Applying novel connectivity networks to wood turtle populations to provide comprehensive conservation management strategies for species at risk

Cindy Bouchard; Étienne Lord; Nathalie Tessier; François-Joseph Lapointe

doi:10.1371/journal.pone.0271797

Abstract

Genetic diversity within and among populations is frequently used in prioritization processes to rank populations based on their vulnerability or distinctiveness, however, connectivity and gene flow are rarely considered within these frameworks. Using a wood turtle (Glyptemys insculpta) population graph, we introduce BRIDES as a new tool to evaluate populations for conservation purpose without focusing solely on individual nodes. BRIDES characterizes different types of shortest paths among the nodes of a subgraph and compares the shortest paths among the same nodes in a complete network. The main objectives of this study were to (1) introduce a BRIDES selection process to assist conservation biologists in the prioritization of populations, and (2) use different centrality indices and node removal statistics to compare BRIDES results and assess gene flow among wood turtle populations. We constructed six population subgraphs and used a stepwise selection algorithm to choose the optimal number of additional nodes, representing different populations, required to maximize network connectivity under different weighting schemes. Our results demonstrate the robustness of the BRIDES selection process for a given scenario, while inconsistencies were observed among node-based metrics. Results showed repeated selection of certain wood turtle populations, which could have not been predicted following only genetic diversity and distinctiveness estimation, node-based metrics and node removal analysis. Contrary to centrality measures focusing on static networks, BRIDES allowed for the analysis of evolving networks. To our knowledge, this study is the first to apply graph theory for turtle conservation genetics. We show that population graphs can reveal complex gene flow dynamics and population resiliency to local extinction. As such, BRIDES offers an interesting complement to node-based metrics and node removal to better understand the global processes at play when addressing population prioritization frameworks.

Citation: Bouchard C, Lord É, Tessier N, Lapointe F-J (2022) Applying novel connectivity networks to wood turtle populations to provide comprehensive conservation management strategies for species at risk. PLoS ONE 17(8): e0271797. https://doi.org/10.1371/journal.pone.0271797

Editor: Angela Marie White, USDA Forest Service, UNITED STATES

Received: August 25, 2021; Accepted: July 7, 2022; Published: August 12, 2022

Copyright: © 2022 Bouchard et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: Data Availability Statement: The R and C++ source codes are available from the Github repository (https://github.com/etiennelord/BRIDES/) with a LGPL-3.0 License. The primary data underlying these analyses will be deposited in DRYAD.

Funding: This work was supported by a Discovery Grant from the Natural Sciences and Engineering Research Council (NSERC Canada) [OGP0155251] to François-Joseph Lapointe. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

The conservation of units below the species level, commonly populations, is widely used in wildlife management, as it provides critical information to support species survival and evidence of within-species differences. Populations show important local adaptive variations that are essential for species adaptation to changing environments [1–3]. Furthermore, they exhibit greater sensitivity to extirpation, which may inform of the general trend at the species level [4]. The threats experienced by animal and plant populations require distinct management practices that must consider the economic, social and cultural aspects of the human society living in closest proximity to them [5–7]. Rule-based methods have been developed by the International Union for Conservation of Nature (IUCN) to categorize populations based on the level of threat they face [8–11]. Such multiple criteria have been used to categorize species at risk and aid governmental and non-governmental organizations to define conservation units, including: Management Units (MUs, [2, 12]), Evolutionary Significant Units (ESUs, [2]) and Designable Units (DUs). The latter is used by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) to define populations or subspecies of wildlife that require protection under the Species at Risk Act (SC 2002, c 29).

Considering that two populations may be ranked in the same category even when faced with entirely different threats, and the limited financial resources available for conservation biology, prioritizing populations arose as an essential concept to target the most effective conservation actions. Multiple prioritization frameworks have been developed to provide objective criteria during decision-making processes, and either estimate species extinction risks [5] or prioritize threatened populations and determine the management actions required to support them [13]. Prioritization frameworks are often reactive, targeting highly vulnerable populations in need of immediate management actions to avoid reducing genetic diversity and local or global species extinction events [14]. However, for species with long generation times, it may be beneficial to react in a proactive fashion and prioritize populations that exhibit low vulnerability to protect their global genetic diversity as a reservoir [15–17].

Population viability analysis (PVA) is frequently used to evaluate extinction risks given its comprehensive and quantitative basis [18]. To complement such models, or, in cases where data are unsuitable, the relative importance of each population can also be estimated to measure the impact of local extinction on species survival. A series of binary questions have been proposed by Allendorf and colleagues [19] for the Pacific salmon (Oncorhynchus spp.) and adapted for all freshwater fish by Clarkson and colleagues [20], thus providing a scoring system for the biological consequence of local population extinctions for the entire species. Although these rankings are intended to represent objective criteria, the list of questions may overlook important features that deserve conservation attention.

Measures of genetic diversity within and among populations are frequently applied to prioritize populations and define conservation units based on past bottleneck events, effective population size and population distinctiveness [21, 22]. To bridge the gap between researchers and practitioners, Ottewell and colleagues [22] developed a simple framework using population differentiation, genetic diversity, and inbreeding coefficients. The quantification and ranking of these parameters can be of great importance to estimate population isolation, predict their trends and target the most informative genetic processes for management purposes. Although many of these estimators allow to understand the distinctiveness and diversity of a population to predict its persistence, only few consider the dynamic structure underlying population networks. Some demographic studies have accounted for population connectivity by using a metapopulation model in their prioritization framework. However, the data requirements of complex metapopulation models greatly limit their application in conservation studies, calling for a short-term decision-making time frame. For example, the incidence function model of metapopulation dynamics [23] requires estimates of patch colonization and extinction rates that are obtained through longitudinal surveys of all available habitat patches [24]. Yet, connectivity among isolated habitat patches is essential to several ecological processes and should be considered in population prioritization models. Seasonal migration, dispersal of offspring, recolonization following local extinction events, range shifts caused by climate change, and gene flow for the transmission of favourable alleles may all be distorted if change occurs within the connectivity pattern [7, 25].

Gene flow is frequently cited as an evolutionary force that maintains connectivity among populations [26, 27], but connectivity, in itself, can potentially operate at different time scales other than demographic processes, such as dispersion and recolonization events [28, 29]. As such, gene flow should also be considered in population prioritization as it offers complementary insights into population connectivity [28, 30]. Theoretical and empirical studies support the use of gene flow as a measure of functional connectivity [29, 31–33], and graph theory is a valuable tool emphasizing the relevance of connectivity in conservation planning [34]. The processes underlying the structure of complex networks have been efficiently characterized in many disciplines [35] and are increasingly applied in ecology and conservation where food webs, metapopulation dynamics or reserve networks can be modelled [36]. More precisely, population graphs have been developed by Dyer and Nason [37] as a network depicting populations (the nodes of the graph) connected by their genetic covariance (the edges of the graph). Gene flow has since been modelled for many plant and animal species in marine, freshwater, and terrestrial ecosystems using networks [38–40]. For management purposes, population prioritization can be facilitated by the ranking of several node-based centrality indices [41], or by using node removal to assess their relative impact on network connectivity [29]. Population prioritization has also been studied with phylogenetic networks [42], for which genetic distinctiveness was used to score each population. Yet, the choice of relevant network metrics is crucial for the prioritization process using such methods [43–45].

The wood turtle (Glyptemys insculpta) is a threatened freshwater turtle species endemic to North America and listed as endangered according to the IUCN red list, following an overall decrease in population size [46]. It is a semi-terrestrial, long-lived species with longevity that may exceed 50 years [47], delayed sexual maturity at 11–22 years [48, 49], and a long generation time estimated at 36–47 years [46]. Canadian wood turtle populations are protected under the Species at Risk Act, but financial support is lacking to ensure the persistence of all populations. In cases where wildlife managers must decide which populations to protect, a sound prioritization framework is required. Yet, considering the cost and short duration of radiotelemetry studies, the temporal and spatial variation in nest predation, and the difficulty of monitoring hatchlings and juveniles, PVA is not easily applicable to wood turtle populations [50]. Population genetics have been previously used to characterize the population structure of wood turtles [51–54] and the importance of peripheral populations for genetic diversity [55]. In addition, landscape genetics has revealed the importance of the watershed structure for wood turtle population differentiation in Canada [56]. None of these previous studies, however, have relied on graph theory as a statistical tool to analyze connectivity and prioritize populations.

Although many studies highlight the importance of connectivity and gene flow for the maintenance of diversity and the persistence of populations in a fragmented landscape, few frameworks consider connectivity as an important feature of the prioritization ranking. In this paper, we apply BRIDES [57] as a new tool to evaluate whole population graphs for conservation purposes, rather than focusing only on individual patches. Thus, the main objectives of our study are (1) to demonstrate the use of the BRIDES selection algorithm to assist conservation biologists and managers during decision-making processes, and (2) to use graph theory to assess gene flow among wood turtle populations. To do this, we used a dataset of 19 wood turtle populations collected in eastern Canada and genotyped by Bouchard and colleagues [56]. Centrality indices and node removal statistics were first estimated from a population graph. We then constructed six population subgraphs based on three different criteria to protect populations in a proactive or reactive fashion, and using BRIDES, selected the optimal number of additional nodes required to maximize network connectivity under different weighting schemes. The results of the three methods were then applied within the perspective of wood turtle conservation.

Materials and methods

The sampling protocol was approved by the Ministère des Forêts, de la Faune et des Parcs du Québec.

Wood turtle population characterization

The original data set from Bouchard and colleagues [56] included 24 sites, but five sites were excluded due to insufficient sample size (N<6). As a result, 327 wood turtles sampled at 19 different sites and genotyped for nine microsatellite loci as described in Bouchard and colleagues [56] were used in the present study. Using microsatellite data, expected heterozygosity (H_E) was estimated using the diveRsity package with R version 3.5.1 [58]. Element occurrence (EO) ranks were used to estimate the persistence probability of a population for a defined period of time (20–100 years), while considering that current environmental conditions prevailed during that time [59]. These EO ranks vary from excellent viability (A) to poor viability (D), and some rank combinations can be used in cases of uncertain estimates. All EO ranks and estimated population sizes were provided by wildlife biologists at the Québec Ministry of Forests, Wildlife and Parks. Although other measures could have been used to demonstrate the application of population graphs, H_E and EO ranks were selected for their frequent use in conservation studies.

Wood turtle population graph

We constructed a population graph based on Dyer and Nason’s [37] conditional genetic distance (cGD) and pruning method using R v. 3.5.1 [60] popgraph [61] and gstudio [62] packages. In this network, nodes representing different sample sites are connected to each other by undirected edges weighted by the genetic distances between them. However, as some edges do not adequately describe overall among-population distances, the network was pruned by removing redundant edges that did not contribute to the overall genetic covariance of the network, as described by Dyer and Nason [37].

Node-based metrics of connectivity and node removal

To evaluate the importance of each node with respect to network connectivity, we used two current methods: node-based metrics [45, 63] and the node removal approach [38, 64]. Several node metrics were computed to estimate their contribution to gene flow and importance in graph connectivity. Degree centrality (DC) is defined as the number of edges connected to a node, whereas betweenness centrality (BC) is the number of shortest paths upon which the node lies. Based on immediate neighbour connections, the eigenvector centrality (EC) may also provide information on both direct and indirect levels of node connectivity. The clustering coefficient (CC) is the probability that two nodes connected to a neighboring node are also connected together. Computing strength as the sum of all edge weights connected to a given node is another centrality measure of interest for conservation biology, since the average inverse edge weight (AIEW) is positively correlated with the number of migrants to and from a focal node [63].

The node removal approach was also used to estimate the effect of local extirpation on the connectivity of the network and to obtain an overview of the resilience of the network to node loss [38]. In this study, we simulated the effect of local extirpation by removing a node and its associated edges from the graph and measuring the average path length in the resulting graph. The process was repeated iteratively by removing each node, one at a time. Differences in average path lengths were estimated by comparing the reduced and complete networks, with larger differences being associated with more important nodes for network connectivity. A cut node is a node whose deletion, along with incident edges, results in a disconnected graph. Node-based metrics and node removal statistics were used to estimate gene flow between wood turtle populations and were compared with results of the BRIDES selection procedure described below.

BRIDES analysis

Network connectivity was explored using BRIDES v1.2 [57], a software that allows the characterization of dynamic networks through the comparison of shortest paths among pairs of nodes of a subgraph with their respective paths among the same nodes in a complete graph (Fig 1A). Both the complete graph and the subgraph are defined by the users and selected as input networks for the BRIDES algorithm. When applied to population genetic data, the paths identified by BRIDES may thus represent actual gene flow between populations, or past gene flow between remnant populations or metapopulations relative to neutral genetic markers.

Download:

Fig 1. Illustration of the BRIDES algorithm and the selection process.

(A) Example of the six types of paths computed in BRIDES by comparing the shortest path between pairs of nodes in subgraph X_n (white nodes) and the shortest path between the same pairs of nodes in the complete graph X_all including all candidate nodes (blue nodes). In X_all, the shortest path between any pair of nodes from X_n must include at least one of these candidate nodes. (B) Each of the three nodes available for the selection process was iteratively added to X_n to create the subgraphs X_n+1 and corresponding scores were computed with respect to the complete graph X_all. The selection process selected node 11 as the best candidate (score = 1) to maximize network connectivity while minimizing the number of additional nodes. The weighted model used for the selection process was (1, 0, 0, 0, 1, 1).

https://doi.org/10.1371/journal.pone.0271797.g001

Precisely, BRIDES is a polynomial-time algorithm used to characterize different types of paths between a subgraph X_n and a complete graph X_all, where all nodes in X_n are a subset of the nodes in X_all. The shortest paths between pairs of nodes in X_n are first computed. Then, the lengths of the paths among the same pairs of nodes are recomputed in X_all by forcing these paths to include at least one node present in the complete graph X_all but absent in the subgraph X_n. The comparison of these corresponding path lengths thus allows six types of paths to be distinguished (Fig 1A): a breakthrough (B) is a path that is impossible in X_n but possible in X_all (path 4–6); a roadblock (R) is a path that is possible in X_n but impossible in X_all (path 1–4); an impasse (I) is a path that is impossible in both X_n and X_all (path 5–6); a detour (D) is a path that is shorter in X_n than in X_all (path 6–7); an equal (E) path has the same length in X_n and X_all (path 6–8); a shortcut (S) is a path that is longer in X_n than in X_all (path 6–9). The results are summarized in a vector [B, R, I, D, E, S] containing the numbers of each type of paths estimated by the algorithm (see Fig 1A).

The six different types of paths are meant to capture specific network properties when forcing subgraph paths to pass through additional nodes. With respect to conservation genetic data, shortcuts are indicative of the addition of new populations that will in turn improve gene flow, whereas detours are associated with populations that will reduce gene flow. Breakthroughs are created by populations connecting two (or more) components of the network, similar to cut nodes that represent stepping-stones among isolated populations. Equal paths are corresponding to redundant connections among populations for dispersal and gene flow. In contrast, impasses and roadblocks are detrimental to network connectivity by separating populations in isolated components. These latter types of paths are to be avoided in the prioritization process.

BRIDES selection algorithm

We implemented an iterative selection algorithm in BRIDES to determine which candidate node should be added to X_n to maximize the connectivity of X_n+1 (Fig 1B). Given that each type of path may have a different impact on the choice of candidate nodes, distinct weighted models could be defined by the user for the selection algorithm. For example, one may consider only breakthroughs, equals and shortcuts to maximize network connectivity. In that case, assuming a weight of 1 for each of the selected types of paths, the corresponding model would be (1, 0, 0, 0, 1, 1), also denoted as B¹E¹S¹. At each step of the algorithm, a selection score is then computed as the sum of weighted numbers of path types in the BRIDES vector. The maximum score that can be achieved is estimated by including all candidate nodes in the subgraph X_n. For comparison purposes, this maximal score is set to one, and suboptimal solutions are scaled by dividing their scores by the maximum value (see example in Fig 1B).

Weighted models and conservation scenarios

In the present study, five distinct weighting schemes were used for the six types of paths (B, R, I, D, E, S) to select optimal solutions (maximum scores) under different scenarios, with positive weights (denoted with superscripts) attributed to preferred types of paths, negative weights (denoted with subscripts) attributed to rejected types of paths, and null weights (zero) attributed to types of paths ignored in the final score computation. The first weighted model, S¹ = (0, 0, 0, 0, 0, 1), only considered shortcuts, whereas the second model, B³S¹ = (3, 0, 0, 0, 0, 1), and third model, B¹S³ = (1, 0, 0, 0, 0, 3), also accounted for shortcuts and breakthroughs, but with distinct weights. The fourth model, B¹R₁D₁S = (1, -1, 0, -1, 0, 1), considered the negative impact of roadblocks and detours with respect to breakthroughs and shortcuts. For the fifth model, B³D¹E²S³ = (3, 0, 0, 1, 2, 3), we used an ecological interpretation of BRIDES in which breakthroughs, detours, equals and shortcuts were all considered to increase the connectivity of the network, but with different weights.

In addition to the five weighting schemes, six scenarios encompassing three different parameters important for management purposes were considered based on their frequent use in conservation studies: genetic diversity (H_E), estimated population size, and persistence probability (EO ranks). These scenarios were used to generate different subgraphs X_n onto which candidate nodes were added in turn to construct subgraphs X_n+1. For each parameter, pairs of networks (subgraphs) were generated by selecting nodes with either the largest or smallest values, representing nodes that prioritize a proactive (P) or a reactive (R) conservation perspective, respectively. For the first pair of networks, populations with higher (Scenario A_P) and lower (Scenario A_R) heterozygosity values (H_E) were selected as possible subgraphs for the prioritization process. For the next pair of networks, nodes with higher (Scenario B_P) and lower (Scenario B_R) estimates of population sizes were selected. Finally, EO ranks were used to select nodes for the fifth (Scenario C_P) and sixth (Scenario C_R) networks based on their persistence probability for a period of time. The five weighting models were then applied to the six scenarios to examine their impact on the node selection process.

Results

Population graph

The wood turtle population graph constructed with cGD resulted in 19 nodes and 37 edges (19% of the edges in a saturated network) after the pruning procedure. Eight nodes were located on the north shore and eleven nodes on the south shore of the St. Lawrence River (Fig 2). Sample sizes and estimated population sizes of different sites ranged respectively from 6 to 56 accounting for 6 to 2000 individuals per node (Table 1).

Download:

Fig 2. Population graph of 19 wood turtle populations based on conditional genetic distance (cGD).

Redundant edges were pruned when they did not contribute to the overall genetic covariance structure. Node sizes are scaled according to the average inverse edge weight (AIEW), where greater AIEW values are reflected through proportinaly larger circular diameters. Populations located on the northern and southern shores of the St. Lawrence River are identified with gray and white nodes, respectively. Source: Esri. " Light Gray Canvas Map" [basemap]. https://www.arcgis.com/home/webmap/viewer.html?layers=ed712cb1db3e4bae9e85329040fb9a49 Orignal data are from Bouchard and colleagues [56].

https://doi.org/10.1371/journal.pone.0271797.g002

Download:

Table 1. Sampled population parameters and corresponding node-based metrics, node removal statistics and BRIDES results.

https://doi.org/10.1371/journal.pone.0271797.t001

Network metrics and node removal

The population graph was analyzed with different node-based metrics to determine which sites contributed most to network connectivity (Table 1). One node (GA3) ranked top in all metrics, whereas two nodes ranked at the very bottom (MA1 and MA2). Above all, these results illustrated a clear separation between sites located on the northern and southern shores of the St. Lawrence River, and the importance of only two edges connecting them. Several nodes with high betweenness centrality values may act as stepping-stones or bridges [38]. Namely, four of the nodes with large betweenness values (MI2, LO1, LA1, SF3) connected the two shores together. In contrast, nodes with a high-clustering coefficient may act as an anchor of interconnected groups of nodes. The maximum value of the clustering coefficient was obtained for two nodes on the north shore (SM1 and MD1), whereas four nodes on the south shore had null values. To further identify nodes that contributed to both direct (immediate connection) and indirect (connection to its neighbors) network connectivity, node eigenvector centrality was computed. Once again, our results show opposite patterns for nodes located on distinct shores, with those on the northern shore exhibiting much larger eigenvector centrality values compared to nodes in the south. Finally, AIEW was calculated to provide estimates of the numbers of migrants to and from a node. Nodes with higher values (AIEW > 0.065) were all located on the south shore of the St. Lawrence River (Table 1). Average path length of the complete network was 2.708. While the removal of a single node (MI2) increased the value up to 3.353 (Table 1), the removal of six individual nodes (MA3, MD1, GA2, GA4, MA24, SM1) provided results close to that of the complete graph (ranging between 2.595 and 2.706), meaning less contributions to network connectivity. There were no cut nodes in our population graph.

BRIDES selection procedure

Using the BRIDES selection process, we explored which nodes were the most important for network connectivity using six subgraphs that were constructed from the complete population graph including 19 wood turtle populations under different conservation scenarios. Depending on the criteria selected, these networks contained between 6 and 13 nodes (Fig 3). Networks based on genetic heterozygosity included populations with H_E values greater than (Scenario A_P) or lower than 0.72 (Scenario A_R). Those based on estimated population sizes were created by including populations with more (Scenario B_P) or less than (Scenario B_R) 75 individuals. Networks based on element occurrence included populations with EO ranks greater than BC (Scenario C_P) or lower than B (Scenario C_R) (Table 1). The five weighting models were then used to examine their impact on the node selection procedure and explore BRIDES flexibility under a wide range of experimental conditions. The identity of the selected nodes was very similar (90%) regardless of the weighted model selected for a given scenario (S1 Fig and S1 Table).

Download:

Fig 3. Different population subgraphs based on distinct node selection parameters using a proactive or reactive conservation approach.

A: Node selection based on heterozygosity (H_E) values. B: Node selection based on estimated population size. C: Node selection based on element occurrence (EO) values. For each of these experimental scenarios, nodes included in the subgraph represented populations with the largest (Proactive) or smallest (Reactive) parameter values. Populations located on the northern and southern shores of the St. Lawrence River are identified with gray and white nodes, respectively.

https://doi.org/10.1371/journal.pone.0271797.g003

Results of the node selection process revealed that regardless of the weighting schemes and conservation scenarios, maximum scores were obtained by adding from one to six nodes to the subgraph X_n (Table 2). A positive relationship was observed between the number of candidate nodes required to reach the maximum score and the complexity of the weighting scheme, with higher complexity models allowing for more diverse path types, but also requiring more candidate nodes (Kendall Tau = 0.634, p < 0.001) (see S1 Fig for examples). Namely, for the first weighting scheme (Model S¹) that only accounted for shortcuts (complexity of 1), three out of six scenarios returned no solution as no shortcuts were possible by adding candidate nodes to the subgraph. In three other cases, one or two candidate nodes were required to reach a maximal score. Models B³S¹ and B¹S³ considered shortcuts as well as breakthroughs (complexity of 2), but with different weighting schemes, and the number of nodes required to reach the maximum score varied from one to three. Model B¹R₁D₁S¹ contained contrasting roadblocks and detours with respect to breakthroughs and shortcuts (complexity of 4), and two to five candidate nodes were required to reach a maximal score. Finally, Model B³D¹E²S³ accounted for breakthroughs, detours, equals, and shortcuts (complexity of 4), and three to six nodes were required to reach a maximal score. In some cases, multiple solutions with equal scores were available at any given step of the procedure. To represent these equally multiple solutions, we depicted alternative paths with different types of edges in corresponding graphs (Fig 4). This redundancy was observed in the results from five of the six subgraph X_n selections for at least one step of the node selection procedure (Table 2).

Download:

Fig 4. BRIDES node selection process for scenario Ap and Model B³D¹E²S³ (3, 0, 0, 1, 2, 3).

All selection steps are shown, even when multiple solutions had identical scores. In such cases, distinct solutions (i.e., MA1 or MA2) are identified by dashed edges connecting the selected nodes. Maximum score (1.00) was reached with the addition of four candidate nodes. At each step of the procedure, the selected nodes and their corresponding links are presented in blue.

https://doi.org/10.1371/journal.pone.0271797.g004

Download:

Table 2. Summary of the BRIDES node selection procedure providing the minimum number of nodes required to reach a maximal score, and the number of multiple solutions with the same score for different scenarios and weighted model schemes.

https://doi.org/10.1371/journal.pone.0271797.t002

Discussion

Properties of the BRIDES selection procedure

BRIDES can repetitively assess the impact of multiple conservation scenarios without considering a priori models of population dynamics. The selection process thus allows the user to test different scenarios and weighting schemes to decide which one is best for population prioritization. When applied jointly with PopGraph, BRIDES will select the minimum number of nodes required to optimize gene flow and network connectivity. Moreover, BRIDES can identify multiple solutions with identical scores, thus allowing for alternative node combinations in the prioritization process. Its node selection process is robust for a given subgraph. In addition, BRIDES is highly flexible and provides the user with the opportunity to determine which types of paths need to be accounted for, and how they should be weighted to compute scores.

The BRIDES selection algorithm ranks all nodes based on their contribution to path lengths in the subgraph X_n with respect to the complete network X_all (Fig 1). Based on hypothetical conservation scenarios, our results highlighted the relative importance of some candidate nodes in the complete network, when their addition increased the numbers of shortcuts or breakthroughs. However, node selection can vary depending on the nodes chosen for the construction of subgraph X_n. When divergent scenarios are considered, BRIDES provides an objective criterion to guide the decision-making process. Yet, considering the high variation observed in the ranking given the subgraph used, the prioritization process should account for sensitivity to missing data and subgraphs must be chosen carefully. Namely, we suggest to experiment with different measures of interest for management purposes. Our results also highlight the need to further evaluate potential sources of uncertainty in network analysis, especially when applied to conservation biology where the impact of missing nodes on path lengths and component connectivity has yet to be assessed [65, 66].

Important ranking information is largely provided by node-based metrics, as nodes can be compared across a multitude of indices operating at different scales and patterns of connectivity. Cross and colleagues [45] argue that this flexibility can be beneficial when top-ranking nodes are first targeted by conservation actions. Yet, as pointed out by Creech and colleagues [29], the choice of a biologically relevant metric to quantify connectivity has a direct effect on the prioritization rankings, and our results revealed such inconsistencies among node-based metrics (Table 1). Similarly, node removal is highly sensitive, and it may produce inconsistent outcomes in habitat connectivity analysis [43, 44]. In agreement with previous studies, we trust that network metrics are valuable tools to understand complex patterns of connectivity and gene flow. On the other hand, we believe that node-based metrics should not be applied as the only prioritization criteria. Contrary to centrality measures focusing on static networks, BRIDES allows the analysis of dynamic networks onto which candidate nodes and edges are added. As shown in our results, the nodes more frequently selected by BRIDES were not always consistent with those selected by node-based metrics and node removal (Table 1). BRIDES identifies six different types of paths, each one capturing specific properties of path lengths within evolving networks. Yet, as it targets nodes based on their connectivity, BRIDES might miss some important peripheral or isolated nodes. For example, isolated nodes in a subgraph may remain disconnected when their selection leads to more detours and roadblocks than the selection of a central node (see S1 Fig, Scenario C_R). For that reason, BRIDES may be used jointly with complementary tools, such as phylogenetic networks developed by Volkman and colleagues [42] to account for genetic distinctiveness and identify peripheral populations.

Prioritization process of wood turtle populations

In the context of wood turtle conservation genetics, the results we obtained with population graphs are consistent with previous landscape genetics analyses [56], however, relationships among populations are much more detailed. As opposed to pairwise methods, population graphs account for genetic covariance among all populations simultaneously [37, 67]. It is already known that the St. Lawrence River acts as a barrier to gene flow among wood turtle populations located on opposite shores [51, 56, 68] but the population graph revealed which populations and corresponding links are more connected across the river (Fig 2). The network also depicted a detailed account of the interaction patterns between populations located on the two shores. We observed four nodes (LA1, LO1, MI2, SF3) linking the two shores, which could be viewed as stepping-stones or bridges between populations [38]. The results obtained after node removal also support the interconnection among populations on the north shore, except for two nodes linking the southern shore (LA1, LO1) that affect average path lengths. This implies that gene flow among north shore populations makes them less susceptible to local extinctions. On the other hand, populations on the south shore are more loosely connected, except for some sites exhibiting high values of AIEW (MA3, MI2). Metapopulation dynamics should be further investigated for these sites since AIEW is an accurate measure of reproductive success and the number of migrants between nodes [63].

The BRIDES algorithm was applied in this study to demonstrate how it may be used to select wood turtle populations with either a reactive or proactive prioritization procedure. To do so, we considered six hypothetical scenarios and five different weighting schemes, and applied a node selection procedure to prioritize populations. For example, the first scenario only included populations with large heterozygosity values. Even when neutral genetic diversity is not correlated to functional diversity [69], heterozygosity may still identify well-connected sites, large population sizes despite isolation, habitat fragmentation, and perturbance. In a proactive conservation plan, such populations with low vulnerability may be selected first to protect the genetic variability of the species before it erodes [15–17]. Namely, the link between the north and south shores of the St. Lawrence River was included in the subgraphs only when a proactive conservation perspective was used. meaning that some of theses populations might be characterized by small H_E, population sizes, and high EO ranks.

Regardless of their genetic diversity, the topological relationships among nodes within the network may be of importance for connectivity [70]. Peripheral populations associated with MA sites are thus of great interest to quantifying functional connectivity. Although MA3 presents high neutral genetic diversity, all MA sites are clustered together in Bouchard and colleagues [56]. Likewise, node-based metrics for these populations were in the low to mid-range, except for MA3, which exhibited a large AIEW value. Removal of these three nodes had no impact of the average path length of the network. These results are in stark contrast with those obtained with BRIDES, where the MA1 was very likely to be selected to maximize network connectivity. Such a discrepancy illustrates that node-based metrics may overlook important nodal features that BRIDES actually considers during its selection process.

In Québec, riverbeds and riverbanks which are part of wood turtle habitats are protected in public land, but stronger management actions are still needed to ensure long-term survival of populations [71]. To do so, one can apply BRIDES to prioritize populations in a reactive fashion. That is that populations with the smallest sizes–those needing high priority management actions such as nest protection from predators or a head-starting program–would first be selected to construct a subgraph (Scenario B_R, Fig 3 and S1 Fig). As the resulting subgraph may be disconnected and these small populations may face extirpation, one could also adopt a proactive perspective to protect larger populations that support them. More precisely, large turtle populations, as a source of genetically similar individuals, may add redundancy and resilience to population extirpation in the network. Namely, BRIDES can be applied to select these candidate populations based on path lengths and other connectivity indices (Model B³D¹E²S³, S1 Fig). In this particular case, populations BE1, MI3 and GA5 (Model B³D¹E²S³, S1 Fig) would be selected by BRIDES to guide the complex decision-making process of population prioritization.

Subgraph constructions in our study were based on proactive and reactive procedures to illustrate the method, but it is entirely up to the user to select populations in the starting subgraphs. The subgraphs could also be based on characteristics independent of the species, such as land ownership or jurisdictions involved allowing for nesting sites or for hibernacula that can be easily protected and managed. Because the choice of populations in subgraphs greatly influences the subsequent selection of candidate nodes, we suggest to use practical conservation criteria to select the nodes of a subgraph, especially when BRIDES is applied for population prioritization.

Uncertainties about genetic connectivity, genetic differentiation, isolation by distance, and population assignment may all be addressed with a single heuristic approach: the network. Moreover, many levels of connectivity can be considered at a different time and spatial scales when using evolving networks, especially those analyzed with BRIDES. Network analysis also offers an alternative model to population genetic analysis when populations are not considered to be at a mutation-drift balance and depart from the Hardy-Weinberg equilibrium [70, 72, 73]. Yet, networks are not free of a priori assumptions that are associated with the choice of a genetic distance, the pruning procedure, and the criteria selected for the construction of the subgraph. H_E, notably, may not be the most relevant criteria since it was very similar in several populations. In the present case, subgraphs were created by removing nodes and their corresponding edges from the complete population graph among 19 wood turtle populations. Considering that the graph pruning method is based on the removal of redundant edges that do not contribute to the overall genetic covariance of the network, population graphs should be re-estimated every time a node is removed.

Given the specific, historic life traits of long-lived organisms with low recruitment and late sexual maturity, the genetic characterization of recent perturbation events can be challenging [74]. As network topology stabilizes rapidly following a perturbation event, network analysis could be part of the solution to detect more precisely the genetic signal [67, 75]. For example, many species with high levels of gene flow could benefit from this network approach, like many turtle and tortoise species [74, 76]. Likewise, it could apply to other long-lived species of vertebrates [77, 78], as well as some plant species without any geographical barriers to gene flow [79, 80]. More precisely, BRIDES represents an interesting avenue to study trophic networks or to compare various scenarios during the design of protected area networks. Managers could benefit from using such tools to understand the impact of the addition or removal of a species within a food web, whether it be through the impact of the introduction of an invasive exotic species or the selection of a biocontrol agent. Namely, a scenario with fewer breakthroughs or shortcuts could be preferred to limit the impact of introduced invasive species. BRIDES could also be applied to optimize the connectivity among isolated habitat patches, with respect to historical levels of connectivity, when protected areas are designed. In a time where environmental challenges are pressing and greater emphasis is given to temporal scales when modeling climate change, population graphs and evolving networks may provide a better understanding of the evolving ecosystem, trophic networks and population structures.

Supporting information

S1 Fig. Results of the BRIDES node selection procedure for two scenarios and three weighted models.

In each case, networks with optimal scores while minimizing the number of candidate nodes are presented. Nodes from the subgraph X_n are depicted by empty circles, whereas candidate nodes and corresponding edges in X_n+1 are depicted by blue circles.

https://doi.org/10.1371/journal.pone.0271797.s001

(TIF)

S1 Table. Results of the BRIDES node selection procedure for six scenarios (A, B, C) and five weighted models (S¹, B³S¹, B¹S³, B¹R₁D₁S¹, B³D¹E²S³).

For each node, the initial presence of a population in the subgraph is identified by (x) and its selection (1) or not (0) in at least one of the optimal solutions is indicated. Scenarios for which no nodes were selected (-) are also shown.

https://doi.org/10.1371/journal.pone.0271797.s002

(PDF)

Acknowledgments

We would like to thank Philippe Lamarre from ministère des Forêts, de la Faune et des Parcs for information and explanation about populations size estimation and element occurrence ranks of the wood turtles in Quebec. Improvements of earlier drafts of the manuscript are the results of comments from C. Davy, S. Joly, P. Legendre and A. Rogic.

References

1. Soulé ME. Viable populations for conservation. Cambridge: Cambridge University Press; 1987
2. Moritz C. Defining “Evolutionarily Significant Units for conservation”. Trends Ecol Evol. 1994;9: 373–375. pmid:21236896
- View Article
- PubMed/NCBI
- Google Scholar
3. Fitzpatrick MC, Keller SR. Ecological genomics meets community-level modelling of biodiversity: mapping the genomic landscape of current and future environmental adaptation. Ecol Lett. 2015;18: 1–16. pmid:25270536
- View Article
- PubMed/NCBI
- Google Scholar
4. Hughes JB, Daily GC, Ehrlich PR. Population diversity: its extent and extinction. Science. 1997;280(5338): 689–692. pmid:9381179
- View Article
- PubMed/NCBI
- Google Scholar
5. Mace GM, Lande R. Assessing extinction threats: toward a reevaluation of IUCN threatened species categories. Conserv Biol. 1991;5: 148–157.
- View Article
- Google Scholar
6. Wood CC, Gross MR. Elemental conservation units: communicating extinction risk without dictating targets for protection. Conserv Biol. 2008;22: 36–47. pmid:18254851
- View Article
- PubMed/NCBI
- Google Scholar
7. Razgour O, Taggart JB, Manel S, Juste J, Ibanez C, Rebelo H, et al. An integrated framework to identify wildlife populations under threat from climate change. Mol Ecol Resour. 2018;18(1): 18–31. pmid:28649779
- View Article
- PubMed/NCBI
- Google Scholar
8. Millsap BA, Gore JA, Runde DE, Cerulean SI. Setting priorities for the conservation of fish and wildlife species in Florida. Wildl Monogr. 1990;3–57.
- View Article
- Google Scholar
9. Master LL. Assessing threats and setting priorities for conservation. Conserv Biol. 1991;5: 559–563.
- View Article
- Google Scholar
10. Mace GM. 1994 IUCN Red List of Threatened Animals. IUCN. Switzerland and Cambridge; 1993.
11. IUCN S.S.C. IUCN Red List categories and criteria: version 3.1. Switzerland; 2001.
12. Palsbøll PJ, Berube M, Allendorf FW. Identification of management units using population genetic data. Trends Ecol Evol. 2007;22(1): 11–16. pmid:16982114
- View Article
- PubMed/NCBI
- Google Scholar
13. Given DR, Norton DA. A multivariate approach to assessing threat and for priority setting in threatened species conservation. Biol Conserv. 1993;64: 57–66.
- View Article
- Google Scholar
14. Brooks TM, Mittermeier RA, da Fonseca GA, Gerlach J, Hoffmann M, Lamoreux JF, et al. Global biodiversity conservation priorities. Science. 2006;313(5783): 58–61. pmid:16825561
- View Article
- PubMed/NCBI
- Google Scholar
15. Crandall KA, Bininda-Emonds OR, Mace GM, Wayne RK. Considering evolutionary processes in conservation biology. Trends Ecol Evol. 2000;15(7): 290–295. pmid:10856956
- View Article
- PubMed/NCBI
- Google Scholar
16. Moritz C. Strategies to protect biological diversity and the evolutionary processes that sustain it. Syst Biol. 2002;51(2): 238–254. pmid:12028731
- View Article
- PubMed/NCBI
- Google Scholar
17. Mace GM, Purvis A. Evolutionary biology and practical conservation: bridging a widening gap. Mol Ecol. 2008;17(1): 9–19. pmid:17696991
- View Article
- PubMed/NCBI
- Google Scholar
18. Coulson T, Mace GM, Hudson E, Possingham H. The use and abuse of population viability analysis. Trends Ecol Evol. 2001;16(5): 219–221. pmid:11301139
- View Article
- PubMed/NCBI
- Google Scholar
19. Allendorf FW, Bayles D, Bottom DL Currens KP, Frissell CA, Hankin D, et al. Prioritizing pacific salmon stocks for conservation. Conserv Biol. 1997;11: 140–152.
- View Article
- Google Scholar
20. Clarkson RW, Marsh PC, Dowling TE. Population prioritization for conservation of imperilled warmwater fishes in an arid‐region drainage. Aquat Conserv: Mar Freshw Ecosyst. 2012;22(4): 498–510.
- View Article
- Google Scholar
21. Petit RJ, El Mousadik A, Pons O. Identifying populations for conservation on the basis of genetic markers. Conserv Biol. 1998;12(4): 844–855.
- View Article
- Google Scholar
22. Ottewell KM, Bickerton DC, Byrne M, Lowe AJ. Bridging the gap: a genetic assessment framework for population‐level threatened plant conservation prioritization and decision‐making. Divers Distrib. 2016;22(2): 174–188.
- View Article
- Google Scholar
23. Hanski I. Metapopulation dynamics. Nature. 1998;396(6706): 41–49.
- View Article
- Google Scholar
24. Moilanen A, Smith AT, Hanski I. Long-term dynamics in a metapopulation of the American pika. Am Nat. 1998;152(4): 530–542. pmid:18811362
- View Article
- PubMed/NCBI
- Google Scholar
25. Hoban S. Integrative conservation genetics: prioritizing populations using climate predictions, adaptive potential and habitat connectivity. Mol Ecol Resour. 2018;18(1): 14–17. pmid:29393598
- View Article
- PubMed/NCBI
- Google Scholar
26. Crooks KR, Sanjayan MA. Connectivity conservation. Cambridge: Cambridge University Press; 2006.
27. Mills L. Conservation of wildlife populations: demography, genetics, and management. Malden: Blackwell Publishing; 2007.
28. Lamy T, Pointier JP, Jarne P, David P. Testing metapopulation dynamics using genetic, demographic and ecological data. Mol Ecol. 2012;21(6): 1394–1410. pmid:22332609
- View Article
- PubMed/NCBI
- Google Scholar
29. Creech TG, Epps CW, Monello RJ, Wehausen JD. Using network theory to prioritize management in a desert bighorn sheep metapopulation. Landsc Ecol. 2014;29(4): 605–619.
- View Article
- Google Scholar
30. Gubili C, Mariani S, Weckworth BV, Galpern P, McDevitt AD, Hebblewhite M, et al. Environmental and anthropogenic drivers of connectivity patterns: a basis for prioritizing conservation efforts for threatened populations. Evol Appl. 2017;10: 199–211. pmid:28127396
- View Article
- PubMed/NCBI
- Google Scholar
31. Hedgecock D, Barber PH, Edmands S. Genetic approaches to measuring connectivity. Oceanogr. 2007;20(3): 70–79.
- View Article
- Google Scholar
32. Lowe WH, Allendorf FW. What can genetics tell us about population connectivity? Mol Ecol. 2010;19: 3038–3051. pmid:20618697
- View Article
- PubMed/NCBI
- Google Scholar
33. Row JR, Doherty KE, Cross TB, Schwartz MK, Oyler‐McCance SJ, Naugle DE, et al. Quantifying functional connectivity: The role of breeding habitat, abundance, and landscape features on range‐wide gene flow in sage‐grouse. Evol Appl. 2018;11(8): 1305–1321. pmid:30151042
- View Article
- PubMed/NCBI
- Google Scholar
34. Kool JT, Moilanen A, Treml EA. Population connectivity: recent advances and new perspectives. Landsc Ecol. 2013;28(2): 165–185.
- View Article
- Google Scholar
35. Newman MEJ. The structure and function of complex networks. SIAM Rev. 2003;45: 167–256.
- View Article
- Google Scholar
36. Matos C, Petrovan SO, Wheeler PM, Ward AI. Landscape connectivity and spatial prioritization in an urbanising world: a network analysis approach for a threatened amphibian. Biol Conserv. 2019;237: 238–247.
- View Article
- Google Scholar
37. Dyer RJ, Nason JD. Population graphs: the graph theoretic shape of genetic structure. Mol Ecol. 2004;13(7): 1713–1727. pmid:15189198
- View Article
- PubMed/NCBI
- Google Scholar
38. Garroway CJ, Bowman J, Carr D, Wilson PJ. Applications of graph theory to landscape genetics. Evol Appl. 2008;1(4): 620–630. pmid:25567802
- View Article
- PubMed/NCBI
- Google Scholar
39. Treml EA, Halpin PN, Urban DL, Pratson LF. Modeling population connectivity by ocean currents, a graph-theoretic approach for marine conservation. Landsc Ecol. 2008;23(1): 19–36.
- View Article
- Google Scholar
40. Marrotte RR, Bowman J, Brown MG, Cordes C, Morris KY, Prentice MB, et al. Multi-species genetic connectivity in a terrestrial habitat network. Mov Ecol. 2017;5(21): 1–11.
- View Article
- Google Scholar
41. Jacoby DMP, Freeman R. Emerging network-based tools in movement ecology. Trends Ecol Evol. 2016;31: 301–314. pmid:26879670
- View Article
- PubMed/NCBI
- Google Scholar
42. Volkmann L, Martyn I, Moulton V, Spillner A, Mooers AO. Prioritizing populations for conservation using phylogenetic networks. PLoS One. 2014;9(2): e88945. pmid:24586451
- View Article
- PubMed/NCBI
- Google Scholar
43. Jordán F, Báldi A, Orci KM, Racz I, Varga Z. Characterizing the importance of habitat patches and corridors in maintaining the landscape connectivity of a Pholidoptera transsylvanica (Orthoptera) metapopulation. Landsc Ecol. 2003;18(1): 83–92.
- View Article
- Google Scholar
44. Laita A, Kotiaho J, Mönkkönen M. Graph-theoretic connectivity measures: what do they tell us about connectivity? Landsc Ecol. 2011;26(7): 951–967.
- View Article
- Google Scholar
45. Cross TB, Schwartz MK, Naugle DE, Fedy BC, Row JR, Oyler‐McCance SJ. The genetic network of greater sage‐grouse: Range‐wide identification of keystone hubs of connectivity. Ecol Evol. 2018;8(11): 5394–5412. pmid:29938061
- View Article
- PubMed/NCBI
- Google Scholar
46. van Dijk PP, Harding J. Glyptemys insculpta (errata version published in 2016) The IUCN Red List of Threatened Species 2011: eT4965A97416259; 2011 [Cited 2021 june 15] https://dx.doi.org/10.2305/IUCN.UK.2011-1.RLTS.T4965A11102820.en.
47. Lovich JE, Ernst CH, McBreen JF. Growth, maturity, and sexual dimorphism in the wood turtle, Clemmys insculpta. Can J Zool. 1990;68: 672–677.
- View Article
- Google Scholar
48. Brooks RJ, Shilton CM, Brown GP, Quinn NWS. Body size, age distribution, and reproduction in a northern population of Wood Turtles (Clemmys insculpta). Can J of Zool. 1992;70: 462–469.
- View Article
- Google Scholar
49. Walde AD, Bider JR, Daigle C, Masse D., Bourgeois J-C, Jutras J, et al. Ecological aspects of a Wood Turtle, Glyptemys insculpta, population at the northern limit of its range in québec. Can Field-Nat. 2003;117(3): 377–388.
- View Article
- Google Scholar
50. Jones MT, Willey LL, Sievert PR, Akre TSB. Status and Conservation of the Wood Turtle in the Northeastern United States. Report submitted to the Northeast Association of Fish and Wildlife Agencies for Regional Conservation Needs Grant;2015. https://rcngrants.org/content/wood-turtle-glyptemys-insculpta-northeastern-united-states-status-assessment-and
51. Tessier N, Rioux Paquette S, Lapointe F-J. Conservation genetics of the wood turtle (Glyptemys insculpta) in Québec, Canada. Can J Zool. 2005;83(6): 765–772.
- View Article
- Google Scholar
52. Castellano CM, Behler JL, Amato G. Genetic diversity and population genetic structure of the wood turtle (Glyptemys inculpta) at Delaware Water Gap National Recreation Area, USA. Conserv Genet. 2009;10: 1783–1788.
- View Article
- Google Scholar
53. Frigden C, Finnegan L, Reaume C, Cebek J, Trottier J, Wilson P. Conservation of wood turtle (Glyptemys insculpta) populations in Ontario, Canada. Herpetol Conserv Biol. 2013;8(2): 351–358.
- View Article
- Google Scholar
54. Willoughby JR, Sundaram M, Lewis TL, Swanson BJ. Population decline in a long-lived species: the Wood Turtle in Michigan. Herpetologica. 2013;69(2): 186–198.
- View Article
- Google Scholar
55. Spradling TA, Tamplin JW, Dow SS, Meyer KJ. Conservation genetics of a peripherally isolated population of the wood turtle (Glyptemys insculpta) in Iowa. Conserv Genet. 2010;11(5): 1667–1677.
- View Article
- Google Scholar
56. Bouchard C, Tessier N, Lapointe F-J. Watersheds influence the wood turtle’s (Glyptemys insculpta) genetic structure. Conserv Genet. 2019;20(3): 653–664.
- View Article
- Google Scholar
57. Lord E, Le Cam M, Bapteste É, Méheust R, Makarenkov V, Lapointe F-J. BRIDES: a new fast algorithm and software for characterizing evolving similarity networks using Breakthroughs, Roadblocks, Impasses, Detours, Equals and Shortcuts. PloS One. 2016;11(8): e0161474. pmid:27580188
- View Article
- PubMed/NCBI
- Google Scholar
58. Keenan K, McGinnity P, Cross TF, Crozier WW, Prodöhl PA. diveRsity: an R package for the estimation of population genetics parameters and their associated errors. Methods Ecol Evol. 2013;4(8): 782–788.
- View Article
- Google Scholar
59. NatureServe. Element occurrence data standard NatureServe, Arlington, Virginia, USA; 2002. http://downloadsnatureserveorg/conservation_tools/element_occurence_data_standardpdf
60. R Core Team. R: a language and environment for statistical computing R Foundation for Statistical Computing. Vienna, Austria; 2018.
61. Dyer RJ. Popgraph: This is an R package that constructs and manipulates population graphs, R Package version 14; 2014
62. Dyer RJ. gstudio: Tools related to the spatial analysis of genetic marker data R package version 150; 2016
63. Koen EL, Bowman J, Wilson PJ. Node‐based measures of connectivity in genetic networks. Mol Ecol Resour. 2016;16(1): 69–79. pmid:25917123
- View Article
- PubMed/NCBI
- Google Scholar
64. Zero VH, Barocas A, Jochimsen DM, Pelletier A, Giroux-Bougard X, Trumbo DR, et al. Complementary network-based approaches for exploring genetic structure and functional connectivity in two vulnerable, endemic ground squirrels. Front Genet. 2017;8(81): 1–13. pmid:28659969
- View Article
- PubMed/NCBI
- Google Scholar
65. Koen EL, Bowman J, Garroway CJ, Wilson PJ. The sensitivity of genetic connectivity measures to unsampled and under-sampled sites. PLoS One. 2013;8(2): e56204. pmid:23409155
- View Article
- PubMed/NCBI
- Google Scholar
66. Naujokaitis-Lewis I, Rico Y, Lovell J, Fortin M-J, Murphy M. Implications of incomplete networks on estimation of landscape genetic connectivity. Conserv Genet. 2013;14: 287–298.
- View Article
- Google Scholar
67. Dyer RJ, Nason JD, Garrick RC. Landscape modelling of gene flow: improved power using conditional genetic distance derived from the topology of population networks. Mol Ecol. 2010;19: 3746–3759. pmid:20723052
- View Article
- PubMed/NCBI
- Google Scholar
68. Amato ML, Brooks RJ, Fu J. A phylogeographic analysis of populations of the wood turtle (Glyptemys insculpta) throughout its range. Mol Ecol. 2008;17(2): 570–581. pmid:17971089
- View Article
- PubMed/NCBI
- Google Scholar
69. Hedrick PW. Conservation genetics: where are we now? Trends in Ecol Evol. 2001;16(11): 629–636.
- View Article
- Google Scholar
70. Dyer RJ. Population graphs and landscape genetics. Annu Rev Ecol Evol Syst. 2015;46: 327–342.
- View Article
- Google Scholar
71. Équipe de rétablissement de cinq espèces de tortues au Québec pour les années 2005 à 2010: la tortue des bois (Glyptemys insculpta), la tortue géographique (Graptemys geographica), la tortue mouchetée (Emydoidea blandingii), la tortue musquée (Sternotherus odoratus) et la tortue ponctuée (Clemmys guttata). Ministère des Ressources Naturelles et de la Faune, Québec;2005
72. Neigel JE. A comparison of alternative strategies for estimating gene flow from genetic markers. Annu Rev Ecol Evol Syst. 1997;28: 105–128.
- View Article
- Google Scholar
73. Rozenfeld AF, Arnaud-Haond S, Hernández-García E, Eguíluz VM, Serrão EA, Duarte CM. Network analysis identifies weak and strong links in a metapopulation system. Proc Natl Acad Sci U S A. 2008;105(48): 18824–18829. pmid:19022909
- View Article
- PubMed/NCBI
- Google Scholar
74. Kuo CH, Janzen FJ. Genetic effects of a persistent bottleneck on a natural population of ornate box turtles (Terrapene ornata). Conserv Genet. 2004;5(4): 425–437.
- View Article
- Google Scholar
75. Dyer RJ. The evolution of genetic topologies. Theor Popul Biol. 2007;71: 71–78. pmid:16919694
- View Article
- PubMed/NCBI
- Google Scholar
76. Dutcher KE, Vandergast AG, Esque TC, et al. Genes in space: what Mojave desert tortoise genetics can tell us about landscape connectivity. Conserv Genet. 2020;21(2): 289–303.
- View Article
- Google Scholar
77. Mcdougall CA, Welsh AB, Gosselin T, et al. Rethinking the influence of hydroelectric development on gene flow in a long-lived fish, the Lake Sturgeon Acipenser fulvescens. PloS one, 2017;12(3): e0174269. pmid:28329005
- View Article
- PubMed/NCBI
- Google Scholar
78. Mr Fuller, Doyle M. Gene flow simulations demonstrate resistance of long-lived species to genetic erosion from habitat fragmentation. Conserv Genet. 2018;19(6): 1439–1448.
- View Article
- Google Scholar
79. Luna‐Ortiz A, Arteaga MC, Bello‐Bedoy R. et al. High genetic diversity and low structure in an endemic long‐lived tree, Yucca capensis (Asparagaceae). Plant Biol. 2022;24(1): 185–191. pmid:34634170
- View Article
- PubMed/NCBI
- Google Scholar
80. Noutsos C, Borevitz JO, Hodges SA. Gene flow between nascent species: genotypic and phenotypic differentiation within and between Aquilegia formosa and A pubescens. Mol Ecol. 2014;23: 5589–98. pmid:25314338
- View Article
- PubMed/NCBI
- Google Scholar

[ref1] 1. Soulé ME. Viable populations for conservation. Cambridge: Cambridge University Press; 1987

[ref2] 2. Moritz C. Defining “Evolutionarily Significant Units for conservation”. Trends Ecol Evol. 1994;9: 373–375. pmid:21236896
View Article
PubMed/NCBI
Google Scholar

[3] View Article

[4] PubMed/NCBI

[5] Google Scholar

[ref3] 3. Fitzpatrick MC, Keller SR. Ecological genomics meets community-level modelling of biodiversity: mapping the genomic landscape of current and future environmental adaptation. Ecol Lett. 2015;18: 1–16. pmid:25270536
View Article
PubMed/NCBI
Google Scholar

[7] View Article

[8] PubMed/NCBI

[9] Google Scholar

[ref4] 4. Hughes JB, Daily GC, Ehrlich PR. Population diversity: its extent and extinction. Science. 1997;280(5338): 689–692. pmid:9381179
View Article
PubMed/NCBI
Google Scholar

[11] View Article

[12] PubMed/NCBI

[13] Google Scholar

[ref5] 5. Mace GM, Lande R. Assessing extinction threats: toward a reevaluation of IUCN threatened species categories. Conserv Biol. 1991;5: 148–157.
View Article
Google Scholar

[15] View Article

[16] Google Scholar

[ref6] 6. Wood CC, Gross MR. Elemental conservation units: communicating extinction risk without dictating targets for protection. Conserv Biol. 2008;22: 36–47. pmid:18254851
View Article
PubMed/NCBI
Google Scholar

[18] View Article

[19] PubMed/NCBI

[20] Google Scholar

[ref7] 7. Razgour O, Taggart JB, Manel S, Juste J, Ibanez C, Rebelo H, et al. An integrated framework to identify wildlife populations under threat from climate change. Mol Ecol Resour. 2018;18(1): 18–31. pmid:28649779
View Article
PubMed/NCBI
Google Scholar

[22] View Article

[23] PubMed/NCBI

[24] Google Scholar

[ref8] 8. Millsap BA, Gore JA, Runde DE, Cerulean SI. Setting priorities for the conservation of fish and wildlife species in Florida. Wildl Monogr. 1990;3–57.
View Article
Google Scholar

[26] View Article

[27] Google Scholar

[ref9] 9. Master LL. Assessing threats and setting priorities for conservation. Conserv Biol. 1991;5: 559–563.
View Article
Google Scholar

[29] View Article

[30] Google Scholar

[ref10] 10. Mace GM. 1994 IUCN Red List of Threatened Animals. IUCN. Switzerland and Cambridge; 1993.

[ref11] 11. IUCN S.S.C. IUCN Red List categories and criteria: version 3.1. Switzerland; 2001.

[ref12] 12. Palsbøll PJ, Berube M, Allendorf FW. Identification of management units using population genetic data. Trends Ecol Evol. 2007;22(1): 11–16. pmid:16982114
View Article
PubMed/NCBI
Google Scholar

[34] View Article

[35] PubMed/NCBI

[36] Google Scholar

[ref13] 13. Given DR, Norton DA. A multivariate approach to assessing threat and for priority setting in threatened species conservation. Biol Conserv. 1993;64: 57–66.
View Article
Google Scholar

[38] View Article

[39] Google Scholar

[ref14] 14. Brooks TM, Mittermeier RA, da Fonseca GA, Gerlach J, Hoffmann M, Lamoreux JF, et al. Global biodiversity conservation priorities. Science. 2006;313(5783): 58–61. pmid:16825561
View Article
PubMed/NCBI
Google Scholar

[41] View Article

[42] PubMed/NCBI

[43] Google Scholar

[ref15] 15. Crandall KA, Bininda-Emonds OR, Mace GM, Wayne RK. Considering evolutionary processes in conservation biology. Trends Ecol Evol. 2000;15(7): 290–295. pmid:10856956
View Article
PubMed/NCBI
Google Scholar

[45] View Article

[46] PubMed/NCBI

[47] Google Scholar

[ref16] 16. Moritz C. Strategies to protect biological diversity and the evolutionary processes that sustain it. Syst Biol. 2002;51(2): 238–254. pmid:12028731
View Article
PubMed/NCBI
Google Scholar

[49] View Article

[50] PubMed/NCBI

[51] Google Scholar

[ref17] 17. Mace GM, Purvis A. Evolutionary biology and practical conservation: bridging a widening gap. Mol Ecol. 2008;17(1): 9–19. pmid:17696991
View Article
PubMed/NCBI
Google Scholar

[53] View Article

[54] PubMed/NCBI

[55] Google Scholar

[ref18] 18. Coulson T, Mace GM, Hudson E, Possingham H. The use and abuse of population viability analysis. Trends Ecol Evol. 2001;16(5): 219–221. pmid:11301139
View Article
PubMed/NCBI
Google Scholar

[57] View Article

[58] PubMed/NCBI

[59] Google Scholar

[ref19] 19. Allendorf FW, Bayles D, Bottom DL Currens KP, Frissell CA, Hankin D, et al. Prioritizing pacific salmon stocks for conservation. Conserv Biol. 1997;11: 140–152.
View Article
Google Scholar

[61] View Article

[62] Google Scholar

[ref20] 20. Clarkson RW, Marsh PC, Dowling TE. Population prioritization for conservation of imperilled warmwater fishes in an arid‐region drainage. Aquat Conserv: Mar Freshw Ecosyst. 2012;22(4): 498–510.
View Article
Google Scholar

[64] View Article

[65] Google Scholar

[ref21] 21. Petit RJ, El Mousadik A, Pons O. Identifying populations for conservation on the basis of genetic markers. Conserv Biol. 1998;12(4): 844–855.
View Article
Google Scholar

[67] View Article

[68] Google Scholar

[ref22] 22. Ottewell KM, Bickerton DC, Byrne M, Lowe AJ. Bridging the gap: a genetic assessment framework for population‐level threatened plant conservation prioritization and decision‐making. Divers Distrib. 2016;22(2): 174–188.
View Article
Google Scholar

[70] View Article

[71] Google Scholar

[ref23] 23. Hanski I. Metapopulation dynamics. Nature. 1998;396(6706): 41–49.
View Article
Google Scholar

[73] View Article

[74] Google Scholar

[ref24] 24. Moilanen A, Smith AT, Hanski I. Long-term dynamics in a metapopulation of the American pika. Am Nat. 1998;152(4): 530–542. pmid:18811362
View Article
PubMed/NCBI
Google Scholar

[76] View Article

[77] PubMed/NCBI

[78] Google Scholar

[ref25] 25. Hoban S. Integrative conservation genetics: prioritizing populations using climate predictions, adaptive potential and habitat connectivity. Mol Ecol Resour. 2018;18(1): 14–17. pmid:29393598
View Article
PubMed/NCBI
Google Scholar

[80] View Article

[81] PubMed/NCBI

[82] Google Scholar

[ref26] 26. Crooks KR, Sanjayan MA. Connectivity conservation. Cambridge: Cambridge University Press; 2006.

[ref27] 27. Mills L. Conservation of wildlife populations: demography, genetics, and management. Malden: Blackwell Publishing; 2007.

[ref28] 28. Lamy T, Pointier JP, Jarne P, David P. Testing metapopulation dynamics using genetic, demographic and ecological data. Mol Ecol. 2012;21(6): 1394–1410. pmid:22332609
View Article
PubMed/NCBI
Google Scholar

[86] View Article

[87] PubMed/NCBI

[88] Google Scholar

[ref29] 29. Creech TG, Epps CW, Monello RJ, Wehausen JD. Using network theory to prioritize management in a desert bighorn sheep metapopulation. Landsc Ecol. 2014;29(4): 605–619.
View Article
Google Scholar

[90] View Article

[91] Google Scholar

[ref30] 30. Gubili C, Mariani S, Weckworth BV, Galpern P, McDevitt AD, Hebblewhite M, et al. Environmental and anthropogenic drivers of connectivity patterns: a basis for prioritizing conservation efforts for threatened populations. Evol Appl. 2017;10: 199–211. pmid:28127396
View Article
PubMed/NCBI
Google Scholar

[93] View Article

[94] PubMed/NCBI

[95] Google Scholar

[ref31] 31. Hedgecock D, Barber PH, Edmands S. Genetic approaches to measuring connectivity. Oceanogr. 2007;20(3): 70–79.
View Article
Google Scholar

[97] View Article

[98] Google Scholar

[ref32] 32. Lowe WH, Allendorf FW. What can genetics tell us about population connectivity? Mol Ecol. 2010;19: 3038–3051. pmid:20618697
View Article
PubMed/NCBI
Google Scholar

[100] View Article

[101] PubMed/NCBI

[102] Google Scholar

[ref33] 33. Row JR, Doherty KE, Cross TB, Schwartz MK, Oyler‐McCance SJ, Naugle DE, et al. Quantifying functional connectivity: The role of breeding habitat, abundance, and landscape features on range‐wide gene flow in sage‐grouse. Evol Appl. 2018;11(8): 1305–1321. pmid:30151042
View Article
PubMed/NCBI
Google Scholar

[104] View Article

[105] PubMed/NCBI

[106] Google Scholar

[ref34] 34. Kool JT, Moilanen A, Treml EA. Population connectivity: recent advances and new perspectives. Landsc Ecol. 2013;28(2): 165–185.
View Article
Google Scholar

[108] View Article

[109] Google Scholar

[ref35] 35. Newman MEJ. The structure and function of complex networks. SIAM Rev. 2003;45: 167–256.
View Article
Google Scholar

[111] View Article

[112] Google Scholar

[ref36] 36. Matos C, Petrovan SO, Wheeler PM, Ward AI. Landscape connectivity and spatial prioritization in an urbanising world: a network analysis approach for a threatened amphibian. Biol Conserv. 2019;237: 238–247.
View Article
Google Scholar

[114] View Article

[115] Google Scholar

[ref37] 37. Dyer RJ, Nason JD. Population graphs: the graph theoretic shape of genetic structure. Mol Ecol. 2004;13(7): 1713–1727. pmid:15189198
View Article
PubMed/NCBI
Google Scholar

[117] View Article

[118] PubMed/NCBI

[119] Google Scholar

[ref38] 38. Garroway CJ, Bowman J, Carr D, Wilson PJ. Applications of graph theory to landscape genetics. Evol Appl. 2008;1(4): 620–630. pmid:25567802
View Article
PubMed/NCBI
Google Scholar

[121] View Article

[122] PubMed/NCBI

[123] Google Scholar

[ref39] 39. Treml EA, Halpin PN, Urban DL, Pratson LF. Modeling population connectivity by ocean currents, a graph-theoretic approach for marine conservation. Landsc Ecol. 2008;23(1): 19–36.
View Article
Google Scholar

[125] View Article

[126] Google Scholar

[ref40] 40. Marrotte RR, Bowman J, Brown MG, Cordes C, Morris KY, Prentice MB, et al. Multi-species genetic connectivity in a terrestrial habitat network. Mov Ecol. 2017;5(21): 1–11.
View Article
Google Scholar

[128] View Article

[129] Google Scholar

[ref41] 41. Jacoby DMP, Freeman R. Emerging network-based tools in movement ecology. Trends Ecol Evol. 2016;31: 301–314. pmid:26879670
View Article
PubMed/NCBI
Google Scholar

[131] View Article

[132] PubMed/NCBI

[133] Google Scholar

[ref42] 42. Volkmann L, Martyn I, Moulton V, Spillner A, Mooers AO. Prioritizing populations for conservation using phylogenetic networks. PLoS One. 2014;9(2): e88945. pmid:24586451
View Article
PubMed/NCBI
Google Scholar

[135] View Article

[136] PubMed/NCBI

[137] Google Scholar

[ref43] 43. Jordán F, Báldi A, Orci KM, Racz I, Varga Z. Characterizing the importance of habitat patches and corridors in maintaining the landscape connectivity of a Pholidoptera transsylvanica (Orthoptera) metapopulation. Landsc Ecol. 2003;18(1): 83–92.
View Article
Google Scholar

[139] View Article

[140] Google Scholar

[ref44] 44. Laita A, Kotiaho J, Mönkkönen M. Graph-theoretic connectivity measures: what do they tell us about connectivity? Landsc Ecol. 2011;26(7): 951–967.
View Article
Google Scholar

[142] View Article

[143] Google Scholar

[ref45] 45. Cross TB, Schwartz MK, Naugle DE, Fedy BC, Row JR, Oyler‐McCance SJ. The genetic network of greater sage‐grouse: Range‐wide identification of keystone hubs of connectivity. Ecol Evol. 2018;8(11): 5394–5412. pmid:29938061
View Article
PubMed/NCBI
Google Scholar

[145] View Article

[146] PubMed/NCBI

[147] Google Scholar

[ref46] 46. van Dijk PP, Harding J. Glyptemys insculpta (errata version published in 2016) The IUCN Red List of Threatened Species 2011: eT4965A97416259; 2011 [Cited 2021 june 15] https://dx.doi.org/10.2305/IUCN.UK.2011-1.RLTS.T4965A11102820.en.

[ref47] 47. Lovich JE, Ernst CH, McBreen JF. Growth, maturity, and sexual dimorphism in the wood turtle, Clemmys insculpta. Can J Zool. 1990;68: 672–677.
View Article
Google Scholar

[150] View Article

[151] Google Scholar

[ref48] 48. Brooks RJ, Shilton CM, Brown GP, Quinn NWS. Body size, age distribution, and reproduction in a northern population of Wood Turtles (Clemmys insculpta). Can J of Zool. 1992;70: 462–469.
View Article
Google Scholar

[153] View Article

[154] Google Scholar

[ref49] 49. Walde AD, Bider JR, Daigle C, Masse D., Bourgeois J-C, Jutras J, et al. Ecological aspects of a Wood Turtle, Glyptemys insculpta, population at the northern limit of its range in québec. Can Field-Nat. 2003;117(3): 377–388.
View Article
Google Scholar

[156] View Article

[157] Google Scholar

[ref50] 50. Jones MT, Willey LL, Sievert PR, Akre TSB. Status and Conservation of the Wood Turtle in the Northeastern United States. Report submitted to the Northeast Association of Fish and Wildlife Agencies for Regional Conservation Needs Grant;2015. https://rcngrants.org/content/wood-turtle-glyptemys-insculpta-northeastern-united-states-status-assessment-and

[ref51] 51. Tessier N, Rioux Paquette S, Lapointe F-J. Conservation genetics of the wood turtle (Glyptemys insculpta) in Québec, Canada. Can J Zool. 2005;83(6): 765–772.
View Article
Google Scholar

[160] View Article

[161] Google Scholar

[ref52] 52. Castellano CM, Behler JL, Amato G. Genetic diversity and population genetic structure of the wood turtle (Glyptemys inculpta) at Delaware Water Gap National Recreation Area, USA. Conserv Genet. 2009;10: 1783–1788.
View Article
Google Scholar

[163] View Article

[164] Google Scholar

[ref53] 53. Frigden C, Finnegan L, Reaume C, Cebek J, Trottier J, Wilson P. Conservation of wood turtle (Glyptemys insculpta) populations in Ontario, Canada. Herpetol Conserv Biol. 2013;8(2): 351–358.
View Article
Google Scholar

[166] View Article

[167] Google Scholar

[ref54] 54. Willoughby JR, Sundaram M, Lewis TL, Swanson BJ. Population decline in a long-lived species: the Wood Turtle in Michigan. Herpetologica. 2013;69(2): 186–198.
View Article
Google Scholar

[169] View Article

[170] Google Scholar

[ref55] 55. Spradling TA, Tamplin JW, Dow SS, Meyer KJ. Conservation genetics of a peripherally isolated population of the wood turtle (Glyptemys insculpta) in Iowa. Conserv Genet. 2010;11(5): 1667–1677.
View Article
Google Scholar

[172] View Article

[173] Google Scholar

[ref56] 56. Bouchard C, Tessier N, Lapointe F-J. Watersheds influence the wood turtle’s (Glyptemys insculpta) genetic structure. Conserv Genet. 2019;20(3): 653–664.
View Article
Google Scholar

[175] View Article

[176] Google Scholar

[ref57] 57. Lord E, Le Cam M, Bapteste É, Méheust R, Makarenkov V, Lapointe F-J. BRIDES: a new fast algorithm and software for characterizing evolving similarity networks using Breakthroughs, Roadblocks, Impasses, Detours, Equals and Shortcuts. PloS One. 2016;11(8): e0161474. pmid:27580188
View Article
PubMed/NCBI
Google Scholar

[178] View Article

[179] PubMed/NCBI

[180] Google Scholar

[ref58] 58. Keenan K, McGinnity P, Cross TF, Crozier WW, Prodöhl PA. diveRsity: an R package for the estimation of population genetics parameters and their associated errors. Methods Ecol Evol. 2013;4(8): 782–788.
View Article
Google Scholar

[182] View Article

[183] Google Scholar

[ref59] 59. NatureServe. Element occurrence data standard NatureServe, Arlington, Virginia, USA; 2002. http://downloadsnatureserveorg/conservation_tools/element_occurence_data_standardpdf

[ref60] 60. R Core Team. R: a language and environment for statistical computing R Foundation for Statistical Computing. Vienna, Austria; 2018.

[ref61] 61. Dyer RJ. Popgraph: This is an R package that constructs and manipulates population graphs, R Package version 14; 2014

[ref62] 62. Dyer RJ. gstudio: Tools related to the spatial analysis of genetic marker data R package version 150; 2016

[ref63] 63. Koen EL, Bowman J, Wilson PJ. Node‐based measures of connectivity in genetic networks. Mol Ecol Resour. 2016;16(1): 69–79. pmid:25917123
View Article
PubMed/NCBI
Google Scholar

[189] View Article

[190] PubMed/NCBI

[191] Google Scholar

[ref64] 64. Zero VH, Barocas A, Jochimsen DM, Pelletier A, Giroux-Bougard X, Trumbo DR, et al. Complementary network-based approaches for exploring genetic structure and functional connectivity in two vulnerable, endemic ground squirrels. Front Genet. 2017;8(81): 1–13. pmid:28659969
View Article
PubMed/NCBI
Google Scholar

[193] View Article

[194] PubMed/NCBI

[195] Google Scholar

[ref65] 65. Koen EL, Bowman J, Garroway CJ, Wilson PJ. The sensitivity of genetic connectivity measures to unsampled and under-sampled sites. PLoS One. 2013;8(2): e56204. pmid:23409155
View Article
PubMed/NCBI
Google Scholar

[197] View Article

[198] PubMed/NCBI

[199] Google Scholar

[ref66] 66. Naujokaitis-Lewis I, Rico Y, Lovell J, Fortin M-J, Murphy M. Implications of incomplete networks on estimation of landscape genetic connectivity. Conserv Genet. 2013;14: 287–298.
View Article
Google Scholar

[201] View Article

[202] Google Scholar

[ref67] 67. Dyer RJ, Nason JD, Garrick RC. Landscape modelling of gene flow: improved power using conditional genetic distance derived from the topology of population networks. Mol Ecol. 2010;19: 3746–3759. pmid:20723052
View Article
PubMed/NCBI
Google Scholar

[204] View Article

[205] PubMed/NCBI

[206] Google Scholar

[ref68] 68. Amato ML, Brooks RJ, Fu J. A phylogeographic analysis of populations of the wood turtle (Glyptemys insculpta) throughout its range. Mol Ecol. 2008;17(2): 570–581. pmid:17971089
View Article
PubMed/NCBI
Google Scholar

[208] View Article

[209] PubMed/NCBI

[210] Google Scholar

[ref69] 69. Hedrick PW. Conservation genetics: where are we now? Trends in Ecol Evol. 2001;16(11): 629–636.
View Article
Google Scholar

[212] View Article

[213] Google Scholar

[ref70] 70. Dyer RJ. Population graphs and landscape genetics. Annu Rev Ecol Evol Syst. 2015;46: 327–342.
View Article
Google Scholar

[215] View Article

[216] Google Scholar

[ref71] 71. Équipe de rétablissement de cinq espèces de tortues au Québec pour les années 2005 à 2010: la tortue des bois (Glyptemys insculpta), la tortue géographique (Graptemys geographica), la tortue mouchetée (Emydoidea blandingii), la tortue musquée (Sternotherus odoratus) et la tortue ponctuée (Clemmys guttata). Ministère des Ressources Naturelles et de la Faune, Québec;2005

[ref72] 72. Neigel JE. A comparison of alternative strategies for estimating gene flow from genetic markers. Annu Rev Ecol Evol Syst. 1997;28: 105–128.
View Article
Google Scholar

[219] View Article

[220] Google Scholar

[ref73] 73. Rozenfeld AF, Arnaud-Haond S, Hernández-García E, Eguíluz VM, Serrão EA, Duarte CM. Network analysis identifies weak and strong links in a metapopulation system. Proc Natl Acad Sci U S A. 2008;105(48): 18824–18829. pmid:19022909
View Article
PubMed/NCBI
Google Scholar

[222] View Article

[223] PubMed/NCBI

[224] Google Scholar

[ref74] 74. Kuo CH, Janzen FJ. Genetic effects of a persistent bottleneck on a natural population of ornate box turtles (Terrapene ornata). Conserv Genet. 2004;5(4): 425–437.
View Article
Google Scholar

[226] View Article

[227] Google Scholar

[ref75] 75. Dyer RJ. The evolution of genetic topologies. Theor Popul Biol. 2007;71: 71–78. pmid:16919694
View Article
PubMed/NCBI
Google Scholar

[229] View Article

[230] PubMed/NCBI

[231] Google Scholar

[ref76] 76. Dutcher KE, Vandergast AG, Esque TC, et al. Genes in space: what Mojave desert tortoise genetics can tell us about landscape connectivity. Conserv Genet. 2020;21(2): 289–303.
View Article
Google Scholar

[233] View Article

[234] Google Scholar

[ref77] 77. Mcdougall CA, Welsh AB, Gosselin T, et al. Rethinking the influence of hydroelectric development on gene flow in a long-lived fish, the Lake Sturgeon Acipenser fulvescens. PloS one, 2017;12(3): e0174269. pmid:28329005
View Article
PubMed/NCBI
Google Scholar

[236] View Article

[237] PubMed/NCBI

[238] Google Scholar

[ref78] 78. Mr Fuller, Doyle M. Gene flow simulations demonstrate resistance of long-lived species to genetic erosion from habitat fragmentation. Conserv Genet. 2018;19(6): 1439–1448.
View Article
Google Scholar

[240] View Article

[241] Google Scholar

[ref79] 79. Luna‐Ortiz A, Arteaga MC, Bello‐Bedoy R. et al. High genetic diversity and low structure in an endemic long‐lived tree, Yucca capensis (Asparagaceae). Plant Biol. 2022;24(1): 185–191. pmid:34634170
View Article
PubMed/NCBI
Google Scholar

[243] View Article

[244] PubMed/NCBI

[245] Google Scholar

[ref80] 80. Noutsos C, Borevitz JO, Hodges SA. Gene flow between nascent species: genotypic and phenotypic differentiation within and between Aquilegia formosa and A pubescens. Mol Ecol. 2014;23: 5589–98. pmid:25314338
View Article
PubMed/NCBI
Google Scholar

[247] View Article

[248] PubMed/NCBI

[249] Google Scholar

Figures

Abstract

Introduction

Materials and methods

Wood turtle population characterization

Wood turtle population graph

Node-based metrics of connectivity and node removal

BRIDES analysis

BRIDES selection algorithm

Weighted models and conservation scenarios

Results

Population graph

Network metrics and node removal

BRIDES selection procedure

Discussion

Properties of the BRIDES selection procedure

Prioritization process of wood turtle populations

Supporting information

S1 Fig. Results of the BRIDES node selection procedure for two scenarios and three weighted models.

S1 Table. Results of the BRIDES node selection procedure for six scenarios (A, B, C) and five weighted models (S1, B3S1, B1S3, B1R1D1S1, B3D1E2S3).

Acknowledgments

References

S1 Table. Results of the BRIDES node selection procedure for six scenarios (A, B, C) and five weighted models (S¹, B³S¹, B¹S³, B¹R₁D₁S¹, B³D¹E²S³).