Conservation unit recommendations
Both species have undergone accentuated population declines in the last
50 years (Yensen 1999; Gavin et al. 1999; Evans-Mack 2003),
suggesting high susceptibility to environmental and landscape changes.
As a result, NIDGS were listed as threatened under the federal
Endangered Species Act (ESA) in 2000 (Clark 2000), and as Critically
Endangered by the International Union for Conservation of Nature (Hafner
1998). SIDGS are listed as Vulnerable by the IUCN (IUCN 2018) and were a
former candidate for listing under the ESA, although they were
ultimately not listed. We made use of a minimally invasive sampling
method, in order to minimize the impact of sampling on populations
(Carroll et al. 2018). Buccal swabs, as used in this study, have
been specifically used in high throughput sequencing (HTS) genotyping
using targeted sequencing (Chang et al. 2007; McMichael et
al. 2009), and have been shown to allow for non-targeted HTS
genotyping, such as RAD sequencing, in amphibians (Peek et al.2019). To our knowledge, this is the first study genotyping buccal swabs
from a mammal with a non-targeted HTS approach, adding to the growing
literature aiming to use less invasive sampling strategies to examine
genomic variation in species of conservation concern (Andrews et
al. 2018; Carroll et al. 2018).
The decline of these two species has resulted in metapopulations
approaching a state of nonequilibrium, where many populations became
small and isolated, and thus especially prone to both diversification
and extinction (Harrison & Taylor 1997; Hanski & Gaggiotti 2004;
Pironon et al. 2017). Metapopulations can retain genetic
variation more readily than simply isolated subpopulations from a once
panmictic population, and can be more resilient to extinction due to
their intrinsic colonization-extinction-recolonization dynamics (Nee &
May 1992; Levin 1995; Gavrilets et al. 2000). However, this
resilience is dependent on how many individuals and how much genetic
diversity each subpopulation maintains (Fahrig & Merriam 1994). NIDGS
populations appear to have a more homogenous metapopulation in the
western portion of their range based on low and mostly non-significant
levels of pairwise F ST, but form a more
fragmented patch network in the eastern portion of their range, based on
the neutral and adaptive loci (Figures 5 and 6, and S9, Supporting
information). Overall we observed similar patterns of genetic structure
as seen in previous studies for both NIDGS and SIDGS (Garner et
al. 2005; Hoisington 2007).
Following the recommendations of Funk et al. (2012), we defined
three evolutionarily significant units (ESUs) for NIDGS (Figure 6A).
Within the ESU1, we identified two management and adaptive units (MUs
and AUs, respectively), corresponding to the separation of Rocky Top
from all other populations. Lower Butter was identified as a separate
conservation unit, ESU2, which was distinctive at both the neutral and
adaptive levels from all other populations. Within the ESU3, of the four
populations sampled, three were considered separate MUs, while Lost
Valley which was represented by a single individual showed mixed
ancestry (Figure 6A). We also identified Tamarack as a distinct AU.
Interestingly, ESU1 and ESU3 are also supported by previous work using
mitochondrial data (Hoisington 2007; Hoisington-Lopez et al.2012). The presence of long branches between haplogroups and absence of
star-like pattern in the mitochondrial network suggests that these
haplogroups diverged in the intermediate to distant past, allowing for
enough time for these populations to become demographically and
adaptively differentiated (Zink & Barrowclough 2008; Hoisington-Lopezet al. 2012). However, the previous mitochondrial work found a
third haplogroup for populations not sampled in this study, and also did
not include Lower Butter. So additional mitochondrial and genomic work
is necessary to validate and potentially add to the conservation unit
assignments found in this study for NIDGS. For SIDGS, we identified two
ESUs due to a clear separation of the populations east of the Weiser
River with those from Olds Ferry (Figures 2 and 6, and S2, Supporting
information). This separation is also evident at the neutral level
(Table 4 and Figures 5 and 6), as Olds Ferry also represents a distinct
MU from the remaining populations. Previous genetic studies of SIDGS
included more sampling locations than our study, and they also detected
strong genetic differentiation for populations east and west of the
Weiser River using mtDNA sequence data and microsatellite loci (Garneret al. 2005; Hoisington 2007; Hoisington-Lopez et al.2012). Paddock was clearly distinct as an AU within SIDGS (Figure 5,
Figure S10B, Supporting information), but it did not show significant
neutral divergence to be considered a distinct MU. There is evidence for
the vulnerability of SIDGS populations due to the construction of roads
using gravity models, and that might explain the significant reduction
in genetic diversity in most populations east of the Weiser River (Table
4). These populations appear to be the most vulnerable, potentially as a
result of low dispersal propensity and population declines associated
with loss of habitat and consequently, connectivity (Barrett 2005; Panek
2005). This isolation could have resulted in further uniqueness in terms
both neutral and adaptive diversity, and could explain the
identification of Paddock as a separate AU for SIDGS.
Our results are important for developing management strategies in
response to land use and habitat changes (Henry & Russello 2013). If
translocations are warranted, they will likely be most effective if
performed between populations within the same ESU and ideally also
within the same AU, to allow for increased connectivity without
compromising potential effects of local adaptation. Protecting AUs is
essential to preserve adaptive diversity of populations which already
reflects local adaptations that might be unique. However, further
sampling is required to validate the results of those populations with
lower number of samples, given that due to their low sample size it is
less likely that adaptive variation would be detected (Lotterhos &
Whitlock 2015). The inclusion of additional populations would also be
ideal, not only for confirming the patterns found in this study, but
also for identifying other conservation units that might warrant special
protection.