Page 10 - Asylv_molars_Pmax_revised_2014_01
P. 10

10	
  

	
  

A	
  latitudinal	
  trend	
  in	
  molar	
  shape	
  was	
  previously	
  described	
  in	
  the	
  European	
  wood	
  mouse	
  (Renaud	
  
and	
  Michaux	
  2007).	
  It	
  was	
  confirmed	
  here	
  on	
  the	
  total	
  mainland	
  data	
  set	
  (multivariate	
  regression	
  of	
  
FCs	
  vs.	
  latitude:	
  N	
  =	
  413,	
  P	
  <	
  0.0001).	
  The	
  latitudinal	
  trend	
  in	
  molar	
  shape	
  was	
  parallel	
  to	
  all	
  
documented	
  Pmax.	
  Although	
  discrete,	
  it	
  corresponds	
  to	
  a	
  similar	
  trend	
  ranging	
  from	
  broad	
  to	
  slender	
  
molars	
  (Fig.	
  2B).	
  

	
  

Allometric	
  variations	
  

Significant	
  molar	
  size	
  variations	
  were	
  documented	
  across	
  geographic	
  groups.	
  Trends	
  in	
  molar	
  size	
  
may	
  contribute	
  to	
  the	
  pattern	
  in	
  molar	
  shape	
  if	
  both	
  are	
  coupled	
  by	
  allometric	
  mechanisms.	
  Overall,	
  
intra-­‐population	
  allometry	
  was	
  very	
  limited,	
  being	
  non-­‐significant	
  in	
  three	
  out	
  of	
  five	
  populations	
  
(Lantabat:	
  P	
  =	
  0.461;	
  Tarquinia:	
  P	
  =	
  0.206;	
  Porquerolles:	
  P	
  =	
  0.085).	
  The	
  two	
  significant	
  intra-­‐
population	
  allometric	
  relationships	
  (Mimizan:	
  P	
  =	
  0.013;	
  Tourch:	
  P	
  <	
  0.001)	
  were	
  not	
  robustly	
  
estimated,	
  not	
  related	
  together	
  and	
  not	
  related	
  to	
  Pmax	
  (Table	
  5;	
  Fig.	
  4).	
  	
  

The	
  inter-­‐population	
  allometric	
  direction	
  was	
  estimated	
  on	
  mainland,	
  based	
  on	
  group	
  means	
  and	
  on	
  
all	
  specimens,	
  both	
  providing	
  robust	
  and	
  similar	
  estimates	
  of	
  the	
  allometric	
  direction	
  (Table	
  5).	
  These	
  
directions	
  were	
  not	
  related	
  to	
  Pmax,	
  and	
  were	
  not	
  or	
  only	
  weakly	
  related	
  to	
  intra-­‐group	
  allometry	
  
(Table	
  5).	
  Overall,	
  this	
  suggests	
  a	
  limited	
  role	
  of	
  allometry	
  in	
  molar	
  shape	
  differences.	
  	
  

	
  

Pmax	
  and	
  insular	
  divergence	
  

Neighboring	
  mainland	
  localities	
  were	
  considered	
  as	
  reference	
  for	
  evaluating	
  the	
  importance	
  of	
  
insular	
  divergence.	
  All	
  islands	
  documented	
  by	
  a	
  single	
  specimen	
  were	
  not	
  further	
  investigated.	
  
Porquerolles	
  emerged	
  as	
  well	
  differentiated	
  among	
  the	
  Hyères	
  islands	
  while	
  Port-­‐Cros	
  was	
  not	
  (FR-­‐SE	
  
vs.	
  Port-­‐Cros:	
  	
  PMANOVA	
  =	
  0.100;	
  vs.	
  Porquerolles:	
  P	
  <	
  0.001).	
  Ibiza	
  was	
  not	
  differentiated	
  from	
  
mainland	
  Spain	
  (vs.	
  SP-­‐E:	
  P	
  =	
  0.080).	
  Among	
  islands	
  related	
  to	
  the	
  Italian	
  clade,	
  Corsica	
  was	
  weakly	
  
differentiated	
  while	
  Sardinia	
  was	
  clearly	
  divergent	
  (IT-­‐CW	
  vs.	
  Corsica:	
  P	
  =	
  0.031;	
  vs.	
  Sardinia:	
  P	
  <	
  
0.001).	
  Among	
  Atlantic	
  French	
  islands,	
  Oléron	
  and	
  Ré	
  were	
  significantly	
  divergent	
  (FR-­‐W	
  vs.	
  Oléron:	
  P	
  
<	
  0.001;	
  vs.	
  Ré:	
  P	
  <	
  0.001).	
  Noirmoutier	
  molars	
  were	
  not	
  significantly	
  differentiated	
  from	
  the	
  
continent	
  (P	
  =	
  0.310).	
  Finally,	
  Sicily	
  constitutes	
  a	
  peculiar	
  case	
  of	
  an	
  old	
  endemic	
  lineage,	
  related	
  to	
  
the	
  Italian	
  clade	
  from	
  which	
  it	
  significantly	
  differ	
  in	
  molar	
  shape	
  (vs.	
  IT:	
  P	
  =	
  0.010).	
  	
  

For	
  the	
  insular	
  populations	
  showing	
  a	
  significant	
  divergence,	
  the	
  direction	
  of	
  differentiation	
  was	
  
compared	
  to	
  Pmax	
  (Table	
  6;	
  Fig.	
  4).	
  In	
  all	
  cases	
  except	
  Sardinia,	
  the	
  insular	
  divergence	
  followed	
  the	
  
   5   6   7   8   9   10   11   12   13   14   15