RESULTS
3. 1 Phylogenetic analysis
The CO1 -based phylogenetic analysis recovered two distinct
clusters, which were called ‘Tsushima’ and ‘Kuroshio’ lineages after the
corresponding ocean currents in the snails’ native range (Figure 2G).
The Hacheon (Korea) and Nemuro (Japan) individuals were part of the
Tsushima lineage, while the Matsushima (Japan) and Elkhorn Slough (USA)
individuals belonged to the Kuroshio lineage. These lineage assignments
agree with a previous study that identified the region of Japan most
likely to be the source of the introduced North American population
(Miura, Torchin, Kuris, Hechinger, & Chiba, 2006).
3.2 Multimodel inference and model
averaging
Based on the criteria of Z -scores > 0 andP -values < 0.05, the average parameter estimates
indicated that all the model terms were more or less positively related
to changes in snail movement distance, in which the es term was
the most important, followed by o (relative importance = 98%),
es × o (89%), and li (85%) (Table 1). This result is
also supported by the AICc scores (low to high) of the models that
included es, o, and li (models i,
ii, and iii, Table 2). In particular, the multimodel
inference test indicated that the most parsimonious model (based on the
lowest AICc score, 3416.97) was es × o + li (model i).
In accordance with the ΔAICc ≤ 6 cut-off rule, the second and third
best-fit models were es × o (model ii) and es
× li + o (model iii). These three models respectively received
76%, 14%, and 8% of the total weight wi . All
the other models, which omitted or included lo and p,
received higher AICc scores and lower weights, indicating that these
models are not important in describing the locomotion of salt-stressed
snails. Therefore, the multimodel analysis supports the conclusion that
the geographic distribution-origin and genetic composition substantially
influence the movement distance of snails in response to salinity
stress.
3.3 Locomotor performance changes upon
variables
The locomotor experiments showed that the intertidal snailBatillaria attramentaria from different locations was able to
acclimate to a range of salinity from 13 to 43 PSU. From these
experiments, we observed significant locomotion differences between the
snail groups exposed to 13 PSU and other treatments of 23, 33, and 43
PSU (LMMModelxi,F ES (1, 278) = 99.47, P< 0.0001, Table 3). Particularly, they significantly reduced
movement distance when transferred from normal salinity condition of 33
PSU to acute low salinity of 3 PSU but not significantly changed their
movement distance when transferred to moderately changed salinities of
23 and 43 PSU (Figure 3A). Post-hoc tests using the Tukey post-hoc
criterion for significance indicated that snails exposed to very low
salinity (13 PSU) moved significantly less than the other treatment
groups ( d 23PSU-13PSU = 0.43 ± 0.04
m1/2, d 33PSU-13PSU = 0.48 ±
0.04 m1/2, d 43PSU-13PSU = 0.45
± 0.04 m1/2, P < 0.0001, Appendix 1:
Table A2, Figure 3A). While, the moderately stressed snails (exposed to
23 and 43 PSU) moved slightly less than the control group (exposed to 33
PSU) but this difference was not statistically significant
(d 33PSU-23PSU = 0.05 ± 0.04
m1/2 and P = 0.5688,d 33PSU-43PSU = 0.03 ± 0.04 m1/2and P = 0.8637, and d 43PSU-23PSU = 0.02 ±
0.04 m1/2 and P = 0.9559).
Notably, when considering all treatment groups, we observed that the
origin factor had significant impacts on locomotor pace of the snails
with F O (1, 278) = 31.68 and P< 0.0001 (Model xii). Subsequent post-hoc test of
this analysis indicated that the native populations moved significantly
more than the introduced population
(d Native-Introduced = 0.28 ± 0.05
m1/2, P < 0.0001, Appendix 1: Table
A2, Figure 3B). Though, we did not record any significant differences in
movement distance between the two CO1 -lineages
(LMMModelxv,F li (1, 278) = 0.84, P = 0.3587,
Table 3) with d Tsushima-Kuroshio = 0.03 ± 0.04
m1/2 and P = 0.3597. Besides, we also found
significant locomotion differences based on location and population
(LMMModel xiii,F lo (1, 278) = 28.53, P< 0.0001 and LMMModel xiv,F p (1, 278) = 5.28, P = 0.0223,
Table 3). Subsequent Tukey post-hoc tests revealed that differences in
locomotor responses among native snail populations were not
statistically significant (d Korea-Japan = 0.06 ±
0.04 m1/2 and P = 0.3198,d Hacheon-Nemuro = -0.03 ± 0.05
m1/2 and P = 0.9426,d Hacheon-Matsushima = -0.09 ± 0.05
m1/2 and P = 0.2163,d Nemuro-Matsushima = -0.06 ± 0.05
m1/2 and P = 0.5226), but differences between
native snail locations and the introduced location were significant
(d Korea-USA = 0.24 ± 0.06 m1/2and P = 0.0001, d Japan-USA = 0.30 ± 0.05
m1/2 and P < 0.0001,d Hacheon-Elkhorn Slough = 0.24 ± 0.06
m1/2 and P = 0.0001,d Matsushima-Elkhorn Slough = 0.33 ± 0.06
m1/2 and P < 0.0001,d Nemuro-Matsushima = 0.27 ± 0.06
m1/2 and P < 0.0001, Appendix 1:
Table A2).
A linear mixed-effect model test of the best model (model i)
showed that the interaction of Origin (o) and Salinity Exposure
(es) (LMMModel i,F es × o (1, 275) = 11.59, P =
0.0008) was significant, and so was the effect of Lineage (li)
(LMMModel i,F li (1, 275) = 5.38, P = 0.0211,
Table 3). This result corresponds to the significant es × o
interaction and li term in the model outputs (Zes × o = 3.41, P < 0.001,
Table 1). However, when implemented separately, only es and
o significantly influenced the movement distance of the snails
independently (LMMModel xi,F ES (1, 278) = 99.47, P< 0.0001 and LMMModel xii,F O (1, 278) = 31.68, P< 0.0001, Table 3), while in contrast, li did not
(LMMModel xv,F li (1, 278) = 0.84, P = 0.36,
Table 3). Detailed differences in locomotion under the effect of the
interaction es × o + li estimated by Tukey post-hoc test can be
found in Appendix 1: Table A3.
3.4 Variation in shell length with distribution and CO1
lineage
We conducted a two-way ANOVA to examine the effect of geographic
distribution and genetic composition on the shell length of all 280
individuals included in the locomotor analyses. We confirmed that
introduced B. attramentaria individuals were significantly longer
than native ones (F o (1, 278) = 133.5,P value < 2e-16, Appendix 1: Table
A4A and Figure A2). Simple main effect analyses showed that the average
shell length of introduced snails was 31% longer
(lN ative = 2.14 cm,l Introduced = 3.12 cm, Appendix 1: Table A4B).
These analyses also revealed that snails from different locations or
populations also exhibited significant differences in shell length withF l (2, 277) = 193.7, p< 2e-16 andF P (3, 276) = 195.1, p< 2e-16 (Appendix 1: Table A4A). In
particular, the snail individuals from Korea were smallest and follow by
the Japan and the USA populations
(l Korea(Hacheon) = 1.68 cm,l Japan = 2.38 cm, in whichl Nemuro = 2.62 cm andl Matsushima = 2.13 cm, andl USA = 3.12 cm, Appendix 1: Table A4B).
Furthermore, shell length also significantly varied with lineageF li (1, 278) = 19.42; p =
1.5e-05 with l Tsushima = 2.15
cm and l Kuroshio = 2.62 cm, respectively
(Appendix 1: Table A4A and B), which is not surprising considering that
one of the two lineages includes the introduced (larger) individuals.