ECOLOGICAL CONCLUSIONS FROM SELECTIONFINDER 186

SelectionFinder   analysis   revealed   an   over-­‐representation   of   haplotypes   that   contain  genes  associated  with  responses  to  biotic  and  abiotic  factors,  indicating   the   predictive   power   of   the   application   for   investigating   adaptation   of   populations   to   local   environmental   conditions,   and   most   importantly,   for   informing  the  design  of  field  experiments  to  test  predictions.  For  example,  the   signatures   of   selection   at   known  R   genes   and   other   defense-­‐related   genes   indicate  that  A.  thaliana  populations  in  the  UK  have  been  affected  by  selection   pressure  from  microbial  activity.    

More   than   sixty   genes   containing   a   LRR   domain   were   found   as   potential   candidates  for  selection.  Of  those,  at  least  12  have  previously  been  reported  to   be   in   some   way   involved   in   the   regulation   of   induced   cell   death   (the   hypersensitive  response)  and/or  other  responses  to  microbial  infection.  While   some   of   the   genes   in   this   class   showed   evidence   of   sweeps   across   several   habitat  types,  the  majority  showed  evidence  of  sweeps  in  a  single  habitat  type.   This   may   suggest   that   local   adaptation   to   specific   habitat   types   occurs   across   the  UK,  and  in  turn  suggests  significant  differences  in  both  the  set  of  pathogenic  

species   and   the   environmental   conditions   encountered   across   different   habitats.    

A   number   of   the   LRR   genes   found   to   possess   signatures   of   selection   control   aspects  of  the  phenotype  relating  to  the  abiotic  conditions  of  the  environment.   For  example,  3  genes  -­‐  AT1G05700,  AT1G17610  and  AT3G20600  (see  (Kreps  et  al.   2002;  Wang  et  al.  2013))  –  have  previously  been  reported  to  confer  tolerance  to   cold   conditions;   intriguingly,   within   each   habitat   type,   only   one   of   these   three   genes   exhibits   signatures   of   selection.   This   may   constitute   an   example   of   local   adaptation.    

A   signature   of   selection   corresponding   to   the   gene   AT2G26290   was   found   in   the   population   in   the   wall/rocky   outcrop   habitat.   The   kinase   produced   by   this   gene   becomes  activated  in  conditions  of  dehydration-­‐  or  salt-­‐stress.  Since  wall/outcrop   habitats  are  likely  to  retain  less  water  than  other  habitats  (due  to  the  relative  lack   of  soil),  it  is  reasonable  to  propose  that  selection  pressures  more  strongly  favour   alleles  conferring  an  improved  tolerance  to  dehydration  in  the  wall/outcrop  habitat   than  in  other  habitats.  

Several   signatures   of   selection   were   discovered   at   loci   associated   with   flowering   time,  particularly  within  the  garden  habitat  type  (see  Appendix  6.2),  and  also  at  a   number   of   LRR   loci   associated   with   growth   and   development   –   for   example,   AT1G75820   (associated   with   root   development   (Stahl   et   al.   2013)),   AT3G49670   (associated  with  anther  and  meristem  development  (DeYoung  et  al.  2006;  Hord  et   al.   2006))   and   AT4G20270   (also   associated   with   meristem   development   and   cold   tolerance  (DeYoung  et  al.  2006)).  The  Ka/Ks  ratios  at  all  but  one  of  the  flowering   time-­‐associated  loci  indicate  strong  positive  selection.  In  conjunction  with  the   conclusions   regarding   optimum   points   on   the   r/K   spectrum   discussed   below,   this  may  suggest  the  presence  of  unique  selection  pressures  for  variation  in  life   history  traits  within  this  habitat  type.    

A  fourth  group  of  selection  signatures  corresponds  to  a  set  of  several  genes  of   the   DAR   family   -­‐  AT5G66610,   AT5G66620,   AT5G66630   and   AT5G66640   -­‐   which   play   a   role   in   the   control   of   organ   and   seed   size   (Li   et   al.   2008).   The   gene  

AT2G39830  also  exhibits  a  similar  signature  of  selection,  and  is  also  a  determinant   of  seed  size  (Li  et  al.  2008).  This  signature  of  selection  does  not  appear  across  all   habitats.   Larger   seeds   require   parents   to   dedicate   a   greater   amount   of   resources   towards   the   growth   of   their   offspring,   which   also   determines   the   number   of   offspring   any   given   individual   is   ultimately   capable   of   producing.   This   may,   therefore,   indicate   that   different   habitats   favour   different   optimal   points   in   the   trade-­‐off   between   investment   of   resources   in   offspring   and   in   growth,   and   consequently   different   points   along   the   r/K   strategy   continuum.   The   degree   of   habitat   variability   is   known   to   affect   the   optimal   point   of   this   trade-­‐off;   experimentation   has   shown   that   a   greater   degree   of   environmental   variability   favours   a   strategy   of   higher   fecundity,   necessitating   a   shift   in   investment  of  resources  away  from  the  growth  and  longevity  of  the  parent  and   towards   investment   in   larger   numbers   of   offspring   (Rose   &   Charlesworth   1980).    

Interestingly,   the   signature   of   selection   for   the  DAR   genes   was   detected   in   garden   habitats   (which   undergo   a   substantial   degree   of   human-­‐caused   disturbance)  but  not  in  low-­‐disturbance  ‘wall/outcrop’  habitats.    A  hypothetical   explanation  may  be  that  A.  thaliana  populations  sampled  from  garden  habitats   have  been  undergoing  selection  at  loci  affecting  seed  production  in  response  to   factors   associated   with   habitat   disturbance   (e.g.,   release   of   nutrients   from   cultivation),   whereas   populations   that   were   sampled   from   wall/outcrop   habitats  are  fully  adapted  for  survival  in  the  harsh,  low  nutrient  conditions  and   consequently  genetically  uniform  at  the  same  loci.  

This   variation   in   selection   for   r/K   strategies   may   explain   the   apparent   migratory   history  of  the  UK  population  inferred  from  PCA  clustering  of  genotypes  in  Chapter  

smaller   numbers   of   seeds   than   the   population-­‐wide   average   (see   Chapter  4.4.1).   Since  human  disturbance  of  the  environment  at  these  sites  is  minimal,  a  population   that  establishes  at  such  sites  could  have  adapted  to  become  highly  specialised  to   the   habitat   type   represented   by   walls   and   rocky   outcrops   (and,   thus,   also   somewhat   genotypically   differentiated   from   mainland   populations).   As   humans   also   created   other,   more   disturbed   habitats   –   represented   by   gardens   –   in   more   recent   times,   there   would   therefore   exist   selection   pressures   in   these   habitats   towards  an  r  strategy,  but  the  populations  in  less  disturbed  habitats  would  remain   unaffected,  and  would  therefore  not  show  any  signatures  of  selection  due  to  their   already  high  degree  of  adaptation  to  those  habitats.  This  plausible  scenario  may  be   tested   by   common   garden   experiments   (see   below.   If   supported   by   further   evidence,  this  demonstrates  that  A.  thaliana  is  a  suitable  case  study  for  the  long-­‐ term   effects   of   human   actions   on   the   selection   pressures   exerted   on   wild   populations  by  their  environments.  

Researchers   wishing   to   further   investigate   instances   of   local   adaptation   should   design  common  garden  experiments  using  pools  of  samples  drawn  from  all  relevant   habitat   types.   For   example,   an   experiment   to   further   investigate   the   potential   multiple   instances   of   local   adaptation   in   favour   of   cold   tolerance   could   involve   sowing  seeds  collected  from  representative  populations  within  each  habitat  type,   at  a  set  of  common  gardens  across  all  3  habitat  types.  Genotypic  assays  of  alleles  at   the   3   listed   loci   taken   over   the   course   of   several   generations   would   then   reveal   which  alleles  are  favoured  in  each  circumstance,  and  would  also  more  clearly  show   the  type  of  selection  occurring  (i.e.,  balancing,  directional,  stabilising,  etc.);  further   hypotheses  may  then  be  proposed  as  to  why  certain  alleles  are  favoured  in  a  given   situation.    

4.4.3  PLANT-­‐PATHOGEN  INTERACTION  CONCLUSIONS  FROM