TURUN YLIOPISTON JULKAISUJA – ANNALES UNIVERSITATIS TURKUENSIS Sarja - ser. AII osa - tom. 288 | Biologica - Geographica - Geologica | Turku 2014 Johanna M. Toivonen DETERMINANTS OF POLYLEPIS (ROSACEAE) FOREST DISTRIBUTION AND TREELINE FORMATION IN THE HIGH ANDES Supervised by Dr. Michael Kessler Institute of Systematic Botany University of Zurich, Switzerland Dr. Kalle Ruokolainen Department of Biology University of Turku, Finland University of Turku Faculty of Mathematics and Natural Sciences Department of Biology Section of Biodiversity and Environmental Science Reviewed by Dr. Paul M. Ramsay Associate Professor of Ecology School of Biological Sciences Plymouth University, UK Prof. William K. Smith Biology Department Wake Forest University Winston-Salem, North Carolina, USA Opponent Dr. Arne Cierjacks Interim Professor Department Biology University of Hamburg, Germany Cover photograph by Johanna M. Toivonen 2006 The originality of this thesis has been checked in accordance with the University of Turku quality assurance system using the Turnitin OriginalityCheck service. ISBN 978-951-29-5744-6 (PRINT) ISBN 978-951-29-5745-3 (PDF) ISSN 0082-6979 Painosalama Oy - Turku, Finland 2014                                       For  an  understanding  of  the  success  and  defeat  of   tree  life  at  high  elevations,     “Trees  gets  small  before  they  get  rare”,   and  finally,   “The  non-­adapted  are  absent”.     Christian  Körner      This  thesis  is  based  on  the  following  publications  and  manuscripts,  referred   to  in  the  text  by  their  Roman  numerals.   I   Toivonen,  J.M.,  Horna,  V.,  Kessler,  M.,  Ruokolainen,  K,  &  Hertel,   D.  2014.  Interspecific  variation  in  functional  traits  in  relation  to   species  climatic  niche  optima  in  Andean  Polylepis  (Rosaceae)  tree   species:  evidence  for  climatic  adaptations.  Functional  Plan  Biology   41:  301-­312.   II   Toivonen,  J.M.,  Gonzales-­Inca,  C.A.,  Ruokolainen,  K.,  Kessler,  M.   Terrain  features  shaping  the  spatial  distribution  patterns  of   Polylepis  forest  stands  in  the  high  Andes  of  southern  Peru.   Manuscript.   III   Kessler,  M.,  Toivonen,  J.M.,  Sylvester,  S.,  Kluge,  J.  &  Hertel,  D.   2014.  Tree  height  of  Polylepis  forests  (Rosaceae)  in  the  high  Andes   of  Peru:  effect  of  elevation,  climatic  conditions  and  human  impact.   Frontiers  in  Plant  Science  5:194.  doi:  10.3389/fpls.2014.00194   (uncorrected  proof).   IV   Toivonen,  J.M.,  Kessler,  M.,  Ruokolainen,  K.  &  Hertel,  D.  2011.   Accessibility  predicts  structural  variation  of  Andean  Polylepis   forests.  Biodiversity  and  Conservation  20:  1789–1802.   Articles  I,  III  and  IV  are  reprinted  with  the  permission  of  CSIRO   Publishing,  Frontiers  Science  Production  Office  and  Springer   Science+Business  Media,  respectively.       Table  of  contribution.     I   II   III   IV   Original  idea   JMT,  DH,  VH   JMT,  CGI   JMT,  MK   JMT,  KR,  MK   Field  work   JMT   JMT   JMT,  SS   JMT   Lab  work   JMT   -­   -­   -­   GIS  analysis   -­   JMT,  CGI   -­   -­   Statistics   JMT   JMT   JK   JMT,  KR   Writing  the   manuscript   JMT,  DH,  KR   JMT   MK,  JMT   JMT,  KR,  MK   Commenting   on  the   manuscript   MK,  VH   KR,  MK,  CGI   DH,  SS,  JK   DH   JMT  =  Johanna  M.  Toivonen,  KR  =  Kalle  Ruokolainen,  MK  =  Michael  Kessler,  DH  =   Dietrich  Hertel,  VH=  Viviana  Horna,  CGI  =  Carlos  Gonzales-­Inca,  SS  =  Steven  Sylvester,   JK  =  Jürgen  Kluge      TABLE  OF  CONTENTS   ABSTRACT  ..........................................................................................................  6   RESUMEN  ............................................................................................................  7   TIIVISTELMÄ  ....................................................................................................  8   1.  INTRODUCTION...........................................................................................  9   1.1.  HIGH  ELEVATION  TREELINES  ..........................................................................  9   1.2.  HIGH  ELEVATION  ADAPTATIONS  IN  TREES  ...............................................  11   1.3.  DISTRIBUTION  OF  HIGH-­ANDEAN  POLYLEPIS  FORESTS  ........................  13   1.3.1.  Environmental  constraints  on  Polylepis  tree  growth  and                        forest  distribution  ........................................................................................  13   1.3.2.  Anthropogenic  influence  on  Polylepis  forest  structure  and                          distribution.......................................................................................................  15   1.4.  THE  AIMS  OF  THE  THESIS  ...............................................................................  17   2.  MATERIAL  AND  METHODS  ................................................................  19   2.1.  STUDY  AREA  AND  SPECIES  ............................................................................  19   2.2.  COMMON  GARDEN  EXPERIMENT  .................................................................  21   2.3.  DETERMINATION  OF  SPECIES  CLIMATIC  NICHE  OPTIMA  .......................  22   2.4.  GIS  AND  REMOTE  SENSING  METHODS  .......................................................  23   2.5.  MEASUREMENTS  OF  FOREST  STRUCTURE  AND  MICROCLIMATE  ........  24   2.6.  NUMERICAL  ANALYSES  ..................................................................................  25   3.  RESULTS  AND  DISCUSSION  ................................................................  27   3.1.  CLIMATIC  ADAPTATIONS  IN  FUNCTIONAL  TRAITS  OF  POLYLEPIS                      SPECIES................................................................................................................  27   3.2.  TERRAIN  FEATURES  AND  POLYLEPIS  FOREST  DISTRIBUTION  ..............  28   3.3.  POLYLEPIS  TREE  HEIGHT,  ELEVATION,  CLIMATE  AND  HUMAN                    IMPACT  ................................................................................................................  29   3.4.  ACCESSIBILITY  AND  STRUCTURAL  VARIATION  OF  POLYLEPIS                      FORESTS  ..............................................................................................................  30   4.  CONCLUSIONS  ..........................................................................................  32   5.  ACKNOWLEDGEMENTS  .......................................................................  34   6.  REFERENCES  .............................................................................................  36       Abstract     6   ABSTRACT   High   elevation   treelines   are   formed   under   common   temperature   conditions   worldwide,  but  the  functional  mechanisms  that  ultimately  constrain  tree  growth  are   poorly   known.   In   addition   to   environmental   constraints,   the   distribution   of   high   elevation   forests   is   largely   affected   by   human   influence.   Andean   Polylepis   (Rosaceae)   forests   are   an   example   of   such   a   case,   forests   commonly   growing   in   isolated  stands  disconnected   from   the   lower  elevation  montane  forests.  There  has   been  ample  discussion  as  to  the  role  of  environmental  versus  anthropogenic  causes   of  this  fragmented  distribution  of  Polylepis  forests,  but  the  importance  of  different   factors   is   still   unclear.   In   this   thesis,   I   studied   functional,   environmental   and   anthropogenic   aspects   determining   Polylepis   forest   distribution.   Specifically,   I   assessed   the   degree   of   genetic   determinism   in   the   functional   traits   that   enable   Polylepis   species   to   grow   in   cold   and   dry   conditions.   I   also   studied   the   role   of   environment   and   human   influence   constraining   Polylepis   forest   distribution.   I   found   evidence   of   genetically   determined   climatic   adaptations   in   the   functional   traits  of  Polylepis.  High  elevation  species  had  reduced  leaf  size  and  increased  root   tip  abundance  compared  to  low  elevation  species.  Thus  these  traits  have  potentially   played   an   important   role   in   species   evolution   and   adaptation   to   high   elevation   habitats,   especially   to   low  temperatures.   I   also  found  reduced  photosynthesis   rate   among  high  elevation   tree   species   compared   to   low  elevation   species,   supporting   carbon  source  limitation  at  treelines.  At  low  elevations,  Polylepis  forest  distribution   appeared  to  be  largely  defined  by  human  influence.  This  suggests  that  the  absence   of   Polylepis   forests   in   large   areas   in   the   Andes   is   the   result   of   several   environmental   and   anthropogenic   constraints,   the   role   of   environment   becoming   stronger   towards   high   elevations.   I   also   show   that   Polylepis   trees   grow   at   remarkably  low  air  and  soil  temperatures  near  treelines,  and  present  new  evidence   of   the   role   of   air   temperatures   in   constraining   tree   growth   at   high   elevations.   I   further  show  that  easily  measurable  indices  of  accessibility  are  related  to  the  degree   of   degradation   of   Polylepis   forest,   and   can   therefore   be   used   in   the   rapid   identification  of  potentially  degraded  Polylepis  forests.  This  is  of  great  importance   for  the  conservation  and  restoration  planning  of  Polylepis  forests  in  the  Andes.  In  a   global  context,  the  results  of  this  thesis  add  to  our  scientific  knowledge  concerning   high  elevation   adaptations   in   trees,   and   increase  our  understanding  of   the   factors   constraining   tree   growth   and   forest   distribution   at   high-­elevation   treelines   worldwide.   Resumen     7   RESUMEN   El   límite   altitudinal   arbóreo   está   formado   bajo   condiciones   similares   de   temperatura   a   nivel   mundial,   pero   los   mecanismos   funcionales   que   finalmente   constriñen   el   crecimiento   de   los   árboles   son   poco   conocidos.   Además   de   los   factores  ambientales,  la  distribución  de  los  bosques  de  alturas  está  afectada  por  las   acciones  humanas.  Los  bosques  andinos  de    Polylepis  (Rosaceae)  son  un  ejemplo   de   este   caso:   estos   bosques   crecen   en   rodales   aislados   y   desconectados   de   los   bosques  montanos  a  menor  altitud.  Ha  habido  una  amplia  discusión  sobre  el  rol  de   los  factores  ambientales  y  antropogénicos  que  causan  esta  distribución  fragmentada   de   los   bosques   de   Polylepis,   pero   la   importancia   de   diferentes   factores   no   está   todavía   clara.   En   esta   tesis   estudié   los   aspectos   ecofisiológicos,   ambientales   y   antropogénicos  que  pueden  determinar  la  distribución  de  los  bosques  de  Polylepis.   Específicamente,   evalué   el   grado   del   determinismo   genético   sobre   las   características   ecofisiológicas   que   permiten   a   las   especies   de  Polylepis  crecer   en   condiciones  frías  y  secas.  Además,  estudié  el  rol  de  los  factores  medio  ambientales   y   antropogénicos   que   restringen   la   distribución  de   los   bosques   de  Polylepis.  Mis   resultados  dan  evidencias  de  adaptaciones  climáticos  genéticamente  determinadas   en  las  características  ecofisiológicas  de   las  especies  de  Polylepis.  Las  especies  de   las  zonas  altas  presentan  hojas  de  tamaño  reducido  e  incremento  en  la  abundancia   de  ápices  radiculares  en  comparación  con  las  especies  de  las  zonas  más  bajas.  Estas   características  parecen  haber  desempeñado  un  rol  importante  en  la  evolución  de  las   especies   y   en   la   adaptación   a   los   hábitats   de   mayor   altitud,   especialmente   a   temperaturas   bajas.   Asimismo,   encontré   una   tasa   reducida   de   fotosíntesis   en   las   especies  de  las  zonas  altas  en  comparación  con  las  especies  de  las  zonas  más  bajas.   Esto   apoya   la   hipótesis   de   limitación   de   adquisición   de   carbono   en   el   límite   altitudinal   arbóreo.   A  menor   altitud,   la   distribución   de   los   bosques   de  Polylepis   parece  estar  fuertemente  afectada  por  las  acciones  antropogénicas.  Esto  sugiere  que   la   ausencia   de   los   bosques   de   Polylepis   en   la   mayor   parte   de   los   Andes   es   el   resultado  de  una  serie  de   limitaciones  ambientales  y  antropogénicas,  donde  el   rol   de  los  factores  ambientales  se  incrementa    hacia  las  zonas  altas.  También  encontré   que   los   árboles   de   Polylepis   cerca   del   límite   altitudinal   arbóreo   crecen   a   considerablemente  bajas  temperaturas  del  aire  y  del  suelo.  Además  presento  nuevas   evidencias  del  rol  de  la  temperatura  del  aire  en  la  limitación  del  crecimiento  de  los   árboles   en   las   zonas   altas.   Adicionalmente,   demuestro   que   los   índices   de   accesibilidad   que   son   fácilmente   medibles   están   relacionados   con   el   grado   de   degradación   de   los   bosques   de   Polylepis   y   pueden   ser   aplicados   para   una   identificación  rápida  de  los  bosques  degradados.  Esto  es  de  gran  importancia  para   la  conservación  y  planificación  de  restauración  de  los  bosques  de  Polylepis  en  los   Andes.   En   el   contexto   mundial,   los   resultados   de   esta   tesis   contribuyen   a   incrementar  el  conocimiento  científico  relacionado  a  las  adaptaciones  de  especies   arbóreas  en  hábitats  montañosos  y  el  entendimiento  de  los  factores  que  limitan  el   crecimiento  de  los  árboles  y  la  distribución  de  los  bosques  en  el  límite  altitudinal   arbóreo  a  nivel  mundial.   Tiivistelmä     8   TIIVISTELMÄ   Vuoristopuurajat   esiintyvät   maailmanlaajuisesti   samanlaisissa   lämpötila-­ olosuhteissa,  mutta  puiden  kasvua  rajoittavat  ekofysiologiset  mekanismit  tunnetaan   kuitenkin   huonosti.   Ympäristöolosuhteiden   lisäksi   ihmisvaikutuksella   on   myös   suuri  merkitys  vuoristometsien  levinneisyyteen.  Andien  Polylepis-­metsät  ovat  yksi   esimerkki   tällaisesta   tapauksesta.  Metsät   kasvavat   eristyneinä   laikkuina   alemman   yhtenäisen   vuoristometsävyöhykkeen   yläpuolella.   Ympäristöolosuhteiden   ja   ihmisvaikutuksen  merkitystä  Polylepis-­metsien  levinneisyydelle  on  tutkittu  paljon,   mutta  eri  tekijöiden  merkitys  metsien  levinneisyydelle  on  edelleen  epäselvä.  Tutkin   väitöskirjassani   puiden   ekofysiologiaan,   ympäristöolosuhteisiin   ja   ihmis-­ vaikutukseen   liittyviä   tekijöitä,   jotka   voivat   selittää   Polylepis-­metsien   levin-­ neisyyttä.   Tulokseni   osoittivat,   että   osa   tutkimistani   Polylepis-­puiden   eko-­ fysiologisista  piirteistä   ilmensi  geneettisesti  määräytyneitä   ilmastollisia   sopeumia:   korkean   paikan   lajeilla   oli   pienemmät   lehdet   ja   enemmän   juurenkärkiä   kuin   matalampien   paikkojen   lajeilla.   Näillä   piirteillä   on   todennäköisesti   ollut   tärkeä   merkitys   lajien   evoluutiossa   ja   sopeutumisessa   ylävuoriston   elinympäristöihin,   erityisesti   mataliin   lämpötiloihin.   Tuloksieni   mukaan   korkean   paikan   lajeilla   oli   myös  alentunut  fotosynteesikapasiteetti  matalampien  paikkojen  lajeihin  verrattuna.   Tämä   tukee   hypoteesia   hiilen   sitomiseen   liittyvistä   rajoitteista   vuoristopuurajalla.   Alemmilla   korkeuksilla   ihmistoiminnalla   näytti   olevan   erityisen   suuri   merkitys   Polylepis-­metsien   levinneisyyteen.   Tuloksieni   perusteella   voidaan   todeta,   että   Polylepis-­metsien   nykylevinneisyys   on   seurausta   useasta   eri   ympäristöön   ja   ihmistoimintaan   liittyvästä   tekijästä,   joista   ympäristön   merkitys   levinneisyyden   selittäjänä   kasvaa   siirryttäessä   kohti   ylävuoristoa.   Tutkimustulokseni   myös   osoittivat,   että   Polylepis-­puut   kasvavat   huomattavan   alhaisissa   lämpötila-­ olosuhteissa  lähellä  puurajaa,  sekä  ilman  että  maaperän  lämpötilan  suhteen.  Lisäksi   tulokseni   osoittivat,   että   erityisesti   ilman   lämpötilalla   on   merkitystä   puiden   kasvulle   vuoristossa.   Havaitsin   myös,   että   helposti   mitattavat,   metsien   saavutettavuuteen   perustuvat   indeksit   selittävät   Polylepis-­metsien   rakennetta,   ja   näin   ollen  myös  metsien   kunnon   heikkenemistä.  Näitä   indeksejä   voidaan   käyttää   tunnistamaan   nopeasti   ja   helposti   alueita,   joilla   Polylepis-­metsät   ovat   huonokuntoisimpia.   Tällä   tiedolla   on   merkitystä   Polylepis-­metsien   suojelu-­   ja   kunnostussuunnitelmille   Andeilla.   Väitöskirjatyöni   tulokset   lisäävät   myös   tietoa   puiden   sopeutumisesta   vuoristo-­olosuhteisiin   ja   auttavat   ymmärtämään   eri   tekijöiden   merkitystä   puiden   kasvulle   ja   metsien   levinneisyydelle   vuoristo-­ puurajoilla  maailmanlaajuisesti.       Introduction     9   1.   INTRODUCTION   High   elevation   treelines   are   formed   at   the   elevation   above   which   tree   growth   is   prevented  by  environmental  constraints,  and  tree  stature  changes  to  shrubs  or  other   low  stature  alpine  vegetation  (Körner  2012).  The  specific  elevation  where  the  treeline   is  formed  varies  according  to  latitude  and  specific  local  features,  but  there  appear  to   be   certain   globally   common   temperature   conditions   for   high   elevation   treelines:   a   mean  growing  season  soil   temperature  ranging  from  5   to  8°C  (Körner  1998,  2003;;   Körner   &   Paulsen   2004;;   Wieser   &   Tausz   2007)   and   averaging   6.7   °C±0.8   SD   (Körner  &   Paulsen   2004).   Nevertheless,   after   more   than   a   hundred   years   of   high   elevation   treeline   studies   (e.g.  Schröter  1908;;  Troll   1961,  1973;;  Tranquillini   1979;;   Lauer  1982;;  Miehe  &  Miehe  1994;;  Körner  1998;;  Jobbagy  &  Jackson  2000),  both  the   physiological   mechanisms   constraining   tree   growth   and   the   functional   adaptations   enabling  trees  to  cope  with  low  temperatures  remain  unclear.   Mountains  are  environmentally  demanding  habitats  for  plants,  and  are  also  subject   to  strong  human  influence  around  the  world.  Activities  associated  with  human  land   use  play  a  significant  role  in  determining  the  spatial  patterns  of  natural  vegetation   in  mountain  areas.  The  high  Andean  tree  genus  Polylepis  (Rosaceae)  provides  an   example   of   this   interplay   between   environmental   and   anthropogenic   factors   affecting   vegetation.   However,   the   relative   importance   of   environmental   versus   anthropogenic   factors   in  determining   the  distribution  of  Polylepis   forests   remains   to   be   clarified.   The   genus   Polylepis   consists   of   several   evergreen   tree   species   adapted   to   different   temperature   and   humidity   conditions,   from   relatively   warm   and  wet   cloud   forests   to   cold   and   dry   high  mountains   (Simpson   1986;;   Schmidt-­ Lebuhn   et   al.   2006).   The   highest   forest   stands   grow   near   5000   m   of   elevation,   forming   one   of   the   highest   alpine   treelines   worldwide.   Forest   stands   with   easy   access,  usually  referring  to  stands  at  lower  elevations,  are  extensively  affected  by   human  land  use  (Kessler  1995,  2000;;  Fjeldså  &  Kessler  1996;;  Hagaman  2006).  For   these  reasons,  Polylepis  forests  form  an   interesting  study  system  for   investigating   the  role  of  functional,  environmental  and  anthropogenic  factors  determining  forest   distribution.   1.1.  High  elevation  treelines   The   high   elevation   treeline   denotes   the   upper   elevational   limit   for   tree   growth,   survival  and  reproduction.  It  usually  forms  a  natural  ecological  transition  zone,  an   ecotone,  with  a  gradual  shift  from  high  statured  trees  to  lower  statured  shrubs  and   other   alpine   plants   (Körner   2003,   2012;;   Wieser   &   Tausz   2007).   Commonly,   “treeline”  refers  to  a  boundary  at  which  tree  height  drops  below  3  m  (Körner  2003;;   Smith   et   al.   2003)   and   closed   forest   changes   to   fragmented   forest   patches   surrounded   by   alpine   pastures   (Tranquillini   1979;;   Holtmeier   2009).   However,   abrupt  high  elevation  treelines  have  also  been  documented  (e.g.  Miehe  et  al.  2007).   Other  high  elevation  tree  boundaries  can  also  be  defined:  “timberline”  commonly   refers  to  the  end  of  closed  forests  with  timber-­size  trees,  while  “tree  species  line”   Introduction     10   refers  to  the  upper  limit  of  tree  existence;;  this  may  conflict  with  the  definition  of  a   tree,   i.e.   a  woody   plant   at   least   3  m   in   height   (Körner   2003;;   Smith   et   al.   2003;;   Körner  &  Paulsen  2004)  (Figure  1).       Figure  1.  Conceptual  representation  of  timberline,   treeline  and  tree  species  line  (modified   from  Körner  &  Paulsen  2004).   Global   data   on   high   elevation   treeline   locations   indicate   an   increasing   latitudinal   trend  in  treeline  elevations  from  the  poles  towards  the  subtropics,  with  a  decline  at   the   equatorial   zone   (e.g.   Troll   1973;;   Körner   1998,   2007).   Continentality   also   affects  treeline  elevation;;  trees  grow  at  higher  elevations  in  the  warmer  and  sunnier   inner  parts  of  mountain  areas  than  in  the  outer  parts,  which  are  exposed  to  strong   winds   and   rains.   This   phenomenon   is   also   known   as   mass   elevation   effect,   discussed  in  detail  for  example  by  Schröter  (1908)  and  Barry  (1981).     Temperature   is   regarded   as   a   main   driver   of   high   elevation   treeline   formation   globally   (Körner   1998,   2003,   2012;;  Wieser   &   Tausz   2007),   but   the   underlying   physiological   mechanisms   of   tree   function   are   still   unclear,   as   is   the   question   whether   soil   or   air   temperatures   are   decisive   for   treeline   formation.   Körner   and   Paulsen   (2004)   found   that   at   high   elevation   treelines   the   mean   soil   temperature   during   the   growing   season   –   defined   as   the   period   during   which   the   soil   temperature   at   10   cm   depth   consistently   exceeds   3.2°C   –   universally   averages   6.7  °C,  but  substantial  regional  and  taxonomic  variation  was  also  observed  (Körner   &   Paulsen   2004).   Limited   water   and   nutrient   uptake   by   the   fine   root   system   stresses   the   role  of   soil   temperature   in   the   formation   of   a   high   elevation   treeline   (Wilson  et  al.  1987;;  Leuschner  et  al.  2007;;  Hertel  et  al.  2008;;  Hertel  and  Schöling   2011;;   Körner   2012),   while   the   strong   atmospheric   coupling   of   high   tree   stature   highlights   the   role   of   air   temperatures   (e.g.  Hadley  &  Smith   1987;;  Wilson  et   al.   1987;;   Grace   et   al.   1989;;   Körner   2003).   Körner   (1998)   and   Smith   et   al.   (2003)   Introduction     11   summarize   the   following   five   main   hypotheses   concerning   potential   functional   explanations  for  high  elevation  treeline  formation  worldwide:       1)   the   stress   hypothesis,   according   to   which   the   growth   in   particular   of   young   trees  is  suppressed  by  physiological   tissue  damage  at  cold  and  dry   conditions     2)   the   disturbance   hypothesis,   according   to   which   the   decisive   factor   limiting   tree   growth   involves   mechanical   damages   due   to   wind,   snow   accumulation,  landslides,  avalanches,  fungi  infection  and  herbivory     3)   the   reproduction   limitation   hypothesis,   emphasizing   the   role   of   decreased   pollination,   seed   development,   seed   dispersal,   germination   and   seedling  establishment     4)   the   negative   carbon   balance   hypothesis,   which   suggests   that   carbon   acquisition   is   insufficient   to   maintain   minimum   growth   after   respiratory   carbon  loss     5)   the   growth   limitation   hypothesis,   according   to   which   impeded   tree   growth  is  attributable  to  reduced  carbon  investment  in  the  development  of   new  tissues  at  low  temperatures.     In   searching   for   a   comprehensive,   globally   applicable   explanation   for   high   elevation  treeline  formation,  the  hypotheses  which  have  received  the  most  attention   are   the   last   two:   the   negative   carbon  balance   (carbon   source)  hypothesis   and   the   growth   limitation   (carbon   sink)   hypothesis,   related   to   reduced   carbon   investment   (Körner  1998;;  2003;;  Smith  et  al.  2003;;  Hoch  &  Körner  2005).  So  far,  the  carbon   sink   hypothesis   has   received  more   support   (Körner   1998,   2003;;  Hoch  &  Körner   2005),  but  there  is  also  some  evidence  of  carbon  source  limitation  at  high  elevation   treelines   (Cabrera   et   al.   1998;;   Wieser   et   al.   2010;;   Wittich   et   al.   2012).   More   evidence  is  needed  to  clarify  the  relative  importance  of  these  two  hypotheses.   1.2.  High  elevation  adaptations  in  trees   High  elevation  trees  show  a  number  of  morphological  and  physiological  responses   to   cope   with   low   temperatures,   in   particular,   but   also   with   other   forms   of   environmental  stress.  Examples  of  such  stress  factors  include  water  shortage  due  to   higher  evaporation  and   low  rainfall   (Leuschner  2000),   increased  solar   radiation  –   especially  ultraviolet  B  radiation  –  due  to  reduced  absorption  from  the  atmosphere   and   clouds,   and   low  CO2   partial   pressure   (Caldwell   1986;;   Caldwell   et   al.   1989;;   Körner   2003).   The   responses   of   trees   to   these   conditions   are   regulated   through   different  adaptive  mechanisms:  reversible  acclimation,  non-­reversible  modification   and   evolutionary   adaptation   (Körner   2003).   Evolutionary   adaptations   are   genetically   determined   and   unresponsive   to   environmental   variations,   whereas   acclimations   are   plastic   changes   that   respond   to   environmental   conditions   (e.g.   Clausen  et  al.  1940;;  Bradshaw  1965;;  Schlichting  1986;;  Sultan  2000).       Introduction     12   In  general,  low  temperatures  have  two  types  of  effect  on  organisms:  gradual  effects,   which  permit  important  metabolic  activities  (photosynthesis,  respiration  and  nutrient   uptake)   but   at   reduced   levels,   and   threshold   effects,   which   kill   the   organism   if   a   critical   threshold   value   is   exceeded   (Körner   2012).   High   elevation   trees   have   different   temperature   threshold   limits  depending  on   their  natural  growth  conditions   and   corresponding   climatic   adaptations.   They   also   have   different   physiological   strategies   for   coping   with   low   temperatures,   including   osmotic   adjustments,   supercooling  capacity  and/or  freezing  tolerance  (Körner  2003;;  Azócar  et  al.  2007).     The  most   commonly   documented   response   to   low   temperatures   and   drought   is   a   reduction  in  aboveground  tree  biomass,  manifested  as  a  decrease  in  tree  height  (e.g.   Wilson   et   al.   1987;;   Young   1993;;   Kessler   et   al.   2007);;   this   correlates   with   a   decrease   in   leaf   size  and   specific   leaf   area   (e.g.  Cordell   et   al.   1998;;  Moser  et  al.   2007;;  Hertel  &  Wesche  2008;;  Macek  et  al.  2009).  These  changes  are  linked  to  the   plant   energy   economy,   but   the   functional  mechanisms   behind   them   are   not   fully   understood.   Körner   (2003)   emphasizes   the   role   of   developmental   controls   over   plant   organs,   in   particular   cell   wall   structures,   as   an   underlying   cause   of   the   reduced  size  of  aboveground  tree  parts.     Contrary  to  aboveground  patterns,  the  size  of  belowground  tree  structures  has  been   reported  to  respond  positively  to  decreasing  temperatures  (Hertel  &  Wesche  2008;;   Moser   et   al.   2011;;  Hertel  &   Schöling   2011)   and   increasing   drought   (Gaul   et   al.   2008;;   Hertel   et   al.   2008).   Positive   responses   of   root   systems   to   decreased   temperatures   are   explained   by   a   low   nutrient   supply   at   low   temperatures   due   to   reduced   water   viscosity   and   liquid   diffusion   and   to   low  microbial   activity   (e.g.,   Sveinbjörnsson  et  al.  1992;;  Tanner  et  al.  1998;;  Hertel  &  Wesche  2008;;  Hertel  &   Schöling  2011).     A  positive  relationship  between  maximum  photosynthesis  rate  and  temperature,  as   a   consequence   of   increased   activity   of   the   RuBisCO   enzyme   at   higher   temperatures,   has   frequently   been   documented   (e.g.   Berry   &   Björkman   1980;;   Cabrera   et   al.   1998;;  Yamori   et   al.   2005;;  Zhang   et   al.   2005;;  Azócar   et   al.   2007).   However,  it  is  not  clear  whether  reduced  carbon  acquisition  (photosynthesis)  limits   tree  growth  at  high  elevations  (Körner  2003;;  Hoch  &  Körner  2005;;  Wieser  et  al.   2010;;  Wittich  et  al.  2012).  There  is  some  evidence  of  decreased  carbon  acquisition   in  trees  with  increasing  elevation  (e.g.  Wieser  et  al.  2010;;  Wittich  et  al.  2012);;  on   the  other  hand,  it  has  been  suggested  that  the  limiting  factor  for  tree  growth  at  high   elevations   involves   metabolic   constraints   related   to   carbon   investment   in   tree   organs   at   low   temperatures   rather   than   constraints   in   carbon   acquisition   (Körner   2003;;  Hoch  &  Körner  2005).  The  higher  efficiency  of  CO2  utilization  (ECU)  per   unit  of  leaf  area  in  high  elevation  trees  compared  to  low  elevation  ones,  measured   in   experiments   carried   out   at   varying  CO2  concentrations,   support   this   view.  The   higher  ECU  is  apparently  caused,  at   least   in  part,  by   increased  leaf   thickness  and   the   consequently   increased   N   concentration   per   leaf   area   (Körner   2013).   It   thus   seems   unclear   what   the   high-­   elevation   adaptations   of   photosynthesis   in   trees   Introduction     13   consist   of,   or   indeed   whether   such   adaptations   occur   at   all.   In   addition   to   temperature,  net  photosynthesis  also  depends  on   irradiance  and  water  availability   (Körner  2003,  2012;;  Wieser  2007).   Variations   in   maintenance   respiration   are   mainly   driven   by   temperature   (Larigauderie  &  Körner  1995;;  Wright  et   al.   2006;;  Wieser  2007).  Respiration  has   been   found   to   be   lower   among   high   elevation   trees   than   low   elevation   ones   measured  in  situ  (Körner  2003;;  Wieser  &  Tausz  2007).  This  indicates  a  decreased   respiratory  carbon  loss  among  high  elevation  trees  due  to  metabolic  adjustments  at   low  temperatures.   1.3.  Distribution  of  high-­Andean  Polylepis  forests     Polylepis   forests   are   found   in   the   Andes   at   high   elevations   from   Argentina   to   Venezuela,   forming   one   of   the   world’s   highest   treelines.   One   of   the   highest   documented   sites   for   Polylepis   forest   is   at   4810   m   in   Volcán   Sajama,   Bolivia   (Hoch   &   Körner   2005).   Polylepis   trees   commonly   grow   in   sparsely   distributed   stands   disconnected   from   lower-­elevation   montane   forests.   There   is   evidence   of   both   natural   and   anthropogenic   causes   for   the   current   fragmented   distribution   of   the   forests.   Palaeoecological   studies,   based   on   fossil   pollen   records   from   Salar   Uyuni  and  Lake  Titicaca  in  Bolivia,  suggest  strong  climate-­related  fluctuations  in   the  extent  of  Polylepis  forest  cover  already  12  000  years  ago,  providing  supporting   evidence   of   the   natural   fragmentation   of  Polylepis   forests   (Gosling   et   al.   2009).   The   identification   and   dating   of   soil   charcoal   in   the   northern   Ecuadorian   Andes   also  supports  fluctuations  in  forest  cover  and  treeline  formation  since  the  beginning   of   the  Holocene  due   to  natural   fires   (di  Pasquale  et  al.  2008).  On  the  other  hand,   pollen   cores   from   the   Cordillera   Urubamba,   in   southern   Peru,   suggest   a   strong   decrease   in   the   extent   of  Polylepis   forest   cover   after   the   arrival  of   the  Spaniards   compared   to  pre-­Hispanic   times  (Chepstow-­Lusty  et  al.  1996;;  Chepstow-­Lusty  &   Winfield  2000).  This  supports  anthropogenic  factors,  in  particular  intensive  human   land  use  and  unsustainable  agro-­forestry  methods,  as  one  of  the  main  causes  for  the   current  fragmented  distribution  of  Polylepis  forests.   1.3.1.  Environmental  constraints  on  Polylepis  tree  growth  and  forest  distribution   Two   main   environmental   factors   controlling   Polylepis   forest   distribution   are   temperature  and  water  availability  (e.g.  Rada  et  al.  1996,  2001;;  Kessler  et  al.  2007;;   Macek  et  al.   2009).  These   factors   are   in   turn  modified  by   elevation,   latitude   and   topography,   and   their   relative   importance   seems   to   vary   among   species   and/or   geographical   regions.   Temperature   and   solar   radiation   are   the   most   important   factors   in   humid   and   cloudy   regions,   while   the   importance   of   water   availability   increases  in  dry  regions  (Braun  1997;;  Hoch  &  Körner  2005;;  Kessler  et  al.  2007).     Species   of   Polylepis   can   grow   in   exceptionally   low   temperature   conditions,   apparently  lower  than  several  other  high-­elevation  tree  species  (Kessler  &  Hohnwald   Introduction     14   1998;;  Lauer  &  Rafiqpoor  2000,  2002;;  Hertel  &  Wesche  2008).  Eventually,  however,   Polylepis   tree  growth   is  constrained  by   low   temperatures   similarly   to   that   of  other   high-­elevation   tree   species  everywhere.  Regeneration  patterns  are   likewise  affected   by   low   temperatures   due   to   energy-­saving   needs;;   thus   reproduction   from   seeds   decreases   and   the   proportion   of   vegetative   reproduction   increases   at   low   temperatures   (Cierjacks   2007;;  Hertel  &  Wesche   2008).  On   the   other   hand,   due   to   strong  solar  insolation,  high  temperatures  may  cause  a  heat  stress  in  Polylepis.  This   can   be   especially   critical   for   seedlings   establishment.   It   has   been   observed   that   Polylepis   seedlings   avoid   the   eastern   exposures   in   Ecuador,   probably   because   of   strong  solar  radiation  especially  in  the  mornings  when  temperatures  are  still  low  after   cold   nights   (Sarmiento   1986;;   Bader   et   al.   2007;;   Bader   &   Ruijten   2008).   These   conditions   can   cause  water   stress   in   seedlings,   as   the   shoots  may   lose  more  water   than  can  be  replaced  by  root  water  uptake  from  frozen  soil  (Körner  2003,  2012;;  Mayr   2007).   Under   these   conditions   photoinhibition   may   also   occur   (Ball   et   al.   1991;;   Germino  &  Smith  1999).   In  Bolivia,   in  contrast,  Polylepis   forests  are   taller  on   the   eastern   slopes   because   the   increased   solar   radiation  may   also   increase   the   average   temperature  (Kessler  et  al.  2007).       Figure  2.  Polylepis  forests  in  nearly  natural  condition  in  the  Cuzco  area,  southeastern  Peru:   a)   P.   racemosa   forests   at   3900-­4000   m   in   the   glacial   U-­valley   of   Urubamba.   b)   P.   subsericans  treeline  stand  at  4200  m  at  the  base  of  a  slope  of  glacial  erosion,  at  the  border   of  a  wetland  in  Cancha  Cancha  valley,  Calca.  c)  Humid  P.  pepei  forest  with  a  thick  cover  of   mosses  at  4100  m  at  Abra  Malaga.  d)  P.  sericea   trees  after  a  snow  fall  at  4200  m  on   the   slope  of  Mt.  Palcay,  in  the  protected  area  of  Machu  Picchu.  Photographs  by  J.M.  Toivonen.   a   b   c   d   Introduction     15   Drought   constrains   Polylepis   tree   growth   and   forest   distribution   in   arid   areas.   However,   Polylepis   forests,   similarly   to   other   treeline   forests   worldwide,   reach   their  highest  elevations  in  dry  regions  (Braun  1997;;  Kessler  et  al.  2007;;  Miehe  et  al.   2007).   This   is   probably   related   to   the   positive   effect   of   high   solar   radiation   and   temperature   conditions   favorable   to   tree   growth.   The   lower   limit   of   annual   precipitation  for  Polylepis  tree  growth,  observed  for  the  drought  tolerant  species  P.   tarapacana  in  central  Bolivia,  is  100-­200  mm  (Fjeldså  &  Kessler  1996).  Fog  may   allow  Polylepis   to   survive   in   areas   of   otherwise   low  water   availability.   In   cloud   forest   areas,   however,   the   distribution   of   Polylepis   is   controlled   by   competition   with  other   tree  taxa.  The  upper  soil  moisture   limit   for  Polylepis   is  determined  by   waterlogging.  Polylepis  avoids  poorly  drained,  flat  valley  bottoms,  where  cold  air   masses   also   tend   to   accumulate   at   night   (Sarmiento   1986;;   Young   1993;;  Kessler   2002;;  Fjeldså  &  Kessler  1996).       In  addition  to  constraints  caused  by  temperature  and  water  availability,  Polylepis   avoids  soils  of  high  salinity  (Fjeldså  &  Kessler  1996).  Geodynamic  disturbances   (landslides,   avalanches,   soil   erosion)   may   also   control   Polylepis   forest   distribution  locally  by  preventing  tree  colonization.  The  distribution  of  Polylepis   forest  stands  is  often  associated  with  rocky  ground.  This  is  presumably  related  to   a  favorable  microenvironment,   in  particular   temperature  (e.g.  Fjeldså  &  Kessler   1996).   1.3.2.  Anthropogenic  influence  on  Polylepis  forest  structure  and  distribution   It  has  been  estimated  that  humans  first   inhabited   the  Andes  approximately  7000– 3000  years  ago  (Baied  &  Wheeler  1993;;  Kessler  &  Driesch  1993).  Since  then,  the   Andes  have  been  subject  to  intensive  human  land  use  (Chepstow-­Lusty  et  al.  1996;;   Chepstow-­Lusty   &   Winfield   2000).   The   principal   land   uses   have   been   timber   extraction,   grazing   by   domestic   animals   and   the   associated   burning   of   pastures   (Ellenberg  1958,  1979;;  Laegaard  1992;;  Kessler  1995;;  Purcell  and  Brelsford  2004).   As   a   consequence,   the  Polylepis   forest   cover  has  declined   considerably.  Human-­ induced   fires   in   particular   have   been   suggested   to   contribute   significantly   to   the   loss  of  Polylepis  forests  (Kessler  2000,  2002),  and  to  account  for  the  restriction  of   forest   remnants   to   habitats   protected   from   fires   (Coblentz   and   Keating   2008).   However,  it  is  not  clear  when  this  major  forest  destruction  may  have  happened.  It   has  been  suggested  that  Polylepis  forest  loss  occurred  at  two  main  times:  the  first   approximately  5000–3000  ago,  the  second  at  the  time  of  the  Spanish  conquest  and   during  the  colonial  period  (Kessler  &  Driesch  1993).  A  recent  study  of  the  genetic   diversity  of  P.  pauta  and  P.  incana  forests  in  Ecuador  supports  this  view,  reflecting   both  recent  and  historical  genetic  isolation  (Hensen  et  al.  2012).  Genetic  studies  of   P.  besseri  forests  in  Bolivia  and  P.  australis  forests  in  Argentina,  however,  do  not   support   the   idea   of   an   early   anthropogenic   forest   decline,   but   indicate   relatively   recent  forest  fragmentation,  occurring  since  the  Spanish  conquest  (Julio  et  al.  2008;;   Gareca  et  al.  2013).   Introduction     16   Currently,  Polylepis   forests   are   among   the  most   gravely   endangered   tropical   and   subtropical  mountain   forest   ecosystems   in   the  world   (UNEP-­WCMC  2004).   The   majority   of   the   ca.   30   species   of   the   genus   are   classified   as   vulnerable   (IUCN   2013).  One  of   the  major   threats   to   remaining  Polylepis   forest   stands  seems   to  be   habitat   degradation,   caused   especially   by   livestock   grazing   but   also   by   wood   harvesting   for   timber   and   firewood   (e.g.   Renison   et   al.   2006,   2010;;   Jameson  &   Ramsay   2007).   Grazing   specifically   affects   regeneration   patterns,   favoring   vegetative   regeneration   and   reducing   regeneration   from   seeds   (Cierjacks   et   al.   2007,   2008;;   Toivonen   et   al.   2011).   Long-­term   grazing   and   substantial   wood   harvesting   ultimately   lead   to   a   decrease   in   forest   biomass   and   to   general   forest   degradation,  manifested  as  a  lack  of  certain  age  and/or  size  cohorts  (Renison  et  al.   2011)  and  as  a  decrease  in  canopy  density  (Jameson  &  Ramsay  2007).     Figure  3.  Anthropogenically  disturbed  Polylepis  forests  in  Cuzco  area,  southeastern  Peru.  a)   P.   racemosa   forests   at   4300   m   near   a   village   and   an   old   copper   mine   in   Mantanay,   Urubamba.  b)  Burnt  P.  sericea  forests  at  3800  m  near  Abra  Malaga.  c)  Grazed  P.  racemosa   forest  with  impeded  regeneration  at  4000  m  at  the  border  of  Lake  Yanacocha,  Urubamba.  d)   An  alpaca  browsing  in  Polylepis  racemosa  forest  stand  at  4100  m  in  Cancha  Cancha  valley,   Calca.  Photographs  by  J.M.  Toivonen.   a   b   c   d   Introduction     17   1.4.  The  aims  of  the  thesis     Andean   Polylepis   forests   are   among   the   highest   mountain   forest   ecosystems   worldwide,   having   a   fragmented   and   often   disconnected   distribution   from   the   lower  elevation  montane  forests.  There  has  been  ample  discussion  as  to  the  causes   of  the  fragmented  distribution  of  Polylepis  forests  (Ellenberg  1979;;  Kessler  1995,   2000,  2002;;  Gosling  et  al.  2009;;  Urrego  et  al.  2011).  In  this  thesis,  my  purpose  is  to   clarify  the  role  of  functional,  environmental  and  anthropogenic  factors  determining   Polylepis   forest   distribution   in   the   Central   and   Southern   Andes.   First,   I   assess,   which   functional   traits   enabling  Polylepis   species   to   grow   at   high   elevations   are   genetically  determined  and  have  been  potentially  crucial  to  the  species’  evolution   and  adaptation  to  low  temperatures  and  drought  (article  I).  Secondly,  I  address  the   question  of   the  causes  of   the  current   fragmented  distribution  of  Polylepis   forests,   by   searching   for   a   consensus   on   the   importance   of   natural   versus   anthropogenic   causes  determining  Polylepis  forest  distribution  (articles  II-­IV).   In   response   to  decreased   temperature  and  humidity   conditions  at   high   elevations,   trees   are   expected   to   manifest   genetically   determined   adaptations   and   phenotypically   plastic   acclimations   in   their   functional   traits.   In   article   I,   my   purpose   was   to   determine   whether   there   is   evidence   of   climatic   adaptations   regarding   these   traits,   by   assessing   the   degree   of   genetic   determinism   in   the   functional  traits  of  nine  Polylepis  tree  species  among  fourteen  important  traits  that   enable   trees   to   withstand   cold   and   dry   conditions.   The   degree   of   genetic   determinism  was  defined  by  relating  interspecific  variation  in  a  functional  trait   to   the   climatic   niche   optima   of   the   species.   The   assumption   was   that   if   this   relationship   was   similar   to   the   expected   climate-­trait   relationship,   based   on   empirical   observations   in   previous   studies   and   on   theoretical   predictions   of   climate-­trait   relationships,   any   variation   observed   in   the   trait  must  be   genetically   determined.  Another  purpose  was  to  determine  whether  variation  in  physiological   traits   is   genetically   more   strongly   controlled   than   that   in   morphological   traits.   Finally,   I   wanted   to   find   out   whether   the   selected   functional   traits   show   a   phylogenetic   signal,   i.e.   whether   closely   related   species   are   functionally   more   similar  than  distantly  related  ones.   For  high  elevation   trees,   the   importance  of   the  micro-­climate   in  defining  suitable   conditions  for  tree  growth  and  seedling  establishment  is  expected  to  increase  with   elevation   as   a   result   of   increasing   environmental   harshness;;   the   anthropogenic   impact,   conversely,   is   expected   to   decrease   with   elevation   because   of   reduced   human   accessibility   and   human   population   density   (Toivonen   et   al.   2011).   Following   this   principle,   my   purpose   in   article   II   was   to   determine   whether   Polylepis   forests   are   associated   with   specific   topographic   positions   in   the   landscape,   and   whether   the   potential   associations   between   terrain   features   and   Polylepis  forest  distribution  change  with  elevation.     Introduction     18   Polylepis   forests   form   one   of   the   highest   treelines   in   the   world.   Their   growth   conditions,   and   the   factors   that  ultimately   constrain   tree  growth,   are  nevertheless   still   poorly   known.   My   purpose   in   article   III   was   to   ascertain   how   tree   height   changes  with   elevation   among   stands  of   five   species  of  Polylepis   under  different   humidity   conditions   and   under   human   impact   versus   natural   conditions,   and   whether   air   or   soil   temperatures   are   the   crucial   determinants   of   tree   height   and   treeline  formation.     Due  to  the  long  history  of  human  settlements  in  the  Andes,  human  influence  was   expected   to   strongly   affect  Polylepis   forest   structure   and   distributions.   In   article   IV,  I  investigated  whether  the  structural  variation  of  Polylepis  forest  stands  can  be   explained   by   human  disturbance;;   the   latter  was   quantified   in   terms   of   indices  of   accessibility,   such   as   the   geographical   distance   between   a   forest   stand   and   the   nearest  human  settlement,  road  and  market  center.     Material  and  Methods     19   2.   MATERIAL  AND  METHODS   2.1.  Study  area  and  species   The  Andes  are  the  longest  continental  mountain  chain  in  the  world  (ca.  7000  km  in   a  north-­south  direction),  ranging  from  Venezuela  to  Argentina.  The  highest  peak  is   Mount  Aconcagua  (6962  m)  in  Argentina,  but  several  other  volcanoes  also  exceed   6000   m   in   elevation.   On   an   evolutionary   time-­scale,   the   Andes   are   a   relatively   young  mountain  chain.  The  uplift  of   the  mountain  chain  started  about  20  million   years   ago   (Burnham  &  Graham  1999),   but   the   emergence   of   the  majority   of   the   current  alpine  habitats  dates  back  only  a  few  million  years  or  even  less  (Gregory-­ Wodzicki   20000).   These   new   high   elevation   habitats   were   inhabited   by   taxa   originating,   at   least   in   part,   by   extensive   adaptive   radiations   of   several   lowland   ancestors  (Hooghiemstra  &  Van  der  Hammen  2004).  The  genus  Polylepis  is  one  of   these   taxa   (Simpson  1986;;  Kerr   2003;;  Schmidt-­Lebuhn   et   al.   2006).  The   current   taxonomical   classification  of  Polylepis   genus   suggests   ca.   30   ecologically   and/or   biogeographically  distinct  species  (Schmidt-­Lebuhn  et  al.  2006).  The  species  show   a  gradual  change  in  their  morphological  characters:  from  tall  trees  with  thin  leaves,   multiple   leaflets   and   multi-­flowered   inflorescences   at   the   upper   parts   of   humid   cloud   forests   to   small-­sized   trees   with   small   and   coriaceous   leaves   and   small   inflorescences  in  high  mountain  habitats  (Simpson  1979,  1986;;  Kessler  1995;;  Kerr   2003;;  Schmidt-­Lebuhn  et  al.  2006).     Article   I   dealt  with   the   climatic   adaptations  of  nine  Polylepis   species  originating   from  different  geographical  regions  in  the  central  and  southern  Andes,  from  a  wide   range  of  elevation  and  climatic  conditions:  from  humid  montane  forests  to  dry  high   elevation  habitats  (Table  1,  Fig.  4).     In   articles   II-­IV,   I   studied   the   natural   and   anthropogenic   constraints   determining   the  Polylepis  forest  distribution  in  the  Cordilleras  Vilcanota  and  Vilcabamba  in  the   Cuzco   region,   Southeast   Peru.  The  area   has   been   subject   to  human   influence   for   thousands   of   years,   due   to   its   close   proximity   to   ancient   centers   of   human   settlement  by   the  Urubamba  River   (Chepstow-­Lusty  et  al.  1996;;  Chepstow-­Lusty   &  Winfield   2000).   Despite   this   long   human   history,   the   area   is   known   to   have   fairly  extensive  Polylepis  forests  at  elevations  between  3300  m  and  4950  m  (Lloyd   &  Marsden  2011;;  Toivonen  et  al.  2011).  The  climate  of   the  region  varies   from  a   semi-­dry   to   a   sub-­humid   tropical   alpine   climate,   with   a   pronounced   wet   season   from  October  to  March.  Diurnal  temperature  fluctuations  are  strong,  especially  in   the  dry  season.  Salient  differences   in  humidity  and   temperature  within   the  region   have  enabled  the  evolution,  diversification  and  specialization  of  a  number  of  plant   and  animal  taxa,  including  the  genus  Polylepis  (Fjeldså  1992;;  Fjeldså  and  Kessler   1996).  Seven  species  of  the  genus  are  found  within  a  radius  of  ca.  30  km  around   the  town  of  Urubamba.  This  is  the  highest  concentration  of  Polylepis  species  found   anywhere   (Fjeldså   and   Kessler   1996).   In   article   II   I   focused   on   monospecific,   elevationally   segregated   forest   stands   of  P.   racemosa   and  P.   subsericans   on   the   Material  and  Methods     20   semi-­dry  slopes  of  the  Cordillera  Urubamba,  part  of  the  Vilcanota  mountain  chain.   The   study   area   is   characterized   by   several   steep   glacial   valleys,   with   smaller   ravines   connected   to   them.   The   highest   snow   peak   of   the   Cordillera,   Nevado   Chicon,  reaches  5400  m  of  elevation.  In  articles  III-­IV  I  studied  the  semi-­dry  and   sub-­humid   forest   stands   of   the   Cordilleras   Urubamba   and   Vilcabamba,   with   the   highest   snow   peak,   Salcantay,   reaching   ca.   6300   m.   In   this   area   species   are   segregated  not  only  by  elevation  but  also  by  humidity.  (Table  1,  Fig.  4).     During   recent   decades   the   human   population   has   increased   greatly   in   the   area,   mainly   in   the   valley   bottom   of   the   Urubamba   River   (the   Sacred   Valley   of   the   Incas),   while   high   elevations   have   remained   sparsely   populated.   At   elevations   above  3800  m  only  a  few  villages  are  found,  but  cattle  and  sheep  grazing  are  still   intensively  practiced  there.     Table  1.  Geographical  regions  (A-­E),  species  and  elevation  ranges  for  all  11  species  used   in  the  four  articles.  Asterisks  indicate  a  treeline  species.   Geographical  region   Species   Elevation  (m)   Article   A.  Sub-­humid  eastern  Andes  of  South   Peru   P.  pauta   2700-­3600   I,  III-­IV     P.  sericea   3700-­4200   III-­IV     P.  pepei*   4100-­4600   III-­IV  B.  Semi-­dry  eastern  Andes  of  South   Peru     P.  microphylla   3300-­4000   I     P.  racemosa   3200-­4200   I-­IV     P.  subsericans*   4200-­4900   I-­IV  C.  Humid  eastern  Andes  of  central   Bolivia   P.  hieronymi   1900-­3300   I     P.  neglecta   2400-­3500   I     P.  tomentella*   3200-­4500   I  D.  Dry  western  Andes  of  central   Bolivia     P.  tarapacana*   4000-­4800   I   E.  Semi-­dry  eastern  Andes  of   Argentina     P.  australis*   1800-­3800   I     Material  and  Methods     21   Figure   4.   Maps   of   study   areas.   Left   side:   Five   regions   (A-­E)   showing   provenances   of   different  Polylepis  species  used  in  article  I.  A)  Sub-­humid  eastern  Andes  of  southern  Peru,   B)  Semi-­dry  eastern  Andes  of  southern  Peru,  C)  Humid  eastern  Andes  of  central  Bolivia,   D)  Dry  western  Andes  of  central  Bolivia,  E)  Semi-­dry  eastern  Andes  of  Argentina.  Right   side:  Study  area  in  Cordilleras  Vilcanota  and  Vilcabamba,  Cuzco,  South  Peru  (articles  II-­ IV).  Black  symbols  indicate  study  sites  at  sub-­humid  areas  and  white  symbols  at  semi-­dry   areas.  Number  of  study  plots  at  each  site   is   indicated  close  to  symbols.  Asterisks  indicate   temperature  data  loggers  (air/soil).   2.2.  Common  garden  experiment   Genetically  determined  traits  are  typically  unresponsive  to  environmental  variation,   whereas  plastic  traits  respond  to  varying  environmental  conditions  (e.g.  Clausen  et   al.  1940;;  Bradshaw  1965;;  Schlichting  1986;;  Sultan  2000).  Genetic  determinism  in   a  given  trait  can  be  studied  by  growing  individuals  of  different  species  in  the  same   environment  and  observing  whether  they  show  interspecific  variation  in  the  trait.  If   they   do,   the   trait   must   be   genetically   determined.   In   article   I,   I   carried   out   a   common   garden   experiment   at   the   greenhouses   of   the   Experimental   Botanical   Garden   of   the   University   of   Göttingen   in   2008,   to   assess   the   degree   of   genetic   determinism  in  the  important  functional  traits  of  nine  Polylepis  species  originating   from   different   climatic   conditions   and   geographic   regions   in   the   Andes.   I   grew   individuals   of   different   Polylepis   species   from   seeds,   seedlings   or   root   suckers   under   constant   greenhouse   conditions   for   18   months   and   measured   several   functional  traits  of  the  species,  including  above-­  and  belowground  morphology  and   photosynthetic  performance,  measured   as   light   response   curves  of  photosynthesis   (Meir  et  al.  2007).  The   traits  measured  were  selected  according   to   their  expected   importance  for  the  species’  ability  to  withstand  cold  and  dry  conditions  (Table  2).               Material  and  Methods     22   Table  2.  Functional  traits  measured     Trait   Unit   Single  leaf  area   cm²   Specific  leaf  area   cm²  g-­1   Specific  root  area   cm²  g-­1   Specific  root  length   m  g-­1   Root  tip  abundance   n  mg-­1   Leaf-­area–based  maximum  net  photosynthesis  rate   µmol  CO2  m-­2  s-­1   Leaf-­mass–based  maximum  net  photosynthesis  rate   nmol  CO2  g-­2  s-­1   Stomatal  conductance  to  water  vapor  in  maximum  light  conditions   mol  H2O  m-­2  s-­1   Dark  respiration  rate     µmol  CO2  m-­2  s-­1   Leaf  transpiration  rate   mol  H2O  m-­2  s-­1   Ratio  of  intercellular  to  ambient  CO2  concentration   mol  CO2  mol  CO2-­1   Quantum  use  efficiency  1   unitless   Light  saturation  point  2   ȝPROSKRWRQVP-­2  s-­1   Light  compensation  point  3   ȝPROSKRWRQVP-­2  s-­1   1  Coefficient  of   initial  slope  of   light   response  curve  of  photosynthesis,   indicating  efficiency  of   light   use  in  CO2  assimilation   2  PAR  (photosynthetically  active  radiation)  value  at  which  90  %  of  maximum  net  photosynthesis   is   reached  and  the  carbon  assimilation  rate  can  no  longer  increase  because  of  limited  carboxylation     3  PAR  value  at  which  photosynthesis  and  respiration  are  balanced,  so  that  the  rate  of  CO2  assimilation   matches  the  rate  of  CO2  released  from  respiration   2.3.  Determination  of  species  climatic  niche  optima   In   this   work,   I   used   a   two   dimensional   approximation   of   the   multidimensional   niche   space   (Hutchinson   1957),   determined   by   two   relevant   and   readily   interpretable  climatic  variables:  mean  annual  temperature  and  annual  precipitation   (I).  Toward  this  end,  I  compiled  climate  data  of  species  occurrence  localities,  using   information   from   several   scientific   publications   documenting   Polylepis   species   locations,   my   own   field   excursions,   and   georeferenced   occurrence   information   from   specimen   location   data   in   the   GBIF   (http://www.gbif.org/)   and   Tropicos   (http://www.tropicos.org/)  databases.     I   extracted   the   climate   data   for   the   species   occurrence   localities   from   WorldClim   global   modeled   climate   data   (http://www.worldclim.org)  at  a  spatial  resolution  of  ~1  km2  (Hijmans  et  al.  2005).   Before  using   all   this   legacy  data,   I  manually   cleaned   it.  For   species  with   a   large   distribution   range   I   only   included   climate   records   near   the   areas   of   the   provenances,  that  I  used  in  the  experiment  to  ensure  a  correspondence  between  the   climate   data   and   the   species   ecophysiological   data.   For   species   with   a   narrow   distribution  range,  I  used  all  climatic  records.  WorldClim  climate  data  includes  19   bioclimatic   variables;;   of   these,   I   used  mean   annual   temperature   and   total   annual   precipitation,  two  easily  interpretable  and  statistically  uncorrelated  variables.  Based   on  Principal  Component  Analysis  (PCA),  these  two  variables  also  captured  a  large   part  of  the  variation  in  climatic  niche  space  based  on  all  19  bioclimatic  variables.   Species   climatic   niche   optima   were   defined   as   species-­specific   means   for   mean   annual  temperature  and  total  annual  precipitation.  The  number  of  climatic  records   used  in  calculating  the  species-­specific  means  varied  from  10  to  49,  except  for  one   Material  and  Methods     23   species   (P.   tomentella),   represented   by   139   records.   The   determination   of   the   climatic  niche  optima  was  based  on  climatic  records  from  the  current  distribution   range  of  the  species.  The  current  distribution  is  almost  certainly  reduced  by  human   activities;;   thus   the   estimated   niche   optima   may   not   reflect   the   whole   ecological   potential  of  the  species.   2.4.  GIS  and  remote  sensing  methods   To  study  the  associations  between  micro-­scale  terrain  features  and  Polylepis  forest   distribution,   I   used   a   remote   sensing   -­based   approach   (II).   I   mapped   Polylepis   forest   patches   in   the   Cordillera   Urubamba   using   a   high-­resolution   aerial   photograph  from  the  year  2010,  with  a  pixel  size  of  2  m  x  2  m  (available  in  Bing   Map  web  mapping  service).  The   image  was   largely  cloud-­free,  allowing  accurate   digitizing   of  Polylepis   forest   stands   on   a  GIS   (Geographic   Information  Systems)   platform.  Parts  that  were  covered  by  clouds  were  mapped  with  the  help  of  Google   Earth  satellite  images  (Google  Earth  Quickbird  image  of  August  2012  with  a  pixel   size  of  2.4  m  x  2.4  m,  accessed  in  September  2013).  Extensive  ground  truthing  was   carried   out   during   several   field   excursions   since   2006.   This   mapping   exercise   resulted   in   much   higher   resolution   data   than   in   previous   studies   examining   associations   between   topography   and   Polylepis   forest   distribution   (e.g.   Braun   1997;;  Bader  et  al.  2007;;  Bader  &  Ruijten  2008;;  Coblentz  &  Keating  2008)  (Fig.  5).     Figure  5.  An  example  of  the  digitized  Polylepis  forest  stands  (red  outline)  in  the  Cordillera   Urubamba.  The  image  is  an  aerial  photograph  from  the  year  2010  with  a  pixel  size  of  2  m  x   2  m  (Bing  Map  web  mapping  service).   I  selected  ten  terrain  features  and  indices,  based  on  their  expected  importance  for   tree   growth   and   seedling   establishment   and   survival   at   tropical   high   elevations   Material  and  Methods     24   (Table   3).   Terrain   features   and   indices   were   calculated   using   the   Aster   Global   Digital  Elevation  Model  (version  2),  with  8m  of  vertical  resolution  and  30  m  x  30   m  of  pixel  size  with  algorithms  available  in  SAGA  GIS  and  ArcGIS.   Table   3.   Selected   terrain   features   and   indices   based   on   Aster   Global   Digital   Elevation   model  (version  2).   Terrain  feature  /  index   Unit   Description   1.  Aspect   Categorical   variable   Exposition  towards  slope  is  facing   2.  Morphometric   protection  index     Unitless   Topographic  sheltering;;  calculation  based   on  pixel  values  within  300  m  radius     3.  Overland  distance  to   the  nearest  river   Meters   Overland  distance  to  the  nearest  river   measured  using  DTM-­based  potential   river  network   4.  Plan  curvature   Unitless   Terrain  surface  curvature  described   perpendicularly  to  slope  steepness.  May   be  convex,  linear  or  concave.   5.  Profile  curvature   Unitless   Terrain  surface  curvature  described   parallel  to  slope  steepness.  May  be   convex,  linear  or  concave.   6.  Slope  steepness   °   Slope  angle  in  degrees   7.  Solar  radiation   Mwh  m-­2  year  -­1   Potential  annual  incoming  solar  radiation   (insolation);;  quantitative  equivalent  of   exposure     8.  Direct  solar  radiation   hours   h  year-­1   Duration  of  potential  annual  direct   incoming  solar  radiation   9.  Topographic  wetness   index   Unitless   Soil  moisture  and  surface  saturation  index;;   calculation  based  on  control  of  local   topography  on  hydrological  processes   10.  Vertical  distance  to   the  nearest  river   Meters   Vertical  distance  to  nearest  river,  measured   using  the  DTM-­based  potential  river   network       2.5.  Measurements  of  forest  structure  and  microclimate   The  fieldwork  for  forest  structure  and  microclimate  measurements  was  carried  out   in  two  campaigns  (III-­IV).  In  2006,   I  established  28  plots  of  10  m  x  10  m  in  the   Cordilleras  Urubamba  and  Vilcabamba,  Cuzco,  Peru,  with  the  aim  of  covering  the   distribution  of  the  five  Polylepis  species  found  in  the  sub-­humid  and  semi-­dry  areas   of  the  cordilleras.  Plots  ranged  from  3070  m  to  4450  m  of  elevation.  Averaging  the   data   between   plots   established   within   the   same   forest   stand   and   differing   only   slightly  in  elevation  resulted  in  a  total  of  24  plots,  of  which  eleven  were  in  semi-­ dry  areas  (six  in  P.  racemosa  stands  and  five  in  P.  subsericans  stands)  and  thirteen   in   sub-­humid  areas   (four   in  P.  pauta   stands,   four   in  P.   sericea   stands,   five   in  P.   pepei   stands).   In   each   plot,   I   measured   tree   circumference   at   breast   height   and   estimated  the  height  of  all  Polylepis  trees  (ı10  cm  of  circumference),  and  counted   Material  and  Methods     25   the  number  of  seedlings  and  root  suckers.  I  also  installed  temperature  dataloggers   (DS  1922  Thermochron  iButtons,  Hubbart  et  al.  2005)  to  measure  air  temperature   in  the  shaded  canopy  and  soil  temperature  in  the  root  zone  at  10  cm  depth  in  eleven   plots.  In  2011,  nineteen  extra  plots  were  established  by  my  collaborators,  with  the   aim  of   covering   the  widest   possible  elevational   range  per   species.   In   these  plots,   the   tree   height   and   circumference   of  Polylepis   trees   (ı10   cm  of   circumference)   were   measured   and   seven   extra   dataloggers   were   installed   to   measure   air   temperature.   In   total,   air   temperature  data  were  successfully   recorded   in  eighteen   and  soil  temperature  in  ten  plots  during  one,  two  or  three  years  (between  June  2006   and   April   2012),   depending   on   the   accessibility   of   the   stand.   The   measurement   readings  were  mostly  taken  at  four  hours  intervals.   2.6.  Numerical  analyses   To   find   out   whether   inter-­specific   variation   in   a   given   trait   had   evolved   as   a   response  to   the  environment,   I   related  species-­specific  mean  values   in   the  trait   to   the  preferred  environmental  conditions  (niche  optima)  of  the  species  (I).  If  they  are   correlated,   the   evolutionary   adaptation   of   the   species   along   the   gradient   of   the   environmental   factor   can   be   expected   to   include   genetically   determined   modifications  in  the  trait.  To  relate  the  species-­specific  means  of  functional  traits  to   the   estimated   climatic   niche   optima   of   the   species,   I   used   ordinary   linear   regressions.  To  illustrate  the  relationships  among  all  functional  traits  and  between   functional   traits   and   climate,   I   conducted   a   redundancy   analysis   (RDA),   with   species-­specific  means   of   all   fourteen   functional   traits   as   response   variables   and   species-­specific   means   for   climatic   niche   variables   as   explanatory   variables.   To   study   the   phylogenetic   signal   in   functional   traits,   I   calculated   phylogenetic   distances   between   the   species   based   on   the   phylogenetic   tree   presented   by   Schmidt-­Lebuhn  et  al.  (2006).  Since   the   tree   is  not   time-­calibrated,   I  calculated  a   phylogenetic  distance  matrix  based  on  the  number  of  bifurcations  between  species.   I  also  calculated  dissimilarity  matrices  for  each  functional  trait  based  on  Euclidean   distances.   I   compared   the  matrices   using   a  Mantel   test   of  matrix   correspondence   (Smouse  et  al.  1986).     To  study  the  associations  between  terrain  features  and  Polylepis  forest  distribution,   as  well  as  possible  elevational  changes  in  these  associations  (II),  I  divided  the  data   into   three   elevation   belts   (3800-­4199  m,   4200-­4599  m,   4600-­4950  m).  To  detect   differences  in  mean  values  and  frequency  distributions  of  terrain  features  between   the  Polylepis  stands  and  the  landscape,  I  calculated  marginality  and  specialization   indices.   Marginality   indices   describe   a   relative   difference   in   a   terrain   feature   between  the  mean  for  forest  sites  and  for  the  landscape  in  relation  to  the  total  range   or  standard  deviation  of  the  terrain  feature  in  the  landscape  (Hirzel  et  al.  2002).  The   specialization   index   was   calculated   as   the   ratio   of   the   standard   deviation   of   variation  in  a  landscape  terrain  feature  to  the  standard  deviation  of  variation  in  that   terrain   feature   at   the   Polylepis   forest   sites.   A   randomly   chosen   set   of   cells   is   expected   to   have   a   specialization   value   of   one.  Values   higher   than   one   indicate   Material  and  Methods     26   some   form   of   specialization.   The   specific   values   for   this   index   depend   on   the   reference   set   (Hirzel   et   al.   2002).   To   illustrate   differences   in   the   frequency   distributions  of  terrain  features  between  the  landscape  and  Polylepis  forest  stands,  I   classified  slope  exposition  (aspect)  into  eight  categories  (north,  northeast,  east  etc.)   and  the  other  nine  terrain  variables  into  twenty  categories  at  equal  intervals.  I  also   used   the   variable   contribution   and   Jackknife   tests   of  Maxent   species   distribution   modeling  (Phillips  et  al.  2006)  to  test  the  importance  of  each  terrain  feature  in  the   prediction  of  Polylepis  forest  cover  by  elevation  belts.     To  study  how  Polylepis  tree  height  changes  with  elevation  with  respect  to  variation   in   temperatures,   humidity   and   human   impact   (III),   a)   tree   height-­dbh   (diameter   at   breast  height)  relationship  was  compared  among  five  species  of  Polylepis  at  natural   and  disturbed  sites,  b)  temperature  conditions  based  on  micro-­climatic  measurements   were   compared   between   sub-­humid   and   semi-­dry   sites,   and   c)   tree   height   was   compared  between  sub-­humid  and  semi-­dry  sites  and  between  natural  and  disturbed   sites.   Due   to   the   small   number   of   trees   per   plot,   data   from   the   plots   of   the   same   species  were  combined  within  200  m  wide  elevation  belts.  In  six  of  the  belts  signs  of   human  impact  were  observed;;  the  other  six  were  considered  unaffected.  Tree  height   was  plotted  against  dbh  and  a  fit  between  linear  and  non-­linear  models  was  compared   with  R2   and   AIC-­values   (Burnham  &  Anderson   2002).   Slenderness,   calculated   as   tree   height   divided   by   dbh   (Wang   et   al.   1998)   and   mean   maximum   tree   height,   calculated  as  the  mean  of  10  %  of  the  tallest  trees,  were  also  quantified.  Air  and  soil   temperatures  were  compared  between  sub-­humid  and  semi-­dry  sites  after  taking  into   account   a   possible   elevation   trend   in   temperatures.   If   a   significant   elevation   trend   was  observed,  residual  variations  were  compared  rather  than  raw  values.     To  study   the  association  between  forest   structure  and  accessibility   (IV),   I   related   forest   structural   variables   to   indices   of   accessibility   using   ordinary   linear   regressions.   As   a   measure   of   accessibility   of   the   forest   stand,   I   measured   the   geographical  distance  between  each  study  plot  and   the  nearest  human  settlement,   road  and  market  center,  using  the  satellite  images  of  Google  Earth.  Because  of  the   commonly  documented  elevational  dependence  of  forest  structural  variables  (e.g.,   Wilson  et  al.  1987;;  Young  1993;;  Paulsen  et  al.  2000;;  Kessler  et  al.  2007),  before   relating  these  variables  to  the  indices  of  accessibility  I  first  removed  the  effect  of   elevation.  This  was  done  by  relating  the  variables  to  elevation  in  a  linear  regression   and  using   the  residual  variation   in  a   regression  with  accessibility   indices.  Human   influence  has  historically  been  stronger   in   the   inner  and  drier  parts  of   the  Andes,   where  the  centers  of  ancient  pre-­Hispanic  cultures  were  located,   than  in  the  more   remote  and  inaccessible  humid  Amazonian  side  of  the  Andes.  I  therefore  analyzed   the   data   from   dry   and   humid   slopes   separately.   Sites   close   to   each   other   are   typically   more   similar   than   sites   lying   far   apart,   i.e.   they   are   spatially   autocorrelated   (Dale   et   al.   2002),   which   can   complicate   the   interpretation   of   the   results.  I  tested  the  spatial  structure  of  the  data  comparing  the  dissimilarity  matrices   of  Euclidean  distances  calculated  for  each  residual  variable  of  forest  structure  and   matrices  of  geographical  distances  between  the  plots  with  a  Mantel  test  of  matrix   correspondence  (Smouse  et  al.  1986).     Results  and  Discussion     27   3.   RESULTS  AND  DISCUSSION   I   found   that   the   ability   of   Polylepis   species   to   inhabit   Andean   high   elevations   appear  to  be  based  on  certain  genetically  determined  functional  adaptations  to  cold   and   dry   conditions   (I),   and   that  Polylepis   forest   distribution   is   constrained   by   a   number   of   natural   and   anthropogenic   factors   (II-­IV).  More   specifically,   I   found   that   topographic  features  play  an  important  role  in  determining  the  current  spatial   pattern   of   Polylepis   forest   distribution,   through   the   formation   of   refugia   from   human  activities  at  lower  elevations  and  of  favorable  micro-­climatic  conditions  at   higher  ones  (II).  We  also  found  remarkably  low  treeline  temperatures  and  high  tree   statures  near  treelines  in  comparison  to  global  means,  and  air   temperatures  rather   than  soil  temperatures  are  suggested  as  the  crucial  aspect  of  temperature  conditions   in   limiting   tree   growth   at   high   elevation   treelines   (III).   I   also   found   that   human   influence,  measured  in  terms  of  forest  stand  accessibility,  predicted  stand  structure   and  regeneration:  more  accessible  stands  were  more  degraded  in  terms  of  structural   variables  and  regeneration  than  less  accessible  ones  (IV).   3.1.  Climatic  adaptations  in  functional  traits  of  Polylepis  species   In   article   I,   I   assessed   the   genetic   determinism   of   important   functional   traits   of   Polylepis   species,   traits   which   enable   the   species   to   inhabit   the   cold   and   dry   conditions   of   the   high   Andes,   and   found   some   significant   relationships   between   interspecific  variation  in  functional  traits  and  the  climatic  niche  optima  of  Polylepis   species.   Species-­specific   means   of   specific   leaf   area   increased   and   root   tip   abundance  decreased  with   increasing  mean  annual   temperature  optima.  This   is   in   line  with  the  well-­known  size  reduction  in  above-­ground  tree  biomass  (e.g.  Cordell   et  al.  1998;;  Moser  et  al.  2007;;  Hertel  &  Wesche  2008;;  Macek  et  al.  2009)  and  the   reduction   in   below-­ground   biomass   (e.g.   Hertel   et   al.   2008;;   Hertel   &   Schöling   2011;;   Moser   et   al.   2011)   with   decreasing   temperature.   I   also   found   a   negative   relationship   between   the   mass-­based   maximum   photosynthesis   rate   and   mean   annual   temperature   optima,   suggesting   that   tree   growth   is   limited   by   carbon   acquisition   (photosynthesis)   at   high   elevations   and   not   only   by   metabolic   constraints  related  to  carbon  investment  (Körner  1998;;  Hoch  &  Körner  2005).  This   finding,   supporting   the  carbon  source   limitation  hypothesis,   is   in   line  with  recent   findings   on   tree   ecophysiology   in   situ   at   high   elevation   treelines   (Wieser   et   al.   2010;;   Wittich   et   al.   2012).   Species-­specific   means   of   light   saturation   and   light   compensation   points   were   also   negatively   related,   and   quantum   use   efficiency   positively  related,  to  mean  annual  precipitation  optima.  This  may  indirectly  reflect   a   response   by   these   photosynthetic   traits   to   light   conditions,   which   are   largely   determined   by   cloud   cover   and   thus   also   by   precipitation.   In   general,   the   relationships   between   the   species-­specific  means   of   functional   traits   and   species   climatic   niche   optima   can   be   interpreted   as   evidence   of   genetically   determined   climatic  adaptations  to  cold  and/or  dry  conditions,  which  have  enabled  species  of   Polylepis  to  colonize  Andean  high  elevations.     Results  and  Discussion     28   However,  some  of  the  climate-­trait  relationships  that  we  expected  based  on  existing   theories   and   previous   empirical   evidence   were   not  manifested   (I).   For   example,   neither   specific   leaf   area   nor   specific   root   area   or   root   length   were   related   to   climatic   niche   variables,   suggesting   a   possible   acclimation   of   these   traits.   It   thus   appears   that   the   functional   traits   that   are   crucial   in   enabling  Polylepis   species   to   inhabit  high  elevations  involve  both  genetic  and  phenotypic  components.     I  also  found  that  among  the  traits  and  species  studied  physiological  traits  were  not   more  strongly  related  to  climatic  niche  optima  than  morphological  ones,  indicating   that  the  former  are  not  genetically  more  strongly  controlled  than  the  latter  (I).  Thus   the   hypothesis   of   the   conservative   inheritance   of   physiological   traits   was   not   supported.     Single   leaf   area   was   the   only   trait   showing   a   significant   phylogenetic   signal;;   in   other  words,   closely   related   species  were  more   similar  with   regard   to   single   leaf   area   than   distantly   related   species   (I).  This   suggests   that  most   of   the  manifested   climate-­trait   relationships   are   indeed   not   caused   purely   by   phylogeny,   but   are   mainly   a   result   of   ecological   specialization   along   an   environmental   gradient.  An   adaptive   trend   towards   dry   environments   has   also   earlier   been   found   within   the   genus  Polylepis  (Simpson  1986;;  Schmidt-­Lebuhn  et  al.  2006,  2010);;  adaptations  to   cold   environments,   on   the   other   hand,   may   have   occurred   several   times   independently,   shown   as   a   fairly   even   and   dispersed   distribution   of   cold   climate   species  in  the  phylogenetic  tree  of  the  genus.     3.2.  Terrain  features  and  Polylepis  forest  distribution   In  article  II,  I  studied  the  association  between  terrain  features  and  Polylepis  forest   distribution.  Human   influence   plays   an   important   role   in   determining   the   natural   distribution  and  growth  conditions  of  Polylepis  forests  (Kessler  1995,  2000,  2002;;   Purcell   &   Brelsford   2004;;   Renison   et   al.   2006;;   Coblentz  &  Keating   2008).   The   associations   that   I   found   are   therefore   interpreted   taking   this   into   account.   My   principal   finding   was   an   elevational   shift   in   the   preferences   of   Polylepis   forest   stands  in  relation  to  terrain  features.  Stands  were  more  frequently  found  in  humid,   sheltered,   concave   positions   at   lower   and   middle   elevations   (3800-­4199   m   and   4200-­4599   m),   and   at   sunnier   and   drier   convex   positions   in   western   and   northwestern  exposures  at  high  elevations  (>4600  m).    Stands  at  lower  elevations   were  also  close  to  rivers  measured  in  overland  distance,  but  high  above  the  level  of   the  rivers  in  terms  of  vertical  distance,  while  at  high  elevations  the  pattern  was  the   opposite.   The   associations   observed   at   low   elevations  may   be   caused   by   human   land   use,   especially   by   fire   regimes   and   grazing.   Topographically   sheltered   and   humid  sites  near  rivers  at  low  and  middle  elevations  may  be  better  protected  from   grassland  burning  for  pasture  than  the  landscape  in  general.  A  strong  difference  in   the   vertical   distance   from   the   nearest   river   between   Polylepis   stands   and   the   landscape  mean  at  low  elevations  may  also  be  a  consequence  of  human  activities,   areas   with   difficult   access   (higher   vertical   distance   from   rivers)   being   better   Results  and  Discussion     29   protected.   However,   there   may   also   exist   a   natural   preference,   Polylepis   stands   avoiding   valley   bottoms   due   to   an   accumulation   of   cold   air   masses   at   night   (Sarmiento   1986;;   Young   1993;;   Kessler   2002;;   Fjeldså   &   Kessler   1996).   The   topographic   associations   that  we   observed   at   high   elevations  were   interpreted   as   better   reflecting   the  natural  preferences  of  Polylepis   than   the  associations   that  we   observed  at  lower  elevations.  This  is  due  to  greater  environmental  harshness  with   increasing  elevation,  in  particular  temperature  constraints  for  tree  growth  (Kessler   et  al.  2007;;  Körner  2012);;   the  human   impact  had  also  been  expected   to  decrease   with   increasing   elevation   (Toivonen   et   al.   2011).   Sunnier   and   drier   convex   positions   at  west   and   northwest   exposures   seemed   to   be   suitable   for  Polylepis   at   high  elevations.  The  elevational  pattern  of  the  associations  between  terrain  features   and   Polylepis   forest   distribution   was   similar   to   the   pattern   of   the   proportional   contribution  of  each   terrain  feature   to   the  performance  of  the  predictive  model  of   forest  stand  distribution.   3.3.  Polylepis  tree  height,  elevation,  climate  and  human  impact     Article   III   dealt   with   the   associations   between   Polylepis   tree   height,   elevation,   climate   conditions   and   human   impact.   The   relationship   between   tree   height   and   diameter  at  breast  height  (dbh)  showed  a  linear  relationship  in  elevation  belts  that   were  affected  by  human  impact,  whereas  at  four  of  the  six  undisturbed  elevational   belts   maximum   tree   height   leveled   off   at   a   certain   dbh.   The   lack   of   tree   height   saturation   at   disturbed   sites   supports   the   earlier   findings   of   lower  maximum   tree   height  at  disturbed  sites  due  to  logging  of  the  biggest  trees  (Toivonen  et  al.  2011).   In   natural   stands,   maximum   tree   height   decreased   gradually   with   elevation   at   humid  sites,  but  at  dry  sites  tree  height  peaked  at  middle  elevations  (III).  This  may   be   caused   by   increasing   drought   towards   low   elevations   at   dry   sites.   Drought-­ related  treelines  have  also  been  documented  in  a  forest-­steppe  ecotone  in  Patagonia   (Hertel  et  al.  2008)  and  in  rain-­shadowed  mountain  valleys  (Holtmeier  2009).     Maximum   tree   height   differed   considerably   between   dry   and   humid   areas.   Especially   near   the   treeline,   trees   grow   taller   in   dry   areas   (III).   This   is   a   well-­ known  phenomenon   (Kessler  et  al.  2007;;  Miehe   et   al.   2007;;  Körner  2012),  most   likely  resulting  from  the  difference  in  solar  radiation  (Lauer  1982).  Dry  and  humid   areas   differed   in   air   temperatures   but   not   in   soil   temperatures,   air   temperatures   being   higher   in   dry   areas.   This   confirms   the   importance   of   solar   radiation   in   determining   tree   height,   and   indicates   that   tree   height   in   our   study   area   is  more   closely  related  to  air  temperatures  than  to  soil  temperatures.  This  can  be  explained   by  a   strong  atmospheric  coupling  of  high   tree   stature;;   it  has  been  shown  that   the   above-­ground  parts  of  high   stature  vegetation,   apical  meristems   in  particular,   are   faced  with  much   lower   temperatures   than   those   of   lower   stature   vegetation   (e.g.   Hadley  &  Smith  1987;;  Wilson  et  al.  1987;;  Grace  et  al.  1989;;  Körner  2003).   Results  and  Discussion     30   Remarkably  tall  trees  were  documented  near  treelines:  9  m  tall  at  4530  m  in  humid   areas  and  13  m  tall  at  4650  m  in  dry  areas  (III).  Such   tall   trees  at  corresponding   elevations  have  not  been  documented  in  any  previous  studies  worldwide  (Hoch  &   Körner  2005;;  Kessler  et  al.  2007;;  Miehe  et  al.  2007;;  Bader  &  Ruijten  2008).  Based   on   satellite   images,   the   highest   elevation   forest   stands   are   located   at   4700  m   in   humid  areas  and  at  4950  m  in  dry  areas.  This  suggests  an  abrupt  decrease   in  tree   height   just   below   4700  m,   also   found   for  Polylepis   stands   in   Bolivia   (Hertel  &   Wesche   2008)   as  well   as   for   certain   other   treeline   species   in   tropical  mountains   (Miehe  &  Miehe  2000;;  Miehe  et  al.  2007).   We   also   found   substantially   low   air   and   soil   temperatures   for   high   elevation   Polylepis  forest  stands  (III).  Mean  growing  season  soil   temperatures  were  5.2  °C   for  P.  pepei  at  4530  m  and  4.6  °C  for  P.  subsericans  at  4650  m,  whereas  the  global   treeline  mean  is  6-­7.5  °C  (Körner  &  Paulsen  2004;;  Körner  2012).  Low  mean  soil   temperatures   in   the   growing   season   have   also   been   documented   for   Polylepis   elsewhere  in  the  Andes:  4.5-­6.0  °C  at  4000-­4100  in  Ecuador  (Lauer  &  Rafiqpoor   2000,  2002)  and  4.7-­5.4  °C  at  4810  m  in  western  Bolivia  (Hoch  &  Körner  2005).   Thus  there  is  increasing  evidence  that  Polylepis  treeline  forests  grow  under  lower   temperature   conditions,   especially   soil   temperatures,   than   other   treeline   forests   worldwide.   This   may   be   due   to   specific   physiological   adaptations,   such   as   decreased   leaf   size   and   increased   root   tip  abundance   (Toivonen  et   al.   2013),   that   enable   the   species  of  Polylepis   to  withstand  especially  demanding  environmental   conditions.  It  may  also  be  possible  that  soil  temperatures  are  in  fact  not  limiting  for   Polylepis,  but  rather  air  temperatures,  which  were  also  shown  to  be  related  to  tree   height   differences   between   dry   and   humid   areas.   Further   research   is   needed   to   confirm   the   crucial   aspects   of   temperature   conditions   and   to   clarify   the   physiological  basis  of  the  specific  adaptations,  to  better  understand  the  limitations   of  tree  growth  at  high  elevations.   3.4.  Accessibility  and  structural  variation  of  Polylepis  forests   In  article  IV  I  studied  the  structural  variation  of  Polylepis  forest  stands  in  relation   to   indices   of   accessibility,   as   a   proxy   for   the   degree   of   human   disturbance,   calculated   as   the   geographical   distance   between   forest   stands   and   the   nearest   human  settlement,  road  and  market  center.  Polylepis  forest  structure  was  explained   by   indices   of   accessibility   in   both   dry   and   humid   areas,   but   the   relationships   differed   between   areas.   In   general,   forest   stands   with   easy   access   (near   human   settlements,   roads   and   market   centers)   were   more   degraded   than   those   with   difficult  access  (far  from  settlements,  roads  and  market  centers).  In  humid  areas  the   distance  from  the  nearest  road  was  clearly  the  most  important  factor  explaining  the   forest   structure,  whereas   in   dry   areas   the  most   important   factor  was   the   distance   from   the   nearest   market   center.   In   both   areas,   distance   from   the   nearest   human   settlement  was  negatively  related  to  the  proportion  of  vegetative  regeneration.  The   success   of   vegetative   regeneration   near   villages   was   presumably   caused   by   a   positive   effect   of   moderate   disturbance,   related   to   trampling   and   grazing   by   Results  and  Discussion     31   domestic  animals,  also  found  by  Cierjacks  et  al.  (2007,  2008).  Likewise  Hagaman   (2006)  found  that  Polylepis  stands  near  villages  were  more  impacted  by  grazing  in   particular  than  stands  far  from  villages.  Polylepis  forest  canopy  cover  and  soil  loss   have   also   been   documented   to   relate   similar   distance   based   indirect  measures   of   human  impact  (Cingolani  et  al.  2008;;  Renison  et  al.  2010).  Forest  structure  was  not   more  closely   related   to   indices  of  accessibility   for  drier  valleys   than  humid  ones,   except  the  stronger  historical  human  influence  in  drier  valleys.  This,  however,  does   not   necessarily   mean   that   there   is   no   difference   in   the   degree   of   degradation   between  dry   and  humid  areas   in   absolute   terms.   In  general,  my   study   shows   that   simple  measures  of  geographical  distances  can  predict  the  variation  in  biomass  and   regeneration   patterns   of   Polylepis   forests.   Thus   these   measures   can   be   used   to   identify  areas  where  Polylepis  forests  are  most  degraded,  or  at  the  greatest  risk  of   degradation,  and  thus  in  urgent  need  of  protection  and  sustainable  management.     Conclusions     32   4.   CONCLUSIONS   Species   of   the   high   Andean   tree   genus   Polylepis   form   special   high   elevation   treeline  habitats,  at  up  to  5000  m  of  elevation  in  the  Central  Andes.  These  habitats   are   of   considerable   importance   for   a   number   of   other   endemic   plant   and   animal   species  (Fjeldså  1992;;  Servat  et  al.  2002;;  Lloyd  and  Marsden  2011).  The  current   distribution  of  Polylepis  forests  is  fragmented  and  disconnected  from  that  of  other,   lower-­elevation  montane  forests.  There  has  been  ample  discussion  as  to  the  causes   of   the   fragmented   distribution   of   the   forests.   In   this   thesis,   my   purpose   was   to   identify  the  functional  traits  that  have  been  crucial  for  Polylepis  in  its  adaptation  to   cold  and  dry  conditions,  and   to  examine   the   role  of  natural   factors  versus  human   influence  in  determining  the  current  distribution  of  Polylepis  forests.     In  answer  to  these  questions,  I  present  four  new  findings  concerning  the  functional,   environmental   and   anthropogenic   factors   determining   Andean   Polylepis   forest   distribution  and  treeline  formation  in  the  Andes.  First,  I  show  that  Polylepis  species   have   certain   specific   genetically   determined   functional   adaptations,   such   as   reduced   leaf   size,   increased   root   tip   abundance,   and   reduced   mass-­based   photosynthetic  rate  (I).  The  first   two  are  advantageous  in  colonizing  cold  and  dry   high  elevation  habitats,  while  the  last  one,  the  reduced  photosynthetic  rate,  supports   the   carbon  source   limitation  hypothesis   for   tree  growth  at   high   elevations.  These   findings  add  to  our  scientific  knowledge  concerning  high  elevation  adaptations  in   trees,   and   have   significant   implications   for   the   discussion   as   to   the   factors   constraining   tree   growth   at   high-­elevation   treelines   globally.   They   can   also   be   applied   as   important   base   line   information   predicting   the   responses   of   high-­ elevation   tree   species   to   global   climate   change.   Due   to   the   complexity   of   interactions   and   trade-­offs   between   functional   traits   and   climate,   however,   such   inferences  must  be  drawn  with  caution.  In  future,  it  would  be  interesting  to  relate   phylogeny,  climatic  niches  and  functional  traits  between  all  species  of  the  genus,  as   well  to  carry  out  in  situ  measurements  of  functional  traits.   Second,   I   show   that   terrain   features   play   an   important   role   in   determining   the   current   distribution   of   Polylepis   forests   through   the   formation   of   refugia   from   human  activities  at  lower  elevations  and  of  favorable  micro-­climatic  conditions  at   higher   ones   (II).   This   result   confirms   the   importance   of   human   influence   in   studying   associations   between   vegetation   and   the   environment.   Even   more   interestingly,   it   evokes   several   new   questions   as   to   the   role   of   natural   factors   constraining   Polylepis   forest   distribution.   Noteworthy   topics   for   future   research   would  include  for  example  the  role  of  the  last  glaciation  and  current  processes  of   glacial   erosion   in   preventing   tree   colonization,   and   aspects   related   to   soils   and   nutrient  uptake.   Third,  remarkably  low  air  and  soil  temperatures  and  high  tree  statures  at  Polylepis   treelines  were   found  out   compared   to   the   global  means,   and   some  new   evidence   was  presented  supporting  the  role  of  air  temperatures  rather  than  soil  temperatures   Conclusions     33   to   constrain   tree   growth   at   high   elevations   (III).   These   findings   have   important   implications   for   the   discussion   concerning   high   elevation   treeline   formation   worldwide.  However,  the  specific  functional  mechanisms  behind  the  link  between   tree  growth  and  air  and/or  soil  temperatures  remain  to  be  studied.   Fourth,   I   showed   that   indices   of   accessibility,   used   as   indirect   measures   of   the   human  impact  on  forest  stands,  predict  stand  structure  and  regeneration:  stands  of   higher  accessibility   (near  human   settlements,   roads   and  market   centers)   are  more   degraded   regarding   structural   variables   and   regeneration   than   stands   of   lower   accessibility  (IV).  This  information  can  be  used  for  the  rapid  identification  of  areas   where  Polylepis   forests   are  most   degraded   or   at   the   greatest   risk   of   degradation,   thus  having  important  practical  applications  for  conservation  and  sustainable  forest   management.     With   these   four   findings,   I   conclude   that   the   success   of   Polylepis   species   in   inhabiting  the  high  Andes  is  based  on  specific  climatic  adaptation,  in  particular  in   leaf  and   root  morphology.  On   the  other  hand,   the  absence  of  Polylepis   forests   in   large  areas  of  the  Andes  is   the  result  of  several  environmental  and  anthropogenic   constraints.   The   role   of   environmental   constraints   determining   Polylepis   forest   distribution   becomes   greater   towards   high   elevations,  while   at   low   elevation   the   distribution  is  largely  defined  by  human  influence.  This  information  has  important   implications   for   our   understanding   of   the   causes   of   the   current   distribution   of   Polylepis  forests,  as  well  as  for  the  conservation  and  restoration  of  Polylepis  forests   in  the  Andes.       Acknowledgements     34   5.   ACKNOWLEDGEMENTS   How  did  I  choose  to  study  Polylepis  forests  –  or  maybe  that  question  should  be  the   other  way  around  –  how  did  Polylepis  choose  me?  After  finishing  my  MSc  degree  I   received  an  excursion  grant  from  the  Finnish  Embassy  in  Lima  to  carry  out  a  small   research  project  on  Polylepis  forests  in  the  natural  protected  area  of  Machupicchu,   in  Cuzco,  Peru.  This  opportunity  was  given   to  me  by   Jukka  Salo,  Mikko  Pyhälä   and   Siegfried  Kastl,   to   whom   I   am   deeply   thankful   for   introducing  me   to   these   fascinating  high  Andean  forests.   The   idea   to   carry  out   a  PhD  project  on  Polylepis   arose  when   I   semi-­accidentally   met  with  Michael  Kessler,  my  ex  situ  supervisor,  in  Cuzco.  Apparently  at  that  time   I   had   already   decided   to   do,   or   at   least   considered   doing,   a   PhD   on   Polylepis,   because  Michael  promised  to  be  my  supervisor.  Thank  you,  Michael,  for  being  so   open  minded.   It  has  been  very   inspiring   to  work  with  you  during  all   these  years,   even  it  was  mostly  long  distance.  I  really  admire  your  creativity  and  speed  of  doing   things.   After   visiting   and   talking   to   you,   any   problems   I   had   amazingly   disappeared:  you  always  had  a  solution  for  everything.     This   project   wouldn’t   have   succeeded   without   my   in   situ   supervisor,   Kalle   Ruokolainen.   I   am   incredibly   thankful   to   you,   Kalle,   for   your   time   and   commitment  to  supervise  my  work,  even  though  the  topic  was  sometimes,  as  you   say,   “at   the   limits   of   your   comfort   zone”.   You   helped   me   to   formulate   any   irrelevant   study   questions   into   sound   research   questions   and   taught   me   to   think   critically.   You   had   the   time   and   patience   to   explain   to   me   the   basics   of   almost   everything,   not   to  mention   teaching  me   basic   survival   skills,   such   as   fishing   by   hand.     I  was   fortunate   to  also  have   two  other  unofficial   supervisors,  Dietrich  Hertel  and   Viviana  Horna.  I  am  very  grateful  to  you,  Dietrich,  for  the  opportunity  to  carry  out   lab   work   under   your   supervision   in   Göttingen.   Thank   you   for   your   support   and   valuable  advice  on  everything,  and  especially  for  elucidating  to  me  the  importance   of   belowground   components   for   plant   life.   I   am   also   deeply   thankful   to   you,   Viviana,   for   introducing   me   to   the   secrets   and   complexity   of   gas   exchange   measurements   and   for   all   your   help   and   advice,   as  well   as   for   your   friendship.   I   also  want  to  thank  my  other  co-­authors  Steven  Sylvester,  Jürgen  Kluge  and  Carlos   Gonzales  Inca,  for  your  input  into  this  work.   Furthermore,  I  am  very  grateful  to  Prof.  Pekka  Niemelä,  for  encouragement  and  a   continuous  belief   in   the   success  of   this  project.   I   also  want   to   thank  Dr.  Paul  M.   Ramsay   and   Prof.   William   K.   Smith   for   reviewing   my   thesis,   and   Dr.   Arne   Cierjacks  for  accepting  the  invitation  to  examine  my  thesis.  I  am  also  thankful  to   Ellen  Valle  for  proof-­reading  my  thesis  at  rather  short  notice.   Acknowledgements     35   I   am   also   very   grateful   to   my   field   assistant,   Lou   (Louella   Puelles   Linares)   for   carrying  out  a  valuable  and   laborious  part  of   this  project  with  me.  Thank  you  for   sharing  many  unforgettable  moments  in  the  field:  freezing  mornings,  unbelievably   beautiful  views  of  the  snow  peaks,  night  climbing  in  moonlight,  escape  from  angry   horseflies,  boiling  water  in  a  can  of  tuna  fish  and  the  satisfaction  of  finishing  a  plot   inventory  on  time,  at  least  sometimes.     My   special   thanks   go   to   the   members   of   Amazon   Research   Team.   It   has   been   incredibly  rewarding  to  work  with  you.  I  want  to  thank  especially  Hanna  Tuomisto   for  her  useful  advice  and  constructive  criticism,  which  helped  me  to  see  my  study   in   the   broader   context   of   tropical   ecology.   I   am   also   very   grateful   to   all   my   colleagues   and   friends   from   the   team:   Glenda,   Gabi,   Gabriel,   Youszef,   Samuli,   Lassi,   Liisa,   Isrrael,  Matti,   Ilari,   Nelly,   Anders,  Mirkka,   Outi,   Sofia,   and   all   the   others.  Thank   you   for   the   shared   every-­day  moments   in   the   office,   during   lunch   and  tea  breaks,  and  also  outside  of  the  office.  You  are  great  people!     I  also  want  to  thank  Sonja  and  Salla,  my  academic  mom-­friends.  We  have  so  much   in   common.   Everything   would   have   been   much   heavier   without   sharing   all   the   stresses,  but  also  the  best  joys  and  laughs,  with  you.  Thank  you  for  your  friendship.   This  work  has  been  financially  supported  by  the  Doctoral  Programme  in  Biology,   Geography  and  Geology  of  the  University  of  Turku  (BGG),  Biological  Interactions   Graduate   School   (BIOINT),   Turku   University   Foundation,   University   of   Turku,   Oskar   Öflund   Foundation,   Finnish   Concordia   Fund   and   Jenny   and   Antti  Wihuri   Foundation,  which  are  hereby  acknowledged.   Last,  but  not  least,  I  want  to  thank  my  extended  family,  especially  my  mom  Leena,   dad   Harri,   my   sisters   and   brothers,   and   Mamma:   you   have   been   an   excellent   support  group!  My  deepest  gratitude  goes   to  Carlos  and  our   lovely  girls,  Samira,   Vilja  and  Selja.  I  am  sorry  for  the  time  that  I  have  stolen  from  us,  especially  during   the  last  months.  Thank  you,  girls,  for  reminding  me  every  day  of  what  is  important   in   life.  Carlos,   I   think   I  am  very  fortunate  for  sharing  academic  and  personal  life   with  you.   I  wouldn’t  have  succeeded   in   this  project  without  your   support.  Thank   you  for  your  immense  patience,  sacrifices  and  love.         References     36   6.   REFERENCES   Azócar,   A.,   Rada,   F.   &   García-­Núñez   2007.   Functional   characteristics   of   the   arborescent   genus   Polylepis  along  a  latitudinal  gradient  in  the  high  Andes.  Interciencia  32:  663-­668.   Bader,  M.Y.  &  Ruijten,  J.J.A.  2008.  A  topography-­based  model  of  forest  cover  at  the  alpine  tree  line   in  the  tropical  Andes.  Journal  of  Biogeography  35:  711-­723.   Bader,  M.Y.,   van  Geloof,   I.  &  Rietkerk,  M.   2007.  High   solar   radiation   hinders   tree   establishment   above  the  alpine  treeline  in  northern  Ecuador.  Plant  Ecology  191:  33–45.   Baied,  C.A.  &  Wheeler,  J.C.  1993.  Evolution  of  high  Andean  puna  ecosystems:  environment,  climate,   and   culture   change   over   the   last   12   000   years   in   the   central   Andes.  Mountain   Research   and   Development  13:  145–156.   Ball,  M.C.,  Hodges,  V.S.  &  Laughlin,  G.P.  1991.  Cold-­induced  photoinhibition  limits  regeneration  of   snow  gum  at  tree-­  line.  Functional  Ecology  5:  663–668.   Barry,  R.G.  1981.  Mountain  weather  and  climate.  Methuen,  London.   Berry,   J.A.  &  Björkman,  O.  1980.  Photosynthetic   response  and  adaptation   to   temperature   in  higher   plants.  Annual  Review  of  Plant  Physiology  31:  491-­543.     Bradshaw,   A.D.   1965.   Evolutionary   significance   of   phenotypic   plasticity   in   plants.   Advances   in   Genetics  13:  115–155.     Braun,  G.  1997.  The  use  of  digital  methods  in  assessing  forest  patterns  in  an  Andean  environment:  the   Polylepis  example.  Mountain  Research  and  Development  17:  253–262.   Burnham,   K.P.   &   Anderson,   D.R.   2002.   Model   selection   and   multimodel   inference:   a   practical   information-­theoretic  approach.  Springer,  New  York.   Burnham,  R.J.  &  Graham,  A.   1999.  The   history   of  Neotropical   vegetation:   new   developments   and   status.  Annals  of  the  Missouri  Botanical  Garden  86(2):  546–589.   Cabrera,   H.M.,   Rada,   F.   &   Cavieres,   L.   1998.   Effects   of   temperature   on   photosynthesis   of   two   morphologically  contrasting  plant  species  along  an  altitudinal  gradient  in  the  tropical  high  Andes.   Oecologia  114:  145-­152.   Caldwell,  M.M.  1968.  Solar  ultraviolet  radiation  as  an  ecological  factor  for  alpine  plants.  Ecological   Monographs  38:  243-­268.     Caldwell,  M.M.,  Teramura,  A.H.  &  Tevini,  M.  1989.  The  changing  solar  ultraviolet  climate  and  the   ecological  consequences  for  higher  plants.  Tree  4:  363-­367.       Chepstow-­Lusty,  A.J.,  Bennett,  K.D.,  Switsur,  V.R.  &  Kendall,  A.  1996.  4000  years  of  human  impact   and  vegetation  change  in  the  central  Peruvian  Andes—with  events  paralleling  the  Maya  record?   Antiquity  70:  824–833.   Chepstow-­Lusty,  A.J.  &  Winfield,  M.  2000.   Inca  agroforestry:   lessons   from   the  past.  Ambio  29(6):   322–328.     Cierjacks,  A.,  Rühr,  N.K.,  Wesche,  K.  &  Hensen,   I.   2008.  Effects   of   altitude   and   livestock   on   the   regeneration  of  two  tree  line  forming  Polylepis  species  in  Ecuador.  Plant  Ecology  194:  207-­221.   Cierjacks,  A.,  Wesche,  K.  &  Hensen,  I.  2007.  Potential  lateral  expansion  of  Polylepis  forest  fragments   in  central  Ecuador.  Forest  Ecology  and  Management  242:  477–486.     Cingolani,  A.M.,  Renison,  D.,  Tecco,  P.A.,  Gurvich,  D.E.  &  Cabido,  M.  2008.  Predicting  cover  types   in   a   mountain   range   with   long   evolutionary   grazing   history:   a   GIS   approach.   Journal   of   Biogeography  35:  538–551.   Clausen,   J.,   Keck,   D.D.   &   Hiesey,  W.M.   1940.   Experimental   studies   on   the   nature   of   species.   1.   Effect   of   varied   environments   on   western   North   American   plants.   Carnegie   Institution   of   Washington  Publication  520,  Washington  DC.   Coblentz,  D.  &  Keating,  P.L.   2008.  Topographic   controls   on   the   distribution   of   tree   islands   in   the   high  Andes  of  south-­western  Ecuador.  Journal  of  Biogeography  35:  2026–2038.   References     37   Cordell,   S.,   Goldstein,   G.,   Mueller-­Dombois,   D.   &   Vitousek,   P.M.   1998.   Physiological   and   morphological  variation  in  Metrosideros  polymorphy,  a  dominant  Hawaiian  tree  species,  along  an   altitudinal  gradient:  the  role  of  phenotypic  plasticity.  Oecologia  113:  188–196.   Dale,   M.R.T.,   Dixon,   P.,   Fortin,   M-­J.,   Legendre,   P.,   Myers,   D.E.   &   Rosenberg,   M.S.   2002.   Conceptual  and  mathematical  relationships  among  methods  for  spatial  analysis.  Ecography  25(5):   558–577.   di  Pasquale,  G.,  Marziano,  M.,  Impagliazzo,  S.,  Lubritto,  C.,  de  Natale,  A.  &  Bader,  M.Y.  2008.  The   Holocene   treeline   in   the   northern   Andes   (Ecuador):   first   evidence   from   soil   charcoal.   Palaeogeography,  Palaeoclimatology,  Palaeoecology  259:  17–34.   Ellenberg,   H.   1958.  Wald   oder   Steppe?   Die   natürliche   Pflanzendecke   der   Anden   Perus.   Umschau   1958:  645–681.   Ellenberg,  H.  1979.  Man’s  influence  on  tropical  mountain  ecosystems  in  South  America.  Journal  of   Ecology  67:  401–416.   Fjeldså,  J.  1992.  Biogeography  of  the  birds  of  the  Polylepis  woodlands  of  the  Andes.  In:  Balslev,  H.   &  Luteyn,   J.L.   (eds)  Páramo.  An  Andean   ecosystem  under   human   influence.  Academic   Press,   Padstow,  Great  Britain.   Fjeldså,   J.  &  Kessler,  M.   1996.  Conserving   the   biological   diversity   of  Polylepis   woodlands   of   the   highlands  of  Peru  and  Bolivia.  A  contribution  to  sustainable  natural  resource  management  in  the   Andes.  NORDECO,  Copenhagen.   Gaul,  D.,  Hertel,  D.,  Borken,  W.,  Matzner,  E.  &  Leuschner,  C.  2008.  Effects  of  experimental  drought   on   the  fine  root  system  of  mature  Norway  spruce.  Forest  Ecology  and  Management  256:  1151-­ 1159.   Germino,   M.J.   &   Smith,   W.K.   1999.   Sky   exposure,   crown   architecture,   and   low-­temperature   photoinhibition  in  conifer  seedlings  at  alpine  treeline.  Plant,  Cell  and  Environment  22:  407–415.   Gareca,  E.E.,  Breyne,  P.,  Vandepitte,  K.,  Cahill,  J.R.A.,  Fernandez,  M.  &  Honnay,  O.  2013.  Genetic   diversity  of  Andean  Polylepis  (Rosaceae)  woodlands  and  inferences  regarding  their  fragmentation   history.  Botanical  Journal  of  the  Linnean  Society  172:  544–554.   Grace,  J.,  Allen,  S.J.  &  Wilson  C.  1989.  Climate  and  the  meristem  temperatures  of  plant  communities   near  the  tree-­line.  Oecologia  79:  198-­204.   Gregory-­Wodzicki,   K.M.   2000.   Uplift   history   of   the   Central   and   Northern   Andes:   a   review.   Geological  Society  of  America  Bulletin  112(7):  1091–1105.   Gosling,  W.D.,  Hanselman,  J.A.,  Knox,  C.,  Valencia,  B.G.  &  Bush,  M.B.  2009.  Long-­term  drivers  of   change  in  Polylepis  woodland  distribution  in  the  central  Andes.  Journal  of  Vegetation  Science  20:   1041-­1052.   Hadley,  J.L.  &  Smith,  W.K.  1987.   Influence  of  krummholz  mat  microclimate  on  needle  physiology   and  survival.  Oecologia  73:  82-­90.     Hagaman,   D.J.   2006.   Conservation   and   use   of   Polylepis   pepei   woodlands   in   the   Apolobamba   mountains  of  Bolivia.  MSc  thesis,  University  of  Wisconsin-­Madison.   Hensen,  I.,  Cierjacks,  A.,  Hirsch,  H.,    Kessler,  M.,  Romoleroux,  K.,  Renison,  D.  &  Wesche,  K.  2012.   Historic  and  recent  fragmentation  coupled  with  altitude  affect  the  genetic  population  structure  of   one  of  the  world's  highest  tropical  tree  line  species.  Global  Ecology  and  Biogeography  21:  455-­ 464.   Hertel,  D.  &  Schöling,  D.  2011.  Norway  spruce  shows  contrasting  changes  in  below-­  versus  above-­ ground   carbon   partitioning   towards   the   alpine   treeline:   evidence   from   a  Central   European   case   study.  Arctic,  Antarctic,  and  Alpine  Research  43:  46-­55.   Hertel,   D.,   Therburg,   A.   &   Villalba,   R.   2008.   Above-­   and   belowground   response   by   Nothofagus   pumilio   to   climatic   conditions   at   the   transition   from   the   steppe-­forest   boundary   to   the   alpine   treeline  in  southern  Patagonia,  Argentina.  Plant  Ecology  &  Diversity  1:  21-­33.   References     38   Hertel,  D.  &  Wesche,  K.   2008.  Tropical  moist  Polylepis   stands   at   the   treeline   in  East  Bolivia:   the   effect  of  elevation  on  stand  microclimate,  above-­  and  below-­ground  structure,  and  regeneration.   Trees  22:  303-­315.   Hijmans,   R.J.,   Cameron,   S.E.,   Parra,   J.L.,   Jones,   P.G.   &   Jarvis,   A.   2005.   Very   high   resolution   interpolated   climate   surfaces   for   global   land   areas.   International   Journal   of   Climatology   25:   1965–1978.   Hirzel,   A.H.,   Hausser,   J.,   Chessel,   D.  &  Perrin,  N.   2002.   Ecological-­niche   factor   analysis:   how   to   compute  habitat-­suitability  maps  without  absence  data?  Ecology    83(7):  2027–2036.   Hoch,   G.  &  Körner,   C.   2005.   Growth,   demography   and   carbon   relations   of  Polylepis   trees   at   the   world’s  highest  treeline.  Functional  Ecology  19:  941-­951.   Holtmeier,  F.K.  2009.  Mountain  timberlines:  Ecology,  patchiness  and  dynamics.  Springer,  Berlin.   Hooghiemstra   H.   &   Van   der   Hammen,   T.   2004.   Quaternary   ice-­age   dynamics   in   the   Colombian   Andes:   developing   an   understanding   of   our   legacy.   Philosophical   Transactions   of   the   Royal   Society  London  B,  359:  173–181.   Hubbart,   J.,   Link,   T.,   Campbell,   C.   &   Cobis,   D.   2005.   Evaluation   of   a   low-­cost   temperature   measurement  system  for  environmental  applications.  Hydrological  Processes  19:  1517-­1523.   Hutchinson,   G.E.   1957.   Concluding   remarks.   Cold   Spring   Harbour   Symposium   on   Quantitative   Biology  22:  415-­427.   IUCN   2013.   The  IUCN   Red   List   of   Threatened   Species.   Version   2013.2.   .  Downloaded  on  21  December  2013.     Jameson,  J.S.  &  Ramsay,  P.M.  2007.  Changes   in  high-­altitude  Polylepis  forest  cover  and  quality   in   the  Cordillera  de  Vilcanota,  Perú,  1956–2005.  Biological  Conservation  138:  38-­46.   Jobbagy,   E.G.  &   Jackson,   R.B.   2000.   Global   controls   of   forest   line   elevation   in   the   northern   and   southern  hemisphere.  Global  Ecology  &  Biogeography  9:  253-­268.     Julio,  N.B.,  Sobral,  A.,  Rondan  Dueñas,  J.,  Di  Rienzo,  J.,  Renison,  D.  &  Hensen,  I.  2008.  RAPD  and   ISSR  markers   indicate   diminished   gene   flow   due   to   recent   fragmentation   of  Polylepis   australis   woodlands  in  central  Argentina.  Biochemical  Systematics  and  Ecology  36:  329–335.   Kerr,   M.S.   2003.   A   phylogenetic   and   biogeographic   analysis   of   Sanguisorbeae   (Rosaceae)   with   emphasis   on   the   Pleistocene   radiation   of   the   high   Andean   genus   Polylepis.   Ph.D.   thesis,   University  of  Maryland.   Kessler,  M.   1995.   Present   and   potential   distribution   of  Polylepis   (Rosaceae)   forests   in   Bolivia.   In:   Churchill,   S.P.  Balslev,  H.,   Forero,   E.   and  Luteyn,   J.L.   (eds),  Biodiversity   and   conservation   of   neotropical  montane  forests,  pp.  281-­294.  New  York  Botanical  Garden,  Bronx.     Kessler,  M.  2000.  Observations  on  a  human-­induced  fire  event  at  a  humid  timberline  in  the  Bolivian   Andes.  Ecotropica  6:89–93.   Kessler,  M.  2002.  The  “Polylepis  problem”:  Where  do  we  stand?  Ecotropica  8:  97-­110.   Kessler,  M.,  Böhner,  J.  &  Kluge,  J.  2007.  Modelling  tree  height  to  assess  climatic  conditions  at  tree   lines  in  the  Bolivian  Andes.  Ecological  Modelling  207:  223-­233.   Kessler  M.  &  Driesch,  P.  1993.  Causas  e  historia  de  la  destrucción  de  bosques  altoandinos  en  Bolivia.   Ecología  en  Bolivia  21:  1–18.   Kessler,   M.   &   Hohnwald,   S.   1998.   Bodentemperaturen   innerhalb   und   außerhalb   bewaldeter   und   unbewaldeter  Blockhalden  in  den  bolivianischen  Hochanden.  Ein  Test  der  Hypothese  von  Walter   und  Medina  (1967).  Erdkunde  52:  54-­62.   Körner,   C.   1998.   A   re-­assessment   of   high   elevation   treeline   positions   and   their   explanation.   Oecologia  115:  445-­459.   Körner,  C.  2003.  Alpine  plant  life,  functional  plant  ecology  of  high  mountain  ecosystems.  Springer,   Berlin.   Körner,  C.  2007.  Climatic  treelines:  conventions,  global  patterns,  causes.  Erdkunde  61:  315-­324.   References     39   Körner,   C.   2012.   Alpine   treelines:   Functional   ecology   of   the   global   high   elevation   tree   limits.   Springer,  Basel.     Körner,  C.  &  Paulsen,  J.  2004.  A  world-­wide  study  of  high  altitude  treeline  temperatures.  Journal  of   Biogeography  31:  713-­732.   Larigauderie,  A.  &  Körner,  C.  1995.  Acclimation  of  leaf  dark  respiration  to  temperature  in  alpine  and   lowland  plant  species.  Annals  of  Botany  76:  245-­252.   Lauer,  W.  1982.  Zur  Ökoklimatologie  der  Kallawaya-­Region  (Bolivien).  Erdkunde  36:  223-­247.   Lauer,  W.  &  Rafiqpoor,  M.D.  2000.  Paramo  de  Papallacta  –  a  physiogeographical  map  1:50.000  of   the  area  around  the  Antisana  (Eastern  Cordillera  of  Ecuador).  Franz  Steiner  Verlag,  Stuttgart.   Lauer,   W.   &   Rafiqpoor,   M.D.   2002.   Die   Klimate   de   Erde.   Eine   Klassifikation   of   Grundlage   der   ökophysiologischen  Merkmale  der  Vegetation.  Franz  Steiner  Verlag,  Stuttgart.   Laegaard,   S.   1992.   Influence   of   fire   in   the   grass   páramo   vegetation   of  Ecuador.   In:  Balslev,  H.  &   Luteyn,  J.L.  (eds)  Páramo.  An  Andean  ecosystem  under  human  influence,  pp  151–170.  Academic   Press,  Padstow,  Great  Britain.     Leuschner,  C.  2000.  Are  high  elevations  in  tropical  mountains  arid  environments  for  plants?  Ecology   81(5):  1425-­1436.     Leuschner,  C.,  Moser,  G.,  Bertsch,  C.,  Röderstein,  M.  &  Hertel,  D.  2007.  Large  altitudinal  increase  in   tree  root/shoot  ratio  in  tropical  mountain  forests  of  Ecuador.  Basic  and  Applied  Ecology  8:  219-­ 230.   Lloyd,  H.  &  Marsden,  S.J.   2011.  Between-­Patch  Bird  Movements  within   a  High-­Andean  Polylepis   Woodland/Matrix  Landscape:   Implications  for  Habitat  Restoration.  Restoration  Ecology  19:  74-­ 82.     Macek,   P.,   Macková,   J.   &   De   Bello,   F.   2009.   Morphological   and   ecophysiological   traits   shaping   altitudinal  distribution  of  three  species  in  the  dry  tropical  Andes.  Acta  Oecologica  35:  778-­785.   Mayr,  S.  2007.  Limits  in  water  relations.  In:  Wieser,  G.  &  Tausz,  M.  (eds.)  Trees  at  their  upper  limit,   pp.  145-­162.  Springer,  Dordrecht,  The  Netherland.   Meir,  P.,  Levy,  P.E.,  Grace,  J.  &  Jarvis,  P.G.  2007.  Photosynthetic  parameters  from  two  contrasting   woody  vegetation  types  in  West  Africa.  Plant  Ecology  192:  277-­287.   Miehe,  G.  &  Miehe,  S.  1994.  Zur  oberen  Waldgrenze   in   tropischen  Gebirgen.  Phytocoenologia  24:   53-­110.   Miehe,  G.  &  Miehe,  S.  2000.  The  forest  boundary  in  tropical  mountains.  Phytocoenologia  24:  53-­110.   Miehe,  G.,  Miehe,  S.,  Vogel,  J.,  Co,  S.  &  Duo,  L.  2007.  Highest  treeline  in  the  northern  hemisphere   found  in  southern  Tibet.  Mountain  Research  And  Development  27:  169-­173.   Moser,  G.,  Hertel,  D.  &  Leuschner,  C.   2007.  Altitudinal   change   in  LAI   and   stand   leaf   biomass   in   tropical  montane  forests:  a  transect  study  in  Ecuador  and  a  pan-­tropical  meta-­analysis.  Ecosystems   10:  924–935.   Moser,  G.,  Leuschner,  C.,  Hertel,  D.,  Graefe,  S.,  Soethe,  N.  &  Iost,  S.  2011.  Elevation  effects  on  the   carbon   budget   of   tropical   mountain   forests   (S   Ecuador):   the   role   of   the   belowground   compartment.  Global  Change  Biology  17:  2211–2226.   Rada,  F.,  Azócar,  A.,  Briceno,  B.,  Gonzalez,  J.  &  García-­Núñez,  C.  1996.  Carbon  and  water  balance   in  Polylepis  sericea,  a  tropical  treeline  species.  Trees:  Structure  and  Function  10:  218–222.   Paulsen,  J.,  Weber,  U.M.  &  Körner,  C.  2000.  Tree  growth  near  treeline:  abrupt  or  gradual  reduction   with  altitude?  Arctic.  Antarctic,  and  Alpine  Research  32:14-­20.     Phillips,   S.J.,     Anderson,   R.P.   &   Schapire,   R.E.   2006.   Maximum   entropy   modeling   of   species   geographic  distributions.    Ecological  Modelling  190:  231-­259.   Purcell,  J.  &  Brelsford,  A.  2004.  Reassessing  the  causes  of  decline  of  Polylepis  a  tropical  subalpine   forest.  Ecotropica  10:155–158.     References     40   Rada,  F.,  García-­Núñez,  C.,  Boero,  C.,  Gallardo,  M.,  Hilal,  M.,  González,  J.,  Prado,  M.,  Liberman-­ Cruz,  M.  &  Azócar,  A.  2001.  Low-­temperature  resistance  in  Polylepis  tarapacana,  a  tree  growing   at  the  highest  altitudes  in  the  world.  Plant,  Cell  &  Environment  24:  377-­381.   Renison,   D.,   Hensen,   I.   &   Suarez,   R.   2011.   Landscape   structural   complexity   of   high-­mountain   Polylepis  australis  forests:  a  new  aspect  of  restoration  goals.  Restoration  Ecology  19(3):  390-­398.   Renison,  D.,  Hensen,   I.,   Suarez,  R.  &  Cingolani,  A.M.  2006.  Cover   and  growth  habit   of  Polylepis   woodlands   and   shrublands   in   the   mountains   of   central   Argentina:   human   or   environmental   influence?  Journal  of  Biogeography  33:  876-­887.     Renison,   D.,   Hensen,   I.,   Suarez,   R.,   Cingolani,   A.M.,   Marcora,   P.   &   Giorgis,   M.A.   2010.   Soil   conservation  in  Polylepis  mountain  forests  of  Central  Argentina:  is  livestock  reducing  our  natural   capital?  Austral  Ecology  35:  435.   Sarmiento,  G.  1986.  Ecological  features  of  climate  in  high  tropical  mountains.  In:  Vuilleumier,  F.  &   Monasterio,  M.  (eds),  High  altitude  tropical  biogeography,  pp.  11–45.  Oxford  Univ.  Press.     Schlichting,  C.D.  1986.  The  evolution  of  phenotypic  plasticity   in  plants.  Annual  Review  of  Ecology   Evolution  and  Systematics  17:  667–693.   Schmidt-­Lebuhn,   A.N.,   Fuchs,   J.,   Hertel,   D.,   Hirsch,   H.,   Toivonen,   J.   &   Kessler,   M.   2010.   An   Andean   radiation:   polyploidy   in   the   tree   genus   Polylepis   (Rosaceae,   Sanguisorbeae).   Plant   Biology  12:  917–926.   Schmidt-­Lebuhn  A.N.,  Kessler  M.,  Kumar  M.  2006.  Promiscuity  in  the  Andes:  species  relationships   in   Polylepis   (Rosaceae,   Sanguisorbeae)   based   on   AFLP   and   morphology.   Systematic   Botany   31(3):  547–559.   Schröter,  C.  1908.  Das  Pflanzenleben  der  Alpen:  Eine  Schilderung  der  Hochgebirgsflora.  Raustein,   Zürich.   Servat  GP,  Mendoza  W,  Ochoa  JA  (2002)  Flora  y  fauna  de  cuatro  bosques  de  Polylepis  (Rosaceae)  en   la  cordillera  del  Vilcanota  (Cusco,  Peru).  Ecología  Aplicada  1(1):  25–35.   Simpson,   B.B.   1979.   A   revision   of   the   genus   Polylepis   (Rosaceae:   Sanguisorbeae).   Smithsonian   Contributions  to  Botany  43:  1-­62.   Simpson,  B.B.  1986.  Speciation  and  specialization  of  Polylepis  in   the  Andes.   In:  Vuilleumier,  F.  &   Monasterio,   M.   (eds.)   High   altitude   tropical   biogeography.   pp.   304–316.   Oxford   University   Press,  NY,  US.     Smith,  W.,  Germino,  M.  J.,  Hancock,  T.  E.  &  Johnson,  D.M.  2003  Another  perspective  on  altitudinal   limits  of  alpine  timberlines.  Tree  Physiology  23:  1101-­1112.   Smouse,  P.E.,  Long,  J.C.  &  Sokal,  R.R.  1986.  Multiple  regression  and  correlation  extensions  of  the   Mantel  test  of  Matrix  correspondence.  Systematic  Zoology  35:  627–632.   Sultan,   S.E.   2000.   Phenotypic   plasticity   for   plant   development,   function   and   lifehistory.  Trends   in   Plant  Science  5:  537–542.   Sveinbjörnsson,  B.,  Nordell,  O.  &  Kauhanen,  H.  1992.  Nutrient  relations  of  mountain  birch  growth  at   and  below  the  elevational  tree-­line  in  Swedish  Lapland.  Functional  Ecology  6:  213-­220.   Tanner,  E.V.J.,  Vitousek,  P.M.  &  Cuevas,  E.  1998.  Experimental  investigation  of  nutrient  limitation   of  forest  growth  on  wet  tropical  mountains.  Ecology  79:  10-­22.   Toivonen,  J.M.,  Horna,  V.,  Kessler,  M.,  Ruokolainen,  K.  &  Hertel,  D.  2014.  Interspecific  variation  in   functional  traits  in  relation  to  species  climatic  niche  optima  in  Andean  Polylepis  (Rosaceae)  tree   species:  evidence  for  climatic  adaptations.  Functional  Plant  Biology  41:  301-­312.   Toivonen,   J.M.,   Kessler,  M.,   Ruokolainen,   K.  &  Hertel,   D.   2011.   Accessibility   predicts   structural   variation  of  Andean  Polylepis  forests.  Biodiversity  and  Conservation  20:  1789-­1802.   Tranquillini,  W.  1979.  Physiological  ecology  of  the  alpine  timberline.  Tree  existence  at  high  altitudes   with  special  reference  to  the  European  Alps.  Ecological  Studies  31:  1-­137.   References     41   Troll,  C.  1961.  Klima  und  Pflanzenkleid  der  Erde  in  dreidimensionaler  Sicht.  Naturwissenschaften  9:   332-­348.   Troll,   C.   1973.   The   upper   timberlines   in   different   climatic   zones.   Arctic,   Antarctic,   and   Alpine   Research  5:  A3-­A18.     UNEP-­WCMC.   2004.   United   Nations   Environment   Programme.   World   Conservation   Monitoring   Center.  http://www.unep-­wcmc.org   Urrego,  D.H.,  Niccum,  B.A.,  La  Drew,  C.F.,  Silman,  M.R.  &  Bush,  M.B.  2011.  Fire  and  drought  as   drivers  of  early  Holocene  tree  line  changes  in  the  Peruvian  Andes.  Journal  of  Quaternary  Science   26:  28-­36.   Wang,  Y.,  Titus,  S.J.  &  LeMay,  V.M.  1998.  Relationships  between  tree  slenderness  coefficients  and   tree  or  stand  characteristics  for  major  species  in  boreal  mixedwood  forests.  Canadian  Journal  of   Forest  Research  28:  1171–1183.   Wieser,  G.  2007.  Limitation  by  an   insufficient  carbon  assimilation  and  allocation.   In:  Wieser,  G.  &   Tausz,  M.  (eds.)  Trees  at  their  upper  limit,  pp.  79-­129.  Springer,  Dordrecht,  The  Netherland.   Wieser,  G.,  Oberhuber,  W.,  Walder,  L.,  Spieler,  D.  &  Gruber,  A.  2010.  Photosynthetic   temperature   adaptation  of  Pinus  cembra  within  the  timberline  ecotone  of  the  Central  Austrian  Alps.  Annals  of   Forest  Science  67:  201.   Wieser,   G.  &   Tausz,  M.   2007.  Current   concepts   for   treelife   limitation   at   the   upper   timberline.   In:   Wieser,   G.   &   Tausz,  M.   (eds.)  Trees   at   their   upper   limit,   pp.   1-­18.   Springer,   Dordrecht,   The   Netherland.   Wilson,  C.,  Grace,  J.,  Allen,  S.,  &  Slack,  F.,  1987.  Temperature  and  stature,  a  study  of  temperatures   in  montane  vegetation.  Functional  Ecology  1:  405-­414.   Wittich,  B.,  Horna,  V.,  Homeier,  J.  &  Leuschner,  C.  2012.  Altitudinal  Change  in  the  Photosynthetic   Capacity  of  Tropical  Trees:  A  Case  Study   from  Ecuador  and  a  Pantropical  Literature  Analysis.   Ecosystems  15:  958-­973.   Wright,  I.J.,  Reich,  P.B.,  Atkin,  O.K.,  Lusk,  C.H.,  Tjoelker,  M.G.  &  Westoby,  M.  2006.  Irradiance,   temperature   and   rainfall   influence   leaf   dark   respiration   in   woody   plants:   evidence   from   comparisons  across  20  sites.  New  Phytologist  169:  309-­319.   Yamori,   W.,   Noguchi,   K.   &   Terashima,   I.   2005.   Temperature   acclimation   of   photosynthesis   in   spinach   leaves:   analyses   of   photosynthetic   components   and   temperature   dependencies   of   photosynthetic  partial  reactions.  Plant  Cell  and  Environment  28:  536-­547.   Young,  K.R.   1993.  Tropical   timberlines:   changes   in   forest   structure   and   regeneration  between   two   Peruvian  timberline  margins.  Arctic  and  Alpine  Research  25:167-­174.   Zhang,  S.B.,  Zhou,  Z.K.,  Hu,  H.,  Xu,  K.,  Yan,  N.  &  Li,  S.Y.  2005.  Photosynthetic  performances  of   Quercus   pannosa   vary   with   altitude   in   the   Hengduan   Mountains,   southwest   China.   Forest   Ecology  and  Management  212:  291-­301.