham et al. 2017). While the effective population sizes only reflect that of the sampled areas, as we sampled a large proportion of known currently occupied sites, thus it is unlikely that these estimates would change dramatically with additional samples. The three smallest populations (Central Coast range, Blue Mountains, and Cataract) also have higher inbreeding coefficients for both F and FIS, lower observed heterozygosity and higher mean kinship than the O’Hares and Cordeaux clusters. These factors combined with small effective population size imply that these populations are at risk of reduced adaptive potential, inbreeding depression, and increased sensitivity to stochastic events (Frankham et al. 2017). Additionally, small population sizes in conjunction with high inbreeding puts populations at risk of mutational meltdown and/or genetic purging (Keller & Waller 2002). Inbreeding can act swiftly whilst mutational meltdown and genetic purging are more often long-term threats to small populations (Keller & Waller 2002).
The coefficient of inbreeding recommended by Frankham et al. (2017), F, uses identity by descent methods to estimate cumulative inbreeding over generations (Frankham et al. 2017). On the other hand, FIS is used more widely in population genetic research, but only detects deviations from random mating in recent generations. In the present study, we have populations in three different states (1) High F and high FIS, (2) Low F and low FIS, and (3) High F but low FIS. The Cordeaux cluster is in State 1 (Low F and FIS) and has the largest population size, indicating that this population may not have faced large population declines or the population maintained a large enough size to prevent changes in random mating patterns. The Central Coast Range, Blue Mountains and Cataract are all in State 2 (High F and FIS) and have small population sizes, indicating that these populations have declined and not recovered. The final population, O’Hares, is in State 3 (High F, but low FIS) and has a small but larger population than those in State 2. The difference in inbreeding values may indicate that the current population is not deviating from random mating, but that in the past the population was small enough for inbreeding to be present. Thus, O’Hares likely experienced decline and has recovered to some extent. There are several examples of chytrid susceptible species that have declined in some parts of their range but persist or have recovered in others (Osborne et al. 1996; Retallick et al. 2004; Wassen 2008; Hamer et al 2010; Mahony et al 2013; Newell et al. 2013; Lips 2016). Habitat quality, changes to disease dynamics, recruitment rates and population connectivity have all been linked to stabilizations and recoveries of frog species (Scheele et al. 2014; Mc Donald et al. 2005; Phillott et al. 2013; Scheele et al. 2017; Mc Knight et al. 2019). As L. littlejohni populations appear to differ in recovery history they may provide a useful model for further investigation into drivers of recovery.
Management Recommendations:
Through the application of genetic methods, conservation managers will be able to make decisions for L. littlejohni with more clarity going forward. The small effective populations, high inbreeding, reduce genetic diversity and high mean kinship detected in some L. littlejohni population indicates that urgent on-ground action is needed to increase population sizes. Determining habitat preferences and protecting vital breeding habitat will be important for supporting successful recruitment within populations. Recommended research includes assessing whether inbreeding depression is present in populations with high Fis and F. Inbreeding depression in other frogs has been linked to reduced sperm quality and reduced offspring survival (Hinkson & Poo 2020; Anderson et al. 2004), which obviously has implications for population viability of threatened species. Additionally, we recommend assessing whether assisted gene flow will aid in genetic rescue or lead to outbreeding depression (Frankham et al. 2019). The Cordeaux and O’Hares clusters are recommended as potential source populations as they contain the highest genetic diversity, lowest inbreeding, and largest populations. Additionally, the metrics for coefficient of inbreeding and mean kinship have both been used in other species to assess or plan translocations (Cowen et al. 2021; Farquharson et al. 2021), thus the values reported here provide a baseline for future conservation management.
Global Significance/Conclusions:
This study demonstrates the power of pairing genetic methods with traditional surveys to study rare and difficult to track species. Without employing genetic methods in the present study, we could have not resolved the issue of low sampling rates and low recapture rates that prevented successful mark-recapture methods. Additionally, the SNP data provided insight into potential inbreeding depression, which highlights an urgent need for conservation actions and further research, that would not have been detected through visual encounters surveys alone. Incorporating genetics into conservation research can have a profound impact both our understanding and our ability to provide adequate management of threatened species.