3.3 Influence of landscape on gene flow and genetic distance
According to the IBD model and the Mantel test in the Shunchang County,
there was no significant (r(FST, IBD) = 0.38, P =
0.075) (Figure 4A), but different landscape types had influence
(Supporting information Figure S3). Among the four groups of resistance
values, only the resistance surface created by the first group (the
resistance value of P. massoniana , P. elliottii ,
mixed-forest with hosts, and road was 1; the medium resistance value of
broad-leaved forest, C. lanceolata , water, and nudation was 8;
the high resistance value of farmland was 64 and that of urban 512)
(Figure 4C) was significantly correlated (r(FST, LCP) =
0.44, P = 0.045) (Figure 4B). There was no correlation with the
other three resistance surfaces, in which the resistance values of host
and road landscapes were higher than those of nonhost landscapes (except
roads). This result suggested that at the fine scale (<10 km),
the genetic differentiation of M. alternatus could not be
determined by distance alone, although landscape types could have an
effect. Among the landscapes, the host landscapes and roads had low
resistance to local dispersal; whereas nonhost landscapes were more
likely to inhibit movement in this species.
The top ten models in the LCTA of the two areas all demonstrated thatP. massoniana , mixed forests with hosts, P. elliottii , and
roads could facilitate the dispersal of M. alternatus (Table 3),
which is also consistent with the LCP model in which the host and road
landscapes had a lower resistance value. In Shunchang, the top model
showed that P. massoniana was the most suitable dispersal habitat
for this species, with the transect width of 600 m and the landscape
variable of P. elliottii , which could facilitate gene flow (Table
3). However, in the fifth to tenth models, except for the eighth model
in which roads had a positive effect, the other models all showed that
both farmland and urban landscapes had a negative effect on gene flow
(Table 3). In Xiapu, the top model suggested that mixed forest with
hosts was the most suitable dispersal habitat for this species, with the
transect width of 600 m and the landscape variable of urban, which had a
negative effect on gene flow with the highest resistance value of
29 (Table 3). The second and third models also showed
that urban landscapes inhibited gene flow (Table 3). In the fifth to
eighth models, P. elliottii was also a suitable habitat for this
species to disperse, whereas C. lanceolata had a negative effect
on gene flow (Table 3). Overall, in the two regions, the host and road
landscapes were likely conducive to the dispersal of M.
alternatus at a fine scale, whereas the nonhost landscapes had
significant negative effects on dispersal and gene flow in this species.